Network literature new book recommendation method

Abstract

The invention discloses a network literature new book recommendation method. The method comprises: obtaining basic information of a network book in a site, and processing according to the basic information to obtain a theme feature distribution vector of the network book; calculating the new book relevancy of each network book according to the topic feature distribution vector, and searching a newbook associated with each network book according to the new book relevancy to form a new book candidate set I; reordering the new book candidate set I to form a new book candidate set II, obtaining the display quantity and reading quantity scores of new books through a background log, and recommending books with strong correlation between historical data and user behavior information and networkbooks with weak correlation as positive / negative samples through the background; reordering all new books in the new book candidate set I according to the scoring data and the positive / negative samples; and generating a new book recommendation candidate list of the user.

Description

Technical field [0001] The present invention relates to the field of personalized recommendation, in particular to a method for recommending new online literature books. Background technique [0002] With the rise of online literature, a large number of new online literature books continue to emerge. As a reading platform, it is very important for users to recommend books that meet the user's reading habits among the massive books. The current recommendation system mostly uses the reading identity among a large number of users, that is, users who have read the same type of book will have similar reading habits, but for new books, only a very small number of users have read the book, and there is a lack of Enough user reading and book content related information, except for well-known authors, most books will be obliterated in a large amount of data. Due to insufficient presentation, it will be slow and inefficient to recommend new books to users [0003] Different from traditiona...

AI Technical Summary

Problems solved by technology

The current recommendation system mostly uses the identity of reading among a large number of users, that is, users who have read the same type of books will have similar reading habits, but for new books, only a very small number of users have read the book, lack of Enough user reading and book content related information, except for well-known authors, most books will be lost in a large amount of data

Due to insufficient display, it will be slow and inefficient to recommend new books to users

Images