Research And Application Of Personalized Recommendation In Publishing Industry

Publishing industry because of its single information judgment,analysis of insufficient data and other issues lead to inventory backlog,facing a lot of difficulties.Therefore,how to make the publishing industry real-time,accurate,personalized access to knowledge publishing guidance data,hot words,in order to solve the publishing house lack of publishing guidance information,inventory backlog and other issues is the big data era to solve the publishing house dilemma of an important research topic.Its research has important application value and theoretical significance.According to the publishing guidance demand of publishing industry,this paper extracts hot words and text keywords from massive data for publishing house through the SW-TextRank keyword extraction algorithm,and then provides real-time,accurate and personalized recommended hot words and keywords for different publishing houses according to the proposed MV-CFIDNN personalized recommendation algorithm.In order to provide publishers with a strong information basis for publishing guidance.Aiming at the problems such as publishing orientation and inaccurate knowledge-oriented analysis leading to inventory backlog and digital publishing transformation demand,this paper puts forward a SW-TextRank keyword extraction algorithm,which extracts hot words and keywords for publishing industry.Based on the shortcomings and defects of the traditional text information feature(keyword)extraction algorithm,the proposed comprehensive weight TextRank algorithm(SW-TextRank)is based on TextRank algorithm,which has the defects and inadequacies of the weight division of information attributes.The weight value of each information word is calculated by using the key index of three text information attribute combined with the empowerment method,and then the comprehensive weight is taken as the original weight of the TextRank algorithm to extract the information feature(keyword)algorithm.And the extracted hot words,keywords through the MV-CFIDNN personalized recommendation algorithm for different publishing houses to provide accurate,real-time,personalized recommendations,in order to improve the publishing house publishing guidance accuracy and so on.According to the traditional recommendation algorithm,the precision of information recommendation is not high,the real time is poor,the personalization is not prominent and so on,and the personalized recommendation algorithm of multi-angle deep neural Network(MV-CFIDNN)is proposed.MV-CFIDNN recommended algorithm model can realize the training and learning of eigenvalue,without artificial feature labeling,at the same time,it can better realize the recommendation of high precision,strong real-time and personalized text information.Combining with the needs of publishing industry,this paper constructs a personalized recommendation system of publishing industry service,which is based on mass data processing,and recommends information hot words and keywords for publishing house through algorithm and model,so as to prejudge publishing orientation and digital transformation for publishing house,reduce inventory,personalized publishing and other services.