Study On Retrieval Results Re-ranking And Literature Authority Computing

Literature search is always an important service in scientific research. It would be significant to build a literature search system with good performance that researchers can effectively get the theses. We lay out novel methods which can be used in literature search. They are study on re-ranking of retrieval system based on the opposite theory and computing of the thesis authority based on HITS.Re-ranking based on opposite theory, commonly called opposite re-ranking, is one of the important ways to improve the accuracy and the real-time performance of the search system. Opposite re-ranking focuses on using the known relevant and irrelevant results to re-rank the order of the search results. Because of the defect of the search engine, it is very difficult to solve the hard issue of shielding re-ranking process from obstinate-negative results. Countering this problem, we bring two re-ranking methods.(1) Opposite re-ranking using hierarchical clustering.(2) Opposite re-ranking based on relevant sub-topic dispelling. Through the experiments of re-ranking based on the corpus, we can see the progress of our method compare with the baseline system.Quantization of the literature authority can be used in many fields, such as literature search, sci-tech novelty retrieval and analysis of the development of scientific research and so on. The existing methods are all controversial, just like Impact Factor and H number. There isn’t a unifying and effective way to judge the literature authority. The paper studies on a new method to compute the literature authority based on HITS. We also designed the experiments by which we can analyze the difference between Impact Factor and our method.