Context-Enhanced Personalized Social Summarization

Proceedings of COLING 2012: Technical Papers, pages 1223–1238, COLING 2012, Mumbai, December 2012. Context­Enhanced Personalized Social Summarization Po Hu1,2, Donghong Ji1, Chong Teng1 and Yujing Guo1 (1) Computer School, Wuhan University, China (2) Computer School, Central China Normal University, China phu@mail.ccnu.edu.cn, donghong_ji2000@yahoo.com.cn, tchong616@126.com, yujingguo.ximo@gmail.com ABSTRACT This work investigates an interesting and challenging task in summarization, i.e., personalized social summarization, which aims to adapt summarization result of a specified document to an intended user based on his interests inferred from social context implicitly. Most existing summarization systems generate a uniform version of summary for different users no matter who is reading or generate personalized summaries employing only the local information in the document and the user profile. This paper proposes a novel unsupervised approach by making use of enhanced social context to aid personalized summary generation. In the proposed method, document expansion, user expansion, and implicit induction of the intended user's interest aspects are achieved simultaneously by adopting a fuzzy tripartite clustering algorithm. And both the informativeness of sentences and the user’s interest aspects are incorporated in a unified ranking process. Preliminary experimental results on a social tagging dataset validate the effectiveness of the proposed approach. KEYWORDS: Personalized social summarization, social context, fuzzy tripartite clustering 1223 1 Introduction With the dramatic growth of the Internet, people are overwhelmed by a large number of accessible documents. In recent years, document summarization has become one of the most important research topics, which aims to address such dilemma by automatically capturing the essential content from document(s) and presenting it to a human reader in a succinct and friendly form. However, most existing summarization methods generate the same summary for different users, regardless of the interests of the readers for whom they are intended. These “one size fits all” methods may perform well in general but may not meet the needs of individuals. Now with the rapid growth of social networking services like Delicious 1, CiteULike 2, and Flickr3, users are no longer passive consumers of web contents. They can create contents and add metadata. Similarly, web documents no longer exist on their own and they are naturally associated with other documents and diverse users. All these information can be considered as the potential data source for document understanding and personalization. For generating a personalized summary, traditional methods usually require that a user explicitly provides his interest aspects, such as specifying the categories he prefers (Díaz and Gervás, 2007) or clicking a subset of sentences in a document according to his interests (Yan et al., 2011). However, most users are reluctant to provide such information, thus it is more meaningful to infer a user’s interests implicitly. To address these concerns, we present an unsupervised approach for personalized summarization. The underlying assumption is that it is beneficial to understand both a single document and a single user better if appropriate social context can be leveraged under some constraints. In this work, the expanded social context used to infer users' interests and enrich document’s content is highly …

Content truncated to 500 words