Today's headline personalized recommendation mechanism is mainly:
Topic similarity recommendation of similar articles: recommend by obtaining similar articles that users have read.
News based on the same city: For users with the same geographical information, they will recommend popular articles that match the city.
Recommendation based on article keywords: for each article, extract keywords as the characteristics to describe the content of the article. Then use the article keywords of the user's action history to make matching recommendations.
General recommendation based on popular articles in the station: according to the reading habits of users in the station, find out popular articles and recommend them to all users who have not read them.
Reading habit recommendation based on social friends: according to users' friends outside the station, get comments or published articles forwarded by friends outside the station for recommendation.
Recommend keywords based on users' long-term interests: recommend keywords by comparing users' short-term and long-term reading interests.
List recommendation based on similar users' reading habits: calculate the similarity of users' behavior in a certain period and make cross recommendation of reading content.
Content recommendation based on site distribution sources: Through the source distribution of articles read by users, calculate 20 news sources that users like to recommend.
Please accept it. Thank you.