Research On Employment Recommendation System Based On Text Matching

Nowadays,the online recruitment method has the advantages of wide coverage and high efficiency,which is welcomed by more job seekers and enterprises.While the development of online recruitment has also brought problems such as information overload.How can job seekers choose the most suitable candidates from among the numerous positions,and how can enterprises find the most suitable employees for their company from a large number of resumes,has become the key issue.The existing work mainly focuses on the modeling of personnel position information matching.However,some of the resume information and position information are structured text,and some are unstructured text.Therefore,this article will match the two parts separately.In this paper,structured text is divided into numerical text,domain knowledge text,and name text for similarity calculation.At this stage,when matching unstructured text,there are problems of not combining text semantics and low efficiency.Therefore,this paper proposes a BERT employment recommendation model based on twin networks.This model comprehensively considers information such as work experience,project experience,and self-evaluation in resumes,as well as job responsibilities and job requirements in recruitment requirements,and constructs a sentence representation that represents the information in resumes.Using the BERT model to extract sentence level features from papers,avoiding the problem of missing semantic features,combining twin networks and the BERT model to extract text features from resume information and recruitment requirements,respectively.After that,calculate the semantic matching degree between resume information and recruitment requirements,sort based on the semantic matching degree,and generate a position recommendation list.Multiple sets of job recommendation experiments were conducted on the dataset of the Zhaopin Recruitment and Job Matching Competition and the dataset of BOSS direct recruitment.The experiment shows that the BERT employment recommendation model based on twin networks proposed in this paper has better effects.This paper designs a job recommendation system based on text matching.This system can provide job seekers with employment information that meets their own characteristics,improving employment rate and employment matching.At the same time,it can also provide data support for enterprise personnel selection,improve enterprise recruitment efficiency,and achieve more accurate personnel position matching.