Professional Documents
Culture Documents
Textbook Applications of Computing and Communication Technologies First International Conference Icacct 2018 Delhi India March 9 2018 Revised Selected Papers Ganesh Chandra Deka Ebook All Chapter PDF
Textbook Applications of Computing and Communication Technologies First International Conference Icacct 2018 Delhi India March 9 2018 Revised Selected Papers Ganesh Chandra Deka Ebook All Chapter PDF
https://textbookfull.com/product/information-communication-and-
computing-technology-5th-international-conference-
icicct-2020-new-delhi-india-may-9-2020-revised-selected-papers-
Soft Computing Systems Second International Conference
ICSCS 2018 Kollam India April 19 20 2018 Revised
Selected Papers Ivan Zelinka
https://textbookfull.com/product/soft-computing-systems-second-
international-conference-icscs-2018-kollam-india-
april-19-20-2018-revised-selected-papers-ivan-zelinka/
https://textbookfull.com/product/information-communication-and-
computing-technology-4th-international-conference-
icicct-2019-new-delhi-india-may-11-2019-revised-selected-papers-
abdullah-bin-gani/
Applications of Computing
and Communication
Technologies
First International Conference, ICACCT 2018
Delhi, India, March 9, 2018
Revised Selected Papers
123
Communications
in Computer and Information Science 899
Commenced Publication in 2007
Founding and Former Series Editors:
Phoebe Chen, Alfredo Cuzzocrea, Xiaoyong Du, Orhun Kara, Ting Liu,
Dominik Ślęzak, and Xiaokang Yang
Editorial Board
Simone Diniz Junqueira Barbosa
Pontifical Catholic University of Rio de Janeiro (PUC-Rio),
Rio de Janeiro, Brazil
Joaquim Filipe
Polytechnic Institute of Setúbal, Setúbal, Portugal
Igor Kotenko
St. Petersburg Institute for Informatics and Automation of the Russian
Academy of Sciences, St. Petersburg, Russia
Krishna M. Sivalingam
Indian Institute of Technology Madras, Chennai, India
Takashi Washio
Osaka University, Osaka, Japan
Junsong Yuan
University at Buffalo, The State University of New York, Buffalo, USA
Lizhu Zhou
Tsinghua University, Beijing, China
More information about this series at http://www.springer.com/series/7899
Ganesh Chandra Deka Omprakash Kaiwartya
•
Applications of Computing
and Communication
Technologies
First International Conference, ICACCT 2018
Delhi, India, March 9, 2018
Revised Selected Papers
123
Editors
Ganesh Chandra Deka Pooja Vashisth
Ministry of Skill Development Department of Computer Science,
Delhi Shyama Prasad Mukherji College
India University of Delhi
Delhi
Omprakash Kaiwartya India
School of Science and Technology
Nottingham Trent University Priyanka Rathee
Nottingham Department of Computer Science,
UK North Campus
University of Delhi
Delhi
India
This Springer imprint is published by the registered company Springer Nature Singapore Pte Ltd.
The registered company address is: 152 Beach Road, #21-01/04 Gateway East, Singapore 189721,
Singapore
Preface
People across globe have understood the immense potential of IT with respect to its
contribution in terms of economic growth, efficient governance, and improving quality
of life in general. The main objective of the International Conference on Applications
of Computing and Communication Technologies 2018 (ICACCT-2018) held at
Shyama Prasad Mukherji College, University of Delhi, on March 9, 2018, was a
humble beginning toward the empowerment of our society through technology to
create a better tomorrow.
The University of Delhi is a premier university in India with a venerable legacy and
international acclaim for the highest academic standards, diverse educational programs,
distinguished faculty, illustrious alumni, varied co-curricular activities, and modern
infrastructure. Over the many years of its existence, the university has sustained the
highest global standards and best practices in higher education. Its long-term
commitment to nation building and unflinching adherence to universal human values
are reflected in its motto: “Nishtha Dhriti Satyam” (Dedication,
Steadfastness, and Truth). Established in 1922 as a unitary, teaching, and residential
university by the Act of the then Central Legislative Assembly, a strong commitment to
excellence in teaching, research, and social outreach has made the university a role
model and trend-setter for other universities. The President of India is the Visitor, the
Vice-President is the Chancellor, and the Chief Justice of the Supreme Court of India is
the Pro-Chancellor of the University. Beginning with three colleges and 750 students, it
has grown as one of the largest universities in India with 16 faculties, over 80 academic
departments, an equal number of colleges, and over seven lakh students.
SPM College is a well-known women’s college of the University of Delhi. It was
established in 1969 in the memory of distinguished academician and statesman Shyama
Prasad Mukherji. It has no affiliation with any sect, religion, political group, or
thinking. The college motto “tejasvi naa vadhi mastu” in Sanskrit has a profound
meaning. It is derived from the Taittiriya Upanisad. It means, “Let our efforts at
learning be luminous (Tejasvi) and filled with joy, and endowed with the force of
purpose (Vadhi Mastu).”
This conference was successful in facilitating academics, researchers, and industry
professionals to deliberate upon the latest issues and advancement in ICT and its
applications. In total, 109 papers were submitted in three tracks. After a thorough
review process, 30 papers were selected for oral presentation during the conference.
After the oral presentation of the papers at the conference, the papers were further
refined to enhance their quality.
VI Preface
This conference proceedings will prove beneficial for academics, researchers, and
practitioners as it contains a wealth of valuable information on the recent developments
in ICT.
General Chairs
Jaime Lloret Polytechnic University of Valencia, Spain
Punam Bedi University of Delhi, India
Pooja Vashisth University of Delhi, India
Omprakash Kaiwartya Nottingham Trent University, UK
Priyanka Rathee University of Delhi, India
Advisory Committee
Abdul Hanan Abdullah Universiti Teknologi Malaysia (UTM), Malaysia
Ao Lei University of Surrey, UK
Bingpeng Zhou Hong Kong University of Sciences Technology, SAR
China
Chandra K. Jaggi University of Delhi, India
D. K. Lobiyal Jawaharlal Nehru University, New Delhi, India
Jaime Lloret Polytechnic University of Valencia, Spain
K. K. Bharadwaj Jawaharlal Nehru University, New Delhi, India
Kusum Deep IIT Roorkee, India
Manoranjan Mohanty New York University Abu Dhabi, UAE
M. D. Asri Bin Ngadi Universiti Teknologi Malaysia (UTM), Malaysia
Mukesh Prasad University of Technology, Sydney, Australia
Paresh Virparia Sardar Patel University, Gujarat, India
Prakash C. Jha University of Delhi, India
Rajiv Ratn Shah Singapore Management University, Singapore
Saroj Kaushik Indian Institute of Technology, Delhi, India
Subhash Bhalla University of Aarhus, Denmark
Tiaohao Guo Queen Mary University of London, UK
Vasudha Bhatnagar University of Delhi, India
Vibhakar Mansotra University of Jammu, India
Xiaokang Zhou Shiaga University, Japan
Xu Zhang Xi’an University of Technology, China
Yue Cao Northumbria University, UK
Student Coordinators
Manisha Vashisth
Pooja Garg
Anchal Dua
Deepali Singhal
Vaishali Wahi
Contents
Abstract. A method had been adopted to predict the trending patterns on the
twitter in near-real time environment. These trending patterns help the compa-
nies to know their customers and to predict their brand awareness. The pattern
was recognized by analyzing the tweets fetched in real time environment and by
examining the most popular hashtags on the twitter platform in past few sec-
onds. The work was implemented using a big data technology ‘Spark Stream-
ing’. These hashtag patterns allow the people to follow the discussions on
particular brand, event or any promotion. These hashtags are used by many
companies as a signature tag to gain popularity of their brand on social net-
working platforms.
1 Introduction
maintained. This count denoted the number of occurrences of the respective hashtag.
The pattern so formed is used further as awareness analysis.
Twitter is a social networking platform [4], where users posts, can retweet others’
post and can interact using twitter messages called tweets. Data across twitter appli-
cation is growing exponentially as the number of users and activities over the web are
increasing rapidly. People take over to twitter to present their opinions related to any
issue. Hashtags is usually a label used by many people to emphasize the theme of their
tweet content [5]. It makes it easier for other users to find and follow the content related
to such themes as highlighted by these tags. This work adopted the approach of
analyzing such hashtags to evaluate the theme or trending pattern being followed
currently by the various people on social sites. Different social networking sites uses
these hashtags to evaluate the inclination of various users and comes out with the trends
being followed. Different companies uses this analysis process to identify the per-
spectives of different people about their brand and predicts the reach of their products
to users. This helps the companies to perform the awareness analysis of their respective
products. Exploiting this approach in this work, the pattern of trending was evaluated
[6]. The work was performed in near-real time. Companies follows the same task for
larger intervals of time to come up with large patterns, but such huge computations are
not possible on a single node (standalone).
2 Problem Definition
This work focused on performing pattern recognition of trends being followed on the
twitter. The tweets are important for analysis because data arrive at a high frequency
and algorithms that process them must do so under very strict constraints of storage and
time. All public tweets posted on twitter are freely available through a set of APIs
provided by Twitter. Using streaming APIs, the streaming context object [7] (created
for every task of streaming in Spark) established the connection of the spark master to
the twitter. The fetching of tweets occurred in near-real time since Spark involves
micro- batch processing wherein the fetched data is stored in the form of batches at
regular intervals of time [8]. The pattern was recognized using the hashtags to predict
the trending pattern. Different companies perform the same task of pattern recognition
and trending analysis for large intervals of time. The gist of this task used by many
companies was being showcased in this work.
Today the data is generating in exponentially huge amount across different plat-
forms. Different companies uses such data to analyze the reach of their product to
people and the views about their product. It is impossible to manually process such a
huge amount of data. This is where the need of automatic analyzing process becomes
evident. That is why, the tool ‘Spark’ fits in here correctly. Spark has in-memory
processing, due to which the processing speed of datasets is quite high. It works using
Resilient Distributed Datasets (RDDs), which have partitions to store the data. This
involves parallel processing. Therefore, Spark core API called Spark Streaming was
used to implement the work.
Twitter allows users to post messages, find and follow hashtags and provides its
data in real time for the researchers to perform any kind of analysis on it. The messages
Trending Pattern Analysis of Twitter Using Spark Streaming 5
known as ‘tweets’ are streamed in near-real time in this work, as it included the micro-
batch processing wherein different batches were formed containing tweets, at regular
intervals. Such a huge amount of data can be efficiently used for social network studies
and analysis to gain useful and meaningful results. A user follows a set of people, and
has a set of followers. When that user sends an update out, that update is seen by all of
his followers. User can also retweet other user’s updates. Users find and follow the
hashtags on twitter. Hashtags are the tags, defined using a symbol of hash, which are
used by people to emphasize the theme of their tweet content [2]. These labels are used
by users to easily find the messages with a specific theme or content. Different social
networking platforms uses these hashtags to predict the trending and to identify the
pattern of the popular hashtags. Companies implement this at large scale to predict the
awareness, about their respective products, among people. Since the tweets are fetched
in real time, therefore an ecosystem having high processing speed was required.
3 Proposed System
• The newly created application generates a unique set of Consumer key and Con-
sumer Secret key.
• Along with this, a set of Access Token key and Access Token Secret key was
generated.
These keys are passed as arguments at run-time while executing the application.
Further these APIs are used to set the system properties so that twitter4j library can use
them to generate the OAuth credentials. This library is used by twitter stream to
establish the connection between twitter and the sink and finally create the stream to
fetch the data. To set the system properties, following command was used:
4 Methodology Adopted
Here setAppName() method defines the application name. setMaster() defines the
cluster on which the spark will run. In this work, local mode was used as the cluster and
2 threads were used to implement it.
Spark Streaming uses this object to establish a connection between the source (from
where the live data has to be streamed) and the sink (where the live data will be stored).
Trending Pattern Analysis of Twitter Using Spark Streaming 7
This object acts as the entry point for all streaming functionality. The Streaming
Context object is created after the initialization of Spark Configuration object and the
Spark Context object which are require to start the spark shell. Spark configuration
object also defines the cluster on which the application should be executed. There are
three possible clusters (master node) available in Spark which are ‘Mesos Cluster’,
‘Yarn Cluster’ and the ‘Local Mode’. This local mode is the standalone mode where
the application executes on the single node.
Here Seconds(1) is the micro-batch interval that is, the batch would be formed at
regular intervals of 1 s.
Here ‘stream’ is the DStream formed for streaming the tweets. The ‘filters’ passed
as an argument in createStream() method is the filter if used to fetch the tweets. This
filter is used to target and fetch only those tweets, which have this value of filter being
used, in the tweet text. Therefore, only the tweets containing this word (value of filter)
would be fetched and stored in the DStream created. TwitterUtils is the class used as
the utility file to perform the real-time streaming of tweets.
8 P. Garg et al.
In this process, on every tweet fetched, split() method was executed using the “ ”
delimiter. This divided each tweet text into words. Finally, the words beginning with #
were retained and rest of the words were discarded. Finally, ‘hashTags’ is the DStream
storing only the words starting with #.
Here the count of each unique hashtag was computed that is, how many number of
times a hashtag was used by different users. Finally the hashtags were sorted, using
sortByKey() method, in decreasing order based on the count value. In this command
the ‘Seconds(5)’ defines the window interval. A window was created which considers a
certain number of RDDs at a time. This window was formed at regular intervals of time
that is, at intervals of 5 s. Now the processing of RDDs would be done only at intervals
of 5 s. Therefore, RDDs lying in a particular window would be considered together
while processing.
Trending Pattern Analysis of Twitter Using Spark Streaming 9
4.7 Results
The application was executed on three consecutive days. The popular hashtags were
identified on all the three days, wherein the hashtags were identified from the tweets in
last 5 s. Figure 2 depicts the popular hashtags on day 1. It also defines the total number
of hashtags in that interval. Figure 3 depicts the popular hashtags that were identified
Trending Pattern Analysis of Twitter Using Spark Streaming 11
on day 2. Similarly, the process was executed on day 3, and the hashtags were iden-
tified as depicted in Fig. 4. The Fig. 5 depicts the bar chart depicting the hashtags
identified versus their respective number of occurrences on each day. Figure 6 depicts
the entire flow of steps being followed to generate the set of popular hashtags.
12 P. Garg et al.
5 Conclusion
Users find and follow the hashtags on twitter to support an issue or a product. Different
companies uses this method of trending as a way to compute the awareness of their
product. They identify, how many people are tweeting about their product using these
hashtags analysis. Companies uses it to find the reach of their product to different users.
Different social networking sites are using this kind of pattern recognition to evaluate
the trending patterns in real-time. Spark has the capability of in-memory processing,
Trending Pattern Analysis of Twitter Using Spark Streaming 13
therefore, this tool was used to implement the system. This work is used for identifying
the fake tweets being made by people to come under the top trending group.
Acknowledgments. We would like to thank our institution Guru Gobind Singh Indraprastha
University for such a great exposure to accomplish such tasks and providing a strong platform to
develop our skills and capabilities.
References
1. Tsai, C.-W., Yang, Y.-L., Chiang, M.-C., Yang, C.-S.: Intelligent big data analysis: a review.
Int. J. Big Data Intell. 1(4), 181–191 (2014)
2. Weller, K., Bruns, A., Burgess, J.E., Mahrt, M., Puschmann, C.: Twitter and society: an
introduction. In: Twitter and Society, vol. 89, pp. xxix–xxxviii. Peter Lang, New York
(2014)
3. Hatua, A., Nguyen, T., Sung, A.: An approach for pattern recognition and prediction of
information diffusion model on Twitter. World Acad. Sci. Eng. Technol. Int. J. Comput. Inf.
Eng. 5(3) (2018)
4. Roberts, H., Sadler, J., Chapman, L.: The value of Twitter data for determining the
emotional responses of people to urban green spaces: a case study and critical evaluation.
Urban Stud. (2018). https://doi.org/10.1177/0042098017748544
5. Türker, İ., Sulak, E.E.: A multilayer network analysis of hashtags in twitter via co-
occurrence and semantic links. Int. J. Mod. Phys. B 32(4), 1850029 (2018)
6. De Cock, B., Pedraza, A.P.: From expressing solidarity to mocking on Twitter: pragmatic
functions of hashtags starting with# jesuis across languages. Lang. Soc. 47, 1–21 (2018)
7. Stypinski, M.: Apache Spark streaming (2017)
8. Shoro, A.G., Soomro, T.R.: Big data analysis: Apache spark perspective. Glob. J. Comput.
Sci. Technol. (2015)
9. Li, H., Dombrowski, L., Brady, E.: Working toward empowering a community: how
immigrant-focused nonprofit organizations use twitter during political conflicts. In:
Proceedings of the 2018 ACM Conference on Supporting Groupwork, pp. 335–346.
ACM (2018)
10. Javed, N., Muralidhara, B.L.: Semantic interpretation of tweets: a contextual knowledge-
based approach for tweet analysis. In: Margret Anouncia, S., Wiil, U.K. (eds.) Knowledge
Computing and Its Applications, pp. 75–98. Springer, Singapore (2018). https://doi.org/10.
1007/978-981-10-6680-1_4
NPCCPM: An Improved Approach Towards
Community Detection in Online Social Networks
1 Introduction
Social interactions have been increasing at a rapid pace, so to detect the subsets of nodes
in the social networks that are intra-densely interactive is an important task of the hour.
Understanding the structure of the network and making decisions in the network has
gained a great attention during last few years. Nodes in networks organize into densely
linked groups that are commonly referred to as network communities, clusters or
modules [1, 2]. The idea of community detection is different from the simple division
of network into sub-groups of members, rather it is a concrete (meaningful) division of
network into sub-groups of members. Detecting these communities in the social
networks is an NP Hard problem.
The understanding and models of network communities has evolved over time [3,
4]. Early works on network community detection were heavily influenced by the
research on the strength of weak ties [5]. This lead researchers to think of networks as
consisting of dense clusters that are linked by a small number of long-range ties [6].
Graph partitioning [2], betweenness centrality [6], as well as modularity [7] based
methods all assume such view of network communities and thus search for edges that
can be cut in order to separate the clusters. Since then a lot of work has been done in the
field and later it was realized that such definition of network communities does not allow
for community overlaps. In many networks a node may belong to multiple communities
simultaneously which leads to overlapping community structure [8–10]. Since the inter‐
actions over the social networks are overlapped in nature [10], so using disjoint
community detection approach doesn’t suit best for social networks because of leaving
out some overlapped natural members of communities. Therefore the natural approach
for discovering the community structure in the social network is overlapped one.
Figure 1 is a simple network with two overlapping communities. In the figure, node 5
and node 6 belong to both the two communities, so they are overlapping nodes.
2 Related Work
Girvan and Newman’s algorithm [6] provides the key idea for community detection i.e.
whatever mechanism we use to detect the communities; the idea is to divide the network
into subsets of nodes, thereby increasing the power of intra-closeness within the groups
than between the groups. Detection of communities in social networks has two
approaches: one is the disjoint community detection - in which communities have no
intersected (common) members and another approach is overlapping community detec‐
tion - in which communities have some intersected members.
Since then much more work has been made in this area [11, 12]. Given a social
network represented by an undirected graph G with nodes n and edges m, the task of
community detection is to find the groups of nodes that are much more like communities.
The term community has made its existence since 1887 [6] but there is no globally
concrete definition of this term yet known but for us it is a group of nodes having better
intracloseness within group than intercloseness between groups [7, 8].
Many traditional community detecting methods hold that each node can only
belong to one community, such as Modularity optimization [13, 14], Hierarchical
clustering [6, 15], Spectral Algorithms [16, 17], methods based on statistical
16 H. A. Khanday et al.
inference [18]. However in some real networks, communities are not independent,
nodes can belong to more than one community, which will lead to overlapping
communities. For example, a researcher may belong to more than one research
group, or a protein may exist in multiple complex systems. Therefore, the identifi‐
cation of overlapping communities is of central importance.
The principal step in the direction of the overlapped community detection approach
has been made by Palla et al. in 2005 whilst proposing Clique Percolation Method
(CPM). CPM is based on the impression that communities tend to form cliques due to
the high density of edges inside. The CPM works by detecting the cliques in the given
social graph of some clique size k, where k is the number of nodes forming the clique.
This CPM is able to discover a community structure that mostly makes the coverage of
a minimum number of nodes under respective community cover, although there can be
a vast percentage of nodes in the network that can be best members of some communities
in the CPM’s detected cover.
Palla and co-workers have designed a software package implementing the CPM,
called cfinder, which is freely available (www.cfinder.org). The algorithm has been
extended to the analysis of weighted, directed and bipartite graphs. For weighted graphs,
in principle one can follow the standard procedure of thresholding the weights, and then
applying the method on the resulting graphs, treating them as unweighted. Farkas et al.
[10] proposed that instead of thresholding the weights of cliques, defines the geometric
mean of the weights of all edges of the clique. The method has been extended to bipartite
graphs by Lehmann et al. [8].
The clique size k parameter is a big disadvantage of CPM because it is unclear
beforehand on what value of k, CPM will result the best meaningful community structure
[1]. There is also an extended version of this CPM that has been named as extended
CPM (ECPM). ECPM uses basic result cover of simple CPM as core cover uses
belonging coefficient to cover all other nodes that have been left uncovered by CPM.
This approach takes much more time to make the coverage of all nodes possible.
3 Proposed Method
Clique Percolation Method (CPM) is the most common technique for detecting over‐
lapped communities. It is based on the concept that the internal edges of a community
are likely to form cliques due to their high density. On the other hand, it is unlikely that
intercommunity edges form cliques. Palla et al. uses the term k-clique to indicate a
complete sub-graph with k vertices. But it is not clear a priori which value of k one has
to choose to detect meaningful structures. Our work tries to resolve the problem of this
clique size parameter k by calculating it dynamically in the given network, such that
there is no difference in making choice about its value beforehand in which case one is
not sure about the reliability of final community structure. We then improve the quality
of CPM by using the output communities of CPM as cores and by using community
conductance [10] as a measure to absorb the rest of the nodes that have been left uncov‐
ered by CPM in efficient amount of time; less its value, better is community like structure.
Another random document with
no related content on Scribd:
always and to remember that the modern woman owes it to herself to go out
of the home and keep abreast with the times?”
But it was not a question. It was a statement. Freda made no reply and
her mother changed the subject with the satisfied air of the sower of seed.
“When you come to Ireland,” she told her father laughingly that night,
“you will sit on the doorstep and learn to smoke a pipe. And Gregory will
be president of the Republic. And I will be—(ask mother)—a model
housewife, chasing the pigs—”
They laughed with an abandonment which indicated some joke deeper
than the banality about the pigs.
“It’s a worthy task,” said her father. “I’ll come—and I’ll enjoy learning
to smoke a pipe and see Gregory run the government—and as for you—
whatever you do you’ll be doing it with spirit.”
She nodded.
“I’ve just begun to break my trail.”
Then the day came when they must leave the little frame house and after
the excitement of getting extremely long railway tickets at the station and
checking all Freda’s luggage through to New York, they said good-by to the
Thorstads and left them standing together, incongruous even in their
farewells to their daughter.
They were to stop at St. Pierre over night. Mrs. Flandon had written to
urge them to do so and Freda would not have refused, if she had been
inclined to, bearing the sense of her obligation to them. She had not told her
father of that. It amused her to think that her father and Gregory each felt
the other responsible for those Fortunatus strings of railway ticket. But she
wanted Gregory to meet the Flandons again that the debt might be more
explainable later on.
St. Pierre was familiar this time when they entered it in mid-afternoon as
she had on that first arrival with her mother. It was pleasant to see Mrs.
Flandon again and to taste just for a moment the comfortable luxury of the
Flandon house. Freda felt in Mrs. Flandon a warmth of friendliness which
made it easy to speak of the money and assure her of Gregory’s ability to
pay it a little later.
“You’re not to bother,” said Helen, “until you’re quite ready. We were
more glad to send it than I can tell you. It’s a hostage to fortune for us.”
Then she changed the subject quickly.
“I wonder if you’ll mind that I asked a few people for dinner to-night.
You married a celebrity and you want to get used to it. So many people
were interested in the news item about your marriage and wanted to meet
Gregory and you. I warned them not to dress so that’s all right.”
“It’s very nice,” said Freda, “I’ll enjoy it and I think—though I never
dare to speak for Gregory—that he will too. I remember having a beautiful
time at dinner here before. When I was here visiting the Brownleys you
asked me—do you remember?”
“I asked the Brownleys to-night. They were in town—all but Allie. I
asked the elder two and Bob and her young man—Ted Smillie, you know.”
She looked at Freda a little quizzically and Freda looked back,
wondering how much she knew.
“Think they’ll want to meet me?” she asked straight-forwardly.
“I do, very much. I think it’s better, Freda, just to put an end to any silly
talk. It may not matter to you but you know I liked your father so much and
it occurred to me that it might matter to him if any untrue gossip were not
killed. And it’s so very easy to kill it.”
“You take a great deal of trouble for me,” protested Freda.
Helen hesitated. She was on the verge of greater confidence and decided
against it.
“Let me do as I please then, will you?” she said smilingly and Freda
agreed.
Helen felt a little dishonest about it. The dinner was another hostage to
fortune. It was gathering up the loose ends neatly—it was brushing out of
sight bits of unsightly thought—establishing a basis which would enable
her later to do other things.
She had an idea that it would please Gage, though he had been non-
committal when she had broached the idea of having Gregory and his wife
for a brief visit. Helen had seen but little of Gage of late. She knew he was
working hard and badly worried about money. They had sold a piece of
property to raise that thousand for the Macmillans and he had told her
definitely of bad times ahead for him. She offered to reduce the expenses of
the household and he had agreed in the necessity. They must shave every
expense. But it invigorated Helen. She had amends to make to Gage and the
more practical the form the easier it was to make them. Neither of them
desired to unnecessarily trouble those dark waters of mental conflict now.
Helen guessed that Gage’s mind was not on her and that the bad tangle of
his business life absorbed him. Brusque, haggard, absorbed, never
attempting or apparently needing affection, he came and went. Never since
Carpenter’s death had they even discussed the question of separation. That
possibility was there. They had beaten a path to it. But hysteria was too
thoroughly weeded out of Gage to press toward it. Without mutual reproach
they both saw that separation in the immediate future was the last
advantageous thing for the work of either of them and flimsy as that
foundation seemed for life together, yet it held them. They turned their
backs upon what they had lost or given up and looked ahead. Helen heard
Gage refer some political question to her for the first time, with a kind of
wonder. She suspected irony, then dropped her own self-consciousness as it
became apparent that he really did not have any twisted motive behind the
query. She began to see that in great measure he had swung loose from her,
substituting some new strength for his dependence on her love. And, when
some moment of emotional sorrow at the loss of their ardors came over her,
she turned as neatly as did he from disturbing thought to the work, which
piled in on her by letter and by conference.
Updated editions will replace the previous one—the old editions will
be renamed.
1.D. The copyright laws of the place where you are located also
govern what you can do with this work. Copyright laws in most
countries are in a constant state of change. If you are outside the
United States, check the laws of your country in addition to the terms
of this agreement before downloading, copying, displaying,
performing, distributing or creating derivative works based on this
work or any other Project Gutenberg™ work. The Foundation makes
no representations concerning the copyright status of any work in
any country other than the United States.
• You pay a royalty fee of 20% of the gross profits you derive from
the use of Project Gutenberg™ works calculated using the
method you already use to calculate your applicable taxes. The
fee is owed to the owner of the Project Gutenberg™ trademark,
but he has agreed to donate royalties under this paragraph to
the Project Gutenberg Literary Archive Foundation. Royalty
payments must be paid within 60 days following each date on
which you prepare (or are legally required to prepare) your
periodic tax returns. Royalty payments should be clearly marked
as such and sent to the Project Gutenberg Literary Archive
Foundation at the address specified in Section 4, “Information
about donations to the Project Gutenberg Literary Archive
Foundation.”
• You comply with all other terms of this agreement for free
distribution of Project Gutenberg™ works.
1.F.
1.F.4. Except for the limited right of replacement or refund set forth in
paragraph 1.F.3, this work is provided to you ‘AS-IS’, WITH NO
OTHER WARRANTIES OF ANY KIND, EXPRESS OR IMPLIED,
INCLUDING BUT NOT LIMITED TO WARRANTIES OF
MERCHANTABILITY OR FITNESS FOR ANY PURPOSE.
Please check the Project Gutenberg web pages for current donation
methods and addresses. Donations are accepted in a number of
other ways including checks, online payments and credit card
donations. To donate, please visit: www.gutenberg.org/donate.
Most people start at our website which has the main PG search
facility: www.gutenberg.org.