• No results found

CHAPTER 9 CONCLUSION

9.4 CONTRIBUTIONS

The work presented in this thesis resulted in a number of original contributions. In this section we will outline the significance of the achievements with respect to the related research areas.

9.4.1 Contribution to User Modelling and User Adaptive Systems

There is a growing interest in providing adaptive support for teams, groups and communities. Along this line, personalisation and adaptation can play a crucial role, as illustrated by recent user-modelling approaches (Cheng and Vassileva, 2005; Song et al., 2005). A number of approaches, such as visualizations, notifications, and community ratings, have been exploited to facilitate community/group awareness, motivate participation, and improve community knowledge sharing. However, existing adaptation techniques focus mainly on supporting individual members, rather than supporting the community to function as a whole (Bretzke and Vassileva, 2003; Farzan et al., 2009; Sankaranarayanan and Vassileva, 2009). We proposed a novel approach for community-tailored support which aimed at facilitating processes related to the effectiveness and sustainability of VCs and is based on a community model derived from analysis of log data.

To the best of our knowledge, there are no such holistic community modelling approaches. More specifically we have proposed (a) a novel framework for holistic adaptive support in virtual communities, (b) a mechanism for extracting and maintaining a semantic community model based on the processes identified, and (c) deployment of the community model to identify static patterns and provide holistic support to a virtual community.

The community and relationship model in (Bretzke and Vassileva, 2003) is the closest to ours but there is a crucial difference. Users’ interests are modelled in (Bretzke and Vassileva, 2003) based on how frequently and how recently users have searched for a specific area from the ACM taxonomy, and user relationships are derived based on any successful download or service that took place between two users. In contrast, our approach employs the metadata of the resources shared in the community along with the ontology and derives a semantically relevant list of interests for every user. Furthermore, the CM extracted in our case is semantically richer and theoretically underpinned. Recently research on modelling communities employed graph theory to model relationships between members (Kay et al., 2006) or members’ interactions in general (Falkowski and Spiliopoulou, 2007). The key contribution of our approach to community modelling is the considering of semantic relationships, i.e. an edge connecting two members represents their semantic similarity to each other, and the relevance of this link to the community’s domain.

9.4.2 Contribution to Computer Supported Cooperative Work

CSCW research has exploited different approaches to facilitate group work and knowledge sharing, such as visualisations, notifications and awareness techniques (Ackerman and McDonald, 1996; Zacklad, 2003; Gouvea et al., 2006; Wang et al., 2007). Most of these approaches have been applied to particular settings where positive aspects have been observed. In this line, we are contributing to the CSCW community with a novel approach for providing semantically enriched community awareness. Semantically enriched algorithms have been developed that inform the generation of personalised notifications to VC members.

Visualisation techniques are another approach for providing awareness of what is happening in a community, and thus, supporting participation and collaboration in a VC. For example, graphical representations are used to make people aware of the relevance to the activity or to the position of a particular member in the group (Kay et al., 2006) or to show the status (or popularity) of a resource (Wang et al., 2007). The key limitation of visualisation techniques is their passive influence on the functioning of the community, e.g. while examining graphical representations members may not be able to see how their contribution could be beneficial for the community. In contrast, our approach provides notification messages that explicitly making aware people of how they relate to others in the community.

9.4.3 Contribution to Research in Social Networks

Analysis of community evolution refers to different approaches for detecting changes over time in large or small people networks represented as graphs. Existing approaches are examining mainly structural changes of social networks (e.g. density, degree distribution, average distance, clustering coefficient) by comparing the characteristics of graph instances at given time points (Leskovec et al., 2005; Falkowski et al., 2006; Asur et al., 2007; Falkowski and Spiliopoulou, 2007; Lin et al., 2007; Palla et al., 2007; Lin et al., 2008). In the context of identifying patterns of changes (evolution) in the community we are contributing to the social networks area with a semantically enriched approach for modeling change patterns in a closely-knit VC.

To the best of our knowledge, there is no other approach which examines community evolution with regard to TM, SMM, CCen. In contrast with existing work that monitors how the network/graph under investigation is evolving over time in order to get an insight of the community (Song et al., 2005; Kumar et al., 2006; Falkowski and Spiliopoulou, 2007), in this work the purpose of detecting changes is to exploit the extracted information in order to provide intelligent support to the community as a whole. Along the same line, a principle difference from the existing work is that we aim to detect change patterns connected to specific processes related to effective functioning and sustainability of a VC and not just to model a VC. In contrast with the existing methods, which consider simple indicators for a relationship (e.g. direct connection), (Leskovec et al., 2005; Falkowski et al., 2006; Falkowski and Spiliopoulou, 2007), we exploit semantic techniques (such as resource meta-data and ontological reasoning) to derive possible relationships between members.