Content Quality Assessment in Social Networks

Digital libraries, knowledge databases and information repositories are moving towards data grids, web 2.0 and the semantic web, i.e. publicly available pieces of information and knowledge assets are structured, evaluated and tagged in such a way that the navigation and processing in the information space becomes computer-supportable. The process of searching, navigating and assessing quality of an information artifact is no longer an isolated activity. The quality of an information artifact is the result of a set of social dynamics like endorsement, citation and ranking. To understand information quality in a social context, human judgement needs to be captured and accessed. Many models for the aggregation and propagation of quality and trust have been proposed both in the research field of multi-agent systems (MAS) and recommender systems. The project aims at selecting and/or developing a generalized model for the propagation and aggregation of quality and trust that both meets human intuition and is mathematically sound. Most importantly, existing social networks such as business networks (e.g. linkedin, xing, viaduc), communities for sharing and recommendation (e.g. dooyoo, youtube, clipfish), etc. are to be used as testbeds for the evaluation of the chosen model/ approach.

The objective of the project is to explore the notion of trust and reputation in social networks and use it to assess the quality of information artefacts. The expected outcome of the project is a set of prototype software modules that implement one or more trust models that can be applied for quality assessment of different kinds of objects. This research is targeted at organizations that build or internally host collaborative social networking portals for knowledge exchange and engineering. These could be anything ranging from Wikis, blog spaces, bookmarking/tagging engines, or a combination of all of them.

Activities in this proposed project are divided into the following:

1. Identification of potential applications The proposed model of trust based quality assessment of information artefacts can be applied in different settings. We plan to narrow down to a set of no more than 3 potential application areas like data grids for academic materials, event posting engines, etc.
2. Needs analysis For each of the identified applications, specific potential end-users will be identified for needs analysis. The outcome of this activity is a model of the social network in this application and the semantics associated with trust and object quality.
3. R&D of the trust model In this phase, the actual trust-based quality assessment model is developed. Ideally, the model would be generic enough with specific additions required for each application area.
4. Prototype development This phase involves the development of the prototype software modules and associated activities like revisions, unit testing, etc.
5. Evaluation This involves evaluation of the software modules at one or more end-user sites and an evaluation feedback from the users

social networks
trust
reputation
recommender systems
trust propagation
web of trust

Profile 1: Experts in Collaboration Models
Profile 2: Experts in Social Networks, Social Network Mining, or Networks
Profile 3: Experts in Trust Models
Profile 4: User Partners with existing Communities and/or Content Collection
Profile 5: Experts in Evaluation/ Usability

Profile 1-3: Contribute to Proposal, Define your own Work Packages (collaboratively)
Profile 4: Contribute to Project Requirements Definition, Run Prototypes on your existing Community and/ or Content Collection
Profile 5: Network Function - Track Module Specifications and Requirements for Usability, Evaluate and Improve Usability - both for User Partners and End Users

Profile 1-3: University, Research Institution/ Unit
Profile 4: Commercial, Governmental or Non-Governmental Organization that is willing to test new approaches for the collaborative evaluation, ranking and recommendation of knowledge assets.
Profile 5: University, Research Institution/ Unit or Specialized Service Provider

Yes