CN106653006A - Search method and device based on voice interaction - Google Patents
Search method and device based on voice interaction Download PDFInfo
- Publication number
- CN106653006A CN106653006A CN201611019821.5A CN201611019821A CN106653006A CN 106653006 A CN106653006 A CN 106653006A CN 201611019821 A CN201611019821 A CN 201611019821A CN 106653006 A CN106653006 A CN 106653006A
- Authority
- CN
- China
- Prior art keywords
- participle
- term
- search
- combination
- user
- 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.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/08—Speech classification or search
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/08—Speech classification or search
- G10L2015/081—Search algorithms, e.g. Baum-Welch or Viterbi
Landscapes
- Engineering & Computer Science (AREA)
- Computational Linguistics (AREA)
- Health & Medical Sciences (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Human Computer Interaction (AREA)
- Physics & Mathematics (AREA)
- Acoustics & Sound (AREA)
- Multimedia (AREA)
- Machine Translation (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention provides a search method and device based on voice interaction. The search method based on voice interaction comprises the steps of acquiring a first search word according to voice data provided by a user and used for searching, and acquiring a plurality of second search words for context information of the first search word; generating multiple segmentation combinations according to the first segmentation result of the first search word and the second segmentation result of each second search word; according to the correlation score of each segmentation combination, selecting a target segmentation combination from the multiple segmentation combinations as a third search word; and searching according to the third search word. The method and the device for searching in combination with the context information of voice search improve the search accuracy and the user search experience.
Description
Technical field
The present invention relates to Internet technical field, more particularly to a kind of searching method and device based on interactive voice.
Background technology
With the continuous development of Internet technology, the input cost of keyboard is also constantly lifted therewith.Phonetic search, based on strong
Big speech recognition capabilities, support quickly to initiate search by voice command, make search faster, more direct, more intelligent.It is related
In technology, traditional phonetic search to receive instruction, feedback command pattern, in search every time, with the last time search by this search
Rope and search for independent next time, i.e. in same retrieval sequence, equal onrelevant between search.
Under this mode, if user is carrying out when questioning closely or supplementing search of correlation to current term, need to repeat
Requirement of main body simultaneously provides complete Search Requirement expression, it is impossible to enough scan for reference to the contextual information of phonetic search, in language
It is not high that sound search expression searches for precision in the case of simplifying.
The content of the invention
It is contemplated that at least solving one of technical problem in correlation technique to a certain extent.
For this purpose, it is an object of the present invention to propose a kind of searching method based on interactive voice, can be with reference to voice
The contextual information of search is scanned for, and lifts search precision, lifts user's search experience degree.
Further object is that proposing a kind of searcher based on interactive voice.
A further object of the present invention is to propose a kind of searcher based on interactive voice.
Further object is that proposing a kind of non-transitorycomputer readable storage medium.
Further object is that proposing a kind of computer program.
To reach above-mentioned purpose, the searching method based on interactive voice that first aspect present invention embodiment is proposed, including：
First term is obtained according to the speech data for search that user provides, and is retrieved as the letter above of first term
Multiple second terms of breath；According to the first participle result and the second participle of each the second term of first term
As a result various participle combinations are generated；According to the relevance score of every kind of participle combination, from various participle combinations mesh is selected
Mark participle group is incorporated as the 3rd term；Scanned for according to the 3rd term.
The searching method based on interactive voice that first aspect present invention embodiment is proposed, by according to the first term
Second word segmentation result of first participle result and each the second term generates various participle combinations；According to the combination of every kind of participle
Relevance score, from selection target participle group in the combination of various participles the 3rd term is incorporated as；Entered according to the 3rd term
Line search, can scan for reference to the contextual information of phonetic search, lift search precision, lift user's search experience
Degree.
To reach above-mentioned purpose, the searcher based on interactive voice that second aspect present invention embodiment is proposed, including：
First acquisition module, the speech data for search for being provided according to user obtains the first term；Second acquisition module,
For being retrieved as multiple second terms of the information above of first term；Generation module, for according to described first
The first participle result of term and the second word segmentation result of each the second term generate various participle combinations；Selecting module,
For the relevance score combined according to every kind of participle, from selection target participle group in various participles combinations the is incorporated as
Three terms；Search module, for being scanned for according to the 3rd term.
The searcher based on interactive voice that second aspect present invention embodiment is proposed, by according to the first term
Second word segmentation result of first participle result and each the second term generates various participle combinations；According to the combination of every kind of participle
Relevance score, from selection target participle group in the combination of various participles the 3rd term is incorporated as；Entered according to the 3rd term
Line search, can scan for reference to the contextual information of phonetic search, lift search precision, lift user's search experience
Degree.
To reach above-mentioned purpose, the searcher based on interactive voice that third aspect present invention embodiment is proposed, its spy
Levy and be, including：
Processor；
For storing the memory of processor executable；
Wherein, the processor is configured to：
First term is obtained according to the speech data for search that user provides, and is retrieved as first term
Information above multiple second terms；
Generated according to the first participle result of first term and the second word segmentation result of each the second term many
Plant participle combination；
According to the relevance score of every kind of participle combination, combine and make from selection target participle in various participle combinations
For the 3rd term；
Scanned for according to the 3rd term.
The searcher based on interactive voice that third aspect present invention embodiment is proposed, by according to the first term
Second word segmentation result of first participle result and each the second term generates various participle combinations；According to the combination of every kind of participle
Relevance score, from selection target participle group in the combination of various participles the 3rd term is incorporated as；Entered according to the 3rd term
Line search, can scan for reference to the contextual information of phonetic search, lift search precision, lift user's search experience
Degree.
To reach above-mentioned purpose, the non-transitorycomputer readable storage medium that fourth aspect present invention embodiment is proposed,
When the instruction in the storage medium is performed by the processor of mobile terminal so that mobile terminal is able to carry out one kind and is based on
The searching method of interactive voice, methods described includes：
First term is obtained according to the speech data for search that user provides, and is retrieved as first term
Information above multiple second terms；
Generated according to the first participle result of first term and the second word segmentation result of each the second term many
Plant participle combination；
According to the relevance score of every kind of participle combination, combine and make from selection target participle in various participle combinations
For the 3rd term；
Scanned for according to the 3rd term.
The non-transitorycomputer readable storage medium that fourth aspect present invention embodiment is proposed, by according to the first retrieval
The first participle result of word and the second word segmentation result of each the second term generate various participle combinations；According to every kind of participle group
The relevance score of conjunction, from selection target participle group in the combination of various participles the 3rd term is incorporated as；According to the 3rd retrieval
Word is scanned for, and can be scanned for reference to the contextual information of phonetic search, lifts search precision, lifts user's search body
Degree of testing.
To reach above-mentioned purpose, the computer program that fifth aspect present invention embodiment is proposed, when the computer
When instruction processing unit in program product is performed, a kind of searching method based on interactive voice is performed, methods described includes：
First term is obtained according to the speech data for search that user provides, and is retrieved as first term
Information above multiple second terms；
Generated according to the first participle result of first term and the second word segmentation result of each the second term many
Plant participle combination；
According to the relevance score of every kind of participle combination, combine and make from selection target participle in various participle combinations
For the 3rd term；
Scanned for according to the 3rd term.
The computer program that fifth aspect present invention embodiment is proposed, by the first participle according to the first term
As a result various participle combinations are generated with the second word segmentation result of each the second term；Commented according to the correlation of every kind of participle combination
Point, it is incorporated as the 3rd term from selection target participle group in the combination of various participles；Scanned for according to the 3rd term, energy
Enough scan for reference to the contextual information of phonetic search, lift search precision, lift user's search experience degree.
The additional aspect of the present invention and advantage will be set forth in part in the description, and partly will become from the following description
Obtain substantially, or recognized by the practice of the present invention.
Description of the drawings
The above-mentioned and/or additional aspect of the present invention and advantage will become from the following description of the accompanying drawings of embodiments
It is substantially and easy to understand, wherein：
Fig. 1 is the schematic flow sheet of the searching method based on interactive voice that one embodiment of the invention is proposed；
Fig. 2 is the schematic flow sheet of the searching method based on interactive voice that another embodiment of the present invention is proposed；
Fig. 3 is the schematic flow sheet of the searching method based on interactive voice that another embodiment of the present invention is proposed；
Fig. 4 is phonetic search interface schematic diagram in the embodiment of the present invention；
Fig. 5 is the structural representation of the searcher based on interactive voice that one embodiment of the invention is proposed；
Fig. 6 is the structural representation of the searcher based on interactive voice that another embodiment of the present invention is proposed.
Specific embodiment
Embodiments of the invention are described below in detail, the example of the embodiment is shown in the drawings, wherein from start to finish
Same or similar label represents same or similar element or the element with same or like function.Below with reference to attached
The embodiment of figure description is exemplary, is only used for explaining the present invention, and is not considered as limiting the invention.Conversely, this
Inventive embodiment includes all changes fallen in the range of the spirit and intension of attached claims, modification and is equal to
Thing.
Fig. 1 is the schematic flow sheet of the searching method based on interactive voice that one embodiment of the invention is proposed.Language should be based on
The searching method of sound interaction is configured in the searcher based on interactive voice illustrate.Searching for interactive voice should be based on
Suo Fangfa can be applied in the search engine of electronic equipment, and search engine is the search information for referring to receiving user's input,
And collect from internet and search information related information, after information is organized and processed, provide the user retrieval clothes
Business, the system that user will be showed with search information related information.
Wherein, electronic equipment is, for example, PC (Personal Computer, PC), and cloud device or movement set
It is standby, mobile device such as smart mobile phone, or panel computer etc..
Referring to Fig. 1, should be based on the searching method of interactive voice includes：
S11：First term is obtained according to the speech data for search that user provides, and is retrieved as the first term
Information above multiple second terms.
Traditional phonetic search to receive instruction, feedback command pattern, in search every time, with the last time search by this search
Rope and search for independent next time, i.e. in same retrieval sequence, equal onrelevant between search, if user is to current retrieval
Word carries out when questioning closely or supplementing search of correlation, need to repeat requirement of main body and provide complete Search Requirement expression, it is impossible to be enough
Scan for reference to the contextual information of phonetic search, search for precision in the case where phonetic search expression formula simplifies not high.
And in embodiments of the invention, understood with the deep layer demand of current search above by search, if historical search word
In the initial participle of the second word segmentation result of first participle result and the second term comprising the first term when combining, then root
Scan for according to the combination of initial participle, can effectively lift search efficiency；If combining without initial participle in historical search word, root
The first participle and the second participle are combined into process according to the type information of the first term, various participle combinations are obtained, according to
Related algorithm calculates the relevance score of above-mentioned every kind of participle combination, and the maximum participle of score value is combined as final retrieval
Word is scanned for, and can be scanned for reference to the contextual information of phonetic search, lifts search precision, lifts user's search body
Degree of testing.
In an embodiment of the present invention, the speech data for search that can be provided with receive user, obtains and voice number
According to corresponding first text, and using corresponding first text as the first term.
Alternatively, user can click on the microphone button in search engine search box, be input into speech data, search engine
In voice acquisition module collect the speech data, obtain the first text corresponding with speech data, and by corresponding first
Text is used as the first term.
It is understood that user can relatively be questioned closely search to current search result or supplement search, when
The information above of front first term can be multiple.
For example, the speech data that user is input in a search engine is " U.S. ", and Search Results displayed page provides the U.S.
Relevant information, when user wants to inquire about the demographic data in the U.S., is input into speech data " population is how many ", with reference to information above, can
Know, the first term is " population is how many ", the second term is " U.S. ", or, user can continue to question closely " where capital is
In ", understanding with reference to information above, the first term is " where capital is ", and the second term is：" U.S. " and " have a large population
It is few ".
S12：Generated according to the first participle result of the first term and the second word segmentation result of each the second term many
Plant participle combination.
In an embodiment of the present invention, the first term and the second term can be carried out at cutting word using correlation technique
Reason, obtains the morphological information of the first term and the second term, i.e. obtain first participle result and the second word segmentation result.
It should be noted that the first term not occurred for history, based on context feature, at cutting word
Reason, can filter out the first identical participle, and can filter out phonetic identical fuzzy phoneme.
In an embodiment of the present invention, it can be determined that with the presence or absence of comprising first participle result and second in historical search word
The initial participle combination of word segmentation result；If there is the combination of the initial participle comprising first participle result and the second word segmentation result,
Initial participle is combined as the participle combination for generating；If not existing initial comprising first participle result and the second word segmentation result
Participle is combined, then extract the fisrt feature information of each first participle in first participle result, and is extracted in the second word segmentation result
The second feature information of each the second participle；According to fisrt feature information and second feature information, the class of the first term is determined
Type information；The first participle and the second participle are combined by process according to the type information of the first term, various participles are obtained
Combination.
In some embodiments, referring to Fig. 2, step S12 is specifically included：
S21：Judge in historical search word with the presence or absence of the initial participle comprising first participle result and the second word segmentation result
Combination, if so, performs S22, otherwise, performs S23.
Alternatively, judge in historical search word with the presence or absence of initial point comprising first participle result and the second word segmentation result
Word combination, when there is the combination of the initial participle comprising first participle result and the second word segmentation result in historical search word, directly
Initial participle is combined as the participle combination for generating, scanned for according to the participle combination for generating, can effectively lift search
Efficiency；When there is no the combination of the initial participle comprising first participle result and the second word segmentation result in historical search word, perform
S23。
S22：Initial participle is combined as the participle combination for generating.
Alternatively, there is the combination of the initial participle comprising first participle result and the second word segmentation result in historical search word
When, directly initial participle is combined as the participle combination for generating, scanned for according to the participle combination for generating, can effectively carry
Rise search efficiency.
S23：The fisrt feature information of each first participle in first participle result is extracted, and is extracted in the second word segmentation result
The second feature information of each the second participle.
In an embodiment of the present invention, fisrt feature information/second feature information includes at least one of：First point
The click frequency, general colloquial style after entity type, weight, the user's history search frequency, the user's history search of word/the second participle
Punishment weight, and editing distance.
Wherein, the entity type of the first participle/the second participle for the first participle/the second participle entity information, such as people
Name, place name etc..
The weights size of the weight of the first participle/the second participle describes its weight in the first term/the second term
The property wanted.
The user's history search frequency is used to determine the relevance score of the every kind of participle combination in subsequent step.
General colloquial punishment weight represents and the extensive normalization of colloquial style is carried out to the first participle/the second participle.By
There are a large amount of colloquial style statements, such as " I wants to search ... " " what meaning is consulted ... is " in speech data, it is therefore desirable to
Carry out the extensive normalization of colloquial style.Specifically, the first participle/the second participle is entered according to artificial constructed colloquial style vocabulary
The extensive normalization of row colloquial style, one punishment weight of each participle correspondence in colloquial style vocabulary, punishment weight is bigger, shows
The spoken words degree of participle is higher.
Editing distance is a kind of common similarity calculating method.
It is alternatively possible to extract the fisrt feature information of each first participle in first participle result, and extract second point
The second feature information of each the second participle in word result, to determine the first inspection according to fisrt feature information and second feature information
The type information of rope word.
S24：According to fisrt feature information and second feature information, the type information of the first term is determined.
Wherein, type information includes：First term be the second term supplement search first kind information, first
Term is the Second Type information for questioning closely search of the second term.
In an embodiment of the present invention, first kind information is：Main body semanteme is constant, demand strong correlation, for example, " everyday
Extremely run down load ", pronouns, general term for nouns, numerals and measure words " extremely running " everyday based on, " download " is demand word；Second Type information is：The semantic change of main body, demand
It is semantic constant.
It is alternatively possible to by entity attribute knowledge base recognize the entity attribute of the first term and the second term into
Point.It should be noted that in natural language the meaning of the same race can with different form of presentations, be also in entity attribute description as
This, i.e., same attribute requirements have different form of presentations, and for example, " population how many ", " how many population ", " population is many
It is few " be all " population " attribute requirements different expression, it is therefore desirable to taken out from different form of presentations by trunk extraction technique
Take identical demand trunk.
Specifically, by judging the entity attribute information and multiple second terms of information above of current first term
Entity attribute correlation, i.e., according to fisrt feature information and second feature information, and then judge that the first term is attribute
Change or Entity Change.
For example, when the first term is：" that body weight", the second term is：When " height of Liu Dehua ", by reality
Body attributes match, it is " Liu Dehua " to obtain entity, and " height " " body weight " is the attribute of " Liu Dehua ", therefore, it is known that first
The type information of term is：Attribute is changed, and main body semanteme is constant, i.e. first kind information.
Or, when the first term is：" that Yao Ming", the second term is：When " height of Liu Dehua ", by reality
Body attributes match, it is " Liu Dehua " " Yao Ming " to obtain entity, and their common attributes are " height ", therefore, it is known that the first inspection
The type information of rope word is：The semantic change of main body, demand semanteme is constant, i.e. Second Type information.
S25：The first participle and the second participle are combined by process according to the type information of the first term, obtain various
Participle is combined.
As a kind of example, when the first term be the second term question closely search Second Type information when, for example,
When the first term is：" that France", the second term is：When " where the capital in the U.S. is ", the various participles for obtaining
Combination can be as shown in table 1.
Table 1
It is initial comprising first participle result and the second word segmentation result by existing in historical search word in the present embodiment
When participle is combined, initial participle is combined as the participle combination for generating, search efficiency can be effectively lifted, by searching in history
When there is no initial participle comprising first participle result and the second word segmentation result in rope word and combining, according to fisrt feature information and
Second feature information, determines the type information of the first term, according to the type information of the first term by the first participle and
Two participles are combined process, obtain various participle combinations, carry out in order to subsequently maximum to relevance score participle combination
Search, can make user express Search Requirement freely so that phonetic search interactive experience is more intelligent, lift search accurate
Degree, lifts user's search experience degree.
In some embodiments, after step S12, also include：
S31：In determining various participle combinations, frequency of occurrence of every kind of participle combination in user's search history, Yi Jiyu
Participle combines the corresponding first user historical search frequency of the first participle and the second user history of corresponding second participle is searched
The rope frequency.
It is alternatively possible to assume that frequency of occurrence of the participle combination in user's search history is Nxyi, combine right with participle
The first user historical search frequency of the first participle answered is Nxi, the second user of the second participle corresponding with participle combination goes through
The history search frequency is Nyi, wherein, i=1,2 ..., M, M are the number of participle combination, and the total search frequency of user's history is N.
S32：Phase is determined according to frequency of occurrence, the first user historical search frequency, and the second user historical search frequency
The scoring of closing property.
In an embodiment of the present invention, can be according to frequency of occurrence Nxyi, first user historical search frequency Nxi, second use
Family historical search frequency Nyi, and preset formula determines relevance score Corr (i), wherein, preset formula is：
Corr (i)=Math.log10 (N/Nxi)*Math.log10(N/Nyi)*Nxyi/(Nxi+Nyi-Nxyi)；
By determining phase according to frequency of occurrence, the first user historical search frequency, and the second user historical search frequency
The scoring of closing property, scans in order to subsequently maximum to relevance score participle combination, and user can be made to express inspection freely
Rope demand so that phonetic search interactive experience is more intelligent, lifts search precision, lifts user's search experience degree.
S13：According to the relevance score of every kind of participle combination, combine and make from selection target participle in the combination of various participles
For the 3rd term.
In an embodiment of the present invention, the combination of relevance score highest participle can be selected from the combination of various participles to make
Combine for target participle, and target participle is combined as the 3rd term.
For example, when the Second Type information for questioning closely search that the first term is the second term, and the first term are：
" that France", the second term is：When " where the capital in the U.S. is ", every kind of participle in table 1 is obtained according to preset formula
As shown in table 2 (the first term is the place of the first kind information for supplementing search of the second term to the relevance score of combination
Reason process is similar to, and here is omitted).
Table 2
The combination of relevance score highest participle can be selected from the combination of various participles：The capital of France is where, to make
Combine for target participle, and target participle is combined as the 3rd term, to be scanned for according to the 3rd term.
S14：Scanned for according to the 3rd term.
Used as a kind of example, referring to Fig. 4, Fig. 4 is phonetic search interface schematic diagram in the embodiment of the present invention, works as user input
Speech data be " U.S. " (the second term) when, Search Results displayed page provide U.S.'s relevant information, when user wants to look into
When asking the demographic data in the U.S., only speech data " population is how many " (the first term) need to be directly inputted, with reference to information above, this
Embodiment can the term of automatic identification first be the second term supplement search first kind information, obtain the 3rd retrieval
Word is " U.S. population is how many ", and triggers Search Results, meets user's request.
In the same manner, user can continue to question closely " where capital is " (the first term), understand with reference to information above, the second inspection
Rope word is：" U.S. " and " population is how many ", the present embodiment can the term of automatic identification first be that the supplement of the second term is searched
The first kind information of rope, obtains the 3rd term for " where U.S. capital is ", displaying Search Results.
Further, in order to meet the colloquial style speech expression of user, user can also naturally question closely " that France
", with reference to information above, it is that the second term questions closely the second of search that the present embodiment can automatically identify the first term
Type information, obtains the 3rd term for " where the capital of France is ", displaying Search Results.
In the present embodiment, by the first participle result according to the first term and the second participle of each the second term
As a result various participle combinations are generated；According to the relevance score of every kind of participle combination, from selection target point in the combination of various participles
Phrase is incorporated as the 3rd term；Scanned for according to the 3rd term, can be entered with reference to the contextual information of phonetic search
Line search, lifts search precision, lifts user's search experience degree.
Fig. 5 is the structural representation of the searcher based on interactive voice that one embodiment of the invention is proposed.Language should be based on
The searcher 500 of sound interaction can pass through being implemented in combination in for software, hardware or both.
Referring to Fig. 5, should be based on the searcher 500 of interactive voice can include：First acquisition module 510, second is obtained
Module 520, generation module 530, selecting module 540, and search module 550.Wherein,
First acquisition module 510, the speech data for search for being provided according to user obtains the first term.
Alternatively, the first acquisition module 510 specifically for：The speech data for search that receive user is provided, obtains
The first text corresponding with speech data, and using corresponding first text as the first term.
Second acquisition module 520, for being retrieved as multiple second terms of the information above of the first term.
Generation module 530, for second point according to the first participle result of the first term and each the second term
Word result generates various participle combinations.
Selecting module 540, for the relevance score combined according to every kind of participle, from selection target in the combination of various participles
Participle group is incorporated as the 3rd term.
Alternatively, selecting module 540 specifically for：Relevance score highest participle group is selected from the combination of various participles
Cooperate to be combined for target participle, and target participle is combined as the 3rd term.
Search module 550, for being scanned for according to the 3rd term.
In some embodiments, referring to Fig. 6, should be based on the searcher 500 of interactive voice can also include：
Alternatively, generation module 530 includes：
Judging submodule 531, for judging historical search word in the presence or absence of comprising first participle result and the second participle
As a result initial participle combination.
First processes submodule 532, for there is the initial participle comprising first participle result and the second word segmentation result
During combination, initial participle is combined as the participle combination for generating.
Extracting sub-module 533, for there is no the initial participle group comprising first participle result and the second word segmentation result
During conjunction, extract the fisrt feature information of each first participle in first participle result, and extract in the second word segmentation result each the
The second feature information of two participles.
Alternatively, fisrt feature information/second feature information includes at least one of：
Point after entity type, weight, the user's history search frequency, the user's history search of the first participle/the second participle
Hit the frequency, general colloquial punishment weight, and editing distance.
Determination sub-module 534, for according to fisrt feature information and second feature information, determining the type of the first term
Information, wherein, type information includes：First term is the first kind information for supplementing search of the second term, the first inspection
Rope word is the Second Type information for questioning closely search of the second term.
Second processing submodule 535, for being entered the first participle and the second participle according to the type information of the first term
Row combined treatment, obtains various participle combinations.
First determining module 560, in determining that various participles are combined, every kind of participle is combined in user's search history
Frequency of occurrence, and the first user historical search frequency with the participle corresponding first participle of combination and corresponding second participle
The second user historical search frequency.
Second determining module 570, for being gone through according to frequency of occurrence, the first user historical search frequency, and second user
The history search frequency determines relevance score.
It should be noted that to the explanation of the searching method embodiment based on interactive voice in earlier figures 1- Fig. 4 embodiments
The searcher 500 based on interactive voice for being also applied for the embodiment is illustrated, it realizes that principle is similar to, and here is omitted.
In the present embodiment, by the first participle result according to the first term and the second participle of each the second term
As a result various participle combinations are generated；According to the relevance score of every kind of participle combination, from selection target point in the combination of various participles
Phrase is incorporated as the 3rd term；Scanned for according to the 3rd term, can be entered with reference to the contextual information of phonetic search
Line search, lifts search precision, lifts user's search experience degree.
It should be noted that in describing the invention, term " first ", " second " etc. are not only used for describing purpose, and not
It is understood that to indicate or implying relative importance.Additionally, in describing the invention, unless otherwise stated, the implication of " multiple "
It is two or more.
In flow chart or here any process described otherwise above or method description are construed as, expression includes
It is one or more for realizing specific logical function or process the step of the module of code of executable instruction, fragment or portion
Point, and the scope of the preferred embodiment of the present invention includes other realization, wherein can not press shown or discussion suitable
Sequence, including according to involved function by it is basic simultaneously in the way of or in the opposite order, carry out perform function, this should be of the invention
Embodiment person of ordinary skill in the field understood.
It should be appreciated that each several part of the present invention can be realized with hardware, software, firmware or combinations thereof.Above-mentioned
In embodiment, the software that multiple steps or method can in memory and by suitable instruction execution system be performed with storage
Or firmware is realizing.For example, if realized with hardware, and in another embodiment, can be with well known in the art
Any one of row technology or their combination are realizing：With for realizing the logic gates of logic function to data-signal
Discrete logic, the special IC with suitable combinational logic gate circuit, programmable gate array (PGA), scene
Programmable gate array (FPGA) etc..
Those skilled in the art are appreciated that to realize all or part of step that above-described embodiment method is carried
Suddenly the hardware that can be by program to instruct correlation is completed, and described program can be stored in a kind of computer-readable storage medium
In matter, the program upon execution, including one or a combination set of the step of embodiment of the method.
Additionally, each functional unit in each embodiment of the invention can be integrated in a processing module, it is also possible to
It is that unit is individually physically present, it is also possible to which two or more units are integrated in a module.Above-mentioned integrated mould
Block both can be realized in the form of hardware, it would however also be possible to employ the form of software function module is realized.The integrated module is such as
Fruit is realized and as independent production marketing or when using using in the form of software function module, it is also possible to be stored in a computer
In read/write memory medium.
Storage medium mentioned above can be read-only storage, disk or CD etc..
In the description of this specification, reference term " one embodiment ", " some embodiments ", " example ", " specifically show
The description of example " or " some examples " etc. means to combine specific features, structure, material or spy that the embodiment or example are described
Point is contained at least one embodiment of the present invention or example.In this manual, to the schematic representation of above-mentioned term not
Necessarily refer to identical embodiment or example.And, the specific features of description, structure, material or feature can be any
One or more embodiments or example in combine in an appropriate manner.
Although embodiments of the invention have been shown and described above, it is to be understood that above-described embodiment is example
Property, it is impossible to limitation of the present invention is interpreted as, one of ordinary skill in the art within the scope of the invention can be to above-mentioned
Embodiment is changed, changes, replacing and modification.
Claims (12)
1. a kind of searching method based on interactive voice, it is characterised in that comprise the following steps：
First term is obtained according to the speech data for search that user provides, and is retrieved as the upper of first term
Multiple second terms of literary information；
Various points are generated according to the first participle result of first term and the second word segmentation result of each the second term
Word combination；
According to the relevance score of every kind of participle combination, from selection target participle group in the various participles combination the is incorporated as
Three terms；
Scanned for according to the 3rd term.
2. the searching method of interactive voice is based on as claimed in claim 1, it is characterised in that described according to the described first retrieval
The first participle result of word and the second word segmentation result of each the second term generate various participle combinations, including：
Judge in historical search word with the presence or absence of the initial participle comprising the first participle result and second word segmentation result
Combination；
If there is the combination of the initial participle comprising the first participle result and second word segmentation result, will be described first
Beginning participle is combined as the participle combination for generating；
If there is no the combination of the initial participle comprising the first participle result and second word segmentation result, institute is extracted
The fisrt feature information of each first participle in first participle result is stated, and is extracted each second point in second word segmentation result
The second feature information of word；
According to the fisrt feature information and the second feature information, the type information of first term is determined, wherein,
The type information includes：First term is the first kind information, described for supplementing search of second term
First term is the Second Type information for questioning closely search of second term；
The first participle and second participle are combined by process according to the type information of first term, are obtained
Various participle combinations.
3. the searching method of interactive voice is based on as claimed in claim 2, it is characterised in that the fisrt feature information/institute
Second feature information is stated including at least one of：
After entity type, weight, the user's history search frequency, the user's history search of the first participle/second participle
The click frequency, general colloquial punishment weight, and editing distance.
4. the searching method of interactive voice is based on as claimed in claim 1, it is characterised in that described according to every kind of participle group
Before the relevance score of conjunction, also include：
In determining the various participles combination, frequency of occurrence of every kind of participle combination in user's search history, and with it is described
Participle combines the corresponding first user historical search frequency of the first participle and the second user history of corresponding second participle is searched
The rope frequency；
It is true according to the frequency of occurrence, the first user historical search frequency, and the second user historical search frequency
The fixed relevance score.
5. the searching method of interactive voice is based on as claimed in claim 1, it is characterised in that described from various participle groups
Selection target participle group is incorporated as the 3rd term in conjunction, including：
The relevance score highest participle combination is selected from various participle combinations as target participle combination,
And the target participle is combined as the 3rd term.
6. the searching method of interactive voice is based on as claimed in claim 1, it is characterised in that the use provided according to user
The first term is obtained in the speech data of search, including：
The speech data for search that receive user is provided, obtains corresponding with the speech data the first text, and by institute
Corresponding first text is stated as first term.
7. a kind of searcher based on interactive voice, it is characterised in that include：
First acquisition module, the speech data for search for being provided according to user obtains the first term；
Second acquisition module, for being retrieved as multiple second terms of the information above of first term；
Generation module, for being tied according to the second participle of the first participle result of first term and each the second term
Fruit generates various participle combinations；
Selecting module, for the relevance score combined according to every kind of participle, from selection target point in various participle combinations
Phrase is incorporated as the 3rd term；
Search module, for being scanned for according to the 3rd term.
8. the searcher of interactive voice is based on as claimed in claim 7, it is characterised in that the generation module includes：
Judging submodule, for judging historical search word in the presence or absence of comprising the first participle result and second participle
As a result initial participle combination；
First processes submodule, for exist it is described initial comprising the first participle result and second word segmentation result
When participle is combined, the initial participle is combined as the participle combination for generating；
Extracting sub-module, for there is no include the first participle result and second word segmentation result initial point
During word combination, the fisrt feature information of each first participle in the first participle result is extracted, and extract second participle
As a result the second feature information of each the second participle in；
Determination sub-module, for according to the fisrt feature information and the second feature information, determining first term
Type information, wherein, the type information includes：First term is that second term supplements the of search
One type information, first term are the Second Type information for questioning closely search of second term；
Second processing submodule, for according to the type information of first term by the first participle and described second point
Word is combined process, obtains various participle combinations.
9. the searcher of interactive voice is based on as claimed in claim 8, it is characterised in that the fisrt feature information/institute
Second feature information is stated including at least one of：
After entity type, weight, the user's history search frequency, the user's history search of the first participle/second participle
The click frequency, general colloquial punishment weight, and editing distance.
10. the searcher of interactive voice is based on as claimed in claim 7, it is characterised in that also included：
First determining module, in determining that various participles are combined, every kind of participle combines going out in user's search history
The existing frequency, and the first user historical search frequency and corresponding second participle with the participle corresponding first participle of combination
The second user historical search frequency；
Second determining module, for according to the frequency of occurrence, the first user historical search frequency, and described second uses
The family historical search frequency determines the relevance score.
11. searchers based on interactive voice as claimed in claim 7, it is characterised in that the selecting module is specifically used
In：
The relevance score highest participle combination is selected from various participle combinations as target participle combination,
And the target participle is combined as the 3rd term.
12. searchers based on interactive voice as claimed in claim 7, it is characterised in that the first acquisition module tool
Body is used for：
The speech data for search that receive user is provided, obtains corresponding with the speech data the first text, and by institute
Corresponding first text is stated as first term.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611019821.5A CN106653006B (en) | 2016-11-17 | 2016-11-17 | Searching method and device based on interactive voice |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611019821.5A CN106653006B (en) | 2016-11-17 | 2016-11-17 | Searching method and device based on interactive voice |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106653006A true CN106653006A (en) | 2017-05-10 |
| CN106653006B CN106653006B (en) | 2019-11-08 |
Family
ID=58807746
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201611019821.5A Active CN106653006B (en) | 2016-11-17 | 2016-11-17 | Searching method and device based on interactive voice |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106653006B (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107133345A (en) * | 2017-05-22 | 2017-09-05 | 北京百度网讯科技有限公司 | Exchange method and device based on artificial intelligence |
| CN107608957A (en) * | 2017-09-06 | 2018-01-19 | 百度在线网络技术（北京）有限公司 | Text modification method, apparatus and its equipment based on voice messaging |
| CN108538291A (en) * | 2018-04-11 | 2018-09-14 | 百度在线网络技术（北京）有限公司 | Sound control method, terminal device, cloud server and system |
| CN112259096A (en) * | 2020-10-23 | 2021-01-22 | 海信视像科技股份有限公司 | Voice data processing method and device |
Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101075435A (en) * | 2007-04-19 | 2007-11-21 | 深圳先进技术研究院 | Intelligent chatting system and its realizing method |
| CN101140587A (en) * | 2007-10-15 | 2008-03-12 | 深圳市迅雷网络技术有限公司 | Searching method and apparatus |
| CN101281745A (en) * | 2008-05-23 | 2008-10-08 | 深圳市北科瑞声科技有限公司 | Interactive system for vehicle-mounted voice |
| US20110144995A1 (en) * | 2009-12-15 | 2011-06-16 | At&T Intellectual Property I, L.P. | System and method for tightly coupling automatic speech recognition and search |
| CN103369398A (en) * | 2013-07-01 | 2013-10-23 | 安徽广电信息网络股份有限公司 | Voice searching method and voice searching system based on television EPG (electronic program guide) information |
| CN103995870A (en) * | 2014-05-21 | 2014-08-20 | 百度在线网络技术（北京）有限公司 | Interactive searching method and device |
| CN103995880A (en) * | 2014-05-27 | 2014-08-20 | 百度在线网络技术（北京）有限公司 | Interactive searching method and device |
| CN104102723A (en) * | 2014-07-21 | 2014-10-15 | 百度在线网络技术（北京）有限公司 | Search content providing method and search engine |
| CN104239459A (en) * | 2014-09-02 | 2014-12-24 | 百度在线网络技术（北京）有限公司 | Voice search method, voice search device and voice search system |
| CN104714954A (en) * | 2013-12-13 | 2015-06-17 | 中国电信股份有限公司 | Information searching method and system based on context understanding |
| CN105279227A (en) * | 2015-09-11 | 2016-01-27 | 百度在线网络技术（北京）有限公司 | Voice search processing method and device of homonym |
-
2016
- 2016-11-17 CN CN201611019821.5A patent/CN106653006B/en active Active
Patent Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101075435A (en) * | 2007-04-19 | 2007-11-21 | 深圳先进技术研究院 | Intelligent chatting system and its realizing method |
| CN101140587A (en) * | 2007-10-15 | 2008-03-12 | 深圳市迅雷网络技术有限公司 | Searching method and apparatus |
| CN101281745A (en) * | 2008-05-23 | 2008-10-08 | 深圳市北科瑞声科技有限公司 | Interactive system for vehicle-mounted voice |
| US20110144995A1 (en) * | 2009-12-15 | 2011-06-16 | At&T Intellectual Property I, L.P. | System and method for tightly coupling automatic speech recognition and search |
| CN103369398A (en) * | 2013-07-01 | 2013-10-23 | 安徽广电信息网络股份有限公司 | Voice searching method and voice searching system based on television EPG (electronic program guide) information |
| CN104714954A (en) * | 2013-12-13 | 2015-06-17 | 中国电信股份有限公司 | Information searching method and system based on context understanding |
| CN103995870A (en) * | 2014-05-21 | 2014-08-20 | 百度在线网络技术（北京）有限公司 | Interactive searching method and device |
| CN103995880A (en) * | 2014-05-27 | 2014-08-20 | 百度在线网络技术（北京）有限公司 | Interactive searching method and device |
| CN104102723A (en) * | 2014-07-21 | 2014-10-15 | 百度在线网络技术（北京）有限公司 | Search content providing method and search engine |
| CN104239459A (en) * | 2014-09-02 | 2014-12-24 | 百度在线网络技术（北京）有限公司 | Voice search method, voice search device and voice search system |
| CN105279227A (en) * | 2015-09-11 | 2016-01-27 | 百度在线网络技术（北京）有限公司 | Voice search processing method and device of homonym |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107133345A (en) * | 2017-05-22 | 2017-09-05 | 北京百度网讯科技有限公司 | Exchange method and device based on artificial intelligence |
| US10664504B2 (en) | 2017-05-22 | 2020-05-26 | Beijing Baidu Netcom Science And Technology Co., Ltd. | Interaction method and apparatus based on artificial intelligence |
| CN107133345B (en) * | 2017-05-22 | 2020-11-06 | 北京百度网讯科技有限公司 | Interaction method and device based on artificial intelligence |
| CN107608957A (en) * | 2017-09-06 | 2018-01-19 | 百度在线网络技术（北京）有限公司 | Text modification method, apparatus and its equipment based on voice messaging |
| CN108538291A (en) * | 2018-04-11 | 2018-09-14 | 百度在线网络技术（北京）有限公司 | Sound control method, terminal device, cloud server and system |
| US11127398B2 (en) | 2018-04-11 | 2021-09-21 | Baidu Online Network Technology (Beijing) Co., Ltd. | Method for voice controlling, terminal device, cloud server and system |
| CN112259096A (en) * | 2020-10-23 | 2021-01-22 | 海信视像科技股份有限公司 | Voice data processing method and device |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106653006B (en) | 2019-11-08 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| KR101581816B1 (en) | Voice recognition method using machine learning | |
| CN111191022B (en) | Commodity short header generation method and device | |
| EP2717176A2 (en) | Method for searching for information using the web and method for voice conversation using same | |
| CN102867512A (en) | Method and device for recognizing natural speech | |
| CN106653006B (en) | Searching method and device based on interactive voice | |
| US8583417B2 (en) | Translation device and computer program product | |
| KR20130082835A (en) | Method and appartus for providing contents about conversation | |
| CN102867511A (en) | Method and device for recognizing natural speech | |
| CN111414763A (en) | Semantic disambiguation method, device, equipment and storage device for sign language calculation | |
| Akhtiamov et al. | Speech and Text Analysis for Multimodal Addressee Detection in Human-Human-Computer Interaction. | |
| US20130173251A1 (en) | Electronic device and natural language analysis method thereof | |
| CN107748784A (en) | A kind of method that structured data searching is realized by natural language | |
| KR101410601B1 (en) | Spoken dialogue system using humor utterance and method thereof | |
| CN110808032A (en) | Voice recognition method and device, computer equipment and storage medium | |
| US20180039632A1 (en) | Information processing apparatus, information processing method, and computer program | |
| KR101695014B1 (en) | Method for building emotional lexical information and apparatus for the same | |
| CN108345612A (en) | A kind of question processing method and device, a kind of device for issue handling | |
| CN104182381A (en) | character input method and system | |
| CN108710653B (en) | On-demand method, device and system for reading book | |
| Henderson et al. | PolyResponse: A rank-based approach to task-oriented dialogue with application in restaurant search and booking | |
| CN102970618A (en) | Video on demand method based on syllable identification | |
| CN109063182B (en) | Content recommendation method based on voice search questions and electronic equipment | |
| CN113361252A (en) | Text depression tendency detection system based on multi-modal features and emotion dictionary | |
| CN112765977A (en) | Word segmentation method and device based on cross-language data enhancement | |
| CN109684357B (en) | Information processing method and device, storage medium and terminal |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |