BIROn - Birkbeck Institutional Research Online

    Detecting implicit cross-communities to which an active user belongs

    Taha, K. and Yoo, Paul and Eddinari, F.Z. (2022) Detecting implicit cross-communities to which an active user belongs. PLoS One , ISSN 1932-6203.

    47567a.pdf - Published Version of Record
    Available under License Creative Commons Attribution.

    Download (8MB) | Preview


    Most realistic social communities are multi-profiled cross-communities constructed from users sharing commonalities that include adaptive social profile ingredients (i.e., natural adaptation to certain social traits). The most important types of such cross- communities are the densest holonic ones, because they exhibit many interesting properties. For example, such a cross-community can represent a portion of users, who share all the following traits: ethnicity, religion, neighbourhood, and age-range. The denser a multi-profiled cross-community is, the more granular and holonic it is and the greater the number of its members, whose interests are exhibited in the common interests of the entire cross-community. Moreover, the denser a cross-community is, the more specific and distinguishable its interests are (e.g., more distinguishable from other cross-communities). Unfortunately, methods that advocate the detection of granular multi-profiled cross-communities have been under-researched. Most current methods detect multi-profiled communities without consideration to their granularities. To overcome this, we introduce in this paper a novel methodology for detecting the smallest and most granular multi-profiled cross-community, to which an active user belongs. The methodology is implemented in a system called ID_CC. To improve the accuracy of detecting such cross-communities, we first uncover missing links in social networks. It is imperative for uncovering such missing links because they may contain valuable information (social characteristics commonalities, cross-memberships, etc.). We evaluated ID_CC by comparing it experimentally with eight methods. The results of the experiments revealed marked improvement.


    Item Type: Article
    School: Birkbeck Faculties and Schools > Faculty of Science > School of Computing and Mathematical Sciences
    Depositing User: Paul Yoo
    Date Deposited: 16 May 2022 11:33
    Last Modified: 09 Aug 2023 12:53


    Activity Overview
    6 month trend
    6 month trend

    Additional statistics are available via IRStats2.

    Archive Staff Only (login required)

    Edit/View Item Edit/View Item