Recommender systems: overview of the characteristics of architecture: user/item engineering -- -- -- -- -- -- -- -- > recall layer > sort layer - > test/evaluation 】 【 cold start problems, real-time 】

I. What is a recommendation system

Recommendation system is a measure used for information overload. Faced with massive data information, it can quickly recommend items that meet the user's characteristics.Especially for some people who have "choice phobia" and no clear needs.

Information consumers: It is very difficult to find the information that interests you from a large amount of information.

Information producers: How to make the information produced by yourself stand out and be loved by the public is also a very difficult thing.

1. Purpose of the recommendation system

• In the Internet era, the amount of information is huge, which leads to information overload for individuals with limited information processing capabilities, and it is difficult for individuals to search for valuable information from the massive amount of information.
• Recommendation system is a measure used for information overload. In the face of massive data information, it can quickly recommend items that meet the user's characteristics and reduce the user's search cost.
• Addresses some people's "choice phobia" and those who don't have clear needs.
• A good recommendation system can allow users to get the items or content they need faster and better.
• Let the website (platform) retain user resources more effectively.

2. Basic idea of ​​recommendation system

• Know what you want, push it accurately.Using the feature information of users and items, recommend those items with features that users like.
• Things cluster together.Using the items that the user has liked, recommend items similar to the items that the user has liked to the user.
• People are divided into groups.Using other users who are similar to the user, recommend items to the user that are liked by other users who have similar interests and hobbies.
  

3. Data analysis of recommendation system

• Metadata for item or content to be recommended, such as keywords, category tags, descriptions, etc.
• System Basic information about the user, such as gender, age, interest tags, etc.
• User's behavior data can be transformed into preferences for items or information. Depending on the application itself, it may include the user's rating of the item, the user's viewing record of the item, the user's purchase record, etc..The preference information of these users can be divided into two categories:
  • Displayed user feedback: This category is the user's natural browsing on the website or outside the use of the website, and explicit feedback information, such as the user's rating of an item, or a comment on an item.
  • Implicit user feedback: This type of data is generated by the user when using the website, which implicitly reflects the user's preferences for items, such as the user purchased an item, the length of time the user viewed a certain content, and so on.

---

---

---

References:
Recommendation System (1): Overview of Recommendation System
Recommendation system overview