earticle

영어

Microblog has been a dominant social-network platform. Many social events spread fast on microblogging platforms, because of the high interactivity and large amounts of users of microblogging platforms. However, users can only search for interested events on microblogging platforms by keywords, yielding high redundancy of search results as well as the lack of complete description on events (owing to the limitation on the length of microblogs). Aiming to solve these problems, we propose a new approach to automatically generate event abstracts for microblog events. We summarize an event via two aspects, namely an event-description set and a user-opinion set. For extracting the event description for an event, we first partition microblogs into a set of short sentences. These sentences are further ranked by computing the relevance to the event based on a graph model and the most relevant sentences are selected as the set of event description. For extracting user opinions on an event, we propose a supervised learning model that is based on the set of long sentences extracted from microblogs. We conduct experiments on a data set consisting of 14 events crawled from the Sina Weibo. The experiment results suggest the effectiveness of the proposed approach.