Recognizing Skill Networks and Their Specif i cCommunication and Connection PracticesSergiu Chelaru, Eelco Herder, Kaweh Djafari Naini, Patrick SiehndelL3S Research CenterLeibniz University Hannover, Germany{chelaru, herder, naini, siehndel}@L3S.deABSTRACTSocial networks are a popular medium for building andmaintaining a professional network. Many studies existon general communication and connection practices withinthese networks. However, studies on expertise search sug-gest the existence of subgroups centered around a particularprofession. In this paper, we analyze commonalities anddif f erences between these groups, based on a set of 94,155public user prof i les. The results conf i rm that such subgroupscan be recognized. Further, the average number of connec-tions dif f ers between groups, as a result of dif f erences in in-tention for using social media. Similarly, within the groups,specif i c topics and resources are discussed and shared, andthere are interesting dif f erences in the tone and wording thegroup members use. These insights are relevant for inter-preting results from social media analyses and can be usedfor identifying group-specif i c resources and communicationpractices that new members may want to know about.Categories and Subject DescriptorsH.5.4 [Hypertext/Hypermedia]: Navigation; H.3.5 [OnlineInformation Services]: Web-based servicesKeywordsskills, expertise, social networks, connections, topics, senti-ment, content1. INTRODUCTIONPeople who work in similar professions typically share par-ticular skills. Further, if people are asked to indicate theirskills, it is expected that the skills they mention vary ingranularity. For example, someone working in public rela-tions may indicate skills in social networking and marketing,but also specif i c skills such as DTP software, writing pressreleases and time management.It is also known that people from dif f erent professions orcultural backgrounds have dif f erent practices in how theyPermission to make digital or hard copies of all or part of this work for personal orclassroom use is granted without fee provided that copies are not made or distributedfor prof i t or commercial advantage and that copies bear this notice and the full cita-tion on the f i rst page. Copyrights for components of this work owned by others thanACM must be honored. Abstracting with credit is permitted. To copy otherwise, or re-publish, to post on servers or to redistribute to lists, requires prior specif i c permissionand/or a fee. Request permissions from permissions@acm.org.HT’14, September 1–4, 2014, Santiago, Chile.Copyright 2014 ACM 978-1-4503-2954-5/14/09 ...$15.00.http://dx.doi.org/10.1145/2631775.2631801 .communicate with one another, the communication mech-anisms that they choose and the topics that they discuss[12]. These dif f erences can also be observed on a more pri-vate, personal level: programmers are usually more informalthan bankers, people working in public relations are typi-cally more active in social media than investors, and pastorswill most likely talk about dif f erent topics than real-estateagents.In this paper, we investigate dif f erences in communitieswithin self-reported skill networks. We are particularly in-terested in discovering dif f erences in their communicationpractices: how well is a professional community connected,how often do people post updates via Twitter or Facebook,what are the topics that they talk about, and what is theoverall tone or sentiment of these communications? Partic-ularly for people who aim to identify and approach expertsfrom a dif f erent profession, who wish to promote their ser-vices in other communities, or who consider a career switch,it is important to know the unwritten rules in a network.For example, what would programmers think of overly pos-itive marketing language? How often can one repeat an an-nouncement? Would it be a good idea to add a personaltouch or will that be considered ‘unprofessional’?Being aware of dif f erences between professional communi-ties is also important for interpreting statistical data fromsocial network analysis. For instance, in some communitiesthe average number of followers is considerably higher thanin other communities. As a consequence, a person from awell-connected community like online marketing with, say,300 followers, may be considered isolated; for a programmer,this is actually a very good number. The same dif f erencesapply for interpreting centrality and other in- and out-degreemeasures.The main contributions of our paper are: we provide anoverview on how skills in professional networks are relatedand categorize these skills into professions. Further, we showto what extent dif f erent professions dif f er from one anotherin terms of connections, topics, sentiment and shared con-tent. Finally, we discuss implications for social network anal-ysis and the design of professional networking sites.The remainder of this paper is structured as follows. Inthe next section we discuss related work, followed by a de-scription of the dataset we used. In Section 4 we discuss thestructure of the skill network derived from LinkedIn prof i lesand how this structure is ref l ected in the professions that weextracted using LDA. The results are presented in four sub-sections, covering: connections between people, topics thatpeople discuss about, subjectivity and polarity of the word-13
