1st International Workshop on Collective Semantics: Collective Intelligence & the Semantic Web (CISWeb), Tenerife 2008
The Web 2.0 has introduced new style of information sharing platforms favoring mass participation of users and resulting overall interestingness over the individual quality of information content and information organization. Dynamic knowledge emerges as the outcome of the interactions of masses of users in social networks (over 40 million in facebook). Thereby, the heterogeneity of data sources (e.g. multimedia, over 1 billion photos in flickr; over 1 million streams/day from YouTube), the scale of information (25% of network traffic is estimated to be YouTube related) and the huge amount of knowledge (100 millions of postings in flickr), pose many difficulties in discovering relevant information and in arriving at a larger picture of the available content.
This gestalt of the problem opens up intriguing ways for the Semantic Web to contribute, because the Semantic Web may:
- provide a language basis for integrating multiple platforms;
- support structuring help from distributed, ad-hoc ontologies;
- target specific topics and subgroups by introducing new ways of exploring the information space.
Since purely manual approaches for information and knowledge (re-)organization are doomed to fail because of the sheer size of the problem, combining even a little semantics from manually and semi-automatically built semantic web resources and semantic web languages with the little structure of information sharing platforms, such as tags, social acquaintances, or favorites, may carry a long way. Introducing semantics can largely impact the way and the effectiveness in which IR can be employed due to the fact that tasks such as semantic mining facilitates searching. On the other hand, analysis of multimedia content (text, images, video) in the social context of these platforms may add an important term that is currently missing from the equation (e.g. with tags being only weakly associated to content itself).
Despite the recent progress in content-based automatic extraction of semantic metadata from multimedia, such techniques are far from being perfect and generic applicable. This can be overcome by annotating the resources taking also into account the social context in such a way that best fits the users’ points of view. In this way, handling of multimedia data becomes also a tag-oriented procedure and the extraction of their context (i.e. semantics) turns into a problem of analyzing the corresponding tags in combination with signal-based content analysis techniques.
This workshop targets this integration arising from the mining of Web2.0 information, multimedia content and knowledge with help of the Semantic Web.