Universität zu Köln

User modeling for exploratory search on the Social Web. Exploiting social bookmarking systems for user model extraction, evaluation and integration

Gontek, Mirko (2011) User modeling for exploratory search on the Social Web. Exploiting social bookmarking systems for user model extraction, evaluation and integration. PhD thesis, Universität zu Köln.

[img]
Preview
PDF
Download (5Mb) | Preview

    Abstract

    Exploratory search is an information seeking strategy that extends be- yond the query-and-response paradigm of traditional Information Retrieval models. Users browse through information to discover novel content and to learn more about the newly discovered things. Social bookmarking systems integrate well with exploratory search, because they allow one to search, browse, and filter social bookmarks. Our contribution is an exploratory tag search engine that merges social bookmarking with exploratory search. For this purpose, we have applied collaborative filtering to recommend tags to users. User models are an im- portant prerequisite for recommender systems. We have produced a method to algorithmically extract user models from folksonomies, and an evaluation method to measure the viability of these user models for exploratory search. According to our evaluation web-scale user modeling, which integrates user models from various services across the Social Web, can improve exploratory search. Within this thesis we also provide a method for user model integra- tion. Our exploratory tag search engine implements the findings of our user model extraction, evaluation, and integration methods. It facilitates ex- ploratory search on social bookmarks from Delicious and Connotea and pub- lishes extracted user models as Linked Data.

    Item Type: Thesis (PhD thesis)
    Creators:
    CreatorsEmail
    Gontek, Mirko
    URN: urn:nbn:de:hbz:38-46163
    Subjects: Data processing Computer science
    Uncontrolled Keywords:
    KeywordsLanguage
    exploratory search, information retrieval, social bookmarking, folksonomies, user modeling, recommender systems, linked data, ontologiesEnglish
    Faculty: Philosophische Fakultät
    Divisions: Philosophische Fakultät > Historisches Seminar
    Language: English
    Date: 10 June 2011
    Date Type: Publication
    Date of oral exam: 16 November 2011
    Full Text Status: Public
    Date Deposited: 21 May 2012 14:24:24
    Referee
    NameAcademic Title
    Thaller, ManfredProf. Dr.
    URI: http://kups.ub.uni-koeln.de/id/eprint/4616

    Actions (login required)

    View Item