Research On Personnel Resume Intelligent Extraction System Based On Conditional Random Fields

With the development of society,the demand of company for talents is increasing.Since resume was the main way for enterprises to understand the talents,finding a quick and accurate method to identify useful information from a large number of resumes,which are in different formats,can greatly help enterprises improve their recruitment efficiency.Traditional way of dealing with resumes is accomplished by people and the whole progress is high-cost and inefficient.By using information extraction technology,the research in this paper can efficiently extract valuable information from a variety of non-structured resumes.After doing structure storage and building up talent pool,users can easily find what they want and reuse the information.In order to extract useful information from resumes,this paper proposed a conditional random field information extraction model,which is based on the context feature.And also a hierarchical conditional random field model is designed according to the unstructured format and irregular distribution characteristics of the information in resumes to deal with hierarchical resume information.Moreover,the deep learning technology is introduced in the process of the conditional random field model training to achieve intelligent extraction.The main work of this paper is as follows:1.By analyzing the task of information extraction from resumes and learning main information extraction methods,this paper proposed a conditional random field model,which is flexible in feature design and can make full use of the context information.2.In the task of information extraction from resumes,traditional linear conditional random field model could not be good for dealing with resumes with hierarchically structured information.To solve this problem,an improved Hierarchical conditional random field algorithm model is proposed.This model,which is based on the characteristics of the information structure in resumes,divides the complex information extraction task into several simple sub tasks,and the processing result is transmitted to the lower level extraction task as the information feature.3.Deep learning technology is used to study the characteristics of resumes.Conditional random field model is a supervised extraction model,which needs the feature information designed by users.This paper introduced a feature learning method from deep learning technology and the feature extraction task did get promoted by using this method to study the representation of data features.The improved conditional random field model is applied in extraction task to extract information from resumes coming from documents and web pages in a variety of formats and structures.Compared with traditional conditional random field algorithm model and the commonly used information extraction tool ICTALAS,the extraction result in this paper showed that the improved conditional random field model is suitable for the intelligent extraction task of resumes.