Search recall method and device, server and storage medium

A technology of indexing and related retrieval, applied in the Internet field, can solve the problems of affecting recall accuracy, lack of semantic generalization ability, etc., and achieve the effect of improving search recall accuracy

Active Publication Date: 2017-12-19
BEIJING BAIDU NETCOM SCI & TECH CO LTD
View PDF5 Cites 58 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] The embodiment of the present invention provides a search and recall method and device, server, and storage medium to...

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Search recall method and device, server and storage medium
  • Search recall method and device, server and storage medium
  • Search recall method and device, server and storage medium

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0026] figure 1 It is a flowchart of the search and recall method provided by Embodiment 1 of the present invention. This embodiment is applicable to the case of search and recall. The method can be executed by a search and recall device, which can be implemented by software and / or hardware. Such as figure 1 As shown, the method specifically includes:

[0027] S110. Obtain the search term input by the user.

[0028] Among them, the search term is the key content entered by the user in the search engine according to the purpose of the search. Specifically, the search term can be a word, a word, or a sentence; it can be Chinese, it can be an English letter, it can be one or more numbers, or it can be a combination of the above forms. Retrieve content.

[0029] S120. Calculate the semantic vector of the search term by using the pre-trained neural network model.

[0030] What needs to be explained here is that in the prior art, search engines usually retrieve relevant results...

Embodiment 2

[0041] figure 2 It is a flow chart of the search and recall method provided by Embodiment 2 of the present invention. Embodiment 2 further optimizes the training process of the neural network model on the basis of Embodiment 1. Such as figure 2 As shown, the method includes:

[0042] S210. From the user log accumulated by the search engine, extract items presented to the user according to each search term, use the items clicked by the user as training positive samples, and use the items that the user does not click as training negative samples.

[0043] S220. From the user log, randomly extract a preset number of entries as random negative samples, and the randomly extracted entries have no clicks under all search terms.

[0044] Among them, random negative samples refer to the entries in the user log that do not have clicks under all the search terms. As supplementary negative samples, they are used to improve the filtering accuracy of irrelevant results and reduce wrong ...

Embodiment 3

[0062] Figure 4 It is a schematic structural diagram of the search and recall device in Embodiment 3 of the present invention. Such as Figure 4 As shown, search recall devices include:

[0063] The retrieval term acquiring module 410 is configured to acquire the retrieval term input by the user.

[0064] The semantic vector calculation module 420 is used to calculate the semantic vector of the search term by using the pre-trained neural network model.

[0065] The document recall module 430 is configured to recall target documents related to the semantic vectors of the search terms from the candidate documents according to the pre-established index, wherein the index is established according to the semantic vectors of the candidate documents, and the semantic vector of the candidate documents The vector is calculated according to the pre-trained neural network model.

[0066] Preferably, the search and recall device further includes: a model training module, configured t...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

PUM

No PUM Login to view more

Abstract

Embodiments of the invention disclose a search recall method and device, a server and a storage medium. The search recall method comprises the following steps of: obtaining a retrieval word input by a user; calculating a semantic vector of the retrieval word by utilizing a pre-trained neural network model; and recalling a target document related to the semantic vector of the retrieval word from a candidate document according to a pre-established index, wherein the index is established according to a semantic vector of the candidate document, and the semantic vector of the candidate document is calculated according to the pre-trained neural network model. According to the method, the problem that existing recall method is lack of semantic generalization ability so as to influence the recall correctness can be solved, so that the search recall correctness is improved.

Description

technical field [0001] The embodiment of the present invention relates to Internet technology, and in particular to a search and recall method and device, a server, and a storage medium. Background technique [0002] With the rapid development of Internet technology, the functions of search engines are becoming more and more powerful. Search is usually divided into recall and sorting according to purpose and performance. The purpose of recall is to screen a small number of relevant results from a large number of candidate webpages / documents based on the search terms entered by the user; the purpose of sorting is to optimize the recalled results , using finer features to calculate their correlation, and based on this as the final order presented to the user. [0003] Usually, search engines use inverted indexes and other matching methods based on word / word granularity to achieve recall, and relevant results are extracted through word / word inverted zippers, and TFIDF (term fr...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

Application Information

Patent Timeline
no application Login to view more
IPC IPC(8): G06F17/30G06N3/04
CPCG06F16/332G06F16/36G06F16/328G06N3/04G06F16/3325G06N3/08G06F16/9535G06F16/2425G06F16/24522
Inventor 李辰姜迪王昕煜魏轶彬王璞何径舟
Owner BEIJING BAIDU NETCOM SCI & TECH CO LTD
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap