Professional Documents
Culture Documents
Vol. 17 (2018)
°c The Author(s)
DOI: 10.1142/S0219622018500499
Knowledge society blockchain is one of the most powerful and recent tools to make the internet
environment safer and reliable. Manufacturing has traditionally been dominated by standard
designs that are mass-produced, due to the fact, that custom production causes additional costs
that make it less a®ordable than mass production. This paper proposes to develop a designer
expert system for IoT installation layout designs, using blockchain distributed system based on
a machine learning, with users entering data to the expert system by a smart bot software. This
expert system will work using extreme learning machine as inference engine; therefore, this is
a shell to develop any expert system with fast learning. The whole system is represented by a
smart contract with a value linked to the value of the expert system, the more this expert system
be quoted on the web, the more the shares of the smart contract will cost.
Keywords: Block-chain; initial coin o®er; expert system; blockchain maker expert system;
extreme learning machine; internet of things.
1. Introduction
Blockchain is drastically changing the way technology and knowledge are produced
in the knowledge society1,2 having an impact of transcendental magnitude, not only
in this ¯eld, but also in the restructuring of the way in which society and the economy
are organized to produce goods and services. So much has been said of the new
industrial revolution from di®erent points of view, in the ¯elds related to the
Information Technologies, (IT): Internet of things,3 the semantic web4 and in the
This is an Open Access article published by World Scienti¯c Publishing Company. It is distributed under
the terms of the Creative Commons Attribution 4.0 (CC-BY) License. Further distribution of this work is
permitted, provided the original work is properly cited.
1
2 ~ et al.
R. Carreno
¯eld of the economy: the way to trade currencies and in the ¯eld of society:
the impact that in short-term is expected to modify the way products and services
are made. In such scenario a new industrial revolution is about to be experienced, this
goes aligned with arti¯cial intelligence and ES (expert systems), thus, modifying the
way of production, moving from mass production to customized production, towards
an unprecedented impact by gradually modifying the way of grouping in society for
this purpose, which will derivate in a de-centralized economy.
by 187.216.238.5 on 09/24/18. Re-use and distribution is strictly not permitted, except for Open Access articles.
customized design for him or the service of the maker, to be selected from the list of
the suggested makers provided by the system. All of the above, content in a whole
system working on blockchain technology basis.
The expert system has as inference module; an ELM extreme learning machine,
which it is an innovative learning method for neural networks.25–31 The essence of the
ELM is that the hidden layer need not be tuned, if hidden layer mapping satis¯es
universal approximation conditions (neural network generalization theory), the
by 187.216.238.5 on 09/24/18. Re-use and distribution is strictly not permitted, except for Open Access articles.
hidden node parameters are not only independent of the training data but also of
each other, although hidden nodes are important and critical. Since hidden layer
needs not be tuned, the learning is faster, thousands of times faster in some cases.
Training for ELM has three steps after given a training set: Assign randomly hidden
Int. J. Info. Tech. Dec. Mak. Downloaded from www.worldscientific.com
node parameters, calculate the hidden layer output matrix, and calculate the output
weights.31,33–35,38
The main contribution of this paper is to provide a no currently existing expert
system to design any drawing and even better, based on a de-centralized system as the
blockchain. This type of expert system can be useful in any transformation industry
process and the blockchain will support the safety and integrity of the data against any
human mistakes or hacker human attack. The transforming industry, make assem-
blies, installations, services, etc. based on drawings, since these drawings have all key
information (know-how, bill of material, materials, assembly indications, technological
indications, etc.) to do the job, therefore the creation of a designer expert system can
boost the industry to an industrial revolution. In this case the expert system is focused
to do sensors installation drawings for smart houses, an example drawing is provided at
Fig. 9 in Sec. 4.
The customer, needing the artifact (a house, a car, a chair, etc.) is the one to input
the characteristic of the required artifact using a bot online, this bot is a smart chat
online asking the customer the speci¯cations required which are the expert system
inputs for the design. The maker is the manufacturer represented by the people with
the expertise to make the artifact.
The expert system works using machine learning software such that the more
inputs from no limit quantity of customers the more the expert system will learn, and it
will have more expertise designing. Thus, a big range of existing choices will be
available; in consequence, this will cut down the price of the ¯nal product by avoiding
the high design cost. The BMES recommends the manufacturer based on an evaluation
list according to a history assessment.
The smart contract in this case is an initial coin o®er (ICO) and can be managed
by blockchain technology in order to make it work similarly to a cryptocurrency, in
other words, the smart contract can have an electronic value online with its price
correlated to the value of the share. This share can be bought by anybody online in
the same way a bitcoin can. Its value will depend not only on its demand but also on
the pro¯tability of the BMES, according to the usage from the customer, who pays a
fee for the service.
4 ~ et al.
R. Carreno
the founder.
(2) An ICO unique initial balance of a thousand cryptocurrency units is assigned to
the whole pioneers group.
(3) ICO is bought/sold by investors.
Int. J. Info. Tech. Dec. Mak. Downloaded from www.worldscientific.com
Since ICO is the crypto-currency to pay the service of the ESS, the process for
ICO transactions is represented in Fig. 1, just to show the °ow of the pay transactions.
The ICO price varies depending on two conditions: the o®er/demand of the ESS
and the o®er/demand of the ICO sale/buy.
A Merkle tree is necessary to simplify the key chain length, such that the hash
Int. J. Info. Tech. Dec. Mak. Downloaded from www.worldscientific.com
transactions can be feasible in a future big data scenario.32 The payment veri¯cation
can be done just checking the block header of the longest block of the Merkle root
proof of work.
The essence of smart contracts is to regulate the transactions of the ICO based on
blockchain technology. Smart contracts can be coded using the Ethereum smart
contract platform at www.ethereum.org with the software solidity and java, in order
to be posted in internet for buy/sell purposes. Once the ICO is released, anybody can
trade ICO and use the ESS.
As shown in Fig. 2, not only a blockchain for ICO is required but also a blockchain
for the expert system data base and a blockchain for the service pay o® the expert
systems. In this way, three blockchain in total are required for a full expert system
working on a machine learning environment, therefore this may help in the birth of a
new industrial revolution.
Stability study
The blockchain maker expert system can be represented as:
X_ t ¼ AX t þ Bw t ;
ð1Þ
Y t ¼ CX t þ Dw t ;
Int. J. Info. Tech. Dec. Mak. Downloaded from www.worldscientific.com
where the matrix order corresponds to the di®erential equation order with:
½nn ½n1 ½nn ½n1
A 2 R ½0;1½ ; B 2 R ½0;1½ ; C 2 R ½1;1½ ; D 2 R ½0;1 ; w t 2 N f w t ; 2w t < 1g:
The control law system V t (liapunov variable) with respect to (1) has the form:
V t ¼ H þ ðE t GY t Þ: ð2Þ
With H þ the pseudo-inverse matrix of H (a matrix to ¯nd that ¯t the Lyapunov
law), the innovation process E t considered in (2), and G, as a unknown matrix
having the form G.
Let the system (2) accomplish with (3), for M to have an attracting region with
the transposed of the derivative, accomplishing with
_ t < 0;
Mt M
T
ð3Þ
where the trajectory region with respect to the gain matrix M t , is described in (4).
M_ t ¼ F t : ð4Þ
With F t , a continuous function bounded by intervals
: with uniform measure t,
^
accomplishing with the innovation process (5), with Y t a state observer of Y_ .
:
Y^ t ¼ Y_ t þ F t : ð5Þ
Considering (2) we have (6).
F t ¼ GY t þ HV t Y^t : ð6Þ
:
The di®erence between Y^ t and F t in (5) is described in (7).
ðFðtÞ Y^ ðtÞÞ ¼ GY t HV t : ð7Þ
The system innovation process E t according to (5), it is described in (8).
:
E t ¼ Y^ t F t : ð8Þ
Finally, (2) in (5) and V t , is developed in (9).
V t ¼ H þ ðE t GY t Þ: ð9Þ
An IoT Expert System Shell in Block-Chain Technology 7
3. Expert System
The expert system is based on the semantic web, internet of things, smart o±ces
(domotic), blockchain, and arti¯cial intelligence concepts, since it is an expert system
to draw sensors installation layouts for smart o±ces with data protected by block-
chain technology, using a neural network and accessed remotely by a user with a
terminal and a server in an internet network. The components of expert systems are
by 187.216.238.5 on 09/24/18. Re-use and distribution is strictly not permitted, except for Open Access articles.
basically four, and for this study they are represented as follows:
(1) Knowledgebase: which is represented, in this case, by the database to be
created with the data customer input, this is the essence of the ontology of the
semantic server. Two types of data will form the database: Objects and attribute:
Objects are the concepts, for example, in the case of building a house, an object can
Int. J. Info. Tech. Dec. Mak. Downloaded from www.worldscientific.com
be the style of the house, such as modern house or rustic house; attributes can be
features of objects such as big windows and aluminum doors for a modern house, and
wood doors with wood walls for a rustic house. Any way the customers will de¯ne the
objects and attributes. Objects can be the attributes of another object, and attri-
butes can be shared to build another object. Every single object with their attributes
will form a single and unique combination called label, forming in this way, a network
of objects and attributes linked, and every label has a key or non-repeatable code
that makes the label a single combination of one object with its attributes. A pon-
deration or weighting will be given to each label by a neural network of the expert
system, in order to ¯nd out the desired output (design) of the customer. Two kinds of
labels are available: input labels and output labels in the creation of a design, but an
output label can be used as an input label for a future design, such that the expert
system can give the known ponderation of the input labels if a customer already gave
that input or it gives a random ponderation to that input label if that label is not
known by the expert system. Based on the explanation above, in the knowledge
database all labels are indistinct, neither input or output labels, they will turn into
input or output label when the process of doing the design starts.
An extreme learning machine neural network,31,33–35,38 with a sigmoid function does
the design process. The combination of the selected inputs of the process of the design is
chosen by the customer using a bot,36 or a new input can be created by the customer using
this bot, such that the knowledge base will be created by the customer, therefore, the ESS
starts working and grows by the inputs of the customer. The software developers only code
the brain of ESS, smart contract and their relationship based on a blockchain technology.
Figure 3 is representing the BMES as the output of a neural network process, with
the following symbols: I as the Input, w the weights, the b the weights of the umbral,
n1 the neurons of hidden layer, n2 the neurons of the output layer, a1 the output of
the hidden layer and a2 the output of the neural network which it is the output of the
BMES, since the expert system is a designer of speci¯cations drawings, the BMES
output it is this speci¯cation drawing.
In this study an ELM neural network is proposed to perform the inference
machine task, the concept of label is proposed here to name the inputs, the label will
8 ~ et al.
R. Carreno
by 187.216.238.5 on 09/24/18. Re-use and distribution is strictly not permitted, except for Open Access articles.
the external data base, of a new input created by the ESS, this label is necessary to
re-locate this input in its new location (the internal data base) with an identi¯er in
order to belong to the metadata (internal database) in case this be used for a new
design process in a new ESS request, please note that the output is also a label since
this will be identi¯ed with a hash structure as well, please see Fig. 3 as reference.
(2) Inference engine; as mentioned above in general terms, makes the design with
the ESS using extreme learning machine neuronal networks as the inference engine.
It is not the scope of this project to build a cloud based on blockchain technology,
however, there are already some cloud services based on blockchain technology,
however, they are not used in this research because they are not needed due to the
fact this blockchain system can support the big data since an innovative metadata
software is developed to convert image ¯les into text data ¯les, making in this way
that block system does not need to support big amount of data.
(3) User interface is the bot which is called zig, a smart chat online used by the
customer. Zig is an automate smart chat online.
(4) Metadata: All assembly artifacts to be designed are expected to be load in the
block chain not as images but as metadata, an application to convert the metadata
into image is built.
A decentralized information °ow is a milestone on this study, please note in Fig. 4,
that not only a decentralized architecture is used for ICO but also for the expert
system facilities.
A °ow chart decision tree is provided in Fig. 5 in order to show the infor-
mation logarithms sequence works in a general basis. Please note that the ELM,
must converge therefore a stability study was performed using Lyapunov
concepts.37
In Fig. 6, the structure of the ESS is represented. The user/client requests a
service from the expert system via a chat-bot. In Fig. 6 we can see how the three
modules work and interact: The human machine interface (HMI) module, the META
DATA module and the inference engine module (ELM neural network). The HMI
module is a smart chat bot asking for the information from the user, then the
inference engine module uses an ELM (extreme learning machine) neural network to
process the information to do the design and the output, W1, W2, W3, . . ., etc. are
the weights of this neural network.
10 ~ et al.
R. Carreno
by 187.216.238.5 on 09/24/18. Re-use and distribution is strictly not permitted, except for Open Access articles.
Int. J. Info. Tech. Dec. Mak. Downloaded from www.worldscientific.com
%Hidden weights
for k=1:L
for j=1:N
W(k,j)=rand;
end
end
y(1:2)=0;x(1:2)=0;
%% T R A I N I N G %%%%%%
for i=1:500
for j=1:N % 1:80 nodes
I(j,i)=0; %%%% input to each hidden node
for k=1:L %1:10 inputs
I(j,i)=I(j,i)+W(k,j)*II(k,i);
end
O(j,i)=(exp(I(j,i))-exp(-I(j,i)))/(exp(I(j,i))+exp(-I(j,i)));
end
Y(i)=y(i); %%%% target output (teacher)
end
An IoT Expert System Shell in Block-Chain Technology 11
for i=1:500
for j=1:N %%%%hidden layer
I(j,i)=0;
for k=1:L
I(j,i)=I(j,i)+W(k,j)*II(k,i);
by 187.216.238.5 on 09/24/18. Re-use and distribution is strictly not permitted, except for Open Access articles.
end
O(j,i)=(exp(I(j,i))-exp(-I(j,i)))/(exp(I(j,i))+exp(-I(j,i)));
end
Int. J. Info. Tech. Dec. Mak. Downloaded from www.worldscientific.com
For N arbitrary distinct samples ðXi; tiÞ, where Xi ¼ ½Xi1; Xi2; . . . ; XinT 2 Rn
~ hidden nodes and activation
and ti¼½ti1;ti2;...;tim 2 Rm , standard SLFN38 with N
function gðxÞ are mathematically modeled as
~ ~
X
N X
N
igiðXjÞ ¼ igðWiXj þ biÞ ¼ Oj; j ¼ 1; . . . ; N ; ð10Þ
i¼1 i¼1
where wi ¼ ½wi1; wi2; . . . ; winT is the weight vector connecting the ith hidden node
and the input nodes, i ¼ ½i1; i2; . . . ; imT is the weight vector connecting the ith
hidden node and output nodes, and bi is the threshold of the ith hidden node. Wi and
xj denotes the inner product of wi and xj.
4. Results
A MATLAB simulation was performed on general algorithm to ensure convergence
on a Lyapunov stability study, please see Fig. 7 as reference.
In order to complete a full stability study, a MATLAB simulation was provided
for main parameters involved in the ESS system evolution, please see Fig. 8 as
reference.
As shown in Fig. 9, the ESS designs are delivered by the Bot designer software
after entering the coordinate to the Bot by the user.
The bot can design a layout for o±ce areas, in Fig. 9 we can see an example with a
lobby area (the area with a receptionist), the cubicles area (the area with small desks
for employees), and the meeting room (the area for meetings), all of them for the ¯rst
°oor in this case, additionally a drawing will be done for every °oor. Please note that
12 ~ et al.
R. Carreno
by 187.216.238.5 on 09/24/18. Re-use and distribution is strictly not permitted, except for Open Access articles.
Int. J. Info. Tech. Dec. Mak. Downloaded from www.worldscientific.com
Fig. 7. ICO price will converge in a point since a volume supply is limited to 10 million.
the vortices of the construction are C1 to C8, which they are two dimension coor-
dinates (X,Y) given by the user of the bot, and 3 dimension coordinates (X,Y,Z) are
S1 to S4 which represent the location of the sensors (temperature, lighting, presence,
camera, etc.) for this smart o±ce. The actual readings of the sensors are sent to
internet and the sensors can be controlled by a remote device, creating in this way an
internet of things (IoT) layout with a domotic o±ce, however, the IoT system is not
the core of this paper but the bot designer is the core.
Figure 10 shows the hashes of the expert systems code. The blockchain is built
concatenating the previous hash with the new hash, in this case the previous and the
new hashes generated with the code of the expert system.
Considering the safety of the data crucial issue to preserve the integrity of this
system since, the code of the expert system bot is protected with hashes; this means
that this code is converted to hashes as shown in Fig. 10.
In Fig. 11 the hashes of the ICO transactions are shown.
An IoT Expert System Shell in Block-Chain Technology 13
by 187.216.238.5 on 09/24/18. Re-use and distribution is strictly not permitted, except for Open Access articles.
Int. J. Info. Tech. Dec. Mak. Downloaded from www.worldscientific.com
Transactions ledger data is the most crucial data to be protected, that is why
these data is converted to hashes to avoid a hack attack. These transactions meet the
blockchain technology; this means that any new block hash if formed with the
information of the previous block and the transaction details information.
5. Conclusion
This research makes the fusion of two cut edge remarkable technologies: the block-
chain and the arti¯cial intelligence with ES, it makes an alliance of both technologies
in order to burst the new industrial revolution into the mass customized production,
creating a new smart contract (ICO) linked to an ESS to design artifacts. An ELM
neural network was proven with stability and good performance results, please see
Fig. 7 as reference, this provides stability to the whole system BMES. The ICO is not
feasible to have a de°ation since it has a limit volume of 40 million units. The
convergence of the ICO is assured, please see Fig. 7, where it is the proof that the
system is steady. A convergence is seen both with the training set and with the
testing set. This paper de¯nitely is looking for the scalability of the IoT usage with a
potential impact in related top applications, just like smart cities,42 medical appli-
cations,43 ¯nancial markets44,45 and a knowledge society46 as well, based on the
expert systems tools and blockchain technology.47,48
6. Discussion
A drawing designer expert system is the core on this paper which can be a designer
for any kind of drawings and schematics, the speci¯c focus for this paper is for
domotic sensors installation drawings in smart building o±ces, but not less inter-
esting than the combination between some innovative concepts: either in the neural
networks topic (The extreme learning machine) and data reliability process (the
blockchain), the purpose to combine these innovative concepts is to be at the cut
edge of the state of the art. However sometimes the innovation cannot be a synonym
An IoT Expert System Shell in Block-Chain Technology 15
(1) In the paper named as \Expert System for Ice Hockey Game Prediction: Data
Mining with Human Judgment"39 an expert system to predict National Hockey
League (NHL) game outcome is described. This expert system employs support
vector machines but not neural networks as this paper does, and the focus is for
prediction, it is not a drawing designer.
(2) In the paper named as \an alternative approach to ¯rms evaluation: expert
systems and fuzzy logic", a di®erent method of ¯rm valuation based on fuzzy
logic and expert systems related to discounted cash °ow techniques is proposed,
but nothing to do with a drawing designer.40
(3) In the paper: expert system approach to assessments of bleeding predispositions
in tonsillectomy/adenoidectomy patients, the purpose is to assess a predisposi-
tion to bleeding in a patient undergoing a tonsillectomy and/or adenoidectomy
using correlations but nothing to do with a drawing designer.41
References
1. A. Dattakumar and R. S. Sharma, Smart cities and knowledge societies: Correlation,
causation or distinct?2016 IEEE International Conference on Management of Innovation
and Technology (ICMIT), 2016, pp. 193–197.
2. P. Weingart, The moment of truth for science: The consequences of the `knowledge
society' for society and science, Beyond the Knowledge Trap, Singapore, 2011, pp. 155–164.
16 ~ et al.
R. Carreno
3. F. K. Flores, Internet of things: Managing wireless sensor network with rest API for smart
homes, Theory and Practice of Computation, Manila (2016), pp. 132–142.
4. M. Kopel, Libraries in the semantic web era, Qualitative and Quantitative Methods in
Libraries, (2010) 115–122.
5. S. Haber and W. S. Stornetta, How to time-stamp a digital document, Journal of
Cryptology 3(2) (1991) 99–111.
6. D. Bayer, S. Haber and W. S. Stornetta, Improving the e±ciency and reliability of digital
time-stamping, In Sequences II: Methods in Communication, Security and Computer
by 187.216.238.5 on 09/24/18. Re-use and distribution is strictly not permitted, except for Open Access articles.
24. R. Carreño et al., A nonlinear acoustic transducer model with variable structure control,
Fractals 25(2) (2017).
25. W. Yu et al., Modeling of gasoline blending via discrete time neural networks, Neural
Networks Procedure 2004 (2004) 6.
26. W. Yu et al., Neural networks for the optimization of crude oil blending, International
Journal of Neural Systems 15(5) (2005) 377–389.
27. W. Yu and X. Li, Automated nonlinear system modeling with multiple fuzzy neural
networks and kernel smoothing, International Journal of Neural Systems 20(5) (2010)
by 187.216.238.5 on 09/24/18. Re-use and distribution is strictly not permitted, except for Open Access articles.
429–435.
28. S. Lu et al., Nonlinear autoregressive and nonlinear autoregressive moving average model
parameter estimation by minimizing hypersurface distance, IEEE Transactions on Signal
Processing 51(12) (2003) 3020–3026.
29. L. Sheng and J. Hwan, A new algorithm for linear and nonlinear ARMA model parameter
Int. J. Info. Tech. Dec. Mak. Downloaded from www.worldscientific.com
44. J. L. Zhao, Overview of business innovation and research opportunities in blockchain and
introduction to the special issue, Financial Innovation 3(9) (2017).
45. Y. Huan et al., Nonlinear manifold learning for early warnings in ¯nancial markets,
European Journal of Operational Research 258 (2017) 692–702.
46. Tobon et al., Sociedad del Conocimiento: Estudio documental desde una perspectiva
humanista y compleja, Paradigma 36(2) (2015) 7–36.
47. Y. Huan and G. Kou, A kernel entropy manifold learning approach for ¯nancial data
analysis, Decision Support Systems 64 (2014) 31–42.
by 187.216.238.5 on 09/24/18. Re-use and distribution is strictly not permitted, except for Open Access articles.
48. J. J. Xu, Are blockchain immune to all malicious attacks?Financial Innovation 2(25)
(2016).
Int. J. Info. Tech. Dec. Mak. Downloaded from www.worldscientific.com