An Adaptive Framework for Discovery andMining of User Profiles from Social Web-based Interest Communities
2013 (English)In: The Influence of Technology on Social Network Analysis and Mining / [ed] Özyer, T.; Rokne, J.; Wagner, G.; Reuser, A., Wien: Springer, 2013, Vol. 23, 497-519 p.Chapter in book (Refereed)
Abstract Within this paper we introduce an adaptive framework for semi- tofully-automatic discovery, acquisition and mining of topic style interest proﬁlesfrom openly accessible social web communities. To do such, we build an adaptivetaxonomy search tree from target domain (domain towards which we are gatheringand processing proﬁles for), starting with generic concepts at root moving down tospeciﬁc-level instances at leaves, then we utilize one of proposed Quest schemesto read the concept labels from the tree and crawl the source social networkrepositories for proﬁles containing matching and related topics. Using machinelearning techniques, cached proﬁles are then mined in two consecutive steps,utilizing a clusterer and a classiﬁer in order to assign and predict correct proﬁlesto their corresponding clustered corpus, which are retrieved later on by an ontology-based recommender to suggest and recommend the community members with theitems of their similar interest. Focusing on increasingly important digital culturalheritage context, using a set of proﬁles acquired from an openly accessible socialnetwork, we test the accuracy and adaptivity of framework. We will show that a tradeoff between schemes proposed can lead to adaptive discovery of highly relevant proﬁles.
Place, publisher, year, edition, pages
Wien: Springer, 2013. Vol. 23, 497-519 p.
, Lecture Notes in Social Networks, 23
interest profile, profile mining, semantic recommender, social web mining, community extraction, social network
IdentifiersURN: urn:nbn:se:kth:diva-118487DOI: 10.1007/978-3-7091-1346-2__22ISBN: 978-3-7091-1345-5OAI: oai:DiVA.org:kth-118487DiVA: diva2:606491
QC 201302192013-02-192013-02-192014-01-24Bibliographically approved