Research And Application Of Personalized Employment Recommendation Algorithm For University Chemical Professionals

The chemical industry is a backbone industry in our country’s national economy,and its development and changes have a huge impact on the national economy.In recent years,due to the rapid innovation of our country’s chemical industry and the outbreak of the novel coronavirus,the operating efficiency of the chemical industry has continued to deteriorate.It greatly increased the difficulty of employment for more and more chemical college graduates.If accurate employment recommendations can be provided to graduates,this problem can be effectively alleviated.Personalized employment recommendation algorithm is an important research direction of recommendation algorithm.Aiming at the problems of poor chemical job recommendation effect and inaccurate recommendation of new users in the current personalized employment recommendation algorithm,a personalized employment recommendation algorithm for chemical professionals in colleges and universities is proposed by this paper.For this reason,a personalized employment recommendation algorithm for chemical professionals in colleges and universities is proposed by this paper.First,the text data of chemical industry are collected by Data collection technology based on Scrapy+Selenium and real-time monitoring data collection technology based on Scrapy+Selenium.Next,a chemical pre-training language model method based on knowledge distillation is proposed to convert text data into vectors.Then,a feature fusion algorithm of CNN(Convolutional Neural Network)and BILSTM(Bi-directional Long Short-Term Memory)is proposed to construct a classification model and locate job categories.Finally,the cosine similarity algorithm is used to recommend personalized positions to talents in chemical universities.Experimental results show that the current lack of language models in the chemical industry is overcomed by the proposed chemical pre-training language model method based on knowledge distillation.Compared with traditional knowledge distillation methods,the model learning ability can be better through the knowledge distillation strategy of multi-layer learning.The accuracy rate of the proposed feature fusion algorithm of CNN and BILSTM reaches 95% in chemical job classification,and its performance and performance are better than the current mainstream algorithms.The algorithm makes full use of the feature that CNN can extract local features and the advantage of BILSTM that it has memory to associate the extracted context features to make the semantic information of the text more comprehensive,so as to improve the accuracy of the text classification task.