Design Of News Recommendation Model Based On Mixed Strategy

With the explosive growth of network information and the development of recommendation technology,news recommendation system is gradually appearing in people's field of vision and playing an increasingly important role.News recommendation system allows users to easily get the news they are interested in,and also improve the click rate of news and the stickiness of users for news websites.The news recommendation algorithm and model in the news recommendation system are very important to the recommendation result.The following work is done for news recommendation model in this paper:(1)A news recommendation model based on LDA is proposed and implemented.The news headlines and news content were analyzed by LDA modeling,then we get the user interest model with two layers,the core interest of the user in news headlines based on the LDA model to represent and a wide interest in the user's news content based on the LDA model,these two model's weights are determined by the experiment,then the final construction of user comprehensive interest model based on integrated LDA is built.(2)A news recommendation model based on named entity set is propose and implement.Feature modeling named entity set of news headlines and news content based on user interest model,generating two layers,to represent the core interests of users of news headlines based on named entity model,and another user interest model named entity based on the content of news to represent the wide intersets,then these two model's weights is established by means of experiment,the final construction of user comprehensive interest model based on named entity set.(3)A news recommendation model based on mixed strategy is implemented.The user interest model and comprehensive LDA comprehensive user interest model named entity set based on the results of the calculation of mixed pipelining and parallel hybrid hybrid strategy,and two kinds of mixed strategy recommendation results and operation,add hot news content,then we get the final recommended list,consider a variety of to enhance the recommendation effect factors.(4)A news recommendation system based on mixed strategy is designed.The package system function requirement analysis,architecture design,database design and model design of the four core aspects of the design,through the analysis of the user record reading news,based on user interest model integrated LDA based on user interest model and comprehensive collection of named entity a preliminary recommendation list,and then use mixed strategy selection and recommendation ranking list,get the final recommendation list.