Summarizing threads in blogs using opinion polarity

The huge amount of data available on the Web needs to be organized in order to be accessible to users in real time. This paper presents a method for summarizing subjective texts based on the strength of the opinion expressed in them. We used a corpus of blog posts and their corresponding comments (blog threads) in English, structured around five topics and we divided them according to their polarity and subsequently summarized. Despite the difficulties of real Web data, the results obtained are encouraging; an average of 79% of the summaries is considered to be comprehensible. Our work allows the user to obtain a summary of the most relevant opinions contained in the blog. This allows them to save time and be able to look for information easily, allowing more effective searches on the Web.

The research of Elena Lloret has been supported by the FPI grant (BES-2007-16268) from the Spanish Ministry of Science and Innovation, under the project TEXT-MESS (TIN2006-15265-C06-01). The research has been also partially founded by the European Commission under FP6project QALL-ME number IST-033860.