| The goal of automatic event summarization is to extract important and relevant information from online news and present an event to the user in a condensed form.Most of the conventional sentence extraction summarization methods either extract important sentences from unstructured documents, or draw relevant sentences according to some specific knowledge. The former methods extract information by maximizing the features of documents, which are domain-independent. The latter methods extract information by maximizing the relevance between documents and some domain knowledge. However, some domain-independent methods may extract important but irrelevant sentences, while some domain-specific methods may extract some relevant but unimportant sentences.This paper proposes a novel sentence extraction method for Chinese breaking news summarization. It assigns a feature score to each sentence by analyzing features of the sentence, such as position, length etc. Then it utilizes fuzzy inference mechanism to infer semantic relevance between each sentence and the event topic. It extracts both important and domain-related sentences to generate a news summary. Based on this, it processes the sentences in all news summaries by using clustering, ranking techniques and finally generates an event summary. Experiment is conducted on our corpus of Chinese breaking news. The results indicate that our approach can effectively summarize Chinese breaking news. In addition, this paper successfully applies the proposed summarization methods to user-interest analysis topic. |
PDF Full Text Request