外文：
皇后大学Introduction to Recommender System
Approaches of Collaborative Filtering: Nearest Neighborhood and Matrix Factorization
supply是什么意思“We are leaving the age of information and entering the age of recommendation.”
Like many machine learning techniques, a recommender system makes prediction bad on urs’ historical behaviors. Specifically, it’s to predict ur preference for a t of items bad on past experience. To build a recommender system, the most two popular approaches are Content-bad and Collaborative Filtering.
没问题的英文Content-bad approach requires a good amount of information of items’ own features, rather than using urs’ interactions and feedbacks. For example, it can be movie attributes such as genre, year, director, actor etc., or textual content of articles that can extracted by applying Natural Language Processing. Collaborative Filtering, on the other hand, doesn’t need anything el except urs’ historical preference on a t of items. Becau it’s bad
on historical data, the core assumption here is that the urs who have agreed in the past tend to also agree in the future. In terms of ur preference, it usually expresd by two categories. Explicit Rating, is a rate given by a ur to an item on a sliding scale, like 5 stars for Titanic. This is the most direct feedback from urs to show how much they like an item. 人教版小学三年级英语上册Implicit Rating, suggests urs preference indirectly, such as page views, clicks, purcha records, whether or not listen to a music track, and so on. In this article, I will take a clo look at collaborative filtering that is a traditional and powerful tool for recommender systems.
sodasNearest Neighborhood
The standard method of Collaborative Filtering is known as Nearest Neighborhood algorithm. There are ur-bad CF and item-bad CF. Let’s first look at Ur-bad CF. We have an n × m matrix of ratings, with ur u挚友的反义词ᵢchina blue, i = 1, ...n and item pⱼ北京小学通州分校, j=1, …老友记第一季第一集m. Now we want to predict the rating r在一起英文ᵢⱼ if target ur i did not watch/rate an item j. The process is to calculate the similarities between target ur i and all other urs, lec
t the top X similar urs, and take the weighted average of ratings from the X urs with similarities as weights.
While different people may have different balines when giving ratings, some people tend to give high scores generally, some are pretty strict even though they are satisfied wi
th items. To avoid this bias, we can subtract each ur’s average rating of all items when computing weighted average, and add it back for target ur, shown as below.
