How exactly is the recommendation question?

The computer is automatically arranged, and the manual is too busy.

Dear users, hello!

I won't say it, it's a conversation between me and a user.

The next person to say is me:

Hello, I would like to know how the content in the recommendation section "Waiting for you" on the right side is recommended to us?

The "questions you expect" in what others say about this type of question are determined based on which category the user scores the most. For example, if you have frequent problems with the computer network, then the "Expected Questions" will also be a problem with the computer network.

If you are specific to a subcategory, e.g. know > computer network > know then the expected questions will also be accurate to that category. "I think it's far-fetched, and I'm also recommending the relevant content of the query, is it collecting personal search information?

Hope. Hello! Yes, I took a look at the questions recommended for me myself. I found that there were many things that were also irrelevant to my field.

Of course, a closer look reveals that there are more problems with that category. The "recommended questions for me" is not only related to the category in the personal field, in addition, it is related to your personal information (your own interests, etc.) and historical operations (including those searched by the engine, which will be counted), in addition, some zero questions will also be randomly displayed in the personal recommendation questions. So it can be said that it is multifaceted. Thank you.

Artificial intelligence (AI) will be a strategic focus for the next decade. The existing search mechanism has basically been solidified, the search engine pattern has basically settled, if you only focus on the present and stop R & D and innovation, sooner or later it will enter an ecological bottleneck, artificial intelligence on text, voice and ** understanding is very detailed, simple search will be a new way of human-computer interaction integration, intended to create an AI search engine.

Introduction

The reason why it's called a simple search is to get everything to zero. Simple search is like a nascent AI bot, using machine learning to understand individual user habits, learn and grow with users, and most importantly, show their own values on the product, that is, simple search, fair information and valuable knowledge.

The commercialization of search engines has undergone two revolutions, the first is hyperchain analysis technology, which improves the algorithm of inverse correlation matching of text limb operation model, and the second is localized language, which solves the problem of language habits and semantic grammar in search algorithms of different countries.

How do you recommend it?

Select the site type.

Choose one of the five categories that match your site type from the five categories: General, Sales, News, Film & TV, and Merchandise, and click Next.

Select a recommendation type.

Choose the one you think has a good user experience from the three categories of pop-up, embedded and side-sliding, and click Next.

Select a template style.

Choose from the three categories of ** type, text type, and ** mixed type, and click Next. Each type has several different ways to display it for webmasters to choose from.

Detailed settings are recommended.

This is the most important step, and the pages are recommended for basic settings, style settings, and effect preview areas. Note that the Recommended Window Presentation Page and the Recommended Content page support wildcards, but if you only want to display a subset of pages and pages, you can also take advantage of the new Recommended Page and the New Exclusion Page.

Turn on recommendations. The official version of the recommendation, the text content takes 1 day, and ** takes 2 days. Here you can click on content management to add filter rules and pinned rules, and you can filter the ** page that you don't want to show, and pin key products.

You can sign in using a Promotions Account or a Webmaster Account.

The recommendation window is displayed on the page, personalized content is recommended for each visitor, and a variety of recommendation methods such as text links and **links are supported. Embedded: Embed personalized recommendation content into the web page to bring a better buried recommendation experience, supporting a variety of recommendation methods including text imitation stool characters and **, which require additional installation**.

Side-sliding: Slide out the recommendation window from the left and right sides of the page, recommend personalized content for each visitor, support text chain, ** and other recommendation methods, and statistics users can open it directly, no additional installation is required**. Radical Ant set the recommended style:

Adjust the size, style, color, and more of the recommended area. The following is a preview of my personal settings, which will take about a day or two to analyze before the effect is displayed.

Click on the recommended content to bring ** page views (PV), which can now be directly viewed in the statistical report, and the "quality" of the traffic brought by the recommendation: the quality of the traffic brought by the recommendation.

You can look at the marketing center.