基于学术文献引文的自动摘要关键技术研究

项目名称： 基于学术文献引文的自动摘要关键技术研究

项目编号： No.61303125

项目类型： 青年科学基金项目

立项/批准年度： 2014

项目学科： 自动化技术、计算机技术

项目作者： 蔡晓妍

作者单位： 西北工业大学

项目金额： 27万元

中文摘要： 学术文献中的引文对于帮助学者了解某篇文献的学术价值及对后续研究的影响具有重要的价值。然而随着学术文献数量的日益庞大，使得学者们快速获取引文信息变得越来越困难。基于此，本项目针对学术文献中的引文进行自动摘要关键技术研究，主要包括：（1）分析学术文献中显示引文和隐式引文的特征，提出基于SVM的引文识别算法，提高引文识别率；（2）探索与引文倾向性相关的各种特征，提出基于随机森林的倾向性分类方法；（3）研究基于学术文献引文的自动摘要方法，提出将引文句聚类和排序过程相互作用的排序策略，用以提升摘要生成质量；（4）研究基于语义的文摘评测方法，引入LDA主题模型衡量机器摘要与人工摘要的语义相似度。本项目的研究将为学术文献引文检索系统的实际应用提供理论和技术上的支持，具有重要的科学意义和研究价值。

中文关键词： 自动摘要；引文识别；引文倾向性分析；；

英文摘要： Citation sentence in scientific paper has an important value to help researchers understanding academic value and the influence of other research on a paper. With the number of the scientific papers is increasing rapidly,which makes it more and more difficult for researchers to quickly get citation information. This project aims at researching on citation-based summarization of scientific papers, including(1) Analyzing explict citation features and implicit citation features of scientific papers,proposing a citation detection method based on SVM in order to improve citation detection rate. (2) Exploring features of citation polarity, and proposing a citation sentiment detection approach based on random forest. (3) Researching citation-based scientific papers summarization methods,and proposing a novel ranking mechanism which can mutually and simultaneously update clustering and ranking process, to improve the quality of the generated summary.(4) Researching evaluation methods of summary based on semantic similarity, and comparing semantic similarity between system generated summary and human generated summary by introducing a LDA model.The purpose of the project is to provide theoretical and technical support for academic literature citation retrieval system, it also has important scientific significance and res

英文关键词： Automatic Summarization；Citation Detection；Citation Polarity Analysis；；