On the Need of Integrating Social Media Channels andOpen Source Software Repositories

Aftab IqbalInsight Centre for Data Analytics

National University of Ireland, Galway (NUIG)IDA Business Park, Lower Dangan, Galway, Ireland

Email: aftab.iqbal@insight-centre.org

Stefan DeckerInsight Centre for Data Analytics

National University of Ireland, Galway (NUIG)IDA Business Park, Lower Dangan, Galway, Ireland

Email: stefan.decker@insight-centre.org

Abstract—The growing interest in the usage of social mediachannels have attracted the open source software communityto adopt an identity in order to disseminate project-related in-formation to a wider audience. We foresee the need to integratesocial media channels and open source software repositories inorder to get an integrated view on the software project notonly from the software development perspective but also fromsocial perspective. Therefore, in this paper we study the usageof Twitter by software developers through harvesting theirproject-related activities on Twitter. In particular, we presentthe most commonly used hashtags by software developers andfurther investigate if project-related hashtags are the mostfrequent and commonly used hashtags by software developers.Based on our findings, we argue that relevant information fromsocial media channels should be integrated with the open sourcesoftware repositories in order to provide a homogeneous viewon a software project.

Keywords-Linked Data; Semantic Web; Social Media; Twit-ter; Software Repositories; Data Integration;

I. INTRODUCTION AND MOTIVATION

The growing interest in the usage of online social mediachannels such as Facebook, Twitter, MySpace, LinkedIn etc.,have attracted millions of users. Users join together on socialmedia channels to share information about their activitiesand opinions on various topics. Topics usually range fromdaily routine activities to current events, media news etc.With the popularity of social media channels, we have seenemerging growth in the usage of these channels by the opensource software communities in order to disseminate project-related information to a wider audience. We have noticedactivities of software developers and users on these channelsdiscussing or sharing their thoughts and opinions about aparticular release or issues relevant to software projects.

Open source software communities are often found toadopt an identity on social media channels (e.g., ApacheSolr/Lucene1 on Twitter, MySQL2 on Facebook) in order todisseminate3 project-related information (release announce-ments, major bug fixes etc.) to a wider audience or gather

1https://twitter.com/SolrLucene2https://www.facebook.com/mysql3for example, https://twitter.com/olamy/status/305334578103582720

feedback/questions posted by the users. Software developerscontributing to open source projects also exists on socialmedia channels. Quite often, they discuss, debate or shareexperiences4 with others relevant to a software project usingproject related hashtags (e.g., #apache, #maven, #hadoopetc.). Hence, the discussions covering open source projectsare not limited to dedicated forums, blogs or mailing lists,there also exists huge amount of information on the socialmedia channels. Therefore, it is worth mentioning that theinformation related to open source projects and softwaredevelopers are distributed on the Web in heterogeneousdata islands [1], i.e., social media channels and softwarerepositories of a project as shown in Figure 1.

Figure 1. A conceptual diagram showing existence of software developerson social media channels and software projects.

Given the fact that the information relevant to softwareprojects are shared by software developers and users onsocial media channels, one cannot ignore the importanceof consuming this information. Therefore, there is a needto bridge the connection between software repositoriesand social media channels (i.e., interlinking project-relatedtweets/posts/hashtags, project IDs [2], software developerIDs [3] etc.). By enabling this connection, we will have an

4for example, https://twitter.com/olamy/status/231031288734285824

integrated view on the software project from the softwaredevelopment and social aspects that can be exploited tosupport certain use case scenarios:

• Mining end-users response on the release or usageof a particular software project. The variance andenormity of information that propagates through socialmedia channels presents an opportunity to harness thedata and build models on top to aggregate the opinionsof the community related to a software project [4];

• Investigating the popularity of a particular softwareproject. The popularity of a software project on socialmedia channels [5] can be investigated by applyingsentiment analysis on social messages [6], [7], [8];

• Analysis of the social behavior of software developersin different communication channels (i.e., social mediachannels and software project repositories) [9].

In the current scope of this paper, we do not take intoaccount all Twitter users who are tweeting about someselected software projects but rather we focus our study tolook into the tweets of only those Twitter users who aredevelopers of those selected software projects. The reasonbehind this choice is to highlight the needs of integratingsocial media channels to software project repositories, whichwill be discussed later in detail through two different re-search questions.

Based on the underlying idea of integrating social mediachannels and software project repositories, the contributionof this work is manifolds. We have identified social mediachannels as a platform, which is used by the open sourcesoftware communities to disseminate project-related infor-mation to a wider audience. Further, we highlight the needsof integrating software project repositories and social mediachannels in order to get an integrated view on the softwareproject not only from the software development perspectivebut also from the social perspective. For integration pur-poses, we need to find out the activities of software devel-opers on social media channels with respect to the softwareprojects they are developing. Therefore, we investigate theusage of different hashtags by software developers on socialmedia channels (Twitter, in particular) in order to discover ifproject-related hashtags are the most frequent and commonlyused hashtags by software developers.

The rest of the paper is organized as follows: in Section IIwe discuss related work in the context of analyzing socialmedia channels with respect to its usage by software devel-opers. We investigate the usage of Twitter as a platform bysoftware developers to disseminate project-related informa-tion in Section III. We report on our findings and furthermotivates the need of integrating social media channels tothe software project repositories in Section IV. Finally, weconclude in Section V.

II. RELATED WORK

Much research has been carried out on analyzing socialmedia channels with different perspectives in mind; privacyissues [10], [11], influential users within a community [12],[13], [14], churns in online social communities [15], trendingtopic classifications [16], [17] etc. In the last few years, re-search has also started to emerge on analyzing the activitiesof software developers with respect to software projects onsocial media channels. Wang et al. [18] investigated howopen source community uses social media channels throughstudying the usage of Twitter by software developers ofDrupal open source content management system. In anotherstudy, Bougie et al. [19] found that software developersextensively leverages Twitter’s capabilities for conversationand information sharing. Tian et al. [20] performed a prelim-inary study on what software developers microblogs aboutby analyzing the content of microblogs from Twitter andfurther categorized them. Similarly, Prasetyo et al. [21]automatically classified tweets regarding their relevance tosoftware engineering. Leif et al. [22] discovered in theirqualitative study that software developers use Twitter to keepthemselves up-to-date with the fast changing developmentlandscape, for learning and building relationships.

In particular, there are no research case studies availableto date (to the best of our knowledge), which investigatesif software developers use social media channels to talkabout or promote software projects they are developing. Thework presented in this paper differs from existing researchin the sense that we investigate the most common andfrequently used hashtags by software developers on socialmedia channels (Twitter in particular) with respect to thesoftware projects they are developing.

III. ACTIVITY OF SOFTWARE DEVELOPERS ON SOCIALMEDIA CHANNELS: PRELIMINARY STUDY

In this section, we investigate the usage of Twitter asa platform by software developers in order to find out ifthey promote or discuss software projects (on Twitter) thatthey are developing. This is required as a preliminary steptowards integrating information coming through Twitter withthe software repositories and to the software developersin particular. Therefore, we crawled tweets of softwaredevelopers of some randomly selected Apache projects. TheApache projects and its associated software developers usedin this study are shown in Table I.

For each Apache project under consideration, we con-sidered only those software developers who have commitrights on the source control repository. For these softwaredevelopers, we manually checked if they also exist onTwitter and using the Twitter account frequently. We foundsome software developers whose profile were set privateby themselves and only authorized followers could viewtheir tweets. We ignored such software developers in theTwitter data crawling and analysis phase. Table I shows

Apache Projects Developers (SVN) Developers (Twitter)Apache Camel [23] 36 24Apache Directory [24] 51 11Apache Felix [25] 47 17Apache Hadoop [26] 97 35Apache Logging [27] 37 7Apache Lucene [28] 51 18Apache Maven [29] 40 10Apache Mina [30] 28 9Apache MyFaces [31] 82 16Apache OfBiz [32] 25 11

Table ISOFTWARE DEVELOPERS CONTRIBUTING TO APACHE PROJECTS AND

ALSO USING TWITTER.

for each Apache project, the number of software developerswho have commit rights to the source control repository andthe software developers actually found on Twitter. Although,not all software developers developing the Apache projects(under consideration for this study) are found on Twitter butthe results in Table I shows good evidence of the existenceof software developers on Twitter, which will be sufficientto address some questions discussed later in this section. Foreach software developer, we obtained their tweets and profileinformation using Twitter API. Table II presents summaryof our crawled Twitter data set by showing total numberof software developers, total number of tweets and distincthashtags used in the tweets by software developers.

Developers Tweets hashtags158 186,709 20,395

Table IITWITTER DATASET SUMMARY.

We crawled data from January 2009 till January 2013.The distribution of tweets per software developer is shownin Figure 2. The average number of tweets each softwaredeveloper posted is 1181.70. The decline in the curve showsthat certain software developers tweeted less on Twitter. Onereason could be that certain software developers joined lateon Twitter than other software developers. We queried ourTwitter dataset in order to find out the oldest and recentjoining date of software developers on Twitter. We found thatthe oldest joining date for a software developer was 2007-03-07 and the recent joining date for a software developer was2012-04-11. The recent joining date of software developersmay be the reason behind less tweets by the softwaredevelopers on Twitter, however there may be cases wheresoftware developers exist on Twitter for a long time but havestopped tweeting shortly after joining Twitter or they maynot be the active users of Twitter.

In the following, we will look deeply into couple of ques-tions that will be discussed based on our crawled Twitter dataset. These questions are designed specifically to investigateif software developers are discussing or promoting softwareprojects on Twitter. This will provide us the basis to argue

100

101

102

103

100 101 102

twe

ets

developers

Mean = 1181.70

Median = 748.5

Figure 2. Distribution of the number of tweets sent per software developerof Apache projects.

the needs of interlinking software project related content,which is available on the social media channels to the in-formation about software projects contained inside softwarerepositories in order to address different use case scenarios,some of which are already highlighted in Section I.

Q1: Are technology oriented hashtags most commonlyused by software developers?

The usage of hashtags are common among the usersof Twitter. A hashtag is a character string preceded bya “#” sign. It is used to categorize the tweets. Hashtagsare commonly used by users to specify category, topic orintended audience to a tweet. The usage of hashtags tomark tweets typically allow others to follow conversationsregarding a particular topic. As mentioned by Evandro et al.in [33], it is used to not only add context and metadatato tweets but also for promotion and publicity purposes.Therefore, we are interested to investigate what kind ofhashtags are mostly used by software developers and if theyuse project-related hashtags in particular, which will makeit easier to integrate tweets relevant to a particular softwareproject.

Based on our initial investigation, we found that softwaredevelopers do use project-related hashtags to create commu-nities of people (i.e., developers, contributors or users of asoftware project), who are interested in a particular softwareproject. For example, “#hadoop” is a hashtag commonlyused to associate tweets related to the Apache Hadoopproject. Therefore, any one interested to know about whatis under discussion on Twitter regarding Apache Hadoopproject can search for tweets with a “#hadoop” hashtag. Thisshows that software project-related information do exist onsocial media channels and should be interlinked in order toprovide a homogeneous overview on a particular softwareproject.

We are typically interested to know about the mostcommonly used hashtags by software developers in ourcrawled Twitter data set. Therefore, we first retrieved a listof distinct hashtags. Then for each hashtag, we queriedthe number of distinct software developers who have usedthat particular hashtag in his/her tweets. We ignored thefrequency of appearance/usage of a hashtag in this case andpresent only the top 15 most commonly used hashtags inTable III. The ranking of hashtags are done based on thenumber of software developers who have used that particularhashtag.

Hashtags Developersapache 147java 139opensource 106maven 95oracle 86hadoop 80fb 78apachecon 74git 73android 71devoxx 69in 68cloud 67eclipse 62scala 61

Table IIIMOST COMMONLY USED HASHTAGS BY SOFTWARE DEVELOPERS.

The result shows that tweeting about work related stuff iscommon among software developers. The usage of project-related hashtags is quite obvious because we have crawleddata set of a specific community of users (i.e., Apachedevelopers) from Twitter. The top 3 hashtags listed inTable III are obvious because of the following 3 reasons: (1)the projects under consideration are Apache projects; (2) theprimary programming language used by the Apache projectsunder consideration is Java; and (3) Apache projects underconsideration are open source. Moreover, most hashtags thatare commonly used by software developers are technology-oriented (cf. Table III). Based on the results, we concludethat software developers do use hashtags to share workrelated information with others on Twitter. Therefore, we arerequired to integrate software project-related information,which are disseminated on social media channels to therespective software projects in order to enable a compre-hensive overview on the software project from differentperspectives (i.e., social and development perspectives).

Q2: Do software developers tweet most frequently aboutsoftware projects?

We now investigate if software developers are tweetingspecifically and most frequently about software projects theyare developing. In other words, we are interested to knowif software developers promotes or communicates aboutthe software projects they are developing on Twitter using

project-related hashtags. As the main purpose behind thispaper is to identify the needs of integrating social mediachannels (Twitter) to the open source software repositories,therefore, it is necessary to know how frequent softwaredevelopers are communicating about software projects onsocial media channels. We believe that if software developersare discussing software projects frequently on social mediachannels then it will create communities of users aroundthose software projects on the social media channels. Withthe establishment of such communities, we will have accessto a large pool of information about those software projectsfrom the social media channel.

Our approach towards computing most frequentlyhashtags are discussed in the following. Let’s assume thatwe have a list of hashtags used by software developers,Hashtag = {h1, h2, h3, · · · , hn}. For each Apacheproject under consideration, we first retrieved a listof software developers, Dev = {d1, d2, d3, · · · , dn}.Then for each software developer, we extracted alloccurrences of hashtags in his/her tweets, Hashtagd ={h1, h4, h1, h3, h7, · · · , hn} , Hashtagd ✓ Hashtag.Often, a software developer use same hashtagmultiple times hence we summed up all occurrenceof similar hashtags used by a software developer,Hashtagfreq = {h1(2), h4(1), h3(1), h7(1), · · · , hn(n)}where Hashtagfreq ✓ Hashtagd ✓ Hashtag. Later, wesummed up all occurrences of same hashtags used by allsoftware developers of an Apache project and ranked theresults based on the most frequently used hashtags. Wehave selected only the top 2 most used hashtags for eachApache project under consideration and presents the resultsin Table IV.

X

d2Dev

Hashtagfreq

Apache Projects Project HashtagUsed

1st MostUsed

2nd MostUsed

Apache Camel X #camel #apacheApache Directory 7 #apachecon #apacheApache Felix X #osgi #camelApache Hadoop X #hadoop #opendataApache Logging X #rtw2012 #yamApache Lucene X #lucene #bbuzzApache Maven X #maven #apacheApache Mina X #fb #nettyApache MyFaces X #myfaces #primefacesApache OfBiz X #rtw2012 #ofbiz

Table IVMOST FREQUENTLY USED HASHTAGS BY SOFTWARE DEVELOPERS OF

APACHE PROJECTS.

In the results (cf. Table IV), we found that hashtag ofApache OfBiz project ranked 2nd but hashtags of ApacheFelix, Apache Logging and Apache Mina projects rankedlower among the most frequently used hashtags by software

developers of these projects respectively. We found anotherinteresting fact in the results of Apache Felix project whereApache Camel project’s hashtag appeared to be the 2nd

most frequently used hashtag by the software developersof Apache Felix project. We investigated further to find thereasons behind it and discovered that 5 out of 17 softwaredevelopers of Apache Felix project (cf. Table I) are alsodevelopers of Apache Camel project. This means that those 5software developers are contributors to both Apache projects.Hence, we believe that those 5 software developers tweetedand communicated more about Apache Camel project thanApache Felix project on Twitter.

Based on the results shown in Table IV, we see thatproject-related hashtags are most frequently used by soft-ware developers of only 5 Apache projects out of 10Apache projects considered for this study. Based on thisobservation, we can say that not all software developerstweet most frequently about software projects. However, thisdoesn’t mean that software developers haven’t used project-related hashtags for other Apache projects considered forthis study. Therefore, we investigated if software developershave ever used project-related hashtag on Twitter and foundthat all Apache projects under consideration were mentionedby software developers on Twitter using their respectivehashtags except the case of Apache Directory project. Wefound no usage of project-related hashtag by the softwaredevelopers of Apache Directory project. In order to find outthe reason behind it, we manually checked the general usageof Apache Directory hashtag (#ApacheDS) by the users onTwitter and found only 6 tweets (between January 2009 andJanuary 2013), which contained that particular hashtag onTwitter. Further investigation revealed that none of the userswho mentioned that specific hashtag was a core developer(i.e., the one who has source control commit rights) ofApache Directory project.

IV. DISCUSSION

In the previous section, we investigated the usage ofTwitter as a platform by software developers to promoteor communicate about the software projects they are devel-oping. Based on the questions investigated, we discoveredthat software developers do communicate about softwareprojects on Twitter using project-related hashtags. However,the frequency of using project-related hashtags vary amongApache projects under consideration. We believe that thesefindings can be made more concrete if applied to largenumber of Twitter data set relevant to open source softwareprojects.

These findings are sufficient enough to argue the impor-tance of integrating Twitter and software projects based onproject-related hashtags or software developer IDs. However,we haven’t yet looked into how frequent software developerstweet about software projects. In particular, it will be inter-esting to know if software developers tweet about software

projects on daily, weekly or monthly basis.In order to calculate the average number of days

taken by a software developer to tweet about a soft-ware project (he/she is developing), we extracted alltweets mentioning a particular project-related hashtagsuch that, tweets = {tweet1, tweet2, · · · , tweetn}. Foreach tweet, we extracted the timestamp value, date ={date1, date2, date3, · · · , daten} where date1 is the times-tamp value of tweet1, date2 is the timestamp value oftweet2 and so on. Later, we computed the difference be-tween two tweets containing project-related hashtag in termsof days, such that day = {day1, day2, day3, · · · , dayn}where day1 = date2 � date1, day2 = date3 � date2 andso on. Finally, we computed the average number of daystaken by software developers to post project-related tweetof Apache projects.

X day

|day|

Apache Projects Tweets containingHashtags

Average Time(#days)

Apache Camel 475 2.68Apache Directory 0 0Apache Felix 22 62.27Apache Hadoop 1,578 0.89Apache Logging 20 44.30Apache Lucene 376 4.38Apache Maven 943 1.75Apache Mina 2 50Apache MyFaces 24 31.29Apache OfBiz 62 18.85

Table VAVERAGE NUMBER OF DAYS TAKEN BY SOFTWARE DEVELOPERS TO

POST A SOFTWARE PROJECT RELATED TWEET.

For each Apache project, we present the total number oftweets in which project-related hashtag was mentioned andthe average number of days taken by a software developer topost about a software project using project-related hashtagin Table V. The time duration of each Apache project varies,which is due to the variation in the usage of project-relatedhashtags. We have ignored keyword-based mentioning ofsoftware projects (e.g., mentioning of “hadoop” in the twittermessage text) while computing the average time, which mayhave reduced the average time of project-related hashtagsusage in our study. Further, the results shows that softwaredevelopers of only four Apache projects tweeted aboutthe project in less than a week time using project-relatedhashtag. For other Apache projects, the average time takenby a software developer to post about a software project isquite high.

The results presented in Table V is based on a Twitterdata set that contain tweets of only software developersof these Apache projects. We haven’t taken into accountthe usage of project-related hashtags by all users (i.e.,

users, bug reporters, patch contributors etc.) on Twitter,which we believe will provide more insights into the usageand promotion of a software project using project-relatedhashtags and may also contradict with the average timepresented in Table V.

A. On The Need of Integrating Social Media Channels andOpen Source Software Repositories

Our findings based on the questions reveal that thereis relevant information about software projects exists onsocial media channels (i.e., Twitter). This information canbe integrated (e.g., by interlinking project-related tweet-s/posts/hashtags, developer IDs [3], project IDs etc.) inorder to get an integrated view on the software project.Once it is integrated, several use case scenarios can beaddressed by exploiting the information. For example, wehave shown elsewhere [9] that one could investigate thesocial behavior of software developers on a project mailinglist and compare it with their social behavior on Twitter.This introduces a new dimension to the analysis of socialdynamics of software developers by also taking into accountsocial media channels. This would enable researchers tomeasure and compare the hierarchy and centralization ofsoftware developers in different communication channels(e.g., mailing lists, Twitter etc.) in contrast to previous stud-ies where researchers have been using only mailing lists, bugrepositories or discussion forums [34], [35]. Furthermore,integrating social messages/posts to project-related artifactswill open up new research challenges allowing to analyzethe impact of end-users response on the success/failure ofan open source project.

V. CONCLUSION

In this paper, we have identified social media channelsas a platform, which is used by the open source softwarecommunity to disseminate project-related information to awider audience. Further, we highlighted the need to integrateproject repositories and social media channels in order toget an integrated view on the software project. In order tosupport our proposition, we studied the usage of project-related hashtags by software developers on Twitter and foundthat software developers do use project-related hashtags tocommunicate about the software projects they are workingon. Given that the ratio of usage of software project re-lated hashtags vary for each Apache software project weconsidered in the study (cf. Table V) but this laid down thebasis to integrate software projects with Twitter based onproject-related hashtags. We further investigated if project-related hashtags are frequently used by software developersof Apache projects. We observed that not all softwaredevelopers tweet most frequently about software projectsbut we found that software developers of Apache projectshave used project-related hashtags in their tweets except thecase of Apache Directory project. As our study consists of

limited number of software projects, it remains to be seenwhether a large number of software developers demonstratesimilar kind of behavior. Therefore, we can not say that allor majority of software developers frequently tweets aboutsoftware projects using project-related hashtags.

While more research is needed to better understand theseand other related questions, our study shows the existence ofsoftware project related information on Twitter. Therefore,this information can be integrated with software repositorieshosting that particular software project in order to supportcertain use case scenarios that are mentioned in the begin-ning of this section.

ACKNOWLEDGMENT

This publication has emanated from research conductedwith the financial support of Science Foundation Ireland(SFI) under Grant Number SFI/12/RC/2289.

REFERENCES

[1] A. Iqbal and M. Hausenblas, “Integrating developer-relatedinformation across open source repositories,” in IEEE 13thInternational Conference on Information Reuse and Integra-tion (IRI), 2012, 2012.

[2] M. Conklin, “Project entity matching across floss reposi-tories,” in 3rd International Conference on Open SourceSystems, Springer. Limerick, IE: Springer, 2007, pp. 45–57.

[3] A. Iqbal and M. Hausenblas, “Interlinking developer identitieswithin and across open source projects: The linked dataapproach,” ISRN Software Engineering, 2013.

[4] S. Asur and B. A. Huberman, “Predicting the future withsocial media,” in Proceedings of the 2010 IEEE/WIC/ACMInternational Conference on Web Intelligence and IntelligentAgent Technology - Volume 01, ser. WI-IAT ’10. Washington,DC, USA: IEEE Computer Society, 2010, pp. 492–499.[Online]. Available: http://dx.doi.org/10.1109/WI-IAT.2010.63

[5] L. de Vries, S. Gensler, and P. S. Leeflang, “Popularityof brand posts on brand fan pages: An investigationof the effects of social media marketing,” Journal ofInteractive Marketing, vol. 26, no. 2, pp. 83 – 91, 2012.[Online]. Available: http://www.sciencedirect.com/science/article/pii/S1094996812000060

[6] B. Pang and L. Lee, “Opinion mining and sentiment analysis,”Found. Trends Inf. Retr., vol. 2, no. 1-2, pp. 1–135, Jan. 2008.[Online]. Available: http://dx.doi.org/10.1561/1500000011

[7] A. Agarwal, B. Xie, I. Vovsha, O. Rambow, andR. Passonneau, “Sentiment analysis of twitter data,” inProceedings of the Workshop on Languages in Social Media,ser. LSM ’11. Stroudsburg, PA, USA: Association for Com-putational Linguistics, 2011, pp. 30–38. [Online]. Available:http://dl.acm.org/citation.cfm?id=2021109.2021114

[8] M. Hu and B. Liu, “Mining and summarizing customerreviews,” in Proceedings of the tenth ACM SIGKDDInternational Conference on Knowledge Discovery andData Mining, ser. KDD ’04. New York, NY, USA:ACM, 2004, pp. 168–177. [Online]. Available: http://doi.acm.org/10.1145/1014052.1014073

[9] A. Iqbal, M. Karnstedt, and M. Hausenblas, “AnalyzingSocial Behavior of Software Developers Across DifferentCommunication Channels,” in 25th International Conferenceon Software Engineering and Knowledge Engineering (SEKE2013), Boston, USA, 2013.

[10] L. Fang and K. LeFevre, “Privacy wizards for socialnetworking sites,” in Proceedings of the 19th internationalconference on World wide web, ser. WWW ’10. New York,NY, USA: ACM, 2010, pp. 351–360. [Online]. Available:http://doi.acm.org/10.1145/1772690.1772727

[11] D. Rosenblum, “What anyone can know: The privacy risksof social networking sites,” Security Privacy, IEEE, vol. 5,no. 3, pp. 40–49, 2007.

[12] M. Trusov, A. V. Bodapati, and R. E. Bucklin, “Determininginfluential users in internet social networks,” Journal ofMarketing Research, vol. XLVII, pp. 643–658, 2010.

[13] N. Agarwal, H. Liu, L. Tang, and P. S. Yu, “Identifyingthe influential bloggers in a community,” in Proceedingsof the 2008 International Conference on Web Searchand Data Mining, ser. WSDM ’08. New York, NY,USA: ACM, 2008, pp. 207–218. [Online]. Available:http://doi.acm.org/10.1145/1341531.1341559

[14] J. Weng, E.-P. Lim, J. Jiang, and Q. He, “Twitterrank:finding topic-sensitive influential twitterers,” in Proceedingsof the third ACM international conference on Web searchand data mining, ser. WSDM ’10. New York, NY,USA: ACM, 2010, pp. 261–270. [Online]. Available:http://doi.acm.org/10.1145/1718487.1718520

[15] M. Karnstedt, T. Hennessy, J. Chan, and C. Hayes, “Churn insocial networks: A discussion boards case study,” in SocialComputing (SocialCom), 2010 IEEE Second InternationalConference on, 2010, pp. 233–240.

[16] H. Becker, M. Naaman, and L. Gravano, “Beyond trendingtopics: Real-world event identification on twitter,” in FifthInternational AAAI Conference on Weblogs and Social Media,2011.

[17] K. Lee, D. Palsetia, R. Narayanan, M. M. A. Patwary,A. Agrawal, and A. Choudhary, “Twitter trending topicclassification,” in Proceedings of the 2011 IEEE 11thInternational Conference on Data Mining Workshops, ser.ICDMW ’11. Washington, DC, USA: IEEE ComputerSociety, 2011, pp. 251–258. [Online]. Available: http://dx.doi.org/10.1109/ICDMW.2011.171

[18] X. Wang, I. Kuzmickaja, K.-J. Stol, P. Abrahamsson, andB. Fitzgerald, “Microblogging in open source software de-velopment: The case of drupal using twitter,” IEEE Software,vol. 99, no. PrePrints, p. 1, 2013.

[19] G. Bougie, J. Starke, M.-A. Storey, and D. M.German, “Towards understanding twitter use in softwareengineering: Preliminary findings, ongoing challenges andfuture questions,” in Proceedings of the 2Nd InternationalWorkshop on Web 2.0 for Software Engineering, ser. Web2SE’11. New York, NY, USA: ACM, 2011, pp. 31–36. [Online].Available: http://doi.acm.org/10.1145/1984701.1984707

[20] Y. Tian, P. Achananuparp, I. Lubis, D. Lo, and E.-P.Lim, “What does software engineering community microblogabout?” in 9th IEEE Working Conference on Mining SoftwareRepositories (MSR), 2012, pp. 247–250.

[21] P. Prasetyo, D. Lo, P. Achananuparp, Y. Tian, and E.-P. Lim,“Automatic classification of software related microblogs,” in28th IEEE International Conference on Software Mainte-nance (ICSM), 2012, pp. 596–599.

[22] L. Singer, F. F. Filho, and M.-A. Storey, “SoftwareEngineering at the Speed of Light: How Develop-ers Stay Current Using Twitter,” http://leif.me/papers/Software-Developers-Twitter-DCS-350-IR.pdf, University ofVictoria, Technical Report, 2013.

[23] http://camel.apache.org/.

[24] http://directory.apache.org/.

[25] http://felix.apache.org/.

[26] http://hadoop.apache.org/.

[27] http://logging.apache.org/.

[28] http://lucene.apache.org/.

[29] http://maven.apache.org/.

[30] http://mina.apache.org/.

[31] http://myfaces.apache.org/.

[32] http://ofbiz.apache.org/.

[33] E. Cunha, G. Magno, G. Comarela, V. Almeida,M. A. Goncalves, and F. Benevenuto, “Analyzing thedynamic evolution of hashtags on twitter: a language-based approach,” in Proceedings of the Workshopon Languages in Social Media, ser. LSM ’11.Stroudsburg, PA, USA: Association for ComputationalLinguistics, 2011, pp. 58–65. [Online]. Available:http://dl.acm.org/citation.cfm?id=2021109.2021117

[34] A. Wiggins, J. Howison, and K. Crowston, “Social dynamicsof floss team communication across channels,” in FourthInternational Conference on Open Source Software, 2008.

[35] K. Crowston, , K. Crowston, and J. Howison, “Hierarchyand centralization in free and open source software teamcommunications,” Knowledge Technology Policy, vol. 18, pp.65–85, 2005.