Research On Job Recommender Algorithm Based On Deep Learning

With the advent of Internet era,more and more candidates hunted their jobs from the Internet,bringing an unprecedented increase of human resources information,which led to the problem of information overload in human resources services.And recommender system quickly became the main tool to solve the information overload problem with the ability to locate and push the content of interest to the user from the overloaded information.Also,in recent years,deep learning has achieved great success in a series of areas such as computer vision,natural language processing and semantic identification.However,there is few published work on deep learning for recommender system.Most researchers at home and abroad focus on the application of traditional algorithms,they still mainly use algorithms like collaborative filtering and content-based filtering,or a combination of them.Few researchs have tried to use novel agorithms on recommender system.Therefore,this thesis tried to study and improve the existing recommender algorithm based on deep learning and apply it to the field of job recommendation,hoping to update the algorithms currently using on recommender system.The main work of this thesis is as follows:(1)This thesis collected candidates and job information from the human resources business system,and performed pre-processing operations on collected data,such as data cleaning and data transforming.Then This thesis performed word segmentation and vectorization operations on the text fields of job items,which obtained a human resources data warehouse for recommender algorithm.(2)This thesis proposed hybrid deep collaborative filtering(HDCF)algorithm based on collaborative deep learning(CDL)algorithm.HDCF used the text attributes and structured attributes of items to construct a hybrid feature as input,which improved the recommendation performance of the algorithm.Also,HDCF mixed the content-based filtering algorithm leading to solving the item cold start problem.(3)Based on the main workflow of recommender system,this thesis designed a structure of job recommendation system,and finally realized a job recommender prototype system based on deep learning,which can better overcome the cold start problem,and provide real-time job recommendation.The HDCF algorithm proposed by this thesis overcame the shortcomings of traditional collaborative filtering algorithms when dealing with sparse data and cold-start items,with the help of the feature extraction ability of deep learning.Experimental results showed HDCF obtained better recommendation performance than traditional recommender algorithms such as PMF and CBF.