CN103020224B - A kind of intelligent search method and device - Google Patents
A kind of intelligent search method and device Download PDFInfo
- Publication number
- CN103020224B CN103020224B CN201210536673.XA CN201210536673A CN103020224B CN 103020224 B CN103020224 B CN 103020224B CN 201210536673 A CN201210536673 A CN 201210536673A CN 103020224 B CN103020224 B CN 103020224B
- Authority
- CN
- China
- Prior art keywords
- keyword
- demand
- word
- terminal
- search
- 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.)
- Active
Links
Landscapes
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The present invention provides a kind of intelligent search method and devices, the method comprise the steps that A. obtains the searching request of user；B. according to described search request in terminal attribute judge the type of the user, when the type of the user is mobile subscriber, execute step C；C. according to described search request in keyword judge whether the demand of the user related to terminal, if it is, by described search request in keyword expansion be keyword relevant to terminal；D. search result is obtained using the keyword relevant to terminal.By the above-mentioned means, the present invention improves the accuracy of search.
Description
[technical field]
The present invention relates to search technique, in particular to a kind of intelligent search method and device.
[background technique]
With the development of mobile internet, also more and more using the user of mobile terminal Internet access, and existing search is drawn
Hold up, the response mode to the user for using all terminals to scan for be it is identical, i.e., either which kind of terminal use user,
Obtained search result is the same.Since existing search engine cannot obtain automatically symbol according to the terminal situation that user uses
The search result of user demand is closed, therefore the accuracy searched for is lower.
[summary of the invention]
Technical problem to be solved by the invention is to provide a kind of intelligent search method and devices, to improve the accurate of search
Property.
The present invention is in order to solve the technical problem and the technical solution adopted is that provide a kind of intelligent search method, comprising: A. is obtained
Take the searching request at family；B. according to described search request in terminal attribute judge the type of the user, as the user
Type be mobile subscriber when, execute step C；C. according to described search request in keyword judge that the demand of the user is
It is no related to terminal, if it is, by described search request in keyword expansion be keyword relevant to terminal；D. it utilizes
The keyword relevant to terminal obtains search result.
One of according to the present invention preferred embodiment, the step C includes: that C1. searches the demand vocabulary obtained in advance, with true
The keyword in described search request is determined in the demand vocabulary with the presence or absence of matching word, if it is, determining the use
The demand at family is related to terminal, executes step C2；C2. by described search request in terminal attribute and described search request in
Keyword merges into keyword relevant to terminal.
One of according to the present invention preferred embodiment, the step of obtaining demand vocabulary in advance include: that S1. obtains search log；
S2. cutting is carried out to described search log as unit of session, wherein each cutting segment corresponds to a conversation recording；S3. needle
To each cutting segment, the keyword pair in the segment is extracted, wherein two keywords of the keyword centering are according in the piece
Appearance sequence in section arranges, and is only contained in the word in first keyword and preset terminal attribute vocabulary in rear keyword；
S4. demand word is chosen, from the first keyword of the keyword pair of extraction to obtain demand vocabulary.
One of according to the present invention preferred embodiment, choosing the mode of demand word in the step S4 includes being greater than conversion ratio
The first keyword of setting value is chosen for demand word, wherein the conversion ratio of each first keyword calculates according to the following formula:
Wherein p (w) indicates that the conversion ratio of first keyword w, t1 (w) indicate the keyword centering extracted
The number of keyword pair comprising w, t2 (w) indicate the number that w occurs in search log.
The present invention is in order to solve the technical problem and the technical solution adopted is that provide a kind of intelligent search device, comprising: receives
Unit, for obtaining the searching request of user；First judging unit, for the terminal attribute judgement in being requested according to described search
The type of the user, when the type of the user is mobile subscriber, triggering second judgment unit is executed；Second judgement is single
Member judges whether the demand of the user is related to terminal for the keyword in requesting according to described search, if it is, will
Keyword expansion in described search request is keyword relevant to terminal；Search unit, for using described with terminal phase
The keyword of pass obtains search result.
Preferred embodiment, the second judgment unit include: matching unit one of according to the present invention, are obtained in advance for searching
The demand vocabulary taken, to determine that the keyword in described search request whether there is matching word in the demand vocabulary, such as
Fruit is, it is determined that the demand of the user is related to terminal, and triggers expanding element execution；Expanding element, for being searched described
Terminal attribute in rope request merges into keyword relevant with terminal to the keyword in described search request.
Preferred embodiment, described device further comprise one of according to the present invention: vocabulary acquiring unit, for obtaining in advance
The demand vocabulary, the vocabulary acquiring unit include: log acquisition unit, for obtaining search log；Log cutting unit,
For carrying out cutting to described search log as unit of session, wherein each cutting segment corresponds to a conversation recording；Word pair
Extraction unit extracts the keyword pair in the segment, wherein two of the keyword centering close for being directed to each cutting segment
Keyword is arranged according to the appearance sequence in the segment, and is only contained in first keyword and preset terminal category in rear keyword
Word in property vocabulary；Demand word selection unit, for choosing demand word in the first keyword of the keyword pair from extraction, with
To demand vocabulary.
One of according to the present invention preferred embodiment, it includes that will convert that the demand selected ci poem, which takes the mode of unit selection demand word,
The first keyword that rate is greater than the set value is chosen for demand word, wherein the conversion ratio of each first keyword is counted according to the following formula
It calculates:
Wherein p (w) indicates that the conversion ratio of first keyword w, t1 (w) indicate the keyword centering extracted
The number of keyword pair comprising w, t2 (w) indicate the number that w occurs in search log.
As can be seen from the above technical solutions, the present invention realizes a kind of end that the search need by user is used with user
End type combines the method to obtain search result, and this method, which does not need the accurate keyword of user's input, to be automatically
User returns to the very high search result of suitability, improves the accuracy of search.
[Detailed description of the invention]
Fig. 1 is the flow diagram of the embodiment of intelligent search method in the present invention；
Fig. 2 is the schematic diagram of the search result got in the present invention using keyword relevant to terminal；
Fig. 3 is the structural schematic block diagram of the embodiment one of intelligent search device in the present invention；
Fig. 4 is the structural schematic block diagram of the embodiment two of intelligent search device in the present invention；
Fig. 5 is the structural schematic block diagram of one embodiment of the second judgment unit 303 in the present invention.
[specific embodiment]
To make the objectives, technical solutions, and advantages of the present invention clearer, right in the following with reference to the drawings and specific embodiments
The present invention is described in detail.
Referring to FIG. 1, Fig. 1 is the flow diagram of the embodiment of intelligent search method in the present invention.As shown in Figure 1, should
Embodiment includes:
Step S101: the searching request of user is obtained.
Step S102: judging the type of user according to the terminal attribute in searching request, when the type of user is mobile uses
When family, step S103 is executed.
Step S103: judge whether the demand of user is related to terminal according to the keyword (query) in searching request, such as
Fruit is, then is keyword relevant to terminal by the keyword expansion in searching request.
Step S104: search result is obtained using keyword relevant to terminal.
Above-mentioned steps are described in detail below.
Several parameters are contained in the searching request of user, one of parameter indicates to be made when user issues searching request
With the attribute of terminal, a parameter indicates the keyword that user is used to obtain search result and uses.Terminal attribute can be end
The information such as model, the manufacturer at end, such as " Samsung " or " iphone5 " information.
When user issues searching request using search client, search client can call the system interface of equipment, from
And it gets terminal attributive information and is encapsulated in the searching request of user.But if search client can not get terminal
Attribute information, then the terminal attribute in the searching request of user can be null value.
After search server receives the searching request of user in step s101, in step s 102, so that it may from
Terminal attribute is extracted in the searching request at family, if the terminal attribute extracted is null value either PC model etc
Information, it may be considered that initiation user's non-moving subscribers of searching request, if the terminal attribute extracted is " iphone5 " etc
Mobile terminal model etc information, then can determine that the initiation user of searching request is mobile subscriber, and execute step
S103。
As an implementation, step S103 is specifically included:
Step S1031: the demand vocabulary obtained in advance is searched, to determine the keyword in searching request in the demand vocabulary
In with the presence or absence of matching word, if it is, determine user demand it is related to terminal, execution step S1032.
Step S1032: the terminal attribute in searching request is merged into the keyword in searching request related with terminal
Keyword.
The mode of acquisition demand vocabulary will be introduced subsequent.The word for including in the demand vocabulary, can represent user
Demand be relevant to terminal, if the keyword in searching request matches with the word in the demand vocabulary, illustrate to use
The demand at family is related to terminal.Such as the keyword in searching request is " mobile phone EMS memory deficiency is what if ", which is needing
Asking in vocabulary has matching word, then may determine that using the demand of the user of the keyword be relevant to terminal, therefore executes
Step S1032.
The keyword in searching request is extended in step S1032, it is a kind of to obtain keyword relevant to terminal
Mode is that the terminal attribute in searching request is incorporated as keyword relevant with terminal to the keyword in searching request.Example
If the terminal attribute in searching request is " HTC ", and the keyword in searching request is " what if is mobile phone EMS memory deficiency ", then may be used
Using by " what if is HTC mobile phone EMS memory deficiency " as keyword relevant to terminal.
Referring to FIG. 2, Fig. 2 is the signal of the search result got in the present invention using keyword relevant to terminal
Figure.As it can be seen that by means of the invention it is also possible to the search need of adaptive user, improves the accuracy of search.
Demand vocabulary in the present invention can be the known vocabulary of third party's offer, in addition, demand vocabulary can also pass through
Mode acquires below.
Specifically, the mode for obtaining demand vocabulary includes step S201, step S202, step S203 and step S204.
Step S201: search log is obtained.Search log is the file for recording multiple user's search behaviors.It is searching for
In log, the search behavior of the same user may be recorded in multiple sessions (session), and session is to indicate one section of user behaviour
Make the unit of time.In a session, it can recorde the same user to be used to obtain multiple keywords that search result uses,
Such as user elder generation search key " what if is mobile phone EMS memory deficiency ", then modifying keyword is that " iphone mobile phone EMS memory is insufficient
What if " scan for, if the time of the two search operations all within a session period, is searching for the same of log
In a conversation recording, the information such as the two keywords and its time searched respectively are just had.
Step S202: being that unit carries out cutting to search log with session (session), wherein each cutting segment is corresponding
One conversation recording.By the narration of front it is known that search log contains multiple conversation recordings, step S202 will then be searched
Each conversation recording of Suo Zhi is as a cutting segment.
Step S203: being directed to each cutting segment, extract the keyword pair in the segment, wherein the keyword centering extracted
Two keywords are arranged according to the appearance sequence in the segment, and are only contained in first keyword and preset in rear keyword
Word in terminal attribute vocabulary.Terminal attribute vocabulary can be collected to obtain by data mining, include in the vocabulary and mobile terminal
The relevant various information of attribute, such as can have the type of the brand name (such as iphone, nokia) of mobile terminal, mobile terminal
Number title (such as I9100, galaxy S3), mobile terminal affiliated system platform title (such as ios, android).In cutting piece
The appearance sequence of Duan Zhong, keyword represent time sequencing when being searched for by user, and first keyword is the pass that user first searches for
Keyword, the keyword searched for after rear keyword is user.If being only contained in rear keyword and first closing in a cutting segment
Word in keyword and terminal attribute vocabulary, then the two there are the first keyword of inclusion relation and in rear keyword extraction
Out, as a keyword pair.Such as have recorded three keywords in a cutting segment respectively sequentially in time, respectively
It is:
Keyword A: mobile phone EMS memory
Keyword B: mobile phone EMS memory is insufficient
Keyword C: Samsung mobile phone EMS memory
It include first keyword A in rear keyword B, and further include word " deficiency " for keyword A and B, due to
" deficiency " is not the word in terminal attribute vocabulary, so keyword A and B are not as keyword pair.For keyword B and C,
Keyword C does not include first keyword B afterwards, therefore keyword B and C are also not as keyword pair.For keyword A and C,
Keyword C includes first keyword A afterwards, and further includes word " Samsung " in rear keyword C, and " Samsung " is terminal attribute word
Word in table, therefore keyword A and C can be extracted as a keyword pair.
Step S204: demand word is chosen, from the first keyword of the keyword pair of extraction to obtain demand vocabulary.As
All first keywords of the keyword centering of extraction can be chosen for demand word by a kind of embodiment.But it is more highly preferred to
Mode, be first to calculate the conversion ratio of each first keyword, and the first keyword that conversion ratio is greater than the set value is chosen for
Demand word.Wherein, the conversion ratio of each first keyword can calculate according to the following formula:
Wherein p (w) indicates that the conversion ratio of first keyword w, t1 (w) indicate that the keyword centering extracted includes the key of w
The number of word pair, t2 (w) indicate the number that w occurs in search log.
The demand word of selection is recorded in database table, so that it may obtain the demand vocabulary in the present invention.Above to this
The mode for obtaining demand vocabulary in invention in advance has carried out complete introduction.
The feasible embodiment of demand another kind whether relevant to terminal that user is judged in step S103 is described below.
Whether the demand in order to judge user in step s 103 is related to terminal, can also advance with training corpus pair
The disaggregated model of one unknown parameter is trained, and utilizes the disaggregated model of known parameters after this training in step s 103
Classify to the keyword in searching request to determine whether user demand is related to terminal.
The acquisition process of the disaggregated model of known parameters is introduced below.
Firstly, a collection of demand keyword relevant to terminal can be marked in search log as training corpus.Then from
Characteristic of division is extracted in the training corpus marked.The finally classification using the training corpus with characteristic of division to unknown parameter
Model is trained, and after training, disaggregated model is just provided with known parameter, can be used for the key in searching request
Word is classified.The disaggregated model of unknown parameter is trained, can be carried out by existing various machine learning methods, herein
Repeat no more its realization process.And in the present embodiment, the type of characteristic of division includes at least following one kind:
Feature one: the text feature of keyword.The text feature of word refers to after segmenting to keyword, word
N-gram information, in the present embodiment, N can be 1 or 2.
Feature two: the title feature of the page is clicked caused by keyword.Feature two can obtain in the following manner: to key
Page title is clicked caused by word carries out cutting, point to match from extraction in participle with the word in preset terminal attribute vocabulary
Word is as feature two.The page is clicked caused by keyword, is to search after user searches for the keyword in the correlation that search engine returns
The page clicked in hitch fruit.The acquisition modes and meaning of terminal attribute vocabulary in terminal attribute vocabulary and preceding embodiment
It is all the same.
Feature three: expansion word feature of the keyword in search log.Feature three can obtain in the following manner: from search
The expansion word of keyword is extracted in log as feature three, wherein the expansion word of keyword X refers to the keyword using X as substring
Y subtracts the remaining character string after X.Such as occur keyword X " surfing Internet with cell phone setting " and keyword Y " connection in search log
Logical surfing Internet with cell phone setting ", since X is the substring of Y, Y subtracts the expansion word that remaining character string " connection " after X is exactly X.
Feature four: conversion ratio feature of the keyword in search log.Feature four can obtain in the following manner: with session
Be that unit carries out cutting to search log, and is directed to keyword X, statistics and X appear in same session in rear keyword
Quantity t1, wherein each going out in search log in the word that rear keyword only includes in X and terminal attribute vocabulary, and statistics X
Existing number t2 obtains conversion ratio feature=t1/t2 of the keyword X in search log.
Complete introduction has been carried out to the acquisition process of the disaggregated model of known parameters above.In this way, in step s 103,
First to the characteristic of division of keyword extraction and training corpus same type in searching request, then will have in searching request
The keyword of characteristic of division is input to the disaggregated model of known parameters, and the key in searching request can be exported by the disaggregated model
Word generic, it will be understood that the category includes that user demand is related to terminal or user demand is uncorrelated with terminal.
Referring to FIG. 3, Fig. 3 is the structural schematic block diagram of the embodiment one of intelligent search device in the present invention.Such as Fig. 3 institute
Show, which includes: receiving unit 301, the first judging unit 302, second judgment unit 303 and search unit 304.
Wherein, receiving unit 301, for obtaining the searching request of user.If containing trepang in the searching request of user
Number, one of parameter indicate that user issues the attribute of searching request when institute using terminal, and a parameter indicates user for obtaining
The keyword for taking search result and using.Terminal attribute can be the information such as the model of terminal, manufacturer, such as " Samsung " or
Information such as " iphone5 ".
When user issues searching request using search client, search client can call the system interface of equipment, from
And it gets terminal attributive information and is encapsulated in the searching request of user.But if search client can not get terminal
Attribute information, then the terminal attribute in the searching request of user can be null value.
First judging unit 302, for judging the type of user according to the terminal attribute in searching request, when the class of user
When type is mobile subscriber, triggering second judgment unit 303 is executed.First judging unit 302 can be from the searching request of user
Terminal attribute is extracted, if the terminal attribute extracted is the information of null value either PC model etc, can be recognized
For initiation user's non-moving subscribers of searching request, if the terminal attribute extracted is the mobile terminal type of " iphone5 " etc
Number etc information, then can determine that the initiation user of searching request is mobile subscriber, thus just triggering second judgment unit 303
It executes.
Second judgment unit 303, for judged according to the keyword in searching request user demand whether with terminal phase
It closes, if it is, being keyword relevant to terminal by the keyword expansion in searching request.Specifically, second judgment unit
303 include matching unit 3031 and expanding element 3032.
Wherein, matching unit 3031, for searching the demand vocabulary obtained in advance, to determine the keyword in searching request
With the presence or absence of matching word in demand vocabulary, if it is, determining that the demand of user is related to terminal, and expanding element is triggered
3032 execute.Expanding element 3032, for by the keyword in the terminal attribute and searching request in searching request merge into
The relevant keyword of terminal.
In the present embodiment, demand vocabulary can be the known vocabulary of third party's offer, the word for including in the demand vocabulary,
It is relevant to terminal that the demand of user, which can be represented, if the keyword in searching request and the word phase in the demand vocabulary
Match, then illustrates that the demand of user is related to terminal.Such as the keyword in searching request is " mobile phone EMS memory deficiency is what if ", it should
Keyword has matching word in demand vocabulary, then matching unit 3031 may determine that the demand of the user using the keyword is
It is relevant to terminal, therefore trigger expanding element 3032 and execute.Expanding element 3032 expands the keyword in searching request
Exhibition, to obtain keyword relevant to terminal, a kind of mode is by the pass in the terminal attribute and searching request in searching request
Keyword is incorporated as keyword relevant to terminal.Such as the terminal attribute in searching request is " HTC ", and in searching request
Keyword is " what if is mobile phone EMS memory deficiency ", then can be by " what if is HTC mobile phone EMS memory deficiency " as relevant to terminal
Keyword.
Search unit 304, for obtaining search result using keyword relevant to terminal.Referring to FIG. 2, Fig. 2 is this
The schematic diagram of the search result got in invention using keyword relevant to terminal.As it can be seen that by means of the present invention, it can
With the search need of adaptive user, the accuracy of search is improved.
Referring to FIG. 4, Fig. 4 is the structural schematic block diagram of the embodiment two of intelligent search device in the present invention.Such as Fig. 4 institute
Show, embodiment two further includes vocabulary acquiring unit 305 compared with embodiment one.Wherein vocabulary acquiring unit 305 includes log again
Acquiring unit 3051, log cutting unit 3052, word are to extraction unit 3053 and demand word selection unit 3054.
Log acquisition unit 3051, for obtaining search log.Searching for log is for recording multiple user's search behaviors
File.In search log, the search behavior of the same user may be recorded in multiple sessions (session), and session is
Indicate the unit of one section of operating time of user.In a session, can recorde the same user makes for obtaining search result
Multiple keywords, such as user elder generation search key " what if is mobile phone EMS memory deficiency ", then modifying keyword is
" what if is iphone mobile phone EMS memory deficiency " scans for, if the time of the two search operations is all a session period
It is interior, then in the same conversation recording of search log, just have the letter such as the two keywords and its time searched respectively
Breath.
Log cutting unit 3052, for carrying out cutting to search log using session as unit, wherein each cutting segment
A corresponding conversation recording.
Word extracts the keyword pair in the segment, wherein keyword for being directed to each cutting segment to extraction unit 3053
Two keywords of centering according in the segment appearance sequence arrange, and rear keyword be only contained in first keyword and
Word in preset terminal attribute vocabulary.
Such as have recorded three keywords in a cutting segment respectively sequentially in time, it is respectively:
Keyword A: mobile phone EMS memory
Keyword B: mobile phone EMS memory is insufficient
Keyword C: Samsung mobile phone EMS memory
It include first keyword A in rear keyword B, and further include word " deficiency " for keyword A and B, due to
" deficiency " is not the word in terminal attribute vocabulary, so keyword A and B are not as keyword pair.For keyword B and C,
Keyword C does not include first keyword B afterwards, therefore keyword B and C are also not as keyword pair.For keyword A and C,
Keyword C includes first keyword A afterwards, and further includes word " Samsung " in rear keyword C, and " Samsung " is terminal attribute word
Word in table, therefore keyword A and C can be extracted as a keyword pair.
Demand word selection unit 3054, for choosing demand word in the first keyword of the keyword pair from extraction, with
To demand vocabulary.As an implementation, demand word selection unit 3054 can by the keyword centering of extraction it is all
First keyword is chosen for demand word.But more preferably mode, demand word selection unit 3054 first calculate each first key
The conversion ratio of word, and the first keyword that conversion ratio is greater than the set value is chosen for demand word.Wherein, each first keyword
Conversion ratio can calculate according to the following formula:
Wherein p (w) indicates that the conversion ratio of first keyword w, t1 (w) indicate that the keyword centering extracted includes the key of w
The number of word pair, t2 (w) indicate the number that w occurs in search log.
Referring to FIG. 5, Fig. 5 is the structural schematic block diagram of another embodiment of the second judgment unit 303 in the present invention.
As shown in figure 5, second judgment unit 303 includes taxon 303a and expanding element 303b in the embodiment.Wherein grouping sheet
First 303a classifies to the keyword in searching request using the disaggregated model of known parameters, to determine that the demand of user is
It is no related to terminal, if it is, triggering expanding element 303b is executed.Expanding element 303b, for by the end in searching request
Attribute is held to merge into keyword relevant with terminal to the keyword in searching request.
The disaggregated model of above-mentioned known parameters is to utilize to be extracted classification of the training corpus to unknown parameter of characteristic of division
What model obtained after being trained.Training corpus can be by marking a collection of demand key relevant to terminal in search log
Word obtains.In the present embodiment, the type of characteristic of division includes at least following one kind:
Feature one: the text feature of keyword.The text feature of word refers to after segmenting to keyword, word
N-gram information, in the present embodiment, N can be 1 or 2.
Feature two: the title feature of the page is clicked caused by keyword.Feature two can obtain in the following manner: to key
Page title is clicked caused by word carries out cutting, point to match from extraction in participle with the word in preset terminal attribute vocabulary
Word is as feature two.Here the acquisition modes and meaning of terminal attribute vocabulary and the terminal attribute vocabulary in preceding embodiment are equal
It is identical.
Feature three: expansion word feature of the keyword in search log.Feature three can obtain in the following manner: from search
The expansion word of keyword is extracted in log as feature three, wherein the expansion word of keyword X refers to the keyword using X as substring
Y subtracts the remaining character string after X.Such as occur keyword X " surfing Internet with cell phone setting " and keyword Y " connection in search log
Logical surfing Internet with cell phone setting ", since X is the substring of Y, Y subtracts the expansion word that remaining character string " connection " after X is exactly X.
Feature four: conversion ratio feature of the keyword in search log.Feature four can obtain in the following manner: with session
Be that unit carries out cutting to search log, for keyword X, statistics and X appear in same session in rear keyword quantity
T1, wherein each in the word that rear keyword only includes in X and terminal attribute vocabulary, and count what X occurred in search log
Number t2 obtains conversion ratio feature=t1/t2 of the keyword X in search log.
The foregoing is merely illustrative of the preferred embodiments of the present invention, is not intended to limit the invention, all in essence of the invention
Within mind and principle, any modification, equivalent substitution, improvement and etc. done be should be included within the scope of the present invention.
Claims (4)
1. a kind of intelligent search method, comprising:
A. the searching request of user is obtained；
B. according to described search request in terminal attribute judge the type of the user, when the type of the user is mobile uses
When family, C is executed；
C. search the demand vocabulary obtained in advance, with determine described search request in keyword in the demand vocabulary whether
In the presence of matching word, if it is, determine that the demand of the user is related to terminal, by described search request in terminal attribute
Keyword relevant with terminal is merged into the keyword in described search request；
D. search result is obtained using the keyword relevant to terminal；
Wherein, demand vocabulary is obtained in advance in the following manner:
S1. search log is obtained；
S2. cutting is carried out to described search log as unit of session, wherein each cutting segment corresponds to a conversation recording；
S3. it is directed to each cutting segment, the keyword pair in the segment is extracted, wherein two keywords of the keyword centering are pressed
The appearance sequence impinged upon in the segment arranges, and is only contained in first keyword and preset terminal attribute vocabulary in rear keyword
In word；
S4. demand word is chosen, from the first keyword of the keyword pair of extraction to obtain demand vocabulary.
2. the method according to claim 1, wherein the mode for choosing demand word in the S4 includes by conversion ratio
The first keyword being greater than the set value is chosen for demand word, wherein the conversion ratio of each first keyword calculates according to the following formula:
Wherein p (w) indicates that the conversion ratio of first keyword w, t1 (w) indicate that the keyword centering extracted includes
The number of the keyword pair of w, t2 (w) indicate the number that w occurs in search log.
3. a kind of intelligent search device, comprising:
Receiving unit, for obtaining the searching request of user；
First judging unit judges the type of the user for the terminal attribute in requesting according to described search, when the use
When the type at family is mobile subscriber, triggering second judgment unit is executed；
Second judgment unit, for searching the demand vocabulary obtained in advance, to determine the keyword in described search request in institute
It states with the presence or absence of matching word in demand vocabulary, if it is, determining that the demand of the user is related to terminal, by described search
Terminal attribute in request merges into keyword relevant with terminal to the keyword in described search request；
Search unit, for obtaining search result using the keyword relevant to terminal；
Vocabulary acquiring unit, for obtaining the demand vocabulary in advance, the vocabulary acquiring unit includes:
Log acquisition unit, for obtaining search log；
Log cutting unit, for carrying out cutting to described search log as unit of session, wherein each cutting segment is corresponding
One conversation recording；
Word extracts the keyword pair in the segment, wherein the keyword centering for being directed to each cutting segment to extraction unit
Two keywords arranged according to the appearance sequence in the segment, and be only contained in first keyword and default in rear keyword
Terminal attribute vocabulary in word；
Demand word selection unit, for choosing demand word in the first keyword of the keyword pair from extraction, to obtain demand word
Table.
4. device according to claim 3, which is characterized in that the demand selected ci poem takes the mode packet of unit selection demand word
It includes the first keyword for being greater than the set value conversion ratio and is chosen for demand word, wherein the conversion ratio of each first keyword is according to following
Formula calculates:
Wherein p (w) indicates that the conversion ratio of first keyword w, t1 (w) indicate that the keyword centering extracted includes
The number of the keyword pair of w, t2 (w) indicate the number that w occurs in search log.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210536673.XA CN103020224B (en) | 2012-12-12 | 2012-12-12 | A kind of intelligent search method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210536673.XA CN103020224B (en) | 2012-12-12 | 2012-12-12 | A kind of intelligent search method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN103020224A CN103020224A (en) | 2013-04-03 |
| CN103020224B true CN103020224B (en) | 2019-01-15 |
Family
ID=47968828
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201210536673.XA Active CN103020224B (en) | 2012-12-12 | 2012-12-12 | A kind of intelligent search method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103020224B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9754034B2 (en) * | 2013-11-27 | 2017-09-05 | Microsoft Technology Licensing, Llc | Contextual information lookup and navigation |
| CN104063455B (en) * | 2014-06-24 | 2017-04-12 | 北京奇虎科技有限公司 | Method and device for acquiring counseling messages of disease based on searching |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1677389A (en) * | 2004-03-31 | 2005-10-05 | 宇龙计算机通信科技(深圳)有限公司 | Mobile internet intelligent information retrieval engine based on key-word retrieval |
| CN101008946A (en) * | 2006-01-26 | 2007-08-01 | 冉征 | Search method of Chinese mobile communication information and device thereof |
| WO2007098710A1 (en) * | 2006-03-03 | 2007-09-07 | China Mobile Communications Corporation | Searching system and method based on personalized information |
| CN101052181A (en) * | 2007-05-22 | 2007-10-10 | 中国移动通信集团浙江有限公司 | Wireless searching system and its wireless searching method |
-
2012
- 2012-12-12 CN CN201210536673.XA patent/CN103020224B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1677389A (en) * | 2004-03-31 | 2005-10-05 | 宇龙计算机通信科技(深圳)有限公司 | Mobile internet intelligent information retrieval engine based on key-word retrieval |
| CN101008946A (en) * | 2006-01-26 | 2007-08-01 | 冉征 | Search method of Chinese mobile communication information and device thereof |
| WO2007098710A1 (en) * | 2006-03-03 | 2007-09-07 | China Mobile Communications Corporation | Searching system and method based on personalized information |
| CN101052181A (en) * | 2007-05-22 | 2007-10-10 | 中国移动通信集团浙江有限公司 | Wireless searching system and its wireless searching method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN103020224A (en) | 2013-04-03 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102866990B (en) | A kind of theme dialogue method and device | |
| CN103760991B (en) | Physical input method and physical input device | |
| CN105183787A (en) | Information input method and apparatus | |
| CN103916436B (en) | Information-pushing method, device, terminal and server | |
| CN103440243B (en) | A kind of teaching resource recommendation method and device thereof | |
| CN102402619A (en) | Search method and device | |
| CN104360994A (en) | Natural language understanding method and natural language understanding system | |
| CN109582969A (en) | Methodology for Entities Matching, device and electronic equipment | |
| CN104468959A (en) | Method, device and mobile terminal displaying image in communication process of mobile terminal | |
| CN102905233A (en) | Method and device for recommending terminal function | |
| CN109739961A (en) | A kind of man-machine language exchange method and device | |
| CN107846426A (en) | The tracking and device of user trajectory in page access | |
| CN105989120B (en) | A kind of personalization content recommendation method and individualized content recommender system | |
| JP2018014765A5 (en) | ||
| KR101696499B1 (en) | Apparatus and method for interpreting korean keyword search phrase | |
| CN108304452A (en) | Article processing method and processing device, storage medium | |
| CN109522405A (en) | File information processing method, electronic equipment and computer-readable deposit medium | |
| CN106095998B (en) | Topic method and device is precisely searched applied to intelligent terminal | |
| CN103020224B (en) | A kind of intelligent search method and device | |
| CN103853771B (en) | A kind of method for pushing and system of search result | |
| CN109558531A (en) | News information method for pushing, device and computer equipment | |
| CN107784024B (en) | Construct the method and device of party's portrait | |
| CN103997492B (en) | A kind of adaption system and method | |
| CN105589935A (en) | Social group recognition method | |
| CN105094810B (en) | Data processing method and device based on CGI(Common gateway interface) plug-in unit |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| EXSB | Decision made by sipo to initiate substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |