A Bird’s Eye View on Social Network Sites and Requirements

Engineering

Nazakat Ali

and Jang-Eui Hong

Department of Computer Science, Chungbuk National University, Cheongju, Republic of Korea

Keywords: Social Network Sites, Requirements Engineering, Overview, User Requirements.

Abstract: Social network sites have become popular and their popularity is growing exponentially every day. From the

requirements engineering point of view, social network sites have provided unprecedented opportunities for

software development organizations to understand the requirements of unknown end-users. Using social

network sites, end-users express their experiences, needs, or concerns about a particular system or a product.

Such information can be useful for software developers to address the concerns of users quickly. To get an

overview of how social network sites are helping requirements engineering and new research trends in this

area, we have surveyed a large number of research papers. We found that social network sites can be a major

source that can be used for requirements elicitation, requirements prioritization, and negotiation. We also

found that the research in this domain is at its beginning stage, but it is rapidly growing with the passage of

time.

1 INTRODUCTION

Social Network Sites (SNS) have become a major

source for millions of users to share their daily

activities, experiences, and opinions for a particular

product through posting. As of year, 2019, 2.77

billion users around the globe use SNS (Statista

2019). As a result, a huge amount of real-time and

highly diverse data is produced per day, which

brought a revolution in many research fields such as

in the data science field. In data science, this data can

be used to predict the political affiliations, forecasting

stock market fluctuations, and marketing trends, etc.

(Conver 2011). Being a communication channel, SNS

have been used in various software development

activities such as in Requirements Engineering (RE).

In RE, SNS have proved their support and improved

several process activities such as requirements

elicitation, negotiation, prioritization (kukreja 2012

and Seyff 2015), and identification and prioritization

of stakeholders (Lim et al. 2010). RE is a critical

activity of the software development lifecycle and

user participation in RE can lead to success in

software projects (Hofman et al. 2001). Therefore, it

seems most suitable to use SNS in order to conduct

https://orcid.org/0000-0002-3875-812X

https://orcid.org/0000-0001-9786-7732

various activities of RE due to the presence of a huge

population on it.

The purpose of this study is to present a bird’s eye

view survey of how SNS benefit RE activities. More

specifically, we are formulating the following

research questions to accomplish our goal.

RQI: What is the current state-of-the-practice of the

SNS (Facebook, Twitter, etc.) as a platform to

conduct RE activities?

RQ2: What RE activities are supported by SNS?

RQ3: Is SNS improving the state-of-the-art of RE?

RQ4: What challenges requirements engineers are

face while using SNS to conduct RE activities?

Based on these formulated research questions, we

have conducted a literature review to summarize the

existing literature in which SNS are considered as a

platform to conduct RE activities. The literature

review is conducted by splitting it through

preparation, data collection, and data analysis phase.

In preparation, a search string was prepared to search

the literature for answering formulated questions,

then searching was performed over a number of

databases such as Scopus, IEEEXplore, Springer

Link, Google Scholar, ACM Digital Library, and

Science Direct. All the identified research articles

Ali, N. and Hong, J.

A Bird’s Eye View on Social Network Sites and Requirements Engineering.

DOI: 10.5220/0008117303470354

In Proceedings of the 14th International Conference on Software Technologies (ICSOFT 2019), pages 347-354

ISBN: 978-989-758-379-7

347

were assessed by means of their title and abstract.

Along with this, references and citations of specific

articles were reviewed to explore more relevant

research articles of our focus. Our intention is not to

conduct an extensive literature review of the targeted

field - rather we are presenting a snapshot of the state-

of-the-art by following some guidelines (study

selection, formulating research questions,

information synthesis, and result reporting ) of

Kitchenham (Kitchenham 2004) to conduct

systematic literature in our targeted field.

The major contributions of this study are as

follows:

 We are presenting an overview of the state-of-

the-art on how SNS supports existing RE

activities. Precisely, we are focusing on how

requirements elicitation, requirements analysis,

and requirements management is supported by

utilizing SNS.

 Since machine learning is utilized to carry out

all these mentioned RE activities, therefore, we

are offering an overview on how machine

learning algorithms (classification, clustering,

regression, etc.) are utilized to get full advantage

of SNS data.

 Furthermore, we analyze existing literature to

uncover research trends and future research

directions on the use of SNS for conducting RE

activities.

The rest of this paper is organized as follows. Section

2 provides necessary background information on SNS

and RE is required to understand the rest of the paper.

Section 3 presents an overview of RE activities where

SNS is used as a platform to conduct RE. In section

4, we summarize our major findings of this research

and answer the formulated questions while section 5

concludes this paper.

2 SOCIAL NETWORK SITES AND

REQUIREMENTS

ENGINEERING

The fundamental goal of a software system is to meet

the requirements of its users by offering

functionalities that can satisfy and fulfill the

expectations of its users. This goal is accomplished

by applying several engineering techniques. RE is an

activity through which user requirements are

identified for a specific domain in a systematic

manner to understand what features a specific product

should have in order to satisfy its users’ expectations.

The success of a software project mainly depends

on its RE process because requirements are the

determining factor of software quality. The empirical

evidence has shown that errors in the requirements

are most common in the software development

lifecycle and it is considered as most time- consuming

and expensive task to correct in later stages (Wohlin,

2005). According to a research report (Standish,

2014), 52.7% of projects met with challenges and

problems and cost 189% of their total estimated

budget. The report shows that only 16.1% of all US

projects completed on-budget, on-time with all

planned features, while 31.1% of projects were never

completed. The investigation report identified poor

requirements as the main source of problems, along

with other factors such as unclear objectives and low

user involvement. Likewise, another survey

(European 1996) which was carried out with 3800

organizations from 17 countries in Europe, the

findings show that half (50%) of the problems were

due to requirements specification and rest half (50%)

problems emerged due to requirements management.

3 EXISTING CONTRIBUTIONS

In this section, we will summarize the existing

literature which has considered SNS as a platform to

conduct RE activities. A comprehensive inclusion

and exclusion criteria were followed to select the

literature. As a whole, Kitchenham et al. guidelines

(Kitchenham, 2004) were partially followed to

conduct this survey.

The success of any software project majorly

depends on the accurate identification of

stakeholders’ expectations and requirements for their

anticipated system. Requirements elicitation through

the manual process is costly in terms of resource and

efforts. Therefore, SNS provided a platform to

conduct requirements elicitation by mining

microblogs e.g., social media, app store reviews, and

requirement documents. The mining task is carried

out by applying various techniques such as Natural

Language Processing (NLP), text mining and

Machine Learning (ML) (Ali, 2016). In current

practice, popular SNS such as Twitter, Facebook, and

Snapchat are used to extract user requirements by

applying ML techniques. User comments, tweets, or

reviews are not structured documents, they contain

useful information along with noise which makes

difficult to extract user requirements with manual

annotations. Therefore, ML techniques are most

favorable to identify user requirements automatically

ICSOFT 2019 - 14th International Conference on Software Technologies

348

or semi-automatically by reducing efforts, cost, and

time significantly.

3.1 RE, ML and Social Network Sites

Guzman et al. (Guzman, 2016) have proposed an

approach that uses Twitter as a platform to elicit user

requirements. Their exploratory study investigated

the content of 10,986,494 tweets about 30 software

application. The authors have categorized tweets into

more 22 categories including feature shortcomings,

bug reports, feature request, and feature strength, etc.

The authors have applied Support Vector Machine

(SVM) and C4.5 ML classifiers to classify the tweets

automatically. Both classifiers had a similar

performance to classify tweets related to non-

technical stakeholders. However, SVM performed

well to classify tweets related to technical

stakeholders and the general public.

Guzman et al. (Guzman, 2017(a)) presented a

similar study for classifying, ranking and grouping

tweets for software evolution. The authors used

Twitter as a platform to elicit user requirements. The

proposed approach classifies tweets into two

categories including improvement request and others.

The authors used 68,108 tweets for their experiment.

Naïve Bayes (NB) and Random Forest (RF)

classifiers were used to classifying the tweets. In

summary, the proposed approach was able to classify

tweets automatically into improvement request and

other categories with F-measure of 0.79.

Williams et al. (Williams, 2017(a)) proposed an

approach which also used Twitter as a platform to

elicit user requirements. The authors first analyzed

4000 tweets of 10 various software systems to

classify tweets into bug reports, user requirements

and miscellaneous and spam. Then, the authors used

NB and SVM to classify the tweets automatically.

The authors used three classification features (bag-of-

words, sentiment and stemming) for their tweets

classification. The study claims that the sentiment

feature did not affect classification results. The

authors used 188,737 unique tweets for their

experimentation.

Xiao et al. (Xiao, 2015) have used popular SNS

such as StackOverflow, a popular Q & A software

community site, where programmers and software

engineers share their knowledge and experience. The

collected data from StackOverflow was used to

extract requirements requests to help software

developers. In their study, SVM classifier was used to

test the data. The authors extracted tagged questions

along with related information to elicit requirement

requests for a relevant software application. Initially,

three ML algorithms including pure SVM, linguistics

rules and SVM with dictionary were selected for

experiments and comparison. SVM with requirement

dictionary is the combination of a dictionary

(keyword sets) and SVM to enhance the performance

of classification. It is concluded from the

experimental result that SVM, with precision 69%,

recall 75% and F-measure 72%, performed better

than traditional linguistic rules. SVM with

requirement dictionary improved performance with

precision 72%, recall 77% and F-measure 74.4%.

Prasetyo et al. (Prasetyo, 2012) performed a

preliminary study to investigate the feasibility of

automatic classification of microblogs into relevant

and irrelevant categories. All those tweets that could

help software developers to understand user needs

were categorized as relevant tweets while rest were

categorized as irrelevant tweets. The authors used 300

tweets of [32] comprising either one of the following

9 hashtags including #csharp, #scrum, #javascript,

#opensource #.net, #jquery, #testing, #azure, and

#java, and tweets were classified into 10 categories

including news, commercials, tools and code, events,

personal, tips, opinions, jobs, Q &A, and

miscellaneous. The aim of this study was to classify

tweets automatically into relevant and irrelevant.

Therefore, the authors have re-labeled data set into

relevant and irrelevant categories. If a tweet was

potentially interesting to a software developer to

develop a target system, otherwise they labeled it as

irrelevant. The result shows that 41% of tweets were

relevant while 53% were irrelevant. For the automatic

classification of tweets, the authors used SVM

classifier.

Guzman et al. (Guzman et al., 2017(b)) used

Twitter to elicit user requirements. This study reports

on the usage characteristics, contents, and automatic

classification potential of user tweets about RE and

software evolution. The authors explored a dataset

containing 10,986,495 tweets about 30 various

software applications by using descriptive statistics,

content analysis, lexical sentiment analysis, and ML.

The authors have used NB, multinomial NB, SVM,

J48, and RF to classify a huge amount of user tweets

according to the relevance to different stakeholders

and user type. The result shows that the classifiers

produced promising results in some cases. The

automated relevance classification was possible with

an F-score ranging from 77% to 52%, while the

identification of tweets tweeted by bots got a

promising F-score of 84%. This study has performed

a fine granular evaluation of user feedback from

Twitter and its comparison with App store reviews.

A Bird’s Eye View on Social Network Sites and Requirements Engineering

349

Deshpande et al. (Deshpande et al., 2018)

proposed a study that assessed user feedback from

Twitter in terms of timing as well as content and

compared with the App store review. This study

employed various text analysis and NLP methods

such as semantic analysis to analyze tweets and app

store reviews. The authors have used SVM and NB

for their classification. The analysis of topics showed

that 72% of tweets were functional requirements

which discussed the behavioral aspects of mobile

apps while in the App store it was 80%.

Singer et al. (Singer et al., 2014)) surveyed 271

GitHub users and interviewed 27 software developers

to know about the importance of Twitter in software

development. Authors found that developers often

rely on various online resources, such as Twitter to

keep themselves informed about their systems. The

authors also found that Twitter helps software

developers to stay aware of industry changes,

technology changes and for learning.

Williams et al. (Williams et al., 2017(b))

presented a preliminary study aiming to detect and

interpret emotions present in software related tweets.

The preliminary study was conducted using a dataset

of 1000 tweets taken from tweets of a broad range of

software applications. At first, the tweets were

manually classified by two annotators with 5 years of

programming into two levels of abstractions,

including its general emotional polarity such as

positive, negative and neutral, and the specific

emotions it conveys. After manual classification

dataset was trained and applied SVM and NB to

extract sentiment automatically. The authors have

seen that emotions were closely related to specific

events such as product release etc. The results also

show that tweets which carry negative sentiment

usually report bugs, report feature shortcoming, etc.

Seyff et al. (Seyff et al., 2018) propose a crowd

and sustainability-focused platform for semi-

automated requirements elicitation, negotiation, and

analysis. This platform enables a diverse and

distributed group of stakeholders or users (crowd) to

communicate and negotiate their requirements. The

proposed conceptual solution includes three

components: 1) CrowdFeed enables users to give

their feedback regarding the software service or

products they use, 2) Requirements and Sustainability

Service classifies, clusters, and analyze the feedback

received from the crowd, 3) Requirements and

Sustainability Integrator supports the visualization

and assessment of effects on sustainability.

3.2 RE and Social Network Sites

Lee et al. (Lee et al., 2011) used both Twitter and

Facebook for requirements elicitation and

prioritization. The proposed approach has three

phases including preparing phase, gathering phase

and refining phase. The preparing phase describes the

preparation for requirements elicitation through SNS.

In the preparation phase, a keyword-based method is

used to approach a large number of user opinions. The

authors first selected domain related keywords e.g.

smartphone app, smartphone OS, smartphone design,

etc. The frequency of selected domain-related

keywords was recorded to select those keywords

which are most discussed on SNS. In the gathering

phase, the SNS and its users were approached, and

their opinions were collected. This phase was further

divided into two parts according to types of access to

the subject and the direction of approach: opinion-

based access and community-based access. Opinion-

based access targeted individual users, not SNS

communities while in community-based access

diverse communities were targeted and their opinions

were collected.

Seyff et al. (Seyff et al., 2015) proposed an

approach for RE that employs Facebook for

elicitation, prioritization, and negotiation of

requirements. The authors have conducted three

exploratory studies using Facebook to see whether the

potential of popular SNS (i.e. Facebook) can allow

and support end users to participate in requirements

elicitation, prioritization, and negotiation. This study

is inspired by EasyWinWin (Gruenbacher, 2000) and

authors foresaw that their proposed lightweight and

end-user focused RE approach had many potential

applications. The authors particularly see its

relevance with new software paradigms such as cloud

computing, mobile computing, and ecosystems,

where potential end users are not within instant reach

and support provided by traditional RE methods is

insufficient. The authors claim that their approach can

provide a new channel for eliciting innovative ideas

and needs as well as getting feedback from

stakeholders who are not directly reachable by

software development organizations or software

development teams. Furthermore, authors see its

applicability within traditional software projects as an

additional means for engaging end users.

Ali et al. (Ali et al., 2017) considered both Twitter

and Facebook to evolve systems. The authors

proposed a cyclic process that elicits user

requirements continuously from SNS. After data

extraction, the corpus is gone through a

comprehensive NLP phase to elicit user requirements.

ICSOFT 2019 - 14th International Conference on Software Technologies

350

In this approach, bigrams and trigrams were exacted

to get the related keywords. Later these bigrams and

trigrams were interpreted as user requirements.

Furthermore, the authors used the ontology-based

similarity algorithm to detect redundant keywords.

After requirements identification, a feature model

was manually built. After every iteration, the feature

model was updated, and the system was reconfigured

to adapt to the changes. In another study (Ali et al.,

2016) authors have elicited domain requirements

from SNS to support software product line evolution.

This study has analyzed Facebook, Twitter, and

LinkedIn and selected Twitter and Facebook as a

platform to elicit user requirements.

Based on the win-win (Gruenbacher, 2000)

negotiation model, Kukreja et al. (Kukreja, 2012)

developed a social network site such as “Winbook”,

which is based on the social networking paradigm.

Winbook is similar to Facebook, but its contents are

organized using color-coded labels similar to Gmail.

It also resembles Excel because it has the ability to

prioritize requirements by performing a number of

sensitivity analyses with respect to business goals.

This study is able to elicit, negotiate and prioritize

user requirements.

Romasha et al. (Romasha et al., 2018) have

conducted a survey to know how SNS empower RE.

In this questionnaire-based survey, among

respondents, 41% were software developers, 31 were

team leaders, and 10% of respondents were a business

analyst. As per the survey result, majority

respondents agreed that RE is a communication

activity and 73.8 % respondent expressed that they

face communication problem to gather requirements.

Among respondents, 57% of respondents reported

that they used Facebook to elicit user requirements

and solve the communication problem.

Maalej et al. (Maalej et al., 2015)] have discussed

the usage of SNS and app stores to elicit user

requirements. The authors discussed how software

development organizations can use user feedback to

identify, prioritize, and manage requirements. The

authors have seen three directions in practice. First,

tools for feedback analytics would help in order to

deal with a large number of user opinions by

classifying, summarizing, and filtering them. Second,

automatically collected opinions, logs, and

interaction traces could help software developers to

understand user feedback and improve feedback

quality. Third, once the feedback is collected and

analyzed, the question arises how can software

development organizations use this information and

integrate it into their software development process to

decide when to release new updates. What feature or

requirements should be added, deleted, or changed?

Groen et al. (Groen et al., 2017) have proposed

CrowdRE approach to elicit user requirements from

the crowd. The authors considered the crowd as a

sender of the feedback and software development

companies as a receiver. The authors dived user

feedback into two categories: pull feedback and push

feedback. Push feedback is when software

development organizations explicitly ask the crowd

for feedback while the push feedback is when the user

in the crowd start a discussion. In both cases, user

feedback was utilized to elicit user requirements.

Additionally, the authors have listed a number of

challenges as well including crowd motivation,

privacy and personalization, and feedback

analyzation issues.

Mughal et al. (Mughal et al., 2018) have proposed

a network-based process using Facebook to reduce

the in-group biasness during RE. The proposed

approach combined hybrid centrality measure and

power, legitimacy, and urgency technique. This study

has focused on stakeholder identification and

prioritization during RE. The main strength of this

article is that it has shown better performance than

partner biasness issue through hybrid centrality

measure and the power, legitimacy, urgency and

urgency model, as claimed. The feasibility of the

proposed social network-based process for

identification and prioritization of requirements of

stakeholders is shown through a controlled

experiment conducted on an example set of 40

stakeholders.

Kanchev et al. (Kanchev et al., 2017) have

proposed a methodology called Canary that generates

a requirements-oriented view of online discussions.

The methodology is a conceptual model that brings

requirements-relevant and argumentation-relevant

aspects of online discussions together. The authors

have claimed that it was feasible to convert online

discussions into instances Canary model by

crowdsourcing annotations of the discussions.

4 RESULTS AND DISCUSSION

Our study focuses on a specific area and points to a

number of trends that we will explain in this section.

Overall, the goal of this study is to provide a state-of-

the-art of use of SNS to conduct various activities of

RE for giving an overview to domain experts and act

as a foundation to this field for researchers.

Furthermore, the results we provide here are derived

from the literature review we conducted. Our survey

A Bird’s Eye View on Social Network Sites and Requirements Engineering

351

Table 1: Contributions and RE activities supported by SNS.

Contribution

Publication

Year

Objective/Purpose ML Tasks

RE Activities

Support

Guzman et al.

(+)

2016

Categorization of tweets into 22 categories including

bug reports, feature shortcoming, feature demand etc.

Classification Elicitation

Guzman et al.

(+)

2017(a)

Classification of tweets into two categories:

improvement request and others.

Classification Elicitation

Williams et al.

(+)

2017(a)

Classification of tweets into bug reports, user

requirements, and miscellaneous

Classification

Wordcloud

Elicitation

4 Xiao et al (-) 2015 Extract requirement requests Classification Elicitation

5 Lee et al. (+) 2011 Requirements elicitation and prioritization -

Elicitation

Prioritization

6 Seyff et al. (-) 2015

Requirements elicitation, negotiation, and

prioritization using SNS

Elicitation

Prioritization

Negotiation

7 Ali et al. (*) 2017

Eliciting user requirements from SNS to evolve

software product lines

Elicitation

Prioritization

Prasetyo et al.

(-)

2012

To know the relevance of tweets with software

systems

Classification Elicitation

Guzman et al.

(+)

2017(b)

searching for relevant information on software

applications within the vast stream of tweets

Classification

Elicitation

Prioritization

Deshpande et

al. (*)

2018

Comparison between App store reviews and Tweeter

messages for requirements

Classification Elicitation

Kukreja et al.

(+)

2012 Developing Winbook -

Elicitation

Prioritization

Negotiation

Romasha et

al. (*)

2018

Conducting a survey that to know how SNS empower

Elicitation

Negotiation

Maalej et al.

(*)

2016

The authors discussed how software development

organizations can use user feedback to identify,

prioritize, and manage requirements

Elicitation

Prioritization

Negotiation

14 Leif et al. (*) 2014

Surveying and interviewing software developers to

know about the importance of Twitter in software

development.

‐ ‐

Williams et al.

(-)

2017(b)

presenting a preliminary study aiming to detect and

interpret emotions present in software related tweets.

Sentiment Elicitation

Groen et al.

(*)

2017

Proposing CrowdRE approach to elicit user

requirements from crowd.

- Elicitation

Mughal et al.

(+)

2018

proposing a network-based process using Facebook to

reduce the in-group biasness during RE

Stakeholders

prioritization in

requirements

elicitation

process

20 Ali et al. (-) 2018

Introducing the idea of a platform for crowd-focused

requirements engineering that supports the evolution

towards sustainability

Multi-label

classification

Intended to

elicit

requirements

Kanchev et al.

(+)

2017

extracting and querying requirements-related

information in online discussions (reddit)

- Elicitation

22 Seyffetal.(‐) 2018

sustainability-focused platform for semi-automated

requirements elicitation, negotiation, and analysis

Classification

Elicitation,

Negotiation

Legends: (+) improves the state of the art; (*) no information how it connects with state of the art; (-) comparable with state

of the art;

is not a fully systematic literature review (as

mentioned in section 1) and our conclusions and

results can be revised or extended by future studies in

this domain. Table 1 provides answers to RQ1

(“What are the current state-of-the-practice that use

SNS as a platform to conduct RE activities?”) and

RQ2 (“What RE activities are supported by SNS?”).

Twenty-one papers (Contributions column in Table

1) were thoroughly examined to know the state-of-

the-art which employs SNS to conduct RE activities.

We have seen that; the majority of studies were initial

proposals with a little academic literature and no

industrial applications in real software projects. ML

tasks and RE support Activities columns, as

mentioned in Table 1, answer our RQ2. The

contributions column in Table I also provides a partial

response to RQ 3 (“Is SNS improving the state-of-

the-art of RE?”). The answer to RQ3 appears to be at

its beginning, given the prevalent lack of comparison

with the state of the art as can be seen in Table 1. The

ICSOFT 2019 - 14th International Conference on Software Technologies

352

RQ3 was answered by comparing proposed

approaches, that consider SNS for RE, with

traditional state-of-the-art. Some additional

information is also given in Table 1 i.e. what kind of

ML models were used in order to elicit user

requirements from the corpus. We have seen that NB

and SVM are mostly used for classification.

The SNS contains a huge amount of data

including user expectations and experiences.

Software developers could use this rich information

to understand the needs, experiences, and sentiment

of their product users. Mining SNS especially user

opinions, may yield useful information for the

software development organization to update their

products. However, extracting user requirements out

of huge data is also a challenging task. The opinions

are usually expressed without considering grammar

format which creates problems lately for corpus

processing. Therefore, we have seen that the collected

data was unstructured. Software developers rely on

user feedback for requirements elicitation,

negotiation, and prioritization. However, the

trustworthiness of comments, tweets, or, in general,

feedback is still a big challenge for the software

development community. Groen et al. (Groen et al.

2017) have mentioned a number of challenges in this

domain. User motivation is one of the big challenges

to get feedback. User privacy and personalization are

other issues which RE community faces while using

SNS as a platform to elicit user requirements. These

observations answer RQ4 (“RQ4: What are the

challenges requirements engineers face while using

SNS to conduct RE activities?”).

Table 1 did not mention the NLP techniques

applied in our surveyed papers. It is obvious that NLP

techniques are widely used by the majority of

researchers in their research that tackles the

application of SNS for RE. This is not astonishing and

even more common in data science. We observed that

NLP was used in the preprocessing phase in order to

bring data into the desired format that can be

consumed by ML algorithms. Majority of the articles

we have gone through in our survey have the opinion

that a large amount of imprecise data produced by

SNS users can bring about enormous benefits to

software development organizations when processed

by applying ML algorithms. The majority of papers

we surveyed on the topic of SNS and RE have to do

with either elicitation or prioritization activities of RE

and a little literature talked about negotiation activity

as well. Figure 1 shows the result of our investigation.

As mentioned above, the research in the domain

of RE and SNS is at its beginning but the research

trend is growing rapidly with the passage of time.

Figure 1: RE activities supported by SNS in our study.

Figure 2 shows the number of publications with

the passage of time which clearly shows the ongoing

research trend on SNS and RE domain. Figure 2 also

shows the number of publications that used ML

techniques to support RE activities.

Figure 2: Number of publications in SNS and RE domain.

5 CONCLUSIONS

Through our bird’s eye view on SNS and RE, we have

seen that in the past few years a research trend has

started to bring these two worlds together. The

investigation shows that studies in this domain are at

the beginning stage. Although, the academicians have

started to explore the potentials of SNS in order to use

it to conduct RE activities.

We have conducted a literature survey to know

how SNS is being used to conduct RE activities. We

formulated four types of research questions (Section

I) and extracted data according to answer these

questions. RQ1 revealed the ongoing research that

attempted to bring together SNS and RE worlds. RQ2

exposed that what kind of RE activities are supported

by SNS and what are the ML algorithms are used to

extract the required information from the SNS data.

In RQ3, we have seen that research in this domain is

in its early stages and need more reliable approaches

that will benefit software projects in the real world.

Finally, RQ4 answered that what challenges are faced

by the research community while bringing two worlds

together.

A Bird’s Eye View on Social Network Sites and Requirements Engineering

353

ACKNOWLEDGMENTS

This research was supported by Next-Generation

Information Computing Development Program

through the NRF of Korea funded by the Ministry of

Science, ICT (NRF-2014M3C4A7030505, NRF-

2017M3C4A7066479).

REFERENCES

Ali, N., Kim, S. and Hong, J.E., 2016. Listen closely,

respond quickly: Enhancing conformity of SPL domain

requirements through SNS. International Conference

on Information Science and Communications

Technologies (ICISCT), pp. 1-5.

Ali, N. and Hong, J.E., 2017. Creating adaptive software

architecture dynamically for recurring new

requirements. International Conference on Open

Source Systems & Technologies (ICOSST) (pp. 67-72.

Conover et al., 2011, O. Predicting the political alignment

of twitter users. In 2011 IEEE third international

conference on privacy, security, risk and trust and 2011

IEEE third international conference on social

computing, pp. 192-199.

Deshpande, G. and Rokne, J., 2018. User Feedback from

Tweets vs App Store Reviews: An Exploratory Study

of Frequency, Timing and Content. 5th International

Workshop on Artificial Intelligence for Requirements

Engineering (AIRE) (pp. 15-21.

European Software Institute., 1996. European user survey

analysis. Report USV_EUR 2.1, ESPITI project.

Guzman, E., Alkadhi, R. and Seyff, N., 2016. A needle in a

haystack: What do twitter users say about software?

In IEEE 24th International Requirements Engineering

Conference (RE), pp. 96-105.

Guzman, E., Ibrahim, M. and Glinz, M., 2017(a). A little

bird told me: Mining tweets for requirements and

software evolution. IEEE 25th International

Requirements Engineering Conference (RE), pp. 11-20.

Guzman, E., Alkadhi, R. and Seyff, N., 2017(b). An

exploratory study of Twitter messages about software

applications. RE, 22(3), pp.387-412.

Gruenbacher P., 2000., Collaborative requirements

negotiation with EasyWinWin," Proceedings 11th

International Workshop on Database and Expert

Systems Applications, pp. 954-958.

Groen, E.C.,et al., 2017. The crowd in requirements

engineering: The landscape and challenges. IEEE

software, 34(2), pp.44-52.

Hofmann, H.F. and Lehner, F., 2001. Requirements

engineering as a success factor in software projects.

IEEE software, (4), pp.58-66.

Kukreja, N., 2012,. Winbook: a social networking-based frame-

work for collaborative requirements elicitation and WinWin

negotiations. In Proceedings of the 34th International

Conference on Software Engineering, pp. 1610-1612.

Kitchenham, B., 2004. Procedures for performing systematic

reviews. Keele, UK, Keele University, pp.1-26.

Kanchev, G.M., Murukannaiah, P.K., Chopra, A.K. and

Sawyer, P., 2017. Canary: Extracting requirements-

related information from online discussions. 25th

International Requirements Engineering Conference

(RE)., pp. 31-40.

Lim, S.L., Quercia, D. and Finkelstein, A., 2010. StakeNet:

using social networks to analyse the stakeholders of

large-scale software projects. In Proceedings of the

International Conference on Software

Engineering-Volume 1 (pp. 295-304).

Lee, Y.K., Kim, N.H., Kim, D., Lee, D.H. and In, H.P.,

2011. Customer Requirements Elicitation based on

Social Network Service. KSII Transactions on Internet

& Information Systems, 5(10).

Maalej, W., Nayebi, M., Johann, T. and Ruhe, G., 2015.

Toward data-driven requirements engineering. IEEE

Software, 33(1), pp.48-54.

Mughal, S., Abbas, A., Ahmad, N. and Khan, S.U., 2018.

A Social Network Based Process to Minimize In-Group

Biasedness During Requirement Engineering. IEEE

Access, 6, pp.66870-66885.

Number of social media users worldwide from 2010 to

2021 (in billions https://www.statista.com/statistics/

278414/number-of-worldwide-social-network-users/

accessed January 23, 2019.

Prasetyo, et al., 2012, September. Automatic classification of

software related microblogs. 28th IEEE International

Conference on Software Maintenance (ICSM) (pp. 596-599

Romasha K, et al., 2018., Social Networking Sites

Empower Requirement Engineering” 31

IBIMA

Conference. pp. 25-26 Milan, Italy.

Singer, L., Figueira Filho, F. and Storey, M.A., 2014.

Software engineering at the speed of light: how

developers stay current using twitter. In Pros of the 36th

International Conference on Software Engineering (pp.

211-221

Seyff, N., Betz, S., et al. 2018, August. Crowd-focused semi-

automated requirements engineering for evolution towards

sustainability. In 2018 IEEE 26th International

Requirements Engineering Conference (RE), pp. 370-375.

Seyff, N., Todoran, I., Caluser, K., Singer, L. and Glinz, M.,

2015. Using popular social network sites to support

requirements elicitation, prioritization and

negotiation. Journal of Internet Services and

Applications, 6(1), p.7.

Standish. C Group., 2014. The standish group report.

Wohlin, C., 2005. Engineering and managing software

requirements. Springer Science & Business Media.

Williams, G. and Mahmoud, A., 2017(a). Mining Twitter

feeds for software user requirements. IEEE 25th

International Requirements Engineering Conference

(RE)(pp. 1-10.

Williams G and A. Mahmoud., 2017(b) Analyzing,

Classifying, and Interpreting Emotions in Software Users'

Tweets, 2nd International Workshop on Emotion

Awareness in Software Engineering (SEmotion), pp. 2-7

Xiao, M., Yin, G., Wang, T., Yang, C. and Chen, M., 2015.

Requirement acquisition from social q&a sites. In

Requirements Engineering in the Big Data Era (pp. 64-74.

ICSOFT 2019 - 14th International Conference on Software Technologies

354