You are on page 1of 16

Technological Forecasting & Social Change 176 (2022) 121431

Contents lists available at ScienceDirect

Technological Forecasting & Social Change


journal homepage: www.elsevier.com/locate/techfore

A systematic idea generation approach for developing a new technology:


Application of a socio-technical transition system
Keeeun Lee a, Sunhye Kim a, Byungun Yoon a, *
a
Department of Industrial & Systems Engineering, School of Engineering, Dongguk University, 3-26, Pil-dong 3 ga, Chung-gu, Seoul, 100-715, Korea

A R T I C L E I N F O A B S T R A C T

Keywords: Socio-technical systems have been highlighted as a new driver of novel idea generation, providing greater in­
Coevolution sights for new product development (NPD) in different fields. This research implements a broad set of indicators
Socio-technical system that measure both the technological and social features for a panel of 50 coevolution cases, and uses a structural
Technology idea generation
vector autoregressive (SVAR) model in order to investigate the long-run relationships and coevolution patterns
Structural vector autoregressive model
between technology and society. Technology evolution patterns were analysed using text data from patents, and
social evolution patterns were investigated using text data from different news outlets. These influential re­
lationships between technical and societal variables adopted from the SVAR model are examined over the course
of this study. In addition, this research identifies various coefficients derived from this influential relationship,
and draws out evolutionary patterns to verify how they have influenced each other over time. The results
indicate that socio-technical systems are generally driven by the coevolution of three main technological vari­
ables (tangibility, safety and diffusion) and five social variables (intellectualization, informatization, polariza­
tion, connectivity and serviceability). This research contributes towards improving the explanatory power of
coevolution patterns between society and technology, and more strongly establishes a set of insightful strategies
for technological development.

1. Introduction and enables the acceptance of new technologies across all levels
(Leavitt, 1965). Academics believe that the convergence of humanities,
Within a highly competitive environment, innovation that succeeds social science, and technological studies can create new areas and
at a low cost while also reducing the risk of investment and failure has methods of study and analysis (Song et al., 2010). Therefore, the inte­
become increasingly important. One of the methods for successful gration between society and technology provides meaningful insights
innovation is encouraging open innovation (Cheng et al., 2010; Hung into the creation of new ideas for innovative technological development.
and Chou, 2013). In addition, companies which believe that there is a Various studies have been conducted to see the various effects of
limit to the convergence of adjacent technologies in science and engi­ society and technology, in addition to the coevolution of two systems.
neering have not taken social phenomena into consideration. In order to Geels & Schot (2007) identified that society and technology exist as
implement adequate response on these changes, the innovation of social-technological systems which cannot exist nor operate separately.
next-generation growth engines is no longer limited to the convergence In addition, changes in the social-technological system are seen as part
of technologies adjacent to science and engineering. Technological of a larger transition process within a dominant social-technological
development officials prefer to incorporate changes which reflect social regime. Kim & Jung (2008) investigated the characteristics of coevolu­
trends in order to make the technology more impactful (Moore, 1996), tion by means of system simulations and analysed the relationships
and cannot ignore that society is developing further due to technological between different technologies and various phenomena in the evolution
advancement. That is, the interaction between elements of social and of society using system dynamics methodology.
technical systems affect the overall performance or output level of an This study believes that the concept of co-evolution between tech­
organization (Bostrom and Heinen, 1977; Trist, 1981). Projecting social nology and society should be included in order to more accurately
factors onto technical factors increases general corporate productivity predict and analyze social phenomena during the generation of ideas for

* Corresponding author.
E-mail address: postman3@dongguk.edu (B. Yoon).

https://doi.org/10.1016/j.techfore.2021.121431
Received 7 July 2020; Received in revised form 27 November 2021; Accepted 9 December 2021
Available online 14 December 2021
0040-1625/© 2021 Elsevier Inc. All rights reserved.
K. Lee et al. Technological Forecasting & Social Change 176 (2022) 121431

efficient technology development. However, the existing research on the technology and society by integrating and modifying several methods
establishment of a technology development strategy does not rigorously such as SVM and DSM. In particular, since DSM is applied to identify the
apply social-technological co-evolution theory. Previous research about sequential patterns of social and technological attributes by modifying a
the creation of technology development strategies mainly intended to way to employ the method, this study can show how to specifically use a
draw up concepts about the development of idea generation by taking general approach in order to tackle a co-evolutionary problem between
into consideration social factors, such as customer value and future technology and society. This research is expected to greatly contribute to
environment change (Lee et al., 2015; Park and Lee, 2011). Although the the improvement of the objectivity and explanatory power of coevolu­
coevolution of socio-technical systems has been widely discussed in the tion patterns between society and technology, and also establishes a
literature, few empirical studies have been conducted, mainly due to an strategy for the technological development of R&D departments and
absence of operational measures. The study of Geels (2002, 2004) and policymakers in the field of business. In addition, it could provide in­
Geels & Schot (2007) presents a socio-technical system theory within a sights into the more optimal operation and implementation of industrial
conceptual framework; however, it does not include any empirical policy which would further foster the emergence of new technologies
study. In addition, case studies were conducted mainly from the view­ that consist of a humanity-focused society.
point of technological changes (Kim et al., 2014; Lee et al., 2013), and The rest of the paper is organized as follows. In Section 2, the
centered upon discussions on ways to juxtapose images of non-objective theoretical background about the multi-faceted relationships between
future societies and technologies (Song, 2012), and could not confirm technology and society is described. Section 3 explains the methodo­
specific coevolution between social and technological aspects. Likewise, logical aspects of this study for empirical analysis. In Section 4, the re­
they unfold the discussion on coevolution issues, using abstract features sults and implications of the statistical analysis are discussed. The
of future society and technologies. concluding remarks and future research are presented in Section 5 and 6.
Thus, this study attempts to overcome the limitations of existing
research and analyze the social-technological systems empirically. This 2. Literature review
research defines three questions for study:
2.1. Theoretical background
(1) What are the aspects of coevolution between social trends and
technology trends? 2.1.1. Socio-technical transition system
(2) How can this coevolution that has been accessed merely from a The socio-technical transition system assumes that society and
conceptual perspective in previous studies, be analysed technology exist as an interactive system consisting of two super do­
empirically? mains, and is based on the premise that they cannot exist separately.
(3) Which technology is most strongly affected by social trends, and That is, society and technology form a multi-party system in which they
which social trend is most strongly affected by technology, and are complementary and in harmony with one another. Changes in the
vice versa? socio-technical system are seen as part of the dominant social-
technological regime transition process (Geels and Schot, 2007). These
As a result, the objective of this study is to reason out coevolutionary changes have a step in which changes in their broader landscape create a
patterns between technical evolution and social changes by identifying chance of opportunity and the social-technical system reconstructs the
the relationships based on the time series of technology and society existing regime. In other studies, the “socio-technical transition system”
growth. refers to a series of systemic innovations which lead to a fundamental
In this paper, a structural vector autoregressive (SVAR) model and shift in socio-technological systems, based on multiple causal relation­
design structure matrix (DSM) are used to overcome the aforementioned ships and coevolution (Geels, 2005; Markard et al., 2012; Rotmans et al.,
limitations of previous literature. The SVAR model investigates long-run 2001). In this study, the socio-technical transition system views tech­
relationships and coevolution patterns between technology and society, nology and society according to the theory that their relationships
and the DSM is a tool which actively used in the software field to derive evolve by a means of interactions that are not separate and fixed, and in
process optimization. Finally, this paper presents a pattern of coevolu­ other words, are constantly in flux. Existing studies have attempted to
tion between technology and society which can be referenced during the analyze this theory empirically and have incorporated this theory into
creation of ideas for technology development through the prediction of many research areas. Papachristos & Kohler (2011) analysed the rela­
future trends. For this, the proposed approach combines the time-series tionship between technology and society using system dynamics meth­
analysis of the SVAR model, which is a multivariate time-series model, odology. Additionally, Shigeto et al. (2014) has established R&D
with regression analysis to find the correlations and Granger causality program scenarios which take into consideration these issues between
between the relevant variables. DSM, which analyses a square matrix of social and technology in order to reduce carbon dioxide.
two factors to make the complex structures of a system simple and easy There are several types of socio-technical transition systems that can
to understand, is highly suitable for expressing coevolution between be used to understand the various stages or characteristics of transition.
complex technologies and social factors. This study proposes a new Geels & Schot (2007) suggested the following types of transition path­
method for generating technological concepts using socio-technical ways: reproduction process, transformation path, de-alignment, tech­
transition system theory. Thus, the suggested methodology was nological substitution, and reconfiguration pathway. Additionally,
designed to identify the needs of technical development at a practical Dalpiaz et al. (2013) proposed three types of reconfiguration in order to
level that can be more practically and usefully used by a firm. simulate a socio-technical system: task reassignment reconfigurations,
The results of this paper indicate that socio-technical systems are push system reconfigurations, and actuate reconfigurations.
driven by the coevolution of technical variables, such as reliability and
safety, as well as social variables, such as connectivity. This research can 2.1.2. Coevolution between technology and society
contribute to existing literature on socio-technical systems from the Coevolution is the theory that technology and society evolve by
theoretical point of view. First, this study is different from previous means of interaction, not separation and fixation. If technological
studies in that it is the first attempt to investigate the co-evolution be­ determinism and social constructivism are of an open-circuit nature,
tween technology and society by proposing a new methodology that coevolution theory is based on feedback vision. Before coevolution
uses text data for quantitative analysis. This gives other researchers research is explained in this chapter, we will first look at the studies of
insight which would allow them to access and apply co-evolution technological evolution and social evolution and then investigate what
analysis within their field of interest. Second, this study proposes a features they consist of.
systematic approach to analysing the co-evolution patterns of First, the research on technological evolution is based on empirical

2
K. Lee et al. Technological Forecasting & Social Change 176 (2022) 121431

evidence as follows. Lee et al. (2011) used 80 USPTO patents from 1984 & Schot (2007), which understood that social and technological changes
to 2009 in the field of Lithography to monitor technological changes exist as connected socio-technical systems. Kim & Jung (2008)
using methods from formal concept analysis (FCA). They compiled di­ researched coevolution by means of system simulations, and then ana­
agrams of patents with specific concepts according to occurrence fre­ lysed the relationship of hyper phenomena within the evolution of so­
quency and the timing of keywords, and based on the prepared ciety and technology. In this case, the technology-driven approach
diagrams, identified technological changes by rebuilding the annual consisted of a focus on technical maturity, exposure to the media, social
graphs and defining cluster names for each group. Chen et al. (2012) visibility, expectation, investment of technology, maturity of social skill,
conducted a yearly study to track technology clusters in order to confirm and the exposure of an excellent case. However, there are limitations in
the course of the technological evolution which occurred as the tech­ building upon data from intuitive information, there are no empirical
nological sector emerged and disappeared, and used data from 15,948 sources to draw from. The simulation analysis has shown that the source
USPTO patents from 1995 to 2009. In this work, the central commission of the hypothesis is different between the maturity of the technology and
network was established in order to map the connections between the social maturity that accommodates it. After these early studies, many
technologies, and the top patents were extracted using the studies focused on case study analysis, rather than merely presenting a
Givan-Newman algorithm for technology clustering. As a result, the theory or model on the co-evolution of technology and society. Park, S.
order of the evolution of technology for smart grids was explained. For analysed the co-evolution of politics and hydrogen energy technology,
example, they explained that wireless communication technology was and Giones, F., & Brem, A collected data on various technological and
evolving through the various stages of cellular networks in 1999, mobile social issues related to the development of drones, and then organized
station, base station, and ad-hoc networks in 2003, frequency-hopping their development flow (Park, S., 2011; Giones, F., and Brem, A., 2017).
spread spectrums in 2006, and Bluetooth and coordinator-based wire­ Rzevski, G. defined the co-evolution of business and society related to
less networks and backbone networks in 2007. information technology as a new industrial revolution stage, and then
Several studies which analysed the patterns of technological evolu­ analysed the co-evolution between them, and Zeng, Y., et al. performed
tion showed visible results through 2D visualization analysis, and patent a case study analysis of the co-evolution of green technology and con­
data and clustering methods have enabled us to identify the evolution sumers’ pro-environmental attitudes over time (Rzevski, G., 2019; Zeng
patterns for each technological field. In particular, the flow of techno­ et al., 2020). As such, many studies have mainly performed case studies,
logical trajectory was accurately identified using citation information and studies which take a systematic approach to how technology and
from patents. However, in other papers, excluding TRIZ-related ones, it society influence each other remain scarce. Table 1 below summarizes
was difficult to understand the evolution patterns and accurately the papers mentioned above.
monitor technological progress. In addition, the analysis of specific
technological fields still experiences limitations in existing studies. As
they do not take into consideration the external influences of technol­ Table 1
ogy, merely monitoring the evolution can create uncertainty in tech­ Research about coevolution between technology and society.
nological forecasting. At this point, it is necessary to attempt to
Topic Paper Main idea Limitation
overcome these limitations through efforts such as taking into consid­
Technology Lee et al., Monitor technological Considering only one-
eration a greater variety of scenarios, in addition to incorporating
evolution 2011 changes using formal way evolution of
quantitative methods. To this end, we aim to learn more fully about concept analysis (FCA) technology
social evolution, which can be seen as an external shock to technological Chen et al., Monitor the Does not consider the
development, and to also investigate patterns of coevolution between 2012 technological evolution impact on society
technologies and societies. in which the
technological sector
Second, existing studies on social evolution using quantitative
emerged using network
analysis include the following. Song et al. (2016) proposed a method for clustring
analyzing the evolution of service concepts based on social trends, called Social Song et al., Proposed a method of Considering only the
a map of service evolution, which took into consideration product evolution 2016 analysing the evolution one-way influence of
of service concepts social
components. In order to extract service and product keywords, 5000
based on social trends
service documents released on the Apple’s App Store web were crawled Technology Geels and Asserts that social and Does not establish the
and scraped for data. Service feature-related keywords were used, such society Schot, technological changes clear relationship
as acquisition, airplane, and amusement. Lee (2018) analysed a total of coevolution 2007 exist as socio-technical between technology
19 detailed research areas by performing modularity analysis and text systems and society
Kim and Analysed the Building data from
mining which examined the citation relationships of papers. Previous
Jung, 2008 relationship of hyper intuitive information
studies, including the above studies, also made use of a clustering-based phenomena in the because there are no
megatrend selection process. The first step in the process is to collect evolution of society and empirical sources
future forecast data (futuristic data), including megatrend and trend technology
information, and to then categorize them by criteria, such as STEEP Park, S., Analyzed the co- Conducted ex post
2011 evolution of politics and case studies
(social, technological, economic, environmental, and political). Next, hydrogen energy Does not propose a
the final major megatrend is selected by examining emerging issues and technology universally applicable
wildcards, while also grouping the collected data into similar­ Giones, F., Organized the coevolution theory or
ity/commonality. Examples of megatrend keywords include changes in and Brem, development flow of analysis methodology
A., 2017 drone technology and
demographic structure, changes in values of life, well-being, changes in
society
consumption conditions, increases in health risk factors, polarization, in Rzevski, G., Defined the co-
addition to changes in the labor market. These conventional trend 2019 evolution of business
analysis methods could confirm how society has changed the frequency and society related to
of certain keywords for each year, but could not derive universal key­ information technology
Zeng et al., Performed a case study
words which could further elucidate how technology has affected soci­ 2020 analysis of the co-
ety. In this study, it is necessary to extract functional keywords for social evolution of green
trends and match them with technical keywords. technology and
Finally, several studies are underway on the coevolution between consumers’ pro-
environmental attitudes
technology and society. One of the main studies of coevolution is Geels

3
K. Lee et al. Technological Forecasting & Social Change 176 (2022) 121431

2.1.3. Linking the technological elements and social elements


AZt = B(L)Zt− 1 + Ct (1)
Since the methods of analysing technologies and social evolution
differ so widely, collecting the data and defining attributes for each
where Zt is the vector of the endogenous variable, Zt-1 is the vector of an
discipline also vary. The technical data are generally extracted from the
endogenous variable containing a lag, and matrix A is square, based on
technical keywords by investigating product documents (Shin and Park,
n × n, when n is the number of variables which represent the structural
2005; Yoon et al., 2020), or the analysis is carried out by examining the
parameters of endogenous variables within the contemporaneous rela­
bibliographic information and text data of patents and research articles
tionship. Matrix C is also a square matrix, representing the contempo­
(Park and Yoon, 2018; Park et al., 2019). A study using the VAR model
raneous response to a variation in disturbances. L is the lag-operator,
for technological development was conducted by Shin & Park (2005),
and B (L) is the polynomial matrix of the pth dimension; 3t consists of
and it was designed to analyze the relationships between product at­
the disturbance terms of each variable, thereby reflecting the external
tributes by extracting keywords from mobile-phone product manuals
variation of the model and having an E (3t) = 0 as white noise in the
and user guides. Murray (2002) analysed the number of patents, the
(n ×1) dimension.
number of citations, the number of articles, and the coevolution of the
patents and articles by looking at the qualitative chart. Kim et al. (2014)
2.2.2. Design structure matrix (DSM)
investigated coevolution by using an inventor’s network analysis, and
DSM is a square matrix with information from the same row and
used time-based numbers and citation information, which are biblio­
column, representing the relationship between the various components
graphic data that is taken from patents and academic papers. Social data
and elements that make up a system. This tool is a widely used modeling
was mainly collected from people’s personal comments, such as reviews
technique for system analysis, particularly for the decomposition and
and online postings. In order to develop user-oriented service maps and
integration of systems. The DSM enables complex relationships between
discover new service opportunities, Kim et al. (2013) drew a list of
components of a system to be presented within a simple matrix, which
keywords from social psychology literature and analysed them using
can then be used in various analysis techniques. Within the DSM, the
product-review data. They used hashtag keywords from social networks,
meaning of each component is displayed in each row and column, and in
such as Twitter and Facebook, to conveniently analyze consumer
the areas other than diagonal components, the level of dependency be­
interaction data, and these hashtags were then used to search for specific
tween components is displayed. The column in the DSM is the input
words in the form of social metadata (He et al., 2016).
source, and the row represents the output source (Browning, 2001).
Earlier, the data for analysing technology was found by using aca­
DSM is classified as binary DSM or numerical DSM according to the
demic papers and patents widely, especially within the patent scope of a
presentation method used. Binary DSM represents the presence of re­
field or a sector. On the other hand, social data have been mainly ana­
lationships between components, and the numerical DSM can show a
lysed for personal contents that people posted on specific sites. However,
greater variety of information than the binary DSM, as well as the
it is not easy to link two datasets because the sources of technological
presence of a greater variety of relationships. For example, if a business
and social data are so different and have few similar keywords. Song
system is configured as a DSM, the numerical DSM can be used to ex­
et al. (2016) used service and product document data, user function,
press the amount of information transferred between the tasks, the
data source, and technical keywords from the product, and ascertained
probability of change, and the intensity of the effects.
that there are many service content-related keywords in the service. By
Relationships between system components can be classified as par­
means of this process, the evolution of the service was analysed by
allel, sequential, or coupled. Parallel relationships are relationships with
focusing on the concept of product elements.
and without any effect between components, and sequential relation­
ships indicate a flow of one-sided influence. A coupled relationship can
be interpreted as linkage where the variables affect and receive each
2.2. Methodology other, causing repeated changes between the relevant components.
Therefore, coupled relationships are considered to be the most impor­
2.2.1. Structural vector autoregressive (SVAR) model tant relationship type when analysing the interactions between systems
The Vector Autoregressive model introduced by Sims (1980) extends and their coevolution throughout a given time period.
the univariate autoregressive model to create a multivariate autore­ DSM can be used to simplify the pattern of information exchange
gressive model, which can perform various works such as time-series between all tasks which occur in a project. In other words, it expresses
correlation, exogeneity, the causal testing of variables, testing hypoth­ the information necessary to initiate a particular task and what infor­
eses derived from theory, and the dynamic structures and policy analysis mation should be provided from the relevant task. The mark at the top of
of models by means of the variance of effect response functions and the diagonal indicates feedback, meaning that all the information
forecast errors. In particular, the VAR model is a typical time-series required for the work is not available at the beginning of a task. Thus,
model used to eliminate the distinction between internal and external the tasks affected by the feedback mark can be viewed as repetitive
variables, and make the most of the information in any given dataset, as work.
the identification problem of the existing structural model remains too
arbitrary. This model is a system of equations, each of which not only 3. Proposed approach
contains lagged dependent variables for each relevant variable, but also
includes lagged dependent variables for all other variables in the model. 3.1. Research concepts
In addition, as the model is small and can be easily applied to most
economic analyses, the VAR model is widely used not only for economic The purpose of this study is to derive the relationship between
forecasting, but also for economic change analysis. However, this model technology and society, and to predict future coevolution patterns be­
was pointed out as being different from those used in traditional eco­ tween technology and society. To achieve this, we take keywords that
nomic theoretical models, and arbitrary in determining the number of were extracted from technological and social data that was collected
lags that determined the selection of variables and the dynamic rela­ from various databases, and make use of the technical and social attri­
tionship between the variables (Cooley and LeRoy, 1985; Leamer, butes defined from the existing literature. The defined technical attri­
1985). For this reason, the SVAR model was then developed in order to butes include keywords that describe the characteristics of the particular
overcome the limitations of the VAR model. technology, and social attributes are identified by keywords that
The SVAR model considers a simultaneous-equation model (1) based describe the characteristics of the society that are affected by or attrib­
on vectors, and consists of an autoregressive process in n multivariate uted to the development of the technology. After defining the attributes,
stationary time series Zt = (Z1t, Z2t, …,Znt), which have p past lags.

4
K. Lee et al. Technological Forecasting & Social Change 176 (2022) 121431

we use the DoV (degree of visibility) to calculate the quantitative value from the websites and patent databases. Nowadays, information on so­
of each attribute. Then, the DoV values of each attribute in each time ciety and technology can be obtained from various online sources,
period are applied to the SVAR in order to analyze the time-series cor­ including official websites, patent site, news, and blogs. Within our
relations between the technical and social attributes. Since the results of study, the data sources were not limited, but we required that social and
SVAR only exhibits one-to-one correlations between single attributes, technological documents most generally need to include substantial,
many-to-many relationships must be analysed in order to more fully see essential information, such as their date of publication, features of the
the coevolution of technology and society. Therefore, when making use society and its technological developments, and information about their
of technology DSM, which is a methodology that is used to prioritize the interactions. Since the social and technological documents remain in the
relationship between elements, we rearrange technology and social at­ form of unstructured data in text format, they need to be transformed
tributes and then identify the sequential relationships between the at­ into structured data for further meaningful analyses. In this study, social
tributes. The final results from applying the Kosaraju algorithm to the documents are transformed into keyword vectors, which consist of basic
DSM matrix, derived from the previous step, is the SCC (strongly con­ information on social features and related technological features, and
nected components) of the technical and social attributes of the coevo­ technological documents are transformed into keyword vectors con­
lution pattern. The Fig. 1 depicts the concepts of this study. sisting of technology information.
Prior to the construction of keyword vectors, the social and techno­
3.2. Database and data collection logical documents are pre-processed. Since the characteristics and for­
mats of the documents depend on their sources, the detailed activities
In this study, data can be divided into technical data and social data for parsing social and technological documents and the extraction of the
in order to detect coevolution between technology and society in each required information needs to be done on a case-by-case basis. This
database. Patent data from USPTO was collected in order to analyze study cannot cover all coevolution cases, but the general process can be
technology, and news text data from the USA’s top news sites, such as described as follows. First, each social document is parsed based on its
USA Today, The New York Times, Los Angeles Times, The Wall Street structure in order to extract the published date and description,
Journal, and The Examiner, were used in order to extract information on including technological information, and the technological documents
social topics and trends. News and blog posts including contents that are parsed based on their given technological features.
offered insights into technological trends, as well as social issues, were Stanfard CoreNLP is a natural language processing software distrib­
also analysed to see if there was a link between technology and society uted by Stanford NLP Group in the UK. Typically, a variety of tasks can
(Kim et al., 2016; Schatzmann et al., 2013). be performed with it, such as named entities recognition, Part-Of-Speech
tagging, word segmentation, reference resolution, sense analysis, in
addition to bootstrapped patterning (Manning et al., 2014). In partic­
3.3. Process
ular, the named entities recognizer can identify words that correspond to
predefined places, times, and units, and uses POS Tagger and the
As Fig. 2 shows, the proposed approach for developing patterns of
grammar information of each word in order to extract keywords. In this
coevolution between society and technology consists of three main
study, the keywords of social and technical documents are extracted and
stages: data collection and pre-processing, analysis of interrelationships
stored in relational databases, patent text data and news text data are
between society and technology, and the detection of promising
collected by means of patent databases, and Web scraping is done from
technology.
PROQUEST. However, web data did not provide long-term news data for
In the data collection and pre-processing stage, news documents
collection methods, limiting the scope of collection in the PROQUEST
describing the target technology are collected from the websites and
database. We then selected a specific technology in the ICT industry as a
then transformed into quantitative data. Based on the quantitative data,
target for analysis. The selection criteria consisted of selecting tech­
within the stage of analysing the interrelationships between society and
nologies that have been investigated in terms of coevolution in existing
technology, coevolution patterns are examined by various methods.
studies.
Promising technology can be developed in their final form on the basis of
the coevolutionary patterns between society and technology.
3.3.2. Extract keywords of society and technology
Existing technological attributes that are generally used in technol­
3.3.1. Collect data
ogy analysis explain how technology should be developed, but they do
After the target field for the analysis on coevolution of technology
not meaningfully describe the components of technology itself. In
and society is chosen, news documents, which represent social data, and
addition, this property has no social perspective, making it difficult to
patent documents which represent technological data, are collected

Fig. 1. Research concept.

5
K. Lee et al. Technological Forecasting & Social Change 176 (2022) 121431

Fig. 2. Research process.

understand the causal relationship between technology and society. keywords that encompass the characteristics of the technology. Since
Since it is difficult to apply the comparison of traditional technological the analysis target is IT technology, IT terms are extracted in a time
attributes such as weight, length, area, and volume to the developments series from the summary and billing information of patent documents
in the IT industry, these technological attributes are not appropriate for based on the IT terminology dictionary. Social attributes are intended to
the purposes of this research. Jeong & Yoon (2015) scored different encompass the characteristics of the society as it has been affected or
kinds of technological components based on the checklist of the influenced by technological development, and the methods use the
SCAMPER method, which consists of concepts and uses such as the following steps. First, this research collected social phenomena changes
ability to substitute, combine, adapt, modify, put to other uses, elimi­ and social properties by drawing from existing literature. Second, we
nate, and reverse, and thereby extracted seven patterns of patent make a checkpoint, and then select social properties according to the
development. To predict the future of active matrix organic given criteria. Checkpoints should be in relation to society, not how the
light-emitting diode (AMOLED), a similar technology, such as light individual changes, and the meanings of the words should not overlap.
emitting diode (LCD) pattern, was analysed and then used. Finally, when taking these checkpoints into consideration, this study
By deriving trend keywords, social and technical attributes can be defines societal attributes as the influencing factors that the final tech­
derived by using topic modeling techniques, which include latent se­ nology can give. In addition, if there is a field that is difficult to obtain
mantic analysis (LDA), bi-term topic model (BTM), and multi-grain LDA through existing research, the researcher selects social attribute key­
(MG- LDA), which derive changes in the year-to-year topics from social words using the terms which appear in the technical terminology dic­
data, and then extract topics that may fall within the previous year as tionary, social terms that can be affected by the technology selected by
social attributes. In this paper, social attributes are defined by taking the researcher, in addition to words from the collected data.
into consideration several criteria, such as an association with a specific The quantitative value of each defined attribute was based on the
technology. However, since the properties derived by topic modeling patent registration of the USPTO database from among the various
methods were too general and were found to not give critical informa­ patent bibliographic criteria, and the number of keywords about news
tion, technology and social properties which were considered important related to the social evolution analysis. For the pre-processing of the
in other literature were collected in this study. data, the text-mining techniques extracted parts of the document’s
As an example of the technical attributes of existing studies, Hyeon contents to create a document-word matrix, and the word frequency of
(2008) suggested that the evolution of the information technology (IT) the corresponding period-specific documents were established as final
field was embedded within three aspects: intelligence in terms of future data. At this time, the number of documents may differ because their
information processing, mobility within respect to networks, and level interests in the field are flexible, and the frequency of the keywords may
of embedment in view of service promotion for usage of the IT as a vary greatly over the course of the year, depending on the number of
whole. Kim et al. (2010) drew from eight attributes of IT convergence documents. Therefore, this research seeks to reduce reliance on the
technology, including, location, time, identity, reliability, connectivity, number of documents by using the following Document Visibility
security, operationality, and autonomy, among others. The trend prop­ indices (DoV) (Yoon, 2012).
erties can be divided into the categories of change, persistence, stability, ( )
TFij
and volume, and the description of each attribute is as follows (Oh et al., DoVij = × {1 − (n − j)}
NNj
2009). Change refers to the degree of variation, such as the difference
between the start and end of a trend curve for a given technology, In this case, TFij refers to the total number of occurrences of a
persistence increases and decreases along a certain direction, stability keyword i in the given period j, and NNj is the total number of news
increases evenly without changing the width of the increase and articles over the period j, when n is the number of periods.
decrease of change, and volume refers to the degree of cumulative fre­
quency values independent of the shape of the trend curve.
To define the technical attributes of this study, we first selected

6
K. Lee et al. Technological Forecasting & Social Change 176 (2022) 121431

3.3.3. Analysing the effects of interrelationships between society and tech­ must be terminated before the component B can be seen as a structure to
nology. To derive the relationship between society and technology, the start. Step 4–2 applies the partitioning algorithm in order to change the
SVAR model was used as the methodology of this study. The SVAR order between the attributes. DSM provides a variety of algorithms
model is a multivariate time-series model that allows for the estimation based on different relationships, such as Partitioning, Tearing, Banding,
of the correlation and causal relationship between variables by and Clustering (Browning, 2001). This study implements the partition­
combining the characteristics of time-series and regression analysis. In ing process, which is often used to prioritize based on the associations
particular, the SVAR model, unlike the existing VAR model, presents between components. The partitioning process is set to rearrange on the
non-recursive general provisions in the current term structure factor, matrix of the DSM for the purpose of aligning all existing associations
and analyses the empirical results of the long-run restrictions model close to or on one particular side of the diagonal. Examining the
rather than that of merely the short-term constraints. The results of this segmented DSM, we can see which attributes appear sequentially, which
analysis are the relationship and significance of the variables in each properties are parallel, and which tasks are interdependent, and thus
interval, which is expressed in lags. The lag in the VAR model can be identify their simultaneous results or coevolutionary properties. By
defined as a time bucket that is divided over the course of a time horizon. means of this step, each attribute can be arranged to gain visibility over
For example, if we set the lag with p= 6, there are 6 sections within the the process of optimizing the matrix, and social and technical attributes
time dimension. can then be relocated to identify the effects and the sequential re­
In order to analyze the effect between attributes, each time-series lationships between attributes. The partitioning process consists of
model must be investigated for error terms or disturbances, which are seven steps which initially enter the matrix in the existing order and
used by the SVAR model in this study, and the data uses the time-specific place the empty line first (Gebala and Eppinger, 1991). Second, the next
document visibility index. empty column is removed and placed at the end of the column, and the
rest of the relationship can then be examined. If there is no other rela­
AZt = A1 Zt− 1 + A2 Zt− 2 + … + Ap Zt− p + et
tionship, the two-way relationship is considered first in order to derive
the cellular automata (CA). After all, considering the remaining re­
where Zt is the (n ×1) vector and et is the (n ×1) structural effect vector;
lationships when there is no next two-way relationship, it is possible to
et assumes that there are n structural impulse terms that are not auto­
identify that, for example, B and G are related, and then eventually
correlations, and that there are no correlations between these given
configure the partitioning process. Finally, Step 4–3 consists of ana­
parts. The variance is given by Var(et ) = E(et et ), which is defined as a

lysing the correlation between each attribute and the focus of the entire
diagonal matrix of et , which consists of the distribution of structural block. In this study, each block is interpreted to describe aspects of the
impulse terms. This function can be expressed as follows. evolutionary process.
AZt = C(L)Zt− 1 + et
3.3.5. Apply and finalize next generation technology and society trend.
C (L) in formula (2) is the matrix finite-order lag polynomial of the This final step seeks to draw up a list of technical ideas that give insight
function of the parallax operator L. A matrix is a current-term structural to the stage of developing ideas in the technological development phase,
coefficient matrix which represents the contemporaneous relationship and to anticipate the final technical ideas and future social direction. As
between variables. a result, the expected outcome is expected to show attributes in tech­
The SVAR model can be used to find the causal relationship, based on nology and society category, so the following technology should be
calculating the Granger causality between the attributes. However, the developed.
VAR model is not adequate for creating, organizing, and managing the Research which elicits promising technologies from existing tech­
cyclical relationships between the technological and social attributes. nologies and features of the next generation of technologies is as follows.
The use of process architecture DSM, presented in the next section, can Lee et al. (2018) defined that emerging technologies use multiple con­
help address these issues in the methodology. tent indicators. Joung & Kim (2017) monitored emerging technologies
using patent data. Thorleuchter & Van den Poel (2015) attempted idea
3.3.4. Identify patterns of coevolution between society and technology. mining from multiple websites. However, existing studies did not take
Using the results that drew one-to-one sequential causal relationships into account the consequence relationship on how the predicted prom­
between social and technical attributes at any given time in Step 3, Step ising technology affects society and its various social needs. In this
4 seeks to identify the relationships of the multiple sequential and respect, the results presented in this study can suggest more
cyclical attributes. As the methodology in this phase, the Design Struc­ multi-faceted and dynamic implications than the results of the previous
ture Matrix (DSM), developed by Steward for the analysis of design study.
specifications, was implemented in the next part of this study. DSM is a
tool used to identify the relationships of complex system components, 4. Results
and is often used to model dependencies among system components
within a matrix in order to define closely related components, or to 4.1. Data
confirm priorities among components (Steward, 1981). A previous study
suggested a methodology which focused on the relationship analysis of In order to illustrate the applicability of the proposed approach, we
the roadmap layer for creating a technological roadmap (Son et al., analysed a case of artificial intelligence (AI) by using artificial intelli­
2018). The study drew up how the links between technologies, products, gence technology and the social phenomena for in-depth analysis in one
and markets were connected sequentially by measuring the proximity field. The concept of artificial intelligence first appeared when Turing
and importance of each element in each layer. (1950) proposed developing a machine or system that could think and
Step 4 of this study proposes a three-step process for eliciting the speak like a person in his paper. Artificial intelligence has emerged as a
coevolution patterns between society and technology using the DSM. key growth engine of the fourth industrial revolution, as well as insti­
First, Step 4–1 applies the DSM matrix to the relationship of mutual gating a global paradigm shift that has become a hot topic in recent
effects between technical and social attributes derived from Step 3. The years, and it remains an area in which active research is being actively
technology and social attributes derived from the previous phase were carried out. AI and the fourth industrial revolution are expected to lead
then configured as activities in DSM. The significance and coefficients of huge megatrends across society, including the weakening of the
the inter-effect relationship derived from the VAR analysis were used to human-labor market, the undermining of human dignity, extreme po­
identify the dependencies between the attributes. At this point, the DSM larization, and the prolonging of human life (Schwab, 2017). In
matrix is only in its sequential form, in which case the A component

7
K. Lee et al. Technological Forecasting & Social Change 176 (2022) 121431

addition, many experts are interested in how AI technology will change and 3 words were newly proposed in this study. Table 2 below shows the
our lives in the near future (Stone et al., 2016). Therefore, the artificial definition of the finally defined attributes and the literature reference.
intelligence technologies that affect society and vice versa serve as There are seven attributes of technology-tangibility, reliability, per­
appropriate data for analysis in this study to see the coevolution of sonality, safety, intelligence, diffusion, and accessibility – in addition to
technology and society. the seven attributes of society – intellectualization, informatization,
stability, efficiency, polarization, connectivity, and serviceability.
The AI-related patents and news data collected in Step 1 were used to
4.2. Process
obtain the values of the attributes of artificial intelligence technology.
To pre-process text data, noise filtering and stemming were used to
4.2.1. Step 1. Collection of data
replace the entire patent document with the stemmed contents, and then
The technological and social data from this study were collected
words were extracted from the sentence in order to identify the quarterly
from different databases. Patent data are collected from USPTO, the U.S.
instances of each year. As a result, 60,882 words were gathered from the
Patent and Trademark Office, and a total of 18,184 registered patents
patent data, and 17,990 words were gathered from the social data. We
were collected from 1st January 1976 to 1st December 2018, using
selected specific words related to 42 technological attributes and 26
keywords such as ’machine learning’ and ’artificial intelligence’ to
social attributes among the words, and then created the final document-
analyze the bulk of information related to artificial intelligence tech­
word matrix, which arranged how many times each word came out in
nology. News data were retrieved by using the ProQuest database to
the quarter of the year. Next, the document visibility index was calcu­
analyze a total of 1951 English web news articles, which were searched
lated on the basis of the document-word matrix.
through with the keywords ’artificial intelligence’ from 1985 to 2017.
Among them, in order to analyze the causal relationship between patent
4.2.3. Step 3. Analyze the effects of interrelationships between society and
data and news data from a common point, data from 1987 to 2017, the
technology
last 30 years, were used for VAR analysis.
The purpose of this step is to analyze the effects of the overall in­
terrelationships between society and technology. For example, the
4.2.2. Step 2. Extraction of keywords of society and technology
tangibility level of technology can increase the accessibility of the
Technological attribute was defined as among the top 7 words that
technology in its early days, whereas the accessibility of the technology
showed high frequency in patent document summaries and claims,
affects the reliability, personality, and intelligence of the technology.
among all words mentioned in existing technological attribute studies or
Accessibility, in particular, continues to affect improvements in the
in social news while also appearing in the IT term dictionary. As a result,
personality and intelligence for all periods, which may be even greater if
5 attributes related to the study of Hyeon (2008) and the study of Kim
there is additional diversity. Before analysing using SVAR, an impulse
et al. (2010) were selected, and 2 attributes were also suggested by the
function is used to analyze the effect between the attributes. Impulse
authors of this study, and a total of 7 were thus derived. For social at­
Response Function was used to examine the influence, direction, and
tributes, the researcher selected 7 words that were judged to be influ­
duration of variables included within the model. As an example, the
encing factors which technology can give to society, from among the
impulse response of tangibility (t1) means a dynamic response in which
words in the news. The final selected words were the same as the 4
the tangibility (t1) changes when a 1 standard deviation shock is given
characteristic words of society that were mentioned in the studies by
to the prediction errors of the 7 descriptive variables and the 7 social
Park (1989), Park (1995), Oh et al. (2009), and Frey & Osborne (2017),

Table 2
Technological and social attributes.
Category Attributes Keywords Definition Author

Technology Tangibility (t1) Reusable, Wearability, Configurate, Decompose, Portable, Attribute that refers to the shape or form of a technology Hyeon (2008)
Embeddedness, Durability
Reliability (t2) Reliability, Usability, Trust, Testability, Credibility Attribute that means a technology precisely performs a Kim et al.
specific action under defined environmental conditions (2010)
Personality (t3) Sensibility, Personality, Tailor, Adjustable, Customize Attribute that means that technology provides desired Suggested
content to specific groups or individuals
Safety (t4) Security, Stability, Safe Attribute that means that there is no risk of danger or Kim et al.
accidents caused by technology (2010)
Intelligence (t5) Learnability, Predictability, Interpretable, Autonomy, Attribute that means that technology includes both Hyeon (2008)
Maintainable, Analyze, Tracking, Reasoning, Automation cognitive and learning abilities to solve a problem
Diffusion (t6) Nomadicity, Extensibility, Mobility, Scalability Attribute that means that spreading widely in fields Hyeon (2008)
where the technology has not spread
Accessible (t7) Download, Accessibility, Compatibility, Ubiquity, Attribute that means that the technology is easy to Suggested
Adaptability, Understandable, Interoperability, Connectivity, understand and easy to access
Integrability
Society Intellectualization Semanticity, Intelligent, Intelligence, White-collar, Attribute that refers to the phenomenon in which Suggested
(s1) Sophisticate, Brainiest sensory and cognitive functions occur in daily life
Informatization (s2) Information-based, Globalization, Knowledge-based A property that refers to a phenomenon in which Park (1989)
information is the core to create value
Stability (s3) Anxiety, Security, Stability, Safe A property that refers to the phenomenon of maintaining Oh et al.
a constant state without changing (2009)
Efficiency (s4) Productivity, Expertise, Professionalism, Adeptness, A property that refers to a phenomenon in which the Suggested
Proficiency desired result is greater than the effort put in
Polarization (s5) Inequality, Polarization A property that refers to a phenomenon that gradually Frey & Osborne
changes and moves away from each other in a specific (2017)
part
Connectivity (s6) Relationship, Cooperation, Harmonization A property that refers to a phenomenon in which the Suggested
importance of connectivity between different parts
increases
Serviceability (s7) Convenience, Comfortable, Serviceability A property that refers to a phenomenon that the pursuit Park (1995)
of convenience

8
K. Lee et al. Technological Forecasting & Social Change 176 (2022) 121431

variables. Fig. 3 presents as an example, each of the responses of the 7 variables. Initially, it has a positive effect on safety and accessibility, and
technological attributes to the shock of the technical attributes, among has a negative effect on intelligence, and in the mid-term, it was found to
the broader results of the shock response analysis. have a negative effect on reliability, a negative effect on personality in
In order to find out whether the VAR model should be constructed as the end, and a positive effect on tangibility. The different technological
a level variable or as a primary variable, the ADF (Augmented Dickey- factors which affect each period can thus be identified. On the other
Fuller) test and the KPS (Kwiatkowski–Phillips–Schmidt–Shin) test hand, accessibility has been shown to have a positive effect over all the
were conducted to check the unit root, which showed that all the unit years on the general improvement of personality and intelligence.
roots existed in the level value of the model variable, and the first dif­ What each society attribute has to do with other society attributes is
ferential value transformed into a stable time series. Information criteria presented in Table 4. Intelligent social phenomena have had an overall
such as AIC (Akaike information criteria), HQIC (Hannan–Quinn infor­ negative effect on stability while contributing positively to the forma­
mation criteria), and SBIC (Schwarz’s Bayesian information criteria) are tion of a serviceable society. In other words, although many social
then calculated to find the lag p of the VAR model. From this result, we changes are caused by intelligence, service ability is seen as having
verify that the greater the time difference, the greater the absolute value increased, and the polarization of social properties has not been signif­
of each information reference value. We judged that it would be icantly affected. The bipolar social phenomenon, which recognizes dif­
reasonable to measure the lag of the VAR model at 5 or 6 by taking into ferences, has a positive effect on the formation of an intelligent society
consideration some of the information basis values, while reflecting the over all of the periods. Accessibility affects intelligence, information,
dynamic movements between the variables as much as possible. Next, and efficient social phenomena, and convenience has a negative effect
the co-integration test of whether there is a relationship between the on stability and accessibility, but it has an overall positive effect on
components, and in the case of p= 5, there was no co-integration relation encouraging the formation of an efficient and diverse society.
at the 1% significance level. Therefore, the final selection of p= 5 in this In order to find the heterogeneous nature of technological and social
paper was chosen for analysis. In other words, there are 5 time lags over relations, the effects of technology on society and the effects of society
the time horizon, meaning 30 years used in the input data can be divided on technology were also analysed. Firstly, technology affects society as
into five sections and each lag has 6 years as a time bucket. shown in the following Table 5. Tangible technology reduces efficiency
Table 3 presents what each technology affects. The tangibility and the functioning of an approachable social phenomenon (connec­
initially has a positive effect on the accessibility of the technology, but a tivity), and while reactivity reduces the efficiency of an effective social
negative effect on its reliability. In the end, it was found to have an phenomenon, it has been shown that it also increases overall accessi­
overall positive effect on safety and intelligence, but an overall negative bility. Personalized technology reduces effective social phenomena, and
effect on performance and accessibility. Diffusion is the only attribute safe technology also has a negative effect on the stability but a positive
among the seven technical variables that affects all of the technical effect on the bipolar phenomenon. Finally, diffusion techniques have an

Fig. 3. Impulse Response Function: technology to technology.

9
K. Lee et al. Technological Forecasting & Social Change 176 (2022) 121431

Table 3
Results of SVAR analysis: technology to technology.
Posterior variable Lags
1 2 3 4 5

Prior variable: Tangibility (t1)


Reliability (t2) − 0.1219 − 0.2518** − 0.1843 − 0.1438 − 0.0090
Personality (t3) 0.0472 − 0.0874 0.0412 − 0.1717 − 0.3103***
Safety (t4) 0.0827 − 0.0262 0.1211 0.1475 0.2401***
Intelligence (t5) − 0.0820 − 0.0517 0.0673 0.0690 0.2521**
Diffusion (t6) 0.0882 0.0907 0.1294 0.0037 0.1270
Accessible (t7) 0.1983* 0.1176 − 0.0354 0.0438 − 0.1918*
Prior variable: Reliability (t2)
Tangibility (t1) 0.0645 0.1730 0.1206 − 0.0058 0.1777*
Personality (t3) 0.1033 − 0.0442 − 0.1834 0.0090 0.0594
Safety (t4) 0.0993 0.0036 − 0.1185 − 0.0542 0.0003
Intelligence (t5) − 0.0112 0.0322 0.1088 0.0617 0.0841
Diffusion (t6) − 0.0121 0.1509 0.0157 0.0528 − 0.0167
Accessible (t7) 0.0002 − 0.0658 − 0.1059 0.0472 0.0721
Prior variable: Personality (t3)
Tangibility (t1) − 0.1506 − 0.0746 0.0764 0.1554 0.0077
Reliability (t2) 0.3076*** 0.3801*** 0.4488*** 0.4074*** 0.4038
Safety (t4) 0.1736* − 0.0218 − 0.0525 0.0090 0.1724**
Intelligence (t5) 0.0319 − 0.0695 0.2249* 0.0490 0.0342
Diffusion (t6) 0.1569* 0.0810 − 0.0067 − 0.0327 − 0.0383
Accessible (t7) − 0.0555 − 0.0095 − 0.1440 − 0.0069 − 0.0227
Prior variable: Safety (t4)
Tangibility (t1) − 0.1750* − 0.0967 − 0.3574*** − 0.2972** − 0.2467***
Reliability (t2) 0.0317 0.0625 0.1176 0.1453 − 0.1008
Personality (t3) 0.1855** 0.3475*** 0.0793 0.0113 − 0.0975
Intelligence (t5) − 0.0474 − 0.0650 0.0578 0.0943 0.0988
Diffusion (t6) 0.0685 − 0.0170 0.1437 0.1060 − 0.0285
Accessible (t7) − 0.0836 − 0.1431 − 0.1396 0.0003 − 0.0001
Prior variable: Intelligence (t5)
Tangibility (t1) − 0.1848* − 0.2783** − 0.0088 − 0.2427* 0.0000
Reliability (t2) − 0.1163 − 0.1383 − 0.1206 − 0.0592 − 0.1089
Personality (t3) 0.0900 0.2562** 0.4234*** 0.5052*** 0.3616***
Safety (t4) 0.0563 0.1256 − 0.0673 − 0.0640 − 0.0249
Diffusion (t6) 0.0956 − 0.0374 0.1172 0.0480 0.1197
Accessible (t7) − 0.6368*** − 0.6000*** − 0.6904*** − 0.2243 − 0.3307***
Prior variable: Diffusion (t6)
Tangibility (t1) 0.0016 − 0.0087 − 0.0937 0.2688* 0.2740**
Reliability (t2) − 0.0723 − 0.3712*** − 0.5809*** − 0.3735*** − 0.0770
Personality (t3) − 0.0681 − 0.0627 0.0205 − 0.3051** − 0.3055***
Safety (t4) 0.2385** 0.2878** 0.1339 0.1381 0.0490
Intelligence (t5) − 0.0464 − 0.3259*** − 0.2442* − 0.0230 0.0337
Accessible (t7) 0.2711** 0.5494*** 0.3802** 0.2486 − 0.1104
Prior variable: Accessible (t7)
Tangibility (t1) − 0.1239 − 0.0207 − 0.1107 − 0.2303* − 0.0853
Reliability (t2) 0.1982** 0.0673 0.1228 − 0.0068 − 0.0462
Personality (t3) 0.1666** 0.1985* 0.5050*** 0.5838*** 0.3110***
Safety (t4) − 0.0657 − 0.1397 − 0.1003 − 0.0410 0.0500
Intelligence (t5) 0.2706*** 0.3121*** 0.7539*** 0.3276** 0.2649***
Diffusion (t6) − 0.0046 − 0.0472 − 0.0210 − 0.0054 0.0258

overall positive effect on bipolar social phenomena, and it was found 4.2.4. Step 4. Identify patterns of coevolution between society and
that intelligence and accessibility do not affect social phenomena. technology
On the other hand, the effects of society on technology are shown as At this stage, the most significant Granger causal results from Step 2
follows in Table 6. The intelligent society and the information society are entered into the DSM, and the sequential relationships of target at­
are not affected by the specific nature of technology. However, they are tributes are rearranged into a matrix. Project DSM software was used for
affected by the development of technology. The higher the social in­ analysis, and causal data can be transformed into a DSM in order to
telligence phenomenon is, the more reliable and secure the technology is account for and visualize the process characteristics. In this study, only 1
in the long-run, while also activating the diffusion of the technology. In or 0 applies to DSM to the extent that the Granger causal relationship by
particular, while the social phenomenon initially negatively affects the attribute is significant at the 90% level. For example, if the T2 attribute
reliability, it contributes to an increase on reliability later. Stable social is used for T3, S1, and S2, an association matrix can then be constructed,
properties seem to have a negative effect on resilient technical proper­ as shown in Fig. 4, by marking 1 at the point where the columns of T2
ties. This polarization has been attributed to less-widespread technol­ technological intersect with the rows of T3, S1, and S2. This then means
ogy, and accessibility has been shown to be positive for the specific type that attribute T2 can be presented after attribute T1 has finished. The
of technology. Social phenomena seeking service ability have a negative strength of the effect flow from attribute T1 to attribute T2 is 0.243 (the
effect on personalization techniques, but they have a positive effect on value in parentheses). After attribute T2 is developed, attribute T3 can
safety and diffusion technologies. It has been revealed that only effi­ be begun by receiving the effectiveness from task T2 with a value of
ciency among social properties does not affect any technological 0.245.
properties. Subsequent application of the Partitioning DSM to this relationship
matrix results in priorities, as shown in Fig. 5, based on the fore

10
K. Lee et al. Technological Forecasting & Social Change 176 (2022) 121431

Table 4
Results of SVAR analysis: society to society.
Posterior variable Lags
1 2 3 4 5

Prior variable: Intellectualization (s1)


Informationization (s2) 0.0191 0.0201 0.0805 0.1236 0.0181
Stability (s3) − 0.0869** − 0.2847 − 0.2395* − 0.1785 − 0.1836*
Efficiency (s4) 0.0092 0.1896 0.1364 − 0.0857 − 0.0381
Polarization (s5) 0.1087 0.1624 0.0233 − 0.2272 − 0.1691
Connectivity (s6) 0.1031 0.1438 0.0418 0.1613 − 0.0737
Serviceability (s7) − 0.0733 0.0379 0.1741 0.5162** 0.0813
Prior variable: Informatization (s2)
Intellectualization (s1) − 0.0101 0.0252 0.0387 − 0.0302 0.0234
Stability (s3) − 0.0904 − 0.2893** − 0.2510** − 0.0644 − 0.2710***
Efficiency (s4) − 0.0985 − 0.1469 0.0056 − 0.1105 − 0.0473
Polarization (s5) − 0.0956 − 0.0772 − 0.0661 − 0.1404 − 0.0577
Connectivity (s6) − 0.0162 0.0136 − 0.0673 0.0290 − 0.0943
Serviceability (s7) 0.1624 0.0331 − 0.0452 − 0.2520 − 0.0575
Prior variable: Stability (s3)
Intellectualization (s1) − 0.0596 0.0505 0.0546 0.2113** 0.0861
Informationization (s2) 0.1239 0.1318 − 0.0542 − 0.0764 0.0558
Efficiency (s4) 0.0029 − 0.0518 − 0.1617 − 0.3712*** − 0.1458
Polarization (s5) − 0.0250 − 0.0393 − 0.0331 0.0972 0.0385
Connectivity (s6) − 0.0217 0.0080 0.1980 0.0368 0.2374**
Serviceability (s7) 0.0007 0.0598 0.1028 0.0801 − 0.0691
Prior variable: Efficiency (s4)
Intellectualization (s1) − 0.0777 − 0.0183 − 0.1207 0.0081 0.0982
Informationization (s2) − 0.0530 − 0.1648 − 0.0746 0.2901** 0.1945**
Stability (s3) 0.0269 0.1103 − 0.0054 0.0914 0.1378
Polarization (s5) − 0.0657 − 0.0880 0.0729 0.1781 0.0088
Connectivity (s6) − 0.0501 − 0.1213 0.0827 0.3455*** 0.2113*
Serviceability (s7) 0.0146 − 0.0700 0.3766* − 0.0750 − 0.2371
Prior variable: Polarization (s5)
Intellectualization (s1) 0.1107* 0.1691** 0.3066*** 0.2753*** 0.1547**
Informationization (s2) 0.0419 − 0.0379 − 0.0548 − 0.0487 0.0001
Stability (s3) 0.0087 0.0520 0.0497 0.0014 − 0.0335
Efficiency (s4) − 0.0526 − 0.0783 − 0.0790 − 0.0584 0.0473
Connectivity (s6) − 0.0030 0.0306 0.0725 − 0.0546 − 0.1164
Serviceability (s7) 0.2932** 0.1306 − 0.0976 − 0.3062** − 0.1489
Prior variable: Connectivity (s6)
Intellectualization (s1) − 0.0371 0.0581 0.1120 0.1651* 0.0479
Informationization (s2) − 0.0321 0.0768 0.2454** 0.1335 0.0787
Stability (s3) 0.1108 − 0.0435 − 0.0770 − 0.0006 0.0114
Efficiency (s4) 0.2679*** 0.2358** 0.2519** 0.2231** 0.0352
Polarization (s5) − 0.0601 0.0275 0.0924 0.1413 − 0.1795
Serviceability (s7) − 0.1302 − 0.0697 − 0.2648 − 0.2617 − 0.0350
Prior variable: Serviceability (s7)
Intellectualization (s1) 0.0094 − 0.0521 − 0.0110 0.0261 0.0618
Informationization (s2) 0.0556 − 0.0058 0.0522 0.0278 − 0.0372
Stability (s3) − 0.0533 − 0.1692** − 0.1179 − 0.0755 − 0.1132**
Efficiency (s4) 0.1545*** 0.2183*** 0.0754 0.0895 0.0523
Polarization (s5) 0.0924 0.0673 0.1990 0.1427** 0.0550
Connectivity (s6) − 0.1812*** − 0.1103 − 0.0515 − 0.0007 0.0654

relationship between the elements. Kosaraju’s algorithm is applied to technology for security appears again after a long period of stagnation.
cluster the DSM in order to make several strongly connected components The second is the emergence of diffusion technology, which imme­
(SCC) into a coevolution pattern. The derived coevolutionary pattern diately manifests itself in society, and is a pattern where diffusion
has ‘diffusion → polarization → safety → stability → intellectualization technology declines after approximately 4 lags (for a period of 24 years).
→ serviceability → efficiency → connectivity → informatization’ and As the diffusion technology declines, the differentiation service declines
‘reliability → tangibility → personality → intelligence → accessible’ again, and then again when the technology spreads again after 4 lags
(for a period of 24 years). In other words, it can be seen that diffusion
5. Implications technologies appear for about 8 lags (for approximately 50 years) with
differentiated services. Although safety technologies continue to in­
There are five major technologies and social coevolution patterns crease their technological prowess, the two technologies differ in terms
that could occur within the next 60 years (for a total of 10 lags) for of the fact that they appear once again after becoming obsolete.
prediction in reference from Table 3 to Table 6. First, the emergence of The third is the accessibility technology, an isolated technology that
safety technologies leads to the emergence of services and then, the directly affects or does not affect services. Whereas other technologies
safety technologies emerge again. With the advent of safety technolo­ temporarily influence the development of other technologies, accessi­
gies, polarization immediately appears, and these services are derived bility technologies continue to affect the development of personalization
directly from intelligent services. After about 4 lags (for a period of 24 and intelligent technologies for a longer period of time. On the other
years), when intelligent services emerge, safety technologies will hand, they have not affected the appearance or decline of the service,
emerge again. After the emergence of these intelligent services, there is a and have not been affected by any other service. In other words,
need to supplement various services, and the result is that the accessibility technology can be viewed as a technological area that is

11
K. Lee et al. Technological Forecasting & Social Change 176 (2022) 121431

Table 5
Results of SVAR analysis: technology to society.
Posterior variable Lags
1 2 3 4 5

Prior variable: Tangibility (t1)


Intellectualization (s1) 0.0003 0.0147 0.0052 0.0731 0.0116
Informationization (s2) − 0.0391 − 0.0665 − 0.1127 − 0.1139 − 0.0330
Stability (s3) − 0.0180 − 0.0649 − 0.0184 0.0004 − 0.0132
Efficiency (s4) − 0.0329 − 0.0070 − 0.0979 − 0.0505 − 0.0998*
Polarization (s5) 0.0981 0.0715 − 0.0178 0.0248 0.0892
Connectivity (s6) − 0.1168* − 0.1046 − 0.0769 − 0.1334* 0.0281
Serviceability (s7) 0.0161 − 0.0236 0.1188 0.0616 − 0.1101
Prior variable: Reliability (t2)
Intellectualization (s1) 0.0001 0.0244 − 0.0242 − 0.0432 − 0.0518
Informationization (s2) − 0.0732 − 0.0199 − 0.0247 − 0.0510 − 0.0347
Stability (s3) 0.0232 − 0.0072 0.0340 0.1000 0.0538
Efficiency (s4) − 0.0902* − 0.1536** − 0.0554 − 0.0086 − 0.0085
Polarization (s5) − 0.0472 0.0880 0.0412 0.0773 − 0.0014
Connectivity (s6) − 0.0116 0.0501 0.1043 0.1511** − 0.0193
Serviceability (s7) 0.0646 0.0312 − 0.1093 − 0.1629 − 0.1521
Prior variable: Personality (t3)
Intellectualization (s1) − 0.0846 − 0.0773 − 0.0445 0.0373 − 0.0053
Informationization (s2) − 0.0494 − 0.0614 − 0.1109 − 0.0911 − 0.0237
Stability (s3) 0.0213 0.0283 0.0589 0.0418 0.0137
Efficiency (s4) − 0.0167 − 0.0273 − 0.1186* − 0.0678 − 0.0572
Polarization (s5) 0.1052 0.0150 − 0.0375 − 0.0135 0.0447
Connectivity (s6) − 0.0613 − 0.0176 − 0.0733 − 0.1354 − 0.0038
Serviceability (s7) 0.0967 − 0.0019 0.0365 − 0.1145 − 0.1379
Prior variable: Safety (t4)
Intellectualization (s1) 0.0535 − 0.0599 − 0.0784 − 0.0897 − 0.0601
Informationization (s2) 0.0561 0.0727 0.0680 0.0954 0.0495
Stability (s3) − 0.0624 − 0.0827 − 0.0930 − 0.1905*** − 0.1411**
Efficiency (s4) − 0.0662 − 0.0567 − 0.0285 0.0556 − 0.0349
Polarization (s5) 0.2161*** 0.1970** 0.1304 − 0.0043 − 0.0895
Connectivity (s6) − 0.0113 0.0470 0.0002 − 0.1176 0.0439
Serviceability (s7) − 0.1512 − 0.0316 0.0089 − 0.0082 − 0.1408
Prior variable: Intelligence (t5)
Intellectualization (s1) − 0.0336 − 0.0227 − 0.0439 − 0.0194 − 0.0395
Informationization(s2) 0.0273 − 0.0053 0.0465 0.0369 0.0154
Stability (s3) − 0.0025 0.0012 − 0.0704 − 0.0231 − 0.0143
Efficiency (s4) − 0.0044 − 0.0213 − 0.0432 − 0.0024 − 0.0101
Polarization (s5) − 0.0782 − 0.1120 − 0.0661 − 0.0349 − 0.0311
Connectivity (s6) − 0.0133 0.0205 0.0399 0.0795 0.0791
Serviceability (s7) − 0.0646 − 0.0571 − 0.1069 − 0.0421 − 0.0207
Prior variable: Diffusion (t6)
Intellectualization (s1) 0.0241 0.0508 − 0.0249 0.0585 0.0205
Informationization (s2) 0.0288 − 0.0032 − 0.0531 − 0.1040 − 0.1011
Stability (s3) 0.1013 0.0422 0.0304 0.0847 0.0676
Efficiency (s4) − 0.0138 − 0.0031 0.0167 − 0.0844 − 0.0492
Polarization (s5) 0.1932** 0.1944* 0.1539 0.1557 0.0634
Connectivity (s6) − 0.0081 − 0.0214 0.0769 0.0234 − 0.0214
Serviceability (s7) − 0.1238 − 0.2609 − 0.1740 − 0.1157 0.0665
Prior variable: Accessible (t7)
Intellectualization (s1) − 0.0452 − 0.0114 − 0.0359 − 0.0510 − 0.0293
Informationization (s2) 0.0204 0.0081 0.0028 0.0151 0.0261
Stability (s3) 0.0042 0.0058 − 0.0150 0.0217 − 0.0040
Efficiency (s4) − 0.0079 − 0.0142 − 0.0047 0.0017 0.0023
Polarization (s5) − 0.0034 0.0254 0.0557 0.0280 0.0058
Connectivity (s6) − 0.0465 − 0.0216 0.0044 − 0.0028 − 0.0186
Serviceability (s7) − 0.0346 − 0.0188 − 0.0167 0.0278 − 0.0338

indirectly affected, rather than directly affected. Meanwhile, services become less reliable because of intelligence and the
The fourth is the influence of information services on technology. emergence of more convenient services for people after about 12 years
Informatization services affect the development of diffusion technolo­ of reliability technology. In other words, intelligent services alone can
gies, but they also affect the decline of reliability technologies. Perhaps suggest that it takes a long period for people to become convenient (Yi
most of all, informatization services have been shown to influence the and Kim, 2019). In addition, with the advent of services with greater
development of tangible technology after a period of around 24 years. It convenience, safety technologies and diffusion technologies will
can be seen as the only service which affects the development of tangible develop rapidly, while personalization technologies will decline over
technology. time. In addition, with the development of reliability technologies, the
Finally, the fifth is the pattern in which changes in popularization technology’s goal is to popularize services 30 years after the gradual
technology and services change as intelligent services begin to emerge. emergence of intelligent services. On the other hand, efficiency services
When intelligent services appear, not only will security technologies and connectivity services increase and then decrease, and continue to
emerge after a long time, but technology to ensure the reliability of show incomplete service changes that do not rise or fall.
intelligence from about six years later, will then continue to evolve over The coevolution pattern derived from the DSM results of this study
a long period of time (approximately 4 lags, around 24 years). has a flow that starts with diffusion as a technological attribute and ends

12
K. Lee et al. Technological Forecasting & Social Change 176 (2022) 121431

Table 6 Table 6 (continued )


Results of SVAR analysis: society to technology. Posterior Lags
Posterior Lags variable 1 2 3 4 5
variable 1 2 3 4 5
Intelligence 0.0124 − 0.0183 0.0897 0.0105 − 0.0791
Prior variable: Intellectualization (s1) (t5)
Tangibility 0.2384 0.3299 0.1145 0.0814 − 0.0411 Diffusion (t6) 0.0108 0.0946 − 0.0871 0.1084 − 0.0369
(t1) Accessible 0.0198 − 0.0175 − 0.1420 − 0.0509 0.0418
Reliability 0.0871 0.4143* 0.4322** 0.3915* 0.4567** (t7)
(t2) Prior variable: Serviceability (s7)
Personality 0.0454 − 0.0785 − 0.2652 − 0.2837 − 0.0880 Tangibility − 0.0294 − 0.0278 − 0.0987 − 0.0529 − 0.0148
(t3) (t1)
Safety (t4) 0.0752 0.2685 0.1705 0.2988* 0.0997 Reliability − 0.0052 − 0.0269 0.0912 0.1953 0.0324
Intelligence 0.0299 0.2346 0.2766 0.2922 0.1927 (t2)
(t5) Personality − 0.1399 − 0.2366** − 0.2497* − 0.1779 − 0.0185
Diffusion (t6) 0.0425 − 0.0336 0.0774 0.0467 0.1663 (t3)
Accessible − 0.0885 − 0.3694 − 0.2296 − 0.1533 − 0.0727 Safety (t4) 0.1648* 0.2649** 0.2052* 0.0843 0.0117
(t7) Intelligence − 0.0110 0.0090 0.0719 0.0351 0.0117
Prior variable: Informatization (s2) (t5)
Tangibility 0.0359 0.1464 0.2796 0.3386* 0.1490 Diffusion (t6) 0.1495* 0.1979* 0.2243** 0.2207** 0.1217
(t1) Accessible 0.0959 0.0877 0.0179 0.0344 − 0.0355
Reliability − 0.2515* − 0.1814 0.0164 0.3204* 0.1853 (t7)
(t2)
Personality − 0.1193 − 0.1130 − 0.1239 − 0.2099 − 0.2080
(t3) with information on the side of social attributes. On the other hand, if it
Safety (t4) 0.0733 − 0.0206 0.1239 − 0.1357 − 0.0639 starts with the technological attribute, reliability, it only evolves into
Intelligence 0.0739 − 0.0022 0.2000 0.1945 0.0721
technology rather than evolution into society, and later ends with the
(t5)
Diffusion (t6) 0.2014 0.3157** 0.3700** 0.0864 0.1174 technological attribute, accessible. The result suggests that, in the
Accessible − 0.0388 0.0367 − 0.2199 − 0.2936 − 0.1377 coevolution between technology and society, the accurate execution of
(t7) technology is important. However, the diffusion of technology can affect
Prior variable: Stability (s3) the evolution of society, and the value of technology can be socially
Tangibility − 0.1012 − 0.0008 0.1688 0.2124 0.2266
(t1)
created in the future. According to Rotmans et al. (2001), technological
Reliability − 0.0738 − 0.2359 − 0.1522 − 0.3315* 0.0536 and social transformation develops through a predevelopment phase, a
(t2) take-off phase, a breakthrough phase, and a stabilization phase. Among
Personality 0.1401 0.0598 − 0.0486 − 0.1023 − 0.0678 these stages, in the breakthrough stage, visible structural changes occur
(t3)
in existing systems based on socio-cultural, economic, ecological, and
Safety (t4) 0.1501 0.2189 0.1699 0.2341 0.1970
Intelligence − 0.0853 − 0.0983 − 0.0116 − 0.0661 0.0996 institutional interactions and feedback mechanisms. In addition, in the
(t5) breakthrough stage, collective learning spreads and internalization
Diffusion (t6) − 0.0404 − 0.0412 − 0.1981 − 0.0837 − 0.1552 processes appear. In this study, it was found that this research can
Accessible 0.1421 0.0621 0.0095 0.0473 − 0.0955 support the existing coevolution research in that social change occurs in
(t7)
Prior variable: Efficiency (s4)
the stage of diffusion. From another perspective, several studies have
Tangibility − 0.0156 0.0428 0.0670 0.3044 0.1631 emphasized the supply side of innovation (Geels, 2002; Geels 2004).
(t1) While technology is an essential factor for the operation of social func­
Reliability − 0.0671 0.0542 − 0.0687 0.0242 − 0.1181 tions such as transportation and communication in modern society,
(t2)
various resources of demand-side other than technology are required to
Personality − 0.0571 0.1366 0.0661 − 0.0991 0.1566
(t3) smoothly perform sub-functions such as production, distribution, and
Safety (t4) 0.1556 − 0.0217 − 0.0270 0.0138 0.0884 utilization of technology. This means that not only technology devel­
Intelligence − 0.0446 − 0.0134 0.0150 0.0147 − 0.1321 opment but also diffusion and safety for consumers should be considered
(t5) for coevolution between technology and society. It was confirmed that
Diffusion (t6) 0.1023 0.0673 0.0839 − 0.2090 0.0557
Accessible 0.0626 0.0005 − 0.0475 − 0.0139 0.0664
this point of view of previous literature is similar to the results of this
(t7) study.
Prior variable: Polarization (s5) Additionally, from a theoretical point of view, the results of this
Tangibility 0.0860 0.0300 0.0822 0.0430 − 0.0295 study hold many differences from those presented in previous studies.
(t1)
First, in the convergence research of technology and humanities, the
Reliability 0.1760 0.1577 0.1214 0.0873 0.1086
(t2) method of solving social issues through technology was the most
Personality − 0.0936 − 0.1558 − 0.0974 − 0.1002 − 0.1839 dominant. For example, the future field of artificial intelligence is spe­
(t3) cifically described as "sharing resource" and "the fusion of virtual space
Safety (t4) 0.0513 0.1532 0.2965 0.1788 0.0470 and actual space" (Lee et al., 2018). However, in this study, rather than
Intelligence 0.0405 0.0978 0.0027 0.1152 0.0810
(t5)
using technology development ideas to solve social issues, the interac­
Diffusion (t6) 0.0570 0.1271 0.0701 0.0216 − 0.1680* tion between technology and society is presented within a macroscopic
Accessible − 0.0630 − 0.0475 0.0297 − 0.0816 − 0.1076 flow. Second, from the point of view of technological and social systems
(t7) research, this study can be seen as similar to previous studies in that it
Prior variable: Connectivity (s6)
also looks at the overall situation from a macroscopic view. However,
Tangibility − 0.0303 0.0504 0.0508 0.0573 0.2504*
(t1) since the papers in the field of technological and social systems include
Reliability 0.1048 0.1003 0.2959 0.0683 − 0.0767 concepts and methods from a wide range of fields, with the market,
(t2) science, industry, technology, policy, and cultural fields as analysis
Personality − 0.0197 − 0.0320 − 0.1064 0.0058 − 0.0582 targets, the concept is more complex than what is present in this study,
(t3)
which only analysed the relationship between technology and society
Safety (t4) 0.0248 0.1242 0.0560 − 0.0343 − 0.0672
(Geels., 2004). Since this study focused solely on the relationship be­
tween technology and society, it was able to better explain the wave

13
K. Lee et al. Technological Forecasting & Social Change 176 (2022) 121431

Fig. 4. Initial model of DSM.

Fig. 5. Result of DSM.

phenomenon derived from the co-evolution process, by expressing the into consideration aspects of coevolution between technology and so­
time delay which has accompanied changes in technology and society ciety. In the case of the existence of disparate elements of technology
over the past 30 years. and society, the SVAR methodology analysed the detailed properties of
technology and society with regards to each other’s long-term evolution.
6. Conclusions In order to derive coevolution aspects between properties, sequential
flows were derived using the DSM methodology. This led to the more
This paper presents data-driven simulation modeling and an analysis nuanced development of specific technologies after specific social phe­
of coevolution between technology and society. This paper proposed a nomena or derived grounds and patterns for predicting social phenom­
process for eliciting opportunities for technological development, taking ena after the development of specific technologies.

14
K. Lee et al. Technological Forecasting & Social Change 176 (2022) 121431

In order to verify the validity of this proposed framework, a case Gebala, D.A., & Eppinger, S.D. (1991). Methods for analyzing design procedures.
Geels, F.W., 2002. Technological transitions as evolutionary reconfiguration processes: a
study was conducted using information drawn from patents and news
multi-level perspective and a case-study. Res. Policy 31 (8–9), 1257–1274.
articles in the AI field. Firstly, existing research and collected news ar­ Geels, F.W., 2004. From sectoral systems of innovation to socio-technical systems:
ticles were used to define the technological and social attributes of both insights about dynamics and change from sociology and institutional theory. Res.
past and current AI technology. Technology, flexibility, utility, person­ Policy 33 (6–7), 897–920.
Geels, F.W., 2005. Processes and patterns in transitions and system innovations: refining
ality, convenience, intelligence, diffusion, and accessibility were derived the co-evolutionary multi-level perspective. Technol. Forecast. Soc. Change 72 (6),
as the main technological attributes, and intellectualization, informati­ 681–696.
zation, stability, efficiency, polarization, connectivity, and serviceability Geels, F.W., Schot, J., 2007. Typology of sociotechnical transition pathways. Res. Policy
36 (3), 399–417.
were defined as the main social attributes. Then, we analysed the rela­ Giones, F., Brem, A., 2017. From toys to tools: the co-evolution of technological and
tionship between the attributes such that the technological attribute entrepreneurial developments in the drone industry. Bus. Horiz. 60 (6), 875–884.
‘tangibility’ held a positive effect on accessibility, the other technological He, S., Zheng, X., Zeng, D., 2016. A model-free scheme for meme ranking in social media.
Decis. Support Syst. 81, 1–11.
attribute. Finally, the DSM was used to derive the coevolutionary Hung, K.P., Chou, C., 2013. The impact of open innovation on firm performance: the
pattern of society and technology. The derived coevolutionary pattern moderating effects of internal R&D and environmental turbulence. Technovation 33
consisted of the following: ‘diffusion → polarization → safety → stability (10–11), 368–380.
Hyeon, C.-.H., 2008. The directions of IT-based convergence policy. Electronics and
→ intellectualization → serviceability → efficiency → connectivity → Telecommunications Trends 23 (2), 1–12.
informatization’ and ‘reliability → tangibility → personality → intelli­ Jeong, Y., Yoon, B., 2015. Development of patent roadmap based on technology roadmap
gence → accessible’. by analyzing patterns of patent development. Technovation 39, 37–52.
Joung, J., Kim, K., 2017. Monitoring emerging technologies for technology planning
In this way, this study contributes towards better methods of
using technical keyword based analysis from patent data. Technol. Forecast. Soc.
deriving the pattern of coevolution between future technology and so­ Change 114, 281–292.
ciety by using text information from technological documents that were Kim, B., Kim, D., Kim, K., Kim, D., 2014. The case study on the niche experimentation in
in the form of patents, and social documents that were represented by offshore wind renewable energy transition. J. Korea Technol. Innov. Soc. 17 (2),
355–379.
news. In addition, there is a contribution point which gave implications Kim, D.-.K., Baek, D.-.H., Jin, H.-.C., 2010. A study on technology characteristics of it
for the future direction of the AI field by means of applications in the AI convergence service and application technology for industrial convergence. JJ. Inf.
field, which has become an issue most recently, and will continue to be Technol. Serv. 9 (2), 1–20.
Kim, J., Han, M., Lee, Y., Park, Y., 2016. Futuristic data-driven scenario building:
important for the far future. incorporating text mining and fuzzy association rule mining into fuzzy cognitive
However, the following limitations may also be pointed out for this map. Expert Syst. Appl. 57, 311–323.
study. Unthinking technologies and social phenomena, such as disrup­ Kim, J., Lee, S., Park, Y., 2013. User-centric service map for identifying new service
opportunities from potential needs: a case of app store applications. Creativity Innov.
tive innovations, are still somewhat difficult to predict, as they have Manag. 22 (3), 241–264.
been used to predict future technological opportunities and social phe­ Kim, S.-.W., Jung, J.-.L., 2008. A system simulation for investigation of IT and society co-
nomena with only limited variables. In addition, due to the character­ evolution dynamics and its policy implications. Korean Syst. Dyn. Rev. 9 (1),
171–197.
istics of DSM in such that it can represent one attribute only once, the Leamer, E.E. (1985). Vector auto regressions for causal inference? paper presented at the
possibility of the attribute that can appear multiple times in one pattern Carnegie-rochester conference series on public policy.
could not be taken into consideration. There is also a limitation in that Leavitt, H.J., 1965. Applied Organizational Change in industry, structural, Technological
and Humanistic Approaches. Handbook of organizations, p. 264.
the interpretation of the resulting patterns can be ambiguous and
Lee, C., Jeon, J., Park, Y., 2011. Monitoring trends of technological changes based on the
quantitative evaluation is therefore impossible. Therefore, further dynamic patent lattice: a modified formal concept analysis approach. Technol.
research is needed to overcome these limitations. Forecast. Soc. Change 78 (4), 690–702.
Lee, C., Kwon, O., Kim, M., Kwon, D., 2018a. Early identification of emerging
technologies: a machine learning approach using multiple patent indicators.
CRediT authorship contribution statement Technol. Forecast. Soc. Change 127, 291–303.
Lee, C., Song, B., Park, Y., 2015. An instrument for scenario-based technology
Keeeun Lee: Conceptualization, Methodology, Data curation, Soft­ roadmapping: how to assess the impacts of future changes on organisational plans.
Technol. Forecast. Soc. Change 90, 285–301.
ware, Formal analysis, Writing – original draft. Sunhye Kim: Method­ Lee, J., Jung, B., Kim, B., 2013. Incident response competence by the security types of
ology, Visualization, Validation, Writing – review & editing. Byungun firms: socio-technical system perspective. J. Inf. Technol. Serv. 12 (1), 289–308.
Yoon: Methodology, Visualization, Investigation, Supervision, Valida­ Lee, L., 2018. An Analysis of Research Trend in the Field of Remanufacturing Using
Network Analysis and Text Mining. Soongsil university, Korea. Unpublished master’s
tion, Writing – review & editing. thesis.
Lee, M., Yun, J.J., Pyka, A., Won, D., Kodama, F., Schiuma, G., Zhao, X., 2018b. How to
Acknowledgments respond to the fourth industrial revolution, or the second information technology
revolution? Dynamic new combinations between technology, market, and society
through open innovation. J. Open Innov.: Technol. Mark. Complex. 4 (3), 21.
This work was supported by the Ministry of Education of the Re­ Manning, C.D., Surdeanu, M., Bauer, J., Finkel, J.R., Bethard, S., & McClosky, D. (2014).
public of Korea and the National Research Foundation of Korea (NRF- The Stanford CoreNLP natural language processing toolkit. Paper Presented At the
Proceedings of 52nd Annual Meeting of the Association For Computational
2020S1A5A2A01040428). linguistics: System Demonstrations.
Markard, J., Raven, R., Truffer, B., 2012. Sustainability transitions: an emerging field of
Reference research and its prospects. Res. Policy 41 (6), 955–967.
Moore, J., 1996. The Death of Competition: Leadership and Strategy in the Age of
Business Ecosystems. HarperCollins, New York.
Bostrom, R.P., Heinen, J.S., 1977. MIS problems and failures: a socio-technical
Murray, F., 2002. Innovation as co-evolution of scientific and technological networks:
perspective. Part I: the causes. MIS Q. 17–32.
exploring tissue engineering. Res. Policy 31 (8–9), 1389–1403.
Browning, T.R., 2001. Applying the design structure matrix to system decomposition and
Oh, H.-.S., Choi, Y.-.J., Shin, W.-.H., Jeong, Y.-.J., Myaeng, S.-.H., 2009. Trend properties
integration problems: a review and new directions. IEEE Trans. Eng. Manage. 48 (3),
and a ranking method for automatic trend analysis. J. KIISE: Softw. Appl. 36 (3),
292–306.
236–243.
Chen, S.-.H., Huang, M.-.H., Chen, D.-.Z., 2012. Identifying and visualizing technology
Papachristos, G., 2011. A system dynamics model of socio-technical regime transitions.
evolution: a case study of smart grid technology. Technol. Forecast. Soc. Change 79
Environ. Innov. Soc. Transit. 1 (2), 202–233.
(6), 1099–1110.
Park, H. S., 1989. The Impact of Information Technology on the Information Society.
Cheng, C., Huizingh, K.R.E., 2010. Open innovation to increase innovation performance:
Special lecture of The Institute of Electronics and Information Engineers, 20-39.
evidence from a large survey. Proceedings of the XXI ISPIM International
Park, G. S., 1995. A Study on the Influence of Science and Technology on Human and
Conference. Bilbao, Spain. June 6–9.
Society. Science & Technology Policy, 5(2), 64-75.
Cooley, T.F., LeRoy, S.F., 1985. A theoretical macro econometrics: a critique. J. Monet.
Park, I., Yoon, B., 2018. Identifying promising research frontiers of pattern recognition
Econ. 16 (3), 283–308.
through bibliometric analysis. Sustainability 10 (11), 4055.
Dalpiaz, F., Giorgini, P., Mylopoulos, J., 2013. Adaptive socio-technical systems: a
Park, I., Yoon, B., Kim, S., & Seol, H. (2019). Technological opportunities discovery for
requirements-based approach. Requirements Eng. 18 (1), 1–24.
safety through topic modeling and opinion mining in the fourth industrial
Frey, C. B., Osborne, M. A., 2017. The future of employment: How susceptible are jobs to
revolution: the case of artificial intelligence. IEEE Trans. Eng. Manage.
computerisation? Technological Forecasting and Social Change, 114, 254-280.

15
K. Lee et al. Technological Forecasting & Social Change 176 (2022) 121431

Park, S., 2011. Co-evolution of policy and emerging technology: hydrogen energy 2030. One Hundred Year Study on Artificial Intelligence: Report of the 2015-2016
policies in Iceland, the United Kingdom, and Korea. Korean J. Policy Stud. 26. Study Panel. Stanford, CA.
Park, Y., Lee, S., 2011. How to design and utilize online customer center to support new Thorleuchter, D., Van den Poel, D., 2015. Idea mining for web-based weak signal
product concept generation. Expert Syst. Appl. 38 (8), 10638–10647. detection. Futures 66, 25–34.
Rotmans, J., Kemp, R., Van Asselt, M., 2001. More evolution than revolution: transition Trist, E., 1981. The evolution of socio-technical systems. Occas. Pap. 2 (1981), 1981.
management in public policy. Foresight-The Journal Of Future Studies, Strategic Turing, I.B.A., 1950. Computing machinery and intelligence-AM Turing. Mind 59 (236),
Thinking And Policy 3 (1), 15–31. 433.
Rzevski, G., 2019. Coevolution of technology, business and society. Manag. Appl. Yi, H., Kim, P., 2019. The effect of consumer’s technology acceptance and resistance on
Complex Sys. 59. intention to use of artificial intelligence (AI). Korean Manag. Rev. 48 (5),
Schatzmann, J., Schäfer, R., Eichelbaum, F., 2013. Foresight 2.0-definition, overview & 1195–1219.
evaluation. Eur. J. Futures Res. 1 (1), 15. Yoon, J., 2012. Detecting weak signals for long-term business opportunities using text
Schwab, K. (2017). The fourth industrial revolution: currency. mining of Web news. Expert Syst. Appl. 39 (16), 12543–12550.
Shigeto, S., Yamagata, Y., Horio, M., 2014. Socio-technological co-evolution approach: Yoon, B., Jeong, Y., Lee, K., Lee, S., 2020. A systematic approach to prioritizing R&D
an endeavor by the JST-RISTEX environment-energy R&D program. Energy Procedia projects based on customer-perceived value using opinion mining. Technovation 98,
61, 1438–1441. 102164.
Shin, J., Park, Y., 2005. Analysis on the dynamic relationship among product attributes: Zeng, Y., Dong, P., Shi, Y., Wang, L., Li, Y., 2020. Analyzing the co-evolution of green
VAR model approach. J. High Technol. Manag. Res. 16 (2), 225–239. technology diffusion and consumers’ pro-environmental attitudes: an agent-based
Sims, C.A., 1980. Macroeconomics and reality. Econom.: J. Econom Soc. 1–48. model. J. Clean. Prod. 256, 120384.
Son, H., Kwon, Y., Park, S.C., Lee, S., 2018. Using a design structure matrix to support
technology roadmapping for product–service systems. Technol. Anal. Strateg.
Keeeun Lee is a researcher in Department of Industrial & Systems Engineering of the
Manag. 30 (3), 337–350.
Dongguk University. Her theme of study has involved SMEs, quality management, open
Song, B., Kang, D., Yoon, B., & Park, Y. (2010). Development of two-layered service
innovation technology forecasting, technology intelligence, data mining and patent
evolution map: structure and development process. Paper Presented At the 2010
analysis.
IEEE International Conference On Industrial Engineering and Engineering
Management.
Song, B., Lee, C., Yoon, B., Park, Y., 2016. Diagnosing service quality using customer Sunhye Kim is currently a master’s degree student with the Department of Industrial &
reviews: an index approach based on sentiment and gap analyses. Serv. Bus. 10 (4), Systems Engineering, Dongguk University, Seoul Rep. of Korea. Her research interests
775–798. include patent analysis, text mining, technology intelligence, and Natural Language
Song, W., 2012. Social and technological system and policy significance. Issues & Policy processing.
(60), 1–13. Seoul: Science and technology policy institute.
Steward, D.V., 1981. The design structure system: a method for managing the design of
Byungun Yoon is a professor with the Department of Industrial & Systems Engineering,
complex systems. IEEE Trans. Eng. Manage. (3), 71–74.
Dongguk University, Seoul Rep. of Korea. His-research interests include patent analysis,
Stone, P., Brooks, R., Brynjolfsson, E., Calo, R., Etzioni, O., Hager, G., Hirschberg, J.,
new technology development methodology, technology intelligence and visualization
Kalyanakrishnan, S., Kamar, E., Kraus, S., Leyton-Brown, K., Parkes, D., Press, W.,
algorithms.
Saxenian, A., Shah, J., Tambe, M., Teller, A., 2016. Artificial intelligence and life in

16

You might also like