Review Content Recommendation Based On Phrase Topic Model And Multi-document Automatic Summarization

Rapid growth of Internet information resources poses a great challenge to research and application of information retrieval. Information retrieval, especially how to deal with the results of information search, receives more and more attention of scholars. How to help uses faster and more efficient access information is a topic worthy of study in depth. Academic literature is a kind of more standard and richer resource, compared to others such as news articles, blog, web pages. How to get information faster and more efficient from amount of academic literature resources is of great value and significance for scholars.In this paper, the Academic system provides a convenient platform of resource utilization for users, with academic resources search, search results’visual analysis and human-machine cooperation review writing. The academic resources search uses open source tool, Lucene to support with building indexes for downloaded resources from Internet academic database. Visual analysis aggregates the metadata of academic resources search list, providing a view of how academic document changes over time, geographic, hotspot and other factors.In human-machine cooperation review writing, uses are provided review reference catalog and content. The reference catalog utilizes carrot2 which is a cluster of search results. The reference content recommendation is the key part. We use phrase topic model to implement topic-focused reference content service. In the paper, SmoothPhraseLDA based PhraseLDA cooperate with SumBasic which is an algorithm for multi-document automatic summarization provide users with reference service.