Content Facets for Individual Information Needs in Media

The amount of content published in traditional media is huge and steadily growing. Additionally, social media gained momentum since people use blogs to share comments and opinions to current events. However, blog content is questionable in respect to quality since blogs are neither reviewed nor edited. From the media consumer perspective, navigating the haystack of information produced by media as well as finding content that meets ones quality demands is challenging. This challenge is the key motivation for this thesis: to support media consumers to filter media content by content facets capturing topical information needs as well as content quality aspects.

For this, two types of content facets are suggested: (i) topic oriented and (ii) topic independent quality related content facets. First, for each facet type (i) and (ii), concrete content facets are proposed. These content facets are then formally defined. Second, a feature study revealed that stylometric features are better suited to assess topic independent content facets, while for topic oriented content facets Bag-of-Words features serve best. Third, the best features have successfully been used to classify traditional and social media content in both types of content facets. To address the problem of lacking training data in classification, this thesis investi- gated whether available classification schemes from traditional media can be mapped onto blogs. The experiments revealed that traditional media correlate content wise with selected blogs; therefore, content facets from traditional media can be applied to blogs. Several proposed content facets have successfully been implemented in APA Labs, a Web-based framework for faceted search in traditional and social me- dia. Consequently, APA Labs supports media consumers to navigate and analyze traditional and social media content. This enables any media consumer to search for information according to their personal information need. This is a substantial improvement to individualize search in media.