Research On Position Recommendation Based On Attention And Convolutional Networks

At present,as the recruitment industry gradually adopts the form of online recruitment,the relevant data information in the field of online recruitment has exploded,so the field of online recruitment has a trouble of information overload as well.Lots of traditional job recommendation systems often use collaborative filtering algorithms to extract information only from job seekers’ interaction history,and use this to implement job recommendation.With the rapid increase of data,problems such as cold start and sparse data appeared,resulting in unsatisfactory recommendation effects.Therefore,it is particularly important to apply personalized recommendation algorithms and deep learning technology to the job recommendation system,and find jobs that match the interests of job seekers from massive job data and recommend them to job seekers.In view of the above problems,this thesis will combine the behavioral operation of job se ekers with the auxiliary information of job seekers and positions,use deep learning technology and attention mechanism to design a personalized job recommendation algorithm,build a userjd data set,and evaluate the performance of the algorithm model raised in this text under the data.At the same time,this thesis also use the proposed algorithm to design a position recommendation system.The specific work is as follows:(1)Since the deep collaborative filtering recommendation framework(Deep CF)integrates the collaborative filtering method based on representation learning and the collaborative filtering method based on matching function learning,the framework can not only extract the complex nonlinear relationship between users and items,but also capture the user the low-order relationship with the item,so the Deep CF framework is used in the field of job recommendation in this paper.(2)Aiming at the problem that the Deep CF framework cannot effectively recommend to new job seekers or newly released positions without implicit feedback data.In the paper,the Deep CF framework is improved,and a deep collaborative person post recommendation algorithm based on auxiliary features and convolutional networks(FCDeep CF)is proposed.Firstly,two parallel convolutional neural networks integrated with attention mechanism are used to extract the text features of job seekers and job,and then the extracted text features and other auxiliary features fusioned with implicit feedback data are used as the input of the Deep CF framework to realize job recommendation for job seekers.Finally,a comparison experiment with other recommendation algorithms is carried out under the userjd data set to verify the effectiveness of this algorithm.(3)When recommending jobs to job seekers who have generated implicit feedback data,the correlation between job seekers’ different interaction history jobs and current generation predicted jobs is different.To solve this problem,on the basis of the Deep CF framework,the attention mechanism is integrated,and an attention-based deep collaborative filtering person post recommendation algorithm(ADeep CF)is proposed.The algorithm takes into account the importance of different historical interactions of job seekers to the current prediction target,and conducts more fine-grained interaction modeling for job seekers and positions.Finally,in order to verify the effectiveness of the algorithm in this paper in improving the recommendation effect,a comparative experiment was carried out with several other algorithms in the userjd dataset.(4)Using the model put forward in this thesis to develop a job recommend dation system.The specific users of this system include job seekers and enterprise users.