Research On Technologies Of Personalized Recommendation In Online Recruitment System

With the rapid development of Internet technology,the scale of online recruitment has rapidly expanded.It has become an important way to recruit and seek employment,but also led to the problem of information overload.This paper presents a new personalized recommendation method,which is to recommend job seekers online recruitment information,thus solving the problem of information overload.This paper analyzed the data characteristics of online recruitment information and resume information of job seekers,and finds that they all contain important short text information.Short texts of 100-200 words,such as "job description" in resume information and "work experience" in resume information,contain a great deal of important information.In this paper,we propose a new keyword extraction method to deal with the large number of important short text data in recruitment and resume information.In this paper,some key techniques such as keyword extraction,word embeddings,graph models,clustering and deep learning are combined,and we presents a new method of keyword extraction based on graph and clustering.Its performance on short text datasets is superior to the current keyword extraction methods.We apply the method of deep learning to the recruitment information recommendation field and get a deep netruel network trainned to automatically score the match between resume information and recruitment information.We use Dropout and stochastic gradient descent to make the training process better and faster.We designed experiments to verify the viability and accuracy of our approach using the real job search and resume information data captured on the Internet.This paper combines the above-mentioned techniques and methods to design and implement a personalized online recruiting information recommendation system for jobseekers.We used the web crawler to obtain real recruitment and resume information data.,designed experiments to verify the feasibility and accuracy of this system and the abovementioned techniques and methods.