Research On Information Hybrid Recommendation Algorithm Based On Attention Degree And Time Series

With the advent of the mobile Internet era,content creators such as media have joined the ranks of information,entertainment,and other content production teams,and contents can be created through the mobile terminal anytime anywhere,a wealth of information.have been generated.Not only that,a lot of data have been generated by people's comments,sharing,clicks,etc.Nowadays,searching for information through search engines has been more difficult than originally,and information needs to be obtained through more efficient methods.The personalized recommendation system was born on the basis of such a demand.It can combine the user's interest preferences,timely recommend information that they are interested in to the user.the user's time will be saved,and the efficiency of the entire society will be improved,at the same time,it also can be used to increase the amount of reading,and thus gain greater benefits by the information distribution platform and content creators.Through the understanding of the current research status at home and abroad,the key technical principles of the traditional recommendation algorithm are in-depth learned and introduced,and their advantages and disadvantages are analysed,and then the recommendation algorithm evaluation indicators are made a brief introduction.The main purpose of this paper is how to improve the recommendation effect of the modern mobile-end information field:This paper has some thoughts and improvements on how to establish a more accurate user modeling and information modeling,quantified the user's information preferences and historical behavior Records,as well as the textual characteristics of information,and they are represented in vector form.Then,according to some characteristics of modern information distribution platform,the traditional collaborative filtering recommendation algorithm is improved based on the attention degree.Combining traditional long-term interest preferences and short-term interest preferences in the process of browsing information,the algorithm was improved separately and a personalized hybrid recommendation algorithm was designed.Finally,the experiment is designed,according to the idea of the paper,using the public data set to experiment,and the parameters defined in this paper are determined.The proposed hybrid recommendation algorithm was compared with other single recommendation algorithms according to the evaluation criteria,and achieved better results.The recommended effect is to verify the superiority of the proposed hybrid recommendation algorithm.