Towards (Semi-) Automatic Moderation of Social Web Annotations

With the rapid growth of social web, users´ ability to publish content (e.g., annotating the multimedia resources in Youtube or Facebook) has created active electronic communities that provide a wealth of information. Subsequently mining and analyzing information content generated by the users are important research topics. Many existing researches have focussed on extraction, transformation, and summarization of information from annotated resources with the objective to classify the content of the annotations (e.g., how many percent of users agree/disagree on a specific topic). However, evaluating the content quality of the annotations is an important issue, which is not being taken into consideration by many of these researches. Unusualness in the sense of remarkable, vulgar, contrastive, unreliable or low-quality content of the social web annotations is not sufficiently analyzed in an automatic way. Therefore the task of (semi-)automatic, semantic moderation of social web annotations becomes increasingly important to identify unusualness among annotations. One of the current technologies, that provides a new framework for modeling and reasoning on social web relations and semantics in machine-processable structures, is the semantic web technology. This paper analyzes different dimensions of social web annotation and defines moderation requirements. Based on this analysis and the semantic web technology, a methodology for (semi-)automatic social web annotation moderation is proposed, which is an efficient extension and mash up of the previously mentioned approaches.