Research On Methods Of Employment Recommendation For College Students Based On Deep Learning

The problems faced by graduates in employment are personal inapplicability of information and information overload.The recommendation system is of practical significance when applied to the employment of graduates.It can recommend a personalized set of enterprises for graduates to reduce their pressure and improve the employment rate and satisfaction.At present,the existing research on employment recommendation for college graduates still has shortcomings in facing the challenge of sparsity and feature extraction.At the same time,it fails to make use of the relevance between students and enterprises effectively.In view of the above problems,the main research contents of this paper are as follows:1.This thesis describes the research background of recommendation algorithm and job recommendation,and introduces in detail the research status of related technologies,as well as the research progress and challenges of job recommendation.2.This thesis proposes an employment recommendation model that incorporates behavioral characteristics.In view of the sparsity of interactive data,the behavior attribute and the description attribute are defined to better mine the characteristics.Then it extracts basic features and sequence features.Finally,each feature domain is subjected to feature interaction and deep fitting of the data to obtain the matching degree between the student and the enterprise,and then make a recommendation.Based on the real data set,the proposed model can better mine the characteristics,and has a good hit rate.3.This thesis proposes an employment recommendation model based on multi-task learning fusion relation graph.To solve the problem of how to use the relationship data effectively,firstly,the construction of the relationship diagram between students and enterprises is carried out.Then two tasks are set,namely graph embedding task and recommendation task,and the interaction module of the two tasks is designed at the same time.Finally,the two tasks are trained jointly,and the recommendation task is used to get the matching degree between students and enterprises.Experimental results on a real dataset show that the proposed model can effectively interact the two tasks.At the same time,the relationship information is mined effectively so as to achieve the purpose of improving the performance of recommendation task.This thesis proposes two employment recommendation models.The former focuses on data sparseness,focusing on the mining of student and enterprise characteristics and the high-level interaction of features.The latter conducts multi-task training for information mining of relational data,so that the graph representation learning can effectively enrich the semantics of the model.Finally,summarize the research work of this thesis,and look forward to the future research direction.