CN103257969A - Related search prompting method and system based on search word pairs - Google Patents
Related search prompting method and system based on search word pairs Download PDFInfo
- Publication number
- CN103257969A CN103257969A CN 201210037573 CN201210037573A CN103257969A CN 103257969 A CN103257969 A CN 103257969A CN 201210037573 CN201210037573 CN 201210037573 CN 201210037573 A CN201210037573 A CN 201210037573A CN 103257969 A CN103257969 A CN 103257969A
- Authority
- CN
- China
- Prior art keywords
- term
- session
- search
- user
- retrieval request
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Abstract
The invention provides a related search prompting method and system based on search word pairs. The prompting method comprises the steps of recording search logs of users and mining search word pairs from the search logs, building indexes of the search word pairs of search words through the association degree to form a search word pair index database, and prompting the users of the search word pairs of the search words in order when the users search with certain search words serving as key words. By means of the prompting method and system, the search efficiency is improved.
Description
Technical field
The present invention relates to the information search technique field, relate in particular to a kind of based on term right relevant search reminding method and system.
Background technology
The way of tradition relevant search is by expansion done in the core keyword of user's input.In Fig. 1, the keyword of user input is " patent ", and the relevant search prompt system is just picked out some words that contain this keyword or displaying done in phrase so.Yet this method has certain limitation, and namely system can only pick out the word that contains same section with the keyword of user's input.If the word of word to be selected and user's input does not have literal contact, system is just helpless.For example " also pearl sound of laughing " and " little swallow ", the method for traditional relevant search for keyword " and pearl sound of laughing ", just can't be pointed out " little swallow ".
Summary of the invention
The object of the present invention is to provide that a kind of it is from user's retrieve log based on term right relevant search reminding method and system, it is right to excavate term, and by setting up the right index of term, the relevant prompting when being used for search improves search efficiency.
For addressing the above problem, the invention provides a kind ofly based on the right relevant search reminding method of term, comprising:
Record each user's term and retrieval request time;
Each term and retrieval request time thereof according to record, obtain other terms as the right correlation degree of the term of a term;
According to described correlation degree to the term of each term to ordering, the term of setting up each term to index to form term to index database;
When the user is key search with certain term that records, right to the term of this term of user prompt by ordering.
Further, recording each user's term and retrieval request during the time, is the described retrieval request of the bound pair time to be session to cut apart with a time predefined also.
Further, obtaining other terms comprises as the step of the right correlation degree of the term of a term:
To the term among each session, that term after its retrieval request time is right as the term of this term;
Term to the term among each session calculates carrying out weights;
The identical term of the identical term among all session to being weighted processing, is obtained other terms as the right correlation degree of the term of a term.
Further, to the term of each term among each session to the formula that carries out weights and calculate be: the distance between the term of the right weights=threshold distance of term/in this session pair and the term.
Accordingly, it is a kind of based on the right relevant search prompt system of term that the present invention also provides, and comprising:
The log analysis module, the term and the retrieval request time that are used for recording each user;
The associative search word is used for each term and retrieval request time thereof according to record to module, obtains other terms as the right correlation degree of the term of a term;
Build the library searching module, be used for according to correlation degree the term of each term ordering, the term of setting up each term to index forming term to index database, and when the user is key search with certain term, right to the term of this term of user prompt by sorting.
Further, described log analysis module is the described retrieval request of the bound pair time to be session to cut apart with a time predefined.
Further, described associative search word obtains other terms to module and comprises as the step of the right correlation degree of the term of a term:
To the term among each session, that term after its retrieval request time is right as the term of this term;
Term to each term among each session calculates carrying out weights;
The identical term of the identical term among all session to being weighted processing, is obtained other terms as the right correlation degree of the term of a term.
Further, described associative search word to module to the term of each term among each session to the formula that carries out weights and calculate is: the distance between the term of the right weights=threshold distance of term/in this session pair and the term.
Compared with prior art, provided by the invention based on the right relevant search prompt system of term, it is right also therefrom to excavate term by the retrieve log of recording user, set up the right index of the term of each term to form term to index database by correlation degree, and when the user is key search with certain term, right to the term of this term of user prompt by ordering, improve search efficiency.
Description of drawings
Fig. 1 is relevant search prompting synoptic diagram of the prior art;
Fig. 2 one embodiment of the invention based on the right relevant search reminding method process flow diagram of term;
Fig. 3 is the configuration diagram based on the right relevant search prompt system of term of one embodiment of the invention.
Embodiment
Below in conjunction with the drawings and specific embodiments being described in further detail based on term right relevant search reminding method and system the present invention's proposition.
As shown in Figure 2, the invention provides a kind ofly based on the right relevant search reminding method of term, may further comprise the steps:
S1, term and the retrieval request time of recording each user;
S2, obtains other terms as the right correlation degree of the term of a term at each term and retrieval request time thereof according to record;
S3, according to described correlation degree to the term of each term to ordering, the term of setting up each term to index to form term to index database;
S4 is when the user is key search with certain term that records, right to the term of this term of user prompt by ordering.
In the present embodiment, step S1 during the time, is the described retrieval request of the bound pair time to be session to cut apart with a time predefined in record each user's term and retrieval request also.Session refers to once complete user search behavior.Comprise request term (query), click the result, browse page turning, conversion term (query) etc.Doing the meaning that session cuts apart with a time predefined is, when the time phase difference time that time and the next one of a last request are asked is not less than described time predefined, just think the continuity that this next one request no longer is a last request (not being a session), but as the beginning of next session.
For example, time predefined is 10 minutes, and term and the retrieval request time of recording each user are following request sequence:
10 requests of user query " little swallow ";
10: 1 request query " little swallow Zhao Wei ";
10: 2 request query " also pearl sound of laughing ";
Request query " opened schoolmate's song " in 10: 13;
Can be registered as after by 10 minutes this request sequence being session and cutting apart: " little swallow ‖ ", " little swallow Zhao Wei | ", " also pearl sound of laughing " be session1, " schoolmate's song | " is session2.
Among the step S2 of present embodiment, obtain the process of the right correlation degree of each term of each term, specifically comprise following process:
A) to the term among each session, that the term after its retrieval request time is right as the term of this term;
B) term to each term among each session calculates carrying out weights, and formula is:
Distance between the term of the right weights=threshold distance of term/in this session pair and the term;
C) to the identical term of the identical term among all session to being weighted processing, obtain other terms as the right correlation degree of the term of a term.
In conjunction with the concrete request sequence among the present embodiment step S1, term is designated as queryA, term after its retrieval request time is designated as queryB, the term of queryA is to being designated as queryAqueryB, in fact the term of queryA is exactly term queryB to queryAqueryB, and then a) the result of step S2 is:
Term (query) " little swallow ", its term are " little swallow Zhao Wei ", " also pearl sound of laughing " to (queryAqueryB);
Term (query) " little swallow Zhao Wei ", its term are " also pearl sound of laughing " to (queryAqueryB);
Term (query) " also pearl sound of laughing " does not have the candidate;
Term (query) " is opened schoolmate's song " does not have the candidate;
The b of step S2) in, will be in Session with term to the distance of relevant two terms (queryA, queryB) of queryAqueryB as basic weights, the distance of two terms i.e. the number of two term queryA, queryB interval term in this Session request sequence, distance is more near, weights are more big, distance is more far away, and weights are more little.
Concrete, for example be that 10 minutes setting threshold distances equal 10 according to time predefined, then,
Term is to weights=10/ of the queryAqueryB distance (described distance＜=10) between queryB and the queryA in this session.
If the distance between queryB and the queryA is 1 o'clock, weights are 10; Distance between queryB and the queryA is 2 o'clock, and weights are 5.Distance between queryB and the queryA＞10 o'clock, weights are 0 (namely thinking uncorrelated).
So, in the request sequence for step S1 record, queryA " little swallow " and queryB " little swallow Zhao Wei " distance is 1, weights are 10, queryA " little swallow " and queryB " also pearl sound of laughing " distance is 2, weights are 5, and the like, obtain the right weights of each term among each session.And the right weights of each term are among the session1 that obtains in the present embodiment:
" little swallow " " little swallow Zhao Wei " 10
" little swallow " " also pearl sound of laughing " 5
" little swallow Zhao Wei " " also pearl sound of laughing " 10
Owing in all session, generally have the identical term of identical term to occurring, the c of step S2) result, identical term to a term queryA among all session of record is weighted summation to the weights of queryAqueryB exactly, to obtain the right correlation degree of final term, can be designated as " term is to the queryAqueryB total weight value ".
The purpose of step S3 is to be key with queryA, sets up queryA to the slide fastener (being that term is to queryAqueryB) of all queryB, then the queryAqueryB total weight value in the slide fastener (being that term is to correlation degree) is sorted from high to low.Like this, we have obtained the term of queryA to the queryAqueryB concordance list.
During retrieval, after the user has imported term queryA, will carry out the relevant search prompting of term queryB to the user, can only show preceding ten queryB that stand out in the queryAqueryB concordance list to the user.
Among the step S3 according to the right correlation degree of term suitable several terms to each term from high to low to ordering after, the term of term " little swallow " to index is:
" little swallow " " little swallow Zhao Wei " 10
" little swallow Zhao Wei " " also pearl sound of laughing " 10
" little swallow " " also pearl sound of laughing " 5
In step S4, when user input " little swallow " during as the key word of retrieval, can point out relevant search in order: " little swallow Zhao Wei ", " also pearl sound of laughing ".
Accordingly, it is a kind of based on the right relevant search prompt system of term that the present invention also provides, and comprising:
The associative search word is used for each term and retrieval request time thereof according to record to module 22, obtains other terms as the right correlation degree of the term of a term;
Build library searching module 23, be used for according to correlation degree the term of each term ordering, the term of setting up each term to index forming term to index database, and when the user is key search with certain term, right to the term of this term of user prompt by sorting.
Further, described log analysis module 21 is the described retrieval request of the bound pair time to be session to cut apart with a time predefined.
Further, described associative search word comprises the step of module 22 other terms of acquisition as the right correlation degree of the term of a term:
To the term among each session, that term after its retrieval request time is right as the term of this term;
The term of each term among each session is calculated carrying out weights, and formula is: the distance between the term of the right weights=threshold distance of term/in this session pair and the term;
The identical term of the identical term among all session to being weighted processing, is obtained other terms as the right correlation degree of the term of a term.
In sum, provided by the invention based on the right relevant search prompt system of term, it is right also therefrom to excavate term by the retrieve log of recording user, set up the right index of the term of each term to form term to index database by correlation degree, and when the user is key search with certain term, right to the term of this term of user prompt by ordering, improve search efficiency.
Obviously, those skilled in the art can carry out various changes and modification to invention and not break away from the spirit and scope of the present invention.Like this, if of the present invention these are revised and modification belongs within the scope of claim of the present invention and equivalent technologies thereof, then the present invention also is intended to comprise these changes and modification interior.
Claims (8)
1. one kind based on the right relevant search reminding method of term, it is characterized in that, comprising:
Record each user's term and retrieval request time;
Each term and retrieval request time thereof according to record, obtain other terms as the right correlation degree of the term of a term;
According to described correlation degree to the term of each term to ordering, the term of setting up each term to index to form term to index database;
When the user is key search with certain term that records, right to the term of this term of user prompt by ordering.
2. as claimed in claim 1ly it is characterized in that based on the right relevant search reminding method of term, record each user's term and retrieval request during the time, is the described retrieval request of the bound pair time to be session to cut apart with a time predefined also.
3. as claimed in claim 2ly it is characterized in that based on the right relevant search reminding method of term, obtain other terms and comprise as the step of the right correlation degree of the term of a term:
To the term among each session, that term after its retrieval request time is right as the term of this term;
Term to each term among each session calculates carrying out weights;
The identical term of the identical term among all session to being weighted processing, is obtained other terms as the right correlation degree of the term of a term.
4. as claimed in claim 3ly it is characterized in that based on the right relevant search reminding method of term, to the term of each term among each session to the formula that carries out weights and calculate be:
Distance between the term of the right weights=threshold distance of term/in this session pair and the term.
5. one kind based on the right relevant search prompt system of term, comprising:
The log analysis module, the term and the retrieval request time that are used for recording each user;
The associative search word is used for each term and retrieval request time thereof according to record to module, obtains other terms as the right correlation degree of the term of a term;
Build the library searching module, be used for according to correlation degree the term of each term ordering, the term of setting up each term to index forming term to index database, and when the user is key search with certain term, right to the term of this term of user prompt by sorting.
6. as claimed in claim 5ly it is characterized in that based on the right relevant search prompt system of term described log analysis module is the described retrieval request of the bound pair time to be session to cut apart with a time predefined.
7. as claimed in claim 6ly it is characterized in that based on the right relevant search prompt system of term that described associative search word obtains other terms to module and comprises as the step of the right correlation degree of the term of a term:
To the term among each session, that term after its retrieval request time is right as the term of this term;
Term to each term among each session calculates carrying out weights;
The identical term of the identical term among all session to being weighted processing, is obtained other terms as the right correlation degree of the term of a term.
8. as claimed in claim 7ly it is characterized in that based on the right relevant search prompt system of term that described associative search word to module to the term of each term among each session to the formula that carries out weights and calculate is:
Distance between the term of the right weights=threshold distance of term/in this session pair and the term.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 201210037573 CN103257969A (en) | 2012-02-17 | 2012-02-17 | Related search prompting method and system based on search word pairs |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 201210037573 CN103257969A (en) | 2012-02-17 | 2012-02-17 | Related search prompting method and system based on search word pairs |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN103257969A true CN103257969A (en) | 2013-08-21 |
Family
ID=48961898
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN 201210037573 Pending CN103257969A (en) | 2012-02-17 | 2012-02-17 | Related search prompting method and system based on search word pairs |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103257969A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109086458A (en) * | 2018-09-12 | 2018-12-25 | 杭州格原信息技术有限公司 | A kind of search engine system applied to reconnaissance projecting trade |
| CN110543484A (en) * | 2019-09-03 | 2019-12-06 | 广州视源电子科技股份有限公司 | prompt word recommendation method and device, storage medium and processor |
-
2012
- 2012-02-17 CN CN 201210037573 patent/CN103257969A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109086458A (en) * | 2018-09-12 | 2018-12-25 | 杭州格原信息技术有限公司 | A kind of search engine system applied to reconnaissance projecting trade |
| CN110543484A (en) * | 2019-09-03 | 2019-12-06 | 广州视源电子科技股份有限公司 | prompt word recommendation method and device, storage medium and processor |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US10496687B2 (en) | Input method, device, and electronic apparatus | |
| US20150032448A1 (en) | Method and apparatus for expansion of search queries on large vocabulary continuous speech recognition transcripts | |
| CN101996195A (en) | Searching method and device of voice information in audio files and equipment | |
| CN106407484A (en) | Video tag extraction method based on semantic association of barrages | |
| CN102156711B (en) | Cloud storage based power full text retrieval method and system | |
| CN104360994A (en) | Natural language understanding method and natural language understanding system | |
| CN103605665A (en) | Keyword based evaluation expert intelligent search and recommendation method | |
| CN103956169A (en) | Speech input method, device and system | |
| NZ601132A (en) | Systems and methods for ranking documents | |
| WO2013173826A3 (en) | Populating and searching a drug informatics database | |
| CN103593371A (en) | Method and device for recommending search keywords | |
| CN103020212A (en) | Method and device for finding hot videos based on user query logs in real time | |
| CN101826099A (en) | Method and system for identifying similar documents and determining document diffusance | |
| CN102339294A (en) | Searching method and system for preprocessing keywords | |
| CN104615734B (en) | A kind of community management service big data processing system and its processing method | |
| CN104199825A (en) | Information inquiry method and system | |
| Korn et al. | Automatically generating interesting facts from wikipedia tables | |
| CN101957860B (en) | Method and device for releasing and searching information | |
| WO2008108297A1 (en) | Homologous search system | |
| CN103152633A (en) | Method and device for identifying key word | |
| Ng | Information fusion for spoken document retrieval | |
| CN103257969A (en) | Related search prompting method and system based on search word pairs | |
| CN104156431A (en) | RDF keyword research method based on stereogram community structure | |
| CN103455964A (en) | Case clue analyzing system and method based on case information | |
| Wu et al. | VIREO@ TRECVid 2021 ad-hoc video search |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C02 | Deemed withdrawal of patent application after publication (patent law 2001) | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20130821 |