科技查新语义角色标注及其在报告自动生成系统中的应用

Semantic Role Labeling of Science and Technology Novelty Retrieval and Its Application in Retrieval Report Automatic Generation System

科技查新工作是高校图书馆的一项重要服务,为实现查新自动化,需要适当的语义化规则。文章采用改良的语义角色标注法,将描述研究的文本内容视为由主动因素、被动因素、研究对象、变化四类角色组成,并以一种包含文本中的语义角色、关键词词组与同义词的五层树结构表达文本的语义。语义化后的查新点可与未处理的文献直接比较,以获知两者是否相关与何处相关,并生成可直接用于查新报告结论部分的文本。这一方法被用于查新报告生成程序从而辅助并减少人工判断。

Science and technology novelty retrieval is an important service in academic libraries.In order to realize the automation of novelty retrieval,we need proper semantic method.In this paper,an improved semantic role labeling is used to describe the research content as four types of roles:active factor,passive factor,research object and change.A five-level tree structure including semantic role,keyword phrase and synonym is used to express the semantics of the text.The semantics of novelty points can be directly compared with the unprocessed documents to find out whether they are related and where they are related.The compared result will be used to generate texts that can be directly used in the conclusion part of the novelty retrieval report.This method is used in the novelty retrieval report generator to assist and reduce manual judgment.