Electronic text is available like never before. In this talk, Hongyan Jing will present research for helping users with navigating, browsing, and understanding the vast amount of electronic documents.

Automatic text summarization can help users grasp the main content of a document in a short time. Commercial companies and research institutes have developed dozens of automatic summarizers to date. One common problem with these automatic summarizers is that they rely on simple extraction of sentences to produce summaries. Such summaries are often inconcise, incoherent, or even misleading. Jing has developed an automatic summarizer that aims to overcome these shortcomings. Rather than simply extracting sentences, this system can do intelligent "editing" to the extracted sentences so that they are more concise and coherent. She used techniques from a number of fields, including information retrieval and natural language processing (both statistical and symbolic techniques).

Jing will also briefly present our work on Information Retrieval, in which morphology and semantics are meaningfully integrated to address the sense ambiguity problem in retrieval. Jing used local context information as well as global corpus information for disambiguating words, particularly for retrieval purpose. The system effectively improves performance over the traditional vector-space model.