Shulou(Shulou.com)06/01 Report--

Now "smart recommendation" has almost become a standard for a content site, in order to improve the retention time of users, we need to find ways to come up with some new tricks. For example, at the end of the user's article, list the articles that the user is interested in and call it: guess you like it.

Now, if the editor gave you $10000 to implement this intelligent recommendation function, what would you do? According to common sense, should the train of thought be like this?

1. Design a tag table with a corresponding tag for each article, so that you can recommend similar articles to readers according to tag.

2. Design a user self-portrait algorithm for each user, analyze each article according to the algorithm, and then recommend the relevant articles to the user.

To be honest, it is not easy to do well, and it is estimated that the most popular AI will be used later. In this way of thinking, then the intelligent recommendation code at least needs to read the database, right? How to get similar article information without reading it?

However, the editor recently found a "smart recommendation" plug-in, the function is very cool, any content site, you only need to copy a line of JS code to your site. There is no need to read the database, as long as the JS code is done.

The editor is puzzled, what kind of thinking do people use to achieve "intelligent recommendation"? Besides, the database of each website is different, and I was really confused at that time when you took all the JS codes. Reader, do you have any ideas?

Internal affairs do not decide to ask baidu, I will enter on Baidu: recommended JS code does not work. A record was found on the segmentfault website:

When the editor saw that "it may take some time for the deployment to take effect," he immediately figured it out. Have you figured it out yet?

It is a universal plug-in, only need to copy its js code to achieve intelligent recommendation, and do not need to read your local database, which recommended data from where? Look at this answer: it may take some time for the deployment to take effect. Actually, the answer is right here.

The thing is, when the user browses the page, the js will send the user's title, url and other relevant information to its server. In essence, it installs a data collector on your website. Although it does not read your database, it will collect your data and store it in its own database, so it only takes a period of time to accumulate relevant data after deployment. Otherwise, it can't be recommended. The recommended record that you display locally is displayed after reading from its own collection library.

In fact, it is just a change of thinking, the database is still needed. But its design idea is really very good, so that one can be omnipotent, and the other is to legally collect data from other websites, which is the best of both worlds.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.