Search method and device and equipment
A search method and technology for generating search, applied in the field of information search, can solve the problems of increasing user burden, affecting user experience, single factors, etc., to achieve the effect of improving search experience and reducing secondary screening
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
Method used
Image
Examples
Embodiment 1
[0028] The embodiment of the present invention can be applied to various search engine servers for search queries, as long as the search engine server can display the corresponding search query information on the search result page according to the user's search request, without specific search query To restrict, such as goods, knowledge, resources or services. The method can be performed by a search device.
[0029] see figure 1 , the search method provided in this embodiment specifically includes: S100-S300.
[0030] S100. Obtain candidate blocks matching the search keyword and entry information in each candidate block.
[0031] Wherein, the search keyword is a general term of a search request entered by the user into the search engine. The candidate block is a block selected from all the candidate blocks according to the search keyword for displaying on the search result page. Alternative blocks are blocks corresponding to different search topics obtained after statisti...
Embodiment 2
[0046] In this embodiment, on the basis of the foregoing embodiments, S200 is described and optimized in detail.
[0047] see image 3 , the search and sorting method provided in Embodiment 2 specifically includes:
[0048] S100. Obtain candidate blocks matching the search keyword and entry information in each candidate block, and execute any one of S210, S221, and S230.
[0049] S210. Based on the search keyword features, the block features of each candidate block, and the entry features of each entry information, sort the acquired candidate blocks and the entry information in each candidate block, and execute S300.
[0050] Wherein, the search keyword feature is a parameter characterizing the characteristics of the search keyword, for example, it may be the number of the search keyword, the field of the search keyword, or the search frequency of the search keyword.
[0051] Specifically, after obtaining the candidate blocks and the entry information of each candidate block...
Embodiment 3
[0067] This embodiment further introduces the above-mentioned page element sorting model on the basis of the above-mentioned embodiments.
[0068] The training data of the page element sorting model provided in various embodiments of the present invention is obtained by extracting features from user behavior logs. For example, by performing feature extraction on each sample log, a piece of sample data can be obtained, expressed as (x, y), where x=(f 1 , f 2 ,..., f n ), which is an extracted feature, y=±1, which indicates whether a certain user behavior occurs, such as whether a conversion or a click occurs, and y is 1 if the behavior occurs, and -1 if the behavior does not occur. The page element ranking model is to determine the weight value of each feature in the input parameter x by predicting whether a certain behavior will occur.
[0069] see Figure 4a , the page element sorting model provided in Embodiment 3 is a four-layer feed-forward neural network model obtaine...
PUM
Login to View More Abstract
Description
Claims
Application Information
Login to View More