Search engine, do you know what I'm thinking?

Search is an information retrieval requirement generated with the increase of the amount of information. When there is a small amount of information, it uses classification, indexing, exact matching, and finally Fuzzy Matching to provide users with retrieval services. In the face of massive amounts of information, simple classification and indexing cannot meet human needs for information. Therefore, when the amount of information is large, how to reduce the user's time cost and make it easier for the user to find the target information is the main factor to measure the availability of a search. The ideal search engine has its own neural units and knowledge base, which can complete basic logical reasoning and understand human thinking, and ultimately meet the advanced needs of human information retrieval.

Next we will analyze the user's search behavior and summarize the types of search requirements. Let's take a look at the interaction process of user search:

1.User input keywords;

2.System Analysis keywords;

3.The system matches information based on keywords;

4.The system returns the search result.

The key here is how the system recognizes the input keywords and then understands the meanings of these words? For example, if a user enters"ABCKeyword. How does a search engine "Guess" keywords?

Scenario1: Enter"ABC", Understood as"ABC". Here, the system does not think about it. The keywords entered by the user can already represent the information that the user eventually needs. This requirement indicates that you already know what you need. It is a explicit requirement called "direct requirement ".

Scenario2 : Enter"ABC ", Initially split into"ABC, AB, BC, AC, a, B, c ", Then remove, extract, and integrate the existing dictionary, and finally the system gets"ABC, AB, . Here, the system carries out a preliminary thought. Through its dictionary knowledge, it analyzes the keywords to find out the three types of information that the user needs. Finally, Through Fuzzy Matching, it finds out the most likely result. This is a correlation requirement called "indirect requirement ".

Scenario3 : Enter"ABC ", In scenarios2 On the basis of, the system also Lenovo"D, E, F "Three and"ABC "A keyword with some internal link. Here, the system understands the potential thinking of users through association, and thinks that users may need or need more"Def Keyword information. This is an implicit requirement, known as "potential demand ".

From the perspective of user search behavior, search engines are a kind of "artificial intelligence" product. From exact match, fuzzy match to Lenovo match, a search engine shows a process from low-level to advanced-level machine learning. The general search engine processing process can be divided into two steps: the first is information capture, processing, processing, organization and storage; the second is keyword splitting, combination, and information matching, extraction, sorting, and presentation. In general search engines, a complex mathematical model is used to calculate the value of a Web page, that is, the weight of a Web page. For example:GoogleOfPRAlgorithm.

The biggest difference between intra-site search and general search engines is that the types of information content are different. The content of the general search engine is a webpage, and the content of intra-Site Search belongs to a specific industry. Therefore, intra-site search requires an algorithm that is different from the algorithm used to determine the weight of a webpage to measure the weight of a specific content.CrAlgorithm "(Content rank).CrThe core of an algorithm is to break down the content from multiple perspectives and abstract the smallest unit that can express the content attributes, that is, semantics or metadata. This process is the process of structuring content information. A large amount of content is processed to form a multi-dimensional content data warehouse. With such a warehouse, data mining can be achieved, finally, it can meet the information retrieval needs of users in different dimensions and provide users with a multi-dimensional user experience.