Recommendation System for use in a bookstore

Question

Hey, I'm trying to learn some of the recommendation algorithms that are used on sites like Amazon.com. Therefore, I have this simple java storage application (spring hibernate postgres), where the book has the attributes title, category, tags, author. For simplicity, the book has no content. The book should be identified by its name, category, author and tags. For every user registering in the application, I should be able to recommend some books. Each user can view the book, add it to the basket and buy it at any time. Therefore, in the database, I remember how many times each user looked at a book, books in his cart and books that the user bought. At the moment there is no rating option, but which can be added too.

So can someone tell me which algorithms I can use to demonstrate recommendations for each user? I want everything to be very simple. This is not a project for sale, but only to expand my knowledge on recommendation algorithms. So suppose that there are about 30 books in total (5 categories and 6 books each). It would be very helpful if someone could also tell me what the attributes should be, which I should use to calculate the similarity between the two users and how to do this with the recommended algorithms.

Thanks in advance. SerotoninChase.

Answer 1

As a specific concrete example, one of the options is the “nearest neighbors K” algorithm.

, , , , . int timesViewed[10], timesViewed[i] - , i.

, , , . c o -1.0 1.0, -1.0 , " c o" 1.0 , c o ".

c , , . K (, 5, 10, 20) ( ), K , 1.0.

timesViewed. , , averageTimesViewed[0] timesViewed[0] K-, . averageTimesViewed[i].

averageTimesViewed, , , , K c . averageTimesViewed, , .

, , , , , - /.

, , (, ..). , , (.. K= ), , , , K .

Answer 2

(Taste framework) .

- , ,

Answer 3

" ". .

Answer 4

You have a huge amount of freedom here. Make a measure of similarity between the two users, and then make a monotonous function that uses the same user ratings as input and returns points for each book. The standard solution is to use matrix multiplication.