| The Internet publishes a large number of news stories every day,resulting in users facing information overload while browsing.Therefore,accurately extracting news of interest to users from thousands of news items has become a pressing problem.However,current news recommendation algorithms have some problems,including the problems of insufficient analysis of time effects and changes in user preferences and the limitations of similar user groups.In this case,it is of practical importance to build a personalized news recommendation system,which can recommend news of interest to users based on their preference characteristics.This thesis aims to solve the above problems and explore how to achieve more accurate news recommendations by improving the existing recommendation algorithms.Therefore,this thesis develops the following research.To address the problem of news timeliness,this research considers the changing relationship between the user’s current browsing time and the click time of the last browsing record and the user’s preference,introduces a time factor to improve the content-based recommendation algorithm and extracts the user’s own preference model;at the same time,considering the influence of news hotness on the user’s preference,uses the popular penalty factor to improve the traditional cosine similarity calculation formula,and proposes a new At the same time,considering the influence of news popularity on users’ preferences,we improve the traditional cosine similarity calculation formula by using the popular penalty factor,and propose a new hybrid similarity calculation method,so as to find the most adjacent users and better extract users’ potential preference models.The experiment proves that the user preference-based recommendation algorithm based on this model outperforms the traditional algorithm in several indexes and can provide users with news contents that match their preferences.Based on the proposed recommendation algorithm based on user preferences,a personalized news recommendation system is designed and implemented,which is divided into four modules: front-end display module,data collection module,data support module and recommendation module,the front-end display module is the part used by users to recommend news for them;the data collection module is divided into crawlers to collect news information and collect user behavior;the data support module The data collection module is divided into crawlers to collect news information and user behavior;the data support module is used for information interaction;the recommendation module recommends data for processing,generates news lists,and sends them to the front-end display to users.By testing the function and performance of the news system,the correctness and practicality of the system are verified.At the same time,this thesis also designs and implements a visual display of the news big data system based on Data V,which greatly improves the interactivity,real-time and flexibility of news data and makes the news big data analysis results intuitive and visible,so as to better realize real-time browsing and monitoring. |
PDF Full Text Request