| The acquisition of scientific and technological personage information is an important part of the construction of science and technology personage graph.How to extract the unstructured resume information of scientific and technological figures accurately and effectively is the key problem in the process of building the graph of scientific and technological figures.At present,information extraction method based on deep learning sequence annotation is the mainstream method to analyze unstructured text content.However,there are still some problems in the existing methods to deal with the resume text of scientific and technological figures,such as insufficient information expression and confusion of similar information classification.In view of the above problems,this paper puts forward the following solutions:Firstly,the traditional word based or word based deep learning method can not effectively express the research field information with more professional terms and complex composition structure in the resume of scientific and technological figures.This paper proposes a deep learning sequence tagging method based on character-word collaboration,which can effectively improve the extraction effect of research domain information.Secondly,it is necessary to distinguish the learning and working experiences of the characters in the extraction of the resumes of scientific and technological figures.However,the time and unit linguistic characteristics contained in the learning and working experiences of the characters are similar,and the existing sequence tagging methods can not accurately distinguish the types of their experiences.This paper proposes a method of classification and correction based on sequence tagging recognition using GBDT,which effectively improves the accuracy of the method Classification accuracy of time and unit information.Finally,this paper integrates the above two methods,designs and implements a resume information extraction system based on deep learning sequence annotation and GBDT classification correction.The system can effectively identify the resume information of scientific and technological figures,and realize the update and query of the atlas of scientific and technological figures. |
