CN108959257A - A kind of natural language analytic method, device, server and storage medium - Google Patents
A kind of natural language analytic method, device, server and storage medium Download PDFInfo
- Publication number
- CN108959257A CN108959257A CN201810713935.2A CN201810713935A CN108959257A CN 108959257 A CN108959257 A CN 108959257A CN 201810713935 A CN201810713935 A CN 201810713935A CN 108959257 A CN108959257 A CN 108959257A
- Authority
- CN
- China
- Prior art keywords
- concepts tab
- concepts
- intention
- sequence
- entity
- 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
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/279—Recognition of textual entities
- G06F40/289—Phrasal analysis, e.g. finite state techniques or chunking
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/30—Semantic analysis
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Machine Translation (AREA)
Abstract
The embodiment of the invention discloses a kind of natural language analytic method, device, server and storage mediums, wherein the described method includes: natural language text to be resolved is carried out word cutting, obtains corresponding multiple word cutting segments;Concept tagging is carried out to each word cutting segment, obtains at least one concepts tab;Permutation and combination is carried out at least one described concepts tab, obtains multiple concepts tab sequences;For each concepts tab sequence, it is intended to knowledge network in conjunction with the entity pre-established and carries out intention derivation, obtain the intention and slot position of the natural language text, wherein, the entity is intended to the entity in knowledge network including multiple and different business scenarios and is intended to set, and it is interrelated to pass through entity between the entity intention set of different business scene.For the embodiment of the present invention by being abstracted to entity, different entities can be multiplexed general intention knowledge, and different business scene need to only introduce the entity mobility models of needs, can quickly derive the true intention of related entities.
Description
Technical field
The present embodiments relate to natural language technical field more particularly to a kind of natural language analytic methods, device, clothes
Business device and storage medium.
Background technique
Natural language understanding technology is the data format that human language text is converted into computer and can identify and understand,
So that in different application systems, computer can provide service for the different demands of user.For example, being produced in human-computer interaction
In product, computer needs the natural language for inputting user to be converted into the data of structuring, and then determines what user was intended by
True intention, for subsequent specific business logic processing.
In the prior art, the corresponding text of natural language usually inputted using simple template matching method from user
The intention of user is identified in information.However, template is according to application scenarios or the business field that specifically carry out natural language understanding
Scape is preset, and each scene requires independently to write corresponding template in advance, cannot be multiplexed between template, then increase exploitation
Cost.
Summary of the invention
The embodiment of the invention provides a kind of natural language analytic method, device, server and storage mediums, existing to solve
Mould corresponding with the application scenarios or business scenario that need to carry out natural language understanding must independently be write by having present in technology
Plate, and the technical issues of cannot be multiplexed between each template.
In a first aspect, the embodiment of the invention provides a kind of natural language analytic methods, comprising:
Natural language text to be resolved is subjected to word cutting, obtains corresponding multiple word cutting segments;
Concept tagging is carried out to each word cutting segment, obtains at least one concepts tab, wherein the concepts tab is used for
Map the abstract concept of word cutting segment;
Permutation and combination is carried out at least one described concepts tab, obtains multiple concepts tab sequences, wherein different
Between concepts tab sequence, the word cutting boundary that the concepts tab in each concepts tab sequence is covered is not overlapped;
For each concepts tab sequence, it is intended to knowledge network in conjunction with the entity pre-established and carries out intention derivation, obtain
The intention and slot position of the natural language text, wherein the entity is intended to include multiple and different business scenarios in knowledge network
Entity be intended to set, and different business scene entity be intended to set between by entity it is interrelated.
Second aspect, the embodiment of the invention also provides a kind of natural language resolvers, comprising:
Word cutting module obtains corresponding multiple word cutting segments for natural language text to be resolved to be carried out word cutting;
Concept tagging module, for obtaining at least one concepts tab to each word cutting segment progress concept tagging,
In, the concepts tab is used to map the abstract concept of word cutting segment;
Composite module, for obtaining multiple concepts tab sequences at least one described concepts tab progress permutation and combination,
Wherein, between different concepts tab sequences, the word cutting boundary that the concepts tab in each concepts tab sequence is covered is not weighed
It closes;
It is intended to derivation module, for being directed to each concepts tab sequence, is intended to knowledge network in conjunction with the entity pre-established
Intention derivation is carried out, the intention and slot position of the natural language text are obtained, wherein includes in the entity intention knowledge network
The entity of multiple and different business scenarios is intended to set, and the entity of different business scene is intended to mutually close between set by entity
Connection.
The third aspect, the embodiment of the invention also provides a kind of servers, comprising:
One or more processors;
Memory, for storing one or more programs,
When one or more of programs are executed by one or more of processors, so that one or more of processing
Device realizes the natural language analytic method as described in any embodiment of the present invention.
Fourth aspect, the embodiment of the invention also provides a kind of computer readable storage mediums, are stored thereon with computer
Program realizes the natural language analytic method as described in any embodiment of the present invention when the program is executed by processor.
A kind of natural language analytic method, device, server and storage medium provided in an embodiment of the present invention, by treating
The natural language text of parsing carries out word cutting, and stamps concepts tab for the word cutting segment obtained, and the concepts tab of acquisition is arranged
Column are combined into multiple concepts tab sequences, are intended to knowledge network based on the entity that pre-establishes, to each concepts tab sequence into
Row is intended to derive, and obtains the corresponding true intention of the natural language text and slot position.It is abstracted from there through to entity, no
Same entity can be multiplexed general intention knowledge, the entity mobility models that each specific business scenario can be needed by introducing,
Quickly derive the true intention of related entities.
Detailed description of the invention
Fig. 1 is a kind of flow diagram for natural language analytic method that the embodiment of the present invention one provides;
Fig. 2 is a kind of flow diagram of natural language analytic method provided by Embodiment 2 of the present invention;
Fig. 3 is a kind of flow diagram for natural language analytic method that the embodiment of the present invention three provides;
Fig. 4 is a kind of structural schematic diagram for natural language resolver that the embodiment of the present invention four provides;
Fig. 5 is a kind of structural schematic diagram for server that the embodiment of the present invention five provides.
Specific embodiment
The present invention is described in further detail with reference to the accompanying drawings and examples.It is understood that this place is retouched
The specific embodiment stated is used only for explaining the present invention rather than limiting the invention.It also should be noted that in order to just
Only the parts related to the present invention are shown in description, attached drawing rather than entire infrastructure.
Embodiment one
Fig. 1 is a kind of flow chart for natural language analytic method that the embodiment of the present invention one provides, and the present embodiment is applicable
In need according to natural language understanding user's true intention of user the case where, this method can be parsed by corresponding natural language
Device executes, which can be realized by the way of software and/or hardware, and is configured on server.Such as Fig. 1 institute
Show, the natural language analytic method provided in the embodiment of the present invention may include:
S110, natural language text to be resolved is subjected to word cutting, obtains corresponding multiple word cutting segments.
In some man-machine interactive products, computer needs the natural language for inputting user to be converted into the number of structuring
According to, and then determine the true intention that user is intended by, for subsequent specific business logic processing.Therefore it needs defeated to user
What is entered is pre-processed from speech speech.
In the present embodiment, carrying out pretreatment for the natural language of user's input includes: to be incited somebody to action based on speech recognition technology
Natural language is converted into corresponding natural language text;Remove the punctuation mark for including in the natural language text, and to text
Letter in this carries out capital and small letter conversion, and capitalization all in text is such as converted into lowercase.Thus to obtain wait solve
The natural language text of analysis.
Word cutting is carried out to natural language text to be resolved, any word cutting algorithm in the prior art is can use and carries out
Word cutting obtains corresponding multiple word cutting segments.For example, by n-gram word cutting algorithm to natural language text to be resolved " bar
The goalkeeper of western team is how high " word cutting is carried out, obtained word cutting segment is respectively " Brazilian team ", " goalkeeper ", " how high ".
S120, concept tagging is carried out to each word cutting segment, obtains at least one concepts tab, wherein the concept mark
Sign the abstract concept for mapping word cutting segment.
In the present embodiment, conceptual analysis is carried out respectively to the multiple word cutting segments obtained in S110, if can not identify
Wherein some corresponding concept of word cutting segment, then skip the word cutting segment, continues to identify next word cutting segment, thus can know
Not Chu at least one corresponding concept of word cutting segment, and concept tagging is carried out to the word cutting segment that has identified, as identified
Participle fragment label concepts tab, since a word cutting segment may correspond to multiple concepts, final available at least one
A concepts tab, wherein the concepts tab is used to map the abstract concept of word cutting segment.Illustratively, word cutting segment " Brazil
The corresponding abstract concept of team " is team, therefore the corresponding concepts tab of word cutting segment " Brazilian team " is that " team " similarly " keeps
The corresponding concepts tab of door member " is " position ", and " how high " corresponding concepts tab is " height " and " integral ".
S130, permutation and combination is carried out at least one described concepts tab, obtains multiple concepts tab sequences, wherein
Between different concepts tab sequences, the word cutting boundary that the concepts tab in each concepts tab sequence is covered is not overlapped.
In the present embodiment, permutation and combination is carried out to all concepts tabs that S120 is obtained, obtains multiple do not conflict generally
Sequence label is read, i.e., each word cutting segment there can only be a concepts tab in the same sequence, for example, word cutting segment " has more
The corresponding concepts tab of height " is " height " and " integral " two concepts tabs, it is therefore desirable to assign to the two concepts tabs not
In same concepts tab sequence.Illustratively, word cutting segment " Brazilian team ", " goalkeeper " and " how high " corresponding concepts tab
Sequence is respectively " team, position, height " and " team, position, integral ".
S140, it is directed to each concepts tab sequence, is intended to knowledge network in conjunction with the entity pre-established and carries out intention derivation,
Obtain the intention and slot position of the natural language text, wherein the entity is intended to include multiple and different business in knowledge network
The entity of scene is intended to set, and interrelated by entity between the entity intention set of different business scene.
In the present embodiment, it is that the entity of real world is passed through one kind that the entity pre-established, which is intended to knowledge network,
To the description language of people close friend, the entity for mapping generation is intended to knowledge network.Specifically, entity is intended in knowledge network include more
The entity of a different business scene is intended to set, and the entity of different business scene is intended to mutually close between set by entity
Connection.Wherein, each entity is intended to intention relevant to entity, that may be present and which concepts tab can be recorded in set
It can be relevant with these intentions.Illustratively, entity this kind of for personage has inquiry personage's gender, age, height, weight etc.
It is intended to.Under football match scene, sportsman is also the entity of figure kind a kind of, necessarily have inquiry sportsman's gender, the age, height,
Weight etc. is intended to, meanwhile, this entity of sportsman also includes the meanings such as distinctive inquiry sportsman uniform number, inquiry sportsman's heavy foot
Figure.Similarly, for this entity of performer, inquiry performer's gender, age, height, weight etc. are intended to, this entity of performer also wraps
The intention of awards is represented containing peculiar inquiry performer's masterpiece or acquisition.People can be passed through for sportsman and the distinctive entity intention of performer
This entity of object is associated, and the two entities can directly be multiplexed the general intention knowledge of people entities.
Therefore, only the concepts tab in each concepts tab sequence need to be intended to knowledge network with entity respectively to match
The true intention and slot position of user can be derived.
In the present embodiment, it by carrying out word cutting to natural language text to be resolved, and is beaten for the word cutting segment obtained
Upper concepts tab is intended to by the concepts tab permutation and combination of acquisition at multiple concepts tab sequences based on the entity pre-established
Knowledge network carries out intention derivation to each concepts tab sequence, obtains the true intention and slot position of the natural language text.
It is abstracted from there through to entity, different entities can be multiplexed general intention knowledge, and each specific business scenario can
Quickly to derive the true intention of related entities by introducing the entity mobility models needed.
Embodiment two
Fig. 2 is a kind of flow diagram of natural language analytic method provided by Embodiment 2 of the present invention.The present embodiment with
It is optimized based on above-described embodiment, as shown in Fig. 2, the natural language analytic method provided in the embodiment of the present invention can wrap
It includes:
S210, natural language text to be resolved is subjected to word cutting, obtains corresponding multiple word cutting segments.
S220, by the combination of adjacent word cutting segment in each word cutting segment and the multiple word cutting segment, and build in advance
Vertical knowledge vocabulary is matched.
In the present embodiment, after obtaining multiple word cutting segments, group is carried out according to the solidification degree between adjacent participle segment
It closes, obtains different portmanteau words, for example, " Beijing " and " road " is two participle segments, " Beijing Road " after combination is used as one
Portmanteau word matches obtained multiple word cutting segments and different portmanteau words with the knowledge vocabulary pre-established respectively.Its
In, knowledge vocabulary includes multiple concepts tabs and the corresponding phrase of each concepts tab, illustratively, in knowledge vocabulary
A concepts tab be " team ", " team " following corresponding phrase includes the balls such as " China Team ", " British team ", " Germany "
The title of team.
If in S230, the knowledge vocabulary, there are matched concepts tabs, using the matched concepts tab as institute
State at least one concepts tab.
Illustratively, if a certain word cutting segment is " China Team ", by the matching result with knowledge vocabulary it is found that word cutting
The corresponding concepts tab of segment " China Team " is " team ".Further for example, " Beijing " corresponding concepts tab in knowledge vocabulary is
" city (city) ", matching result is not present in " road " in knowledge vocabulary, then to " road ", this word cutting segment is not labeled,
And if there is also " Beijing Road " this words in knowledge vocabulary, and it corresponds to " road (road) " this concepts tab, that
, can also to " Beijing Road " on the composite marking of this adjacent word cutting segment " road " concepts tab.It thus can will be in knowledge
There are the combinations of all word cutting segments of matched concepts tab and adjacent word cutting segment to be all labeled in vocabulary.
S240, permutation and combination is carried out at least one described concepts tab, obtains multiple concepts tab sequences, wherein
Between different concepts tab sequences, the word cutting boundary that the concepts tab in each concepts tab sequence is covered is not overlapped.
In the present embodiment, same general due to cannot all put the corresponding multiple concepts tabs of the same word cutting segment into
It reads in sequence label, therefore, the word cutting boundary that the concepts tab in each concepts tab sequence is covered is not overlapped.That is, such as
Two fruit some word cutting segment corresponding A, B concepts tabs, then the two concepts tabs A and B will not appear in simultaneously it is same
In a concepts tab sequence.For another example for some complicated cases, the concepts tab in a certain concepts tab sequence includes
" city " and " road ", respectively corresponds the combination " Beijing Road " of word cutting segment " Beijing " and adjacent word cutting segment, and " Beijing "
" Beijing Road " the two words word cutting boundary " north " having the same, therefore, " city " and " road " the two concepts tabs
It will not appear in simultaneously in same concept sequence label.According to mentioned above principle, at least one described concepts tab is arranged
Column combination, obtains multiple concepts tab sequences.
S250, for multiple concepts tabs in each concepts tab sequence, according to identical as the natural language text
Order of representation, successively by single concepts tab and adjacent multiple concepts tabs, respectively with the entity be intended to knowledge network
It is matched.
Illustratively, on the basis of example 1, natural language text " goalkeeper of Brazilian team is how high " is by cutting
Corresponding concepts tab sequence is respectively " team, position, height " and " team, position, integral " after word and concept tagging.It will
" team, position, height " is intended to knowledge network with entity and is matched, to derive the corresponding user's meaning of the concepts tab sequence
Figure;" team, position, integral " is intended to knowledge network with entity again to match, it is corresponding to derive the concepts tab sequence
User be intended to.
What needs to be explained here is that according to order of representation identical with natural language text, it is therefore an objective to correctly derive user
It is intended to, because the communicative habits of Human Natural Language are not met if the sequence of concepts tab is upset, naturally, to random ordering
Concepts tab carry out derive will not derive correctly be intended to result.
If S260, being matched to intention, the intention is obtained in the entity and is intended to corresponding entity in knowledge network, it will
It the corresponding concepts tab of the entity and not yet carries out matched concepts tab and merges into new concepts tab sequence, and is new to this
Concepts tab sequence executes above-mentioned matching operation, finish until all financial resourcess concept tag match in current concepts sequence label or
Until person's matching is less than intention.
Illustratively, for first concepts tab sequence " team, position, height ", according to concepts tab " team " and
" position " and entity are intended to the matching result of knowledge network, determine that user has " inquiry sportsman " to be intended to, then obtain the intention in reality
Body is intended to corresponding entity, i.e. sportsman in knowledge network, and returns to the corresponding concepts tab of sportsman " sportsman ";According to concepts tab
" sportsman " derives that user has " inquiry personage " to be intended to, then obtains the intention in entity and be intended to corresponding entity in knowledge network,
That is personage, and return to the corresponding concepts tab of entity personage " personage " by concepts tab " personage " and not yet carries out matched general
It reads label " height " and merges into new concepts tab sequence, and be intended to knowledge network with entity and matched, obtain " inquiry personage
Height " is intended to.Concepts tab all uses at this time, can not continue to derive, it is determined that is finally intended to " inquiry personage's height " meaning
Figure.
Illustratively, for second concepts tab sequence " team, position, integral ", according to concepts tab " team " and
The process that " position " obtains user's intention is identical as the derivation process according to first concepts tab sequence, and difference is, is pushing away
After export user has " inquiry personage " to be intended to and return to concepts tab " personage ", by concepts tab " personage " and not yet match
Concepts tab " integral " merge into new concepts tab sequence, and continue to be matched with entity intention knowledge network, as a result
Less than intention, then " inquiry personage is intended to " obtained previous step returns for matching, i.e., what is finally determined is intended to " inquiry personage's meaning
Figure ".
S270, by being all intended to of obtaining of matching in above-mentioned matching process, according to concept corresponding when being matched to intention
Coverage rate of the label in current concepts sequence label determines final intention, and using remaining concepts tab as slot position.
Illustratively, the coverage rate is to derive user to be intended to word cutting segment corresponding to the concepts tab actually used
The quotient of total number of word and natural language text total number of word to be resolved, coverage rate is higher, and it is more accurate that determining user is intended to.For example, the
In one concepts tab sequence, three concepts tabs are all used during deriving user and being intended to, and three concepts tabs pair
The word cutting segment " Brazilian team " answered, " goalkeeper ", " how high " totally 9 words, natural language text has 10 words, therefore coverage rate is
9/10=0.9;Similarly, it is general only to have used " team " and " position " two when being matched to intention for second concepts tab sequence
Label is read, therefore coverage rate is 6/10=0.6.
Therefore what is finally determined is intended to be intended to according to the user that first concepts tab sequence determines, i.e., finally determines
It is intended to " inquiry personage is intended to ", slot position title is respectively " team ", " position ", and corresponding slot position value is respectively Brazilian team, keeps
Men Yuan.
In the present embodiment, adjacent word cutting segment is subjected to group after carrying out word cutting to natural language text to be resolved
Close, obtain portmanteau word, respectively word cutting segment and portmanteau word stamps concepts tab, and by the concepts tab permutation and combination of acquisition at
Multiple concepts tab sequences are intended to knowledge network based on the entity that pre-establishes and are matched, and according to when being matched to intention pairs
Coverage rate of the concepts tab answered in current concepts sequence label determines final be intended to.It is abstracted from there through to entity,
Different entities can be multiplexed general intention knowledge, and each specific business scenario can be known by introducing the entity needed
Know, quickly derives the true intention of related entities.
Embodiment three
Fig. 3 is a kind of flow diagram for natural language analytic method that the embodiment of the present invention three provides.The present embodiment with
It is optimized based on above-described embodiment, as shown in figure 3, the natural language analytic method provided in the embodiment of the present invention can wrap
It includes:
S310, natural language text to be resolved is subjected to word cutting, obtains corresponding multiple word cutting segments.
S320, concept tagging is carried out to each word cutting segment, obtains at least one concepts tab, wherein the concept mark
Sign the abstract concept for mapping word cutting segment.
S330, permutation and combination is carried out at least one described concepts tab, obtains multiple concepts tab sequences, wherein
Between different concepts tab sequences, the word cutting boundary that the concepts tab in each concepts tab sequence is covered is not overlapped.
S340, to each concepts tab sequence, according to concepts tab coverage rate, concepts tab number and concepts tab point
Value is ranked up, and from multiple concepts tab sequences after sequence, chooses the N number of concepts tab sequence for coming front, wherein general
The priority for reading tag coverage, concepts tab number and concepts tab score value successively reduces, and N is positive integer.
In the present embodiment, for multiple concepts tab sequences of acquisition, need to screen it, it illustratively, can be according to
It is ranked up according to the multiple concepts tab sequences of concepts tab coverage rate, concepts tab number and concepts tab score value to acquisition,
Therefrom select the forward N number of sequence that sorts.Wherein, the highest priority of concepts tab coverage rate, therefore preferentially pass through concept mark
It signs coverage rate and determines sequence, the concepts tab in concepts tab coverage rate, that is, sequence is corresponding in natural language text to be resolved
The quotient of the total number of word of the number of words and natural language text of word cutting segment, coverage rate is higher, and sequence is more forward.When coverage rate is identical,
Compare concepts tab number again, i.e., the number of concepts tab is more in sequence, sorts more forward.If concepts tab number also phase
Together, then compare concepts tab score value again, wherein concepts tab score value can be recorded in previously according to the statistical result of historical data
In knowledge vocabulary, therefore, when carrying out concept tagging, available to arrive its corresponding score value, the score value the high, sorts more forward.
It is known that the to be resolved natural language text more for number of words passes through arrangement after marking concepts tab
It combines there are many kinds of available concepts tab sequence number possibility, if be all intended to each concepts tab sequence
It derives, efficiency will necessarily be very low, and can then exclude some sequences for being intended to inaccuracy by sequence, in front to sequence
Some sequences carry out intention derivation, then optimal intention derivation result is therefrom selected, to improve efficiency and system resource
Utilization rate.
S350, to each of N number of concepts tab sequence concepts tab sequence, in conjunction with the entity meaning pre-established
Figure knowledge network carries out intention derivation, obtains the intention and slot position of the natural language text.
To each of N number of concepts tab sequence concepts tab sequence, respectively with entity be intended to knowledge network into
Row matching is determined final according to coverage rate of the concepts tab corresponding when being matched to intention in current concepts sequence label
Intention.
In the present embodiment, according to concepts tab coverage rate, concepts tab number and concepts tab score value to the more of acquisition
A concepts tab sequence is ranked up, and is therefrom selected the forward N number of sequence that sorts and is intended to derive to user, i.e., is first sieved
Choosing carries out intention derivation again, can be promoted and derive the efficiency that user is intended to, while improve the accuracy of derivation.
Example IV
Fig. 4 is a kind of structural schematic diagram for natural language resolver that the embodiment of the present invention four provides.As shown in figure 4,
The device includes:
Word cutting module 410 obtains corresponding multiple word cutting pieces for natural language text to be resolved to be carried out word cutting
Section;
Concept tagging module 420, for obtaining at least one concepts tab to each word cutting segment progress concept tagging,
Wherein, the concepts tab is used to map the abstract concept of word cutting segment;
Composite module 430 obtains multiple concepts tab sequences for carrying out permutation and combination at least one described concepts tab
Column, wherein between different concepts tab sequences, the word cutting boundary that the concepts tab in each concepts tab sequence is covered is not
It is overlapped;
It is intended to derivation module 440, for being directed to each concepts tab sequence, is intended to knowledge knowledge network in conjunction with the entity pre-established
Network carries out intention derivation, obtains the intention and slot position of the natural language text, wherein the entity is intended to wrap in knowledge network
The entity for including multiple and different business scenarios is intended to set, and mutual by entity between the entity intention set of different business scene
Association.
Client provided in this embodiment carries out concept to the word cutting segment that word cutting module obtains by concept tagging module
Mark, and by composite module product concept sequence label, the concepts tab sequence of acquisition is carried out by being intended to derivation module
It is intended to derive, obtains the practical intention of user.
On the basis of the various embodiments described above, the concept tagging module is specifically used for:
By the combination of adjacent word cutting segment in each word cutting segment and the multiple word cutting segment, and pre-establish
Knowledge vocabulary is matched;
If there are matched concepts tabs in the knowledge vocabulary, as at least one described concepts tab.
On the basis of the various embodiments described above, the intention derivation module is specifically used for:
For multiple concepts tabs in each concepts tab sequence, according to expression identical with the natural language text
Sequentially, successively by single concepts tab and adjacent multiple concepts tabs, it is intended to knowledge network progress with the entity respectively
Match;
If being matched to intention, the intention is obtained in the entity and is intended to corresponding entity in knowledge network, by the reality
It the corresponding concepts tab of body and not yet carries out matched concepts tab and merges into new concepts tab sequence, and the concept new to this
Sequence label executes above-mentioned matching operation, until all financial resourcess concept tag match in current concepts sequence label finish or
Until less than intention;
By in above-mentioned matching process matching obtain be all intended to, exist according to concepts tab corresponding when being matched to intention
Coverage rate in current concepts sequence label determines final intention, and using remaining concepts tab as slot position.
On the basis of the various embodiments described above, described device further include:
Sequence obtains module, for each concepts tab sequence, foundation concepts tab coverage rate, concepts tab number
It is ranked up with concepts tab score value, from multiple concepts tab sequences after sequence, chooses the N number of concepts tab for coming front
Sequence, wherein the priority of concepts tab coverage rate, concepts tab number and concepts tab score value successively reduces, and N is positive whole
Number;
Correspondingly, the intention derivation module is used for:
To each of N number of concepts tab sequence concepts tab sequence, know in conjunction with the entity intention pre-established
Know network and carry out intention derivation, obtains the intention and slot position of the natural language text.
On the basis of the various embodiments described above, described device further include:
Preprocessing module, for being pre-processed to the natural language text to be resolved, wherein the pretreatment is extremely
Few includes removing punctuation mark, capital and small letter conversion.
Natural language resolver provided by the embodiment of the present invention can be performed it is provided by any embodiment of the invention from
Right language analytic method, has the corresponding functional module of execution method and beneficial effect.
Embodiment five
Fig. 5 is the structural schematic diagram for the server that the embodiment of the present invention five provides.Fig. 5, which is shown, to be suitable for being used to realizing this hair
The block diagram of the exemplary servers 12 of bright embodiment.The server 12 that Fig. 5 is shown is only an example, should not be to the present invention
The function and use scope of embodiment bring any restrictions.
As shown in figure 5, server 12 is showed in the form of universal computing device.The component of server 12 may include but not
Be limited to: one or more processor or processing unit 16, memory 28 connect different system components (including memory 28
With processing unit 16) bus 18.
Bus 18 indicates one of a few class bus structures or a variety of, including memory bus or Memory Controller,
Peripheral bus, graphics acceleration port, processor or the local bus using any bus structures in a variety of bus structures.It lifts
For example, these architectures include but is not limited to industry standard architecture (ISA) bus, microchannel architecture (MAC)
Bus, enhanced isa bus, Video Electronics Standards Association (VESA) local bus and peripheral component interconnection (PCI) bus.
Server 12 typically comprises a variety of computer system readable media.These media can be and any can be serviced
The usable medium that device 12 accesses, including volatile and non-volatile media, moveable and immovable medium.
Memory 28 may include the computer system readable media of form of volatile memory, such as random access memory
Device (RAM) 30 and/or cache memory 32.Server 12 may further include other removable/nonremovable, easy
The property lost/nonvolatile computer system storage medium.Only as an example, storage system 34 can be used for reading and writing it is immovable,
Non-volatile magnetic media (Fig. 5 do not show, commonly referred to as " hard disk drive ").Although being not shown in Fig. 5, can provide for pair
The disc driver of removable non-volatile magnetic disk (such as " floppy disk ") read-write, and to removable anonvolatile optical disk (such as
CD-ROM, DVD-ROM or other optical mediums) read-write CD drive.In these cases, each driver can pass through
One or more data media interfaces is connected with bus 18.Memory 28 may include at least one program product, the program
Product has one group of (for example, at least one) program module, these program modules are configured to perform the function of various embodiments of the present invention
Energy.
Program/utility 40 with one group of (at least one) program module 42 can store in such as memory 28
In, such program module 42 include but is not limited to operating system, one or more application program, other program modules and
It may include the realization of network environment in program data, each of these examples or certain combination.Program module 42 is usual
Execute the function and/or method in embodiment described in the invention.
Server 12 can also be logical with one or more external equipments 14 (such as keyboard, sensing equipment, display 24 etc.)
Letter, can also be enabled a user to one or more equipment interact with the server 12 communicate, and/or with make the server
The 12 any equipment (such as network interface card, modem etc.) that can be communicated with one or more of the other calculating equipment communicate.
This communication can be carried out by input/output (I/O) interface 22.Also, server 12 can also pass through network adapter 20
With one or more network (such as local area network (LAN), wide area network (WAN) and/or public network, such as internet) communication.
As shown, network adapter 20 is communicated by bus 18 with other modules of server 12.It should be understood that although not showing in figure
Out, can in conjunction with server 12 use other hardware and/or software module, including but not limited to: microcode, device driver,
Redundant processing unit, external disk drive array, RAID system, tape drive and data backup storage system etc..
The program that processing unit 16 is stored in memory 28 by operation, thereby executing various function application and data
Processing, such as realize natural language analytic method provided by the embodiment of the present invention, comprising:
Natural language text to be resolved is subjected to word cutting, obtains corresponding multiple word cutting segments;
Concept tagging is carried out to each word cutting segment, obtains at least one concepts tab, wherein the concepts tab is used for
Map the abstract concept of word cutting segment;
Permutation and combination is carried out at least one described concepts tab, obtains multiple concepts tab sequences, wherein different
Between concepts tab sequence, the word cutting boundary that the concepts tab in each concepts tab sequence is covered is not overlapped;
For each concepts tab sequence, it is intended to knowledge network in conjunction with the entity pre-established and carries out intention derivation, obtain
The intention and slot position of the natural language text, wherein the entity is intended to include multiple and different business scenarios in knowledge network
Entity be intended to set, and different business scene entity be intended to set between by entity it is interrelated.
Embodiment six
A kind of storage medium comprising computer executable instructions is provided in the embodiment of the present invention, the computer is executable
Instruction by computer processor when being executed for executing a kind of natural language analytic method applied to terminal, this method packet
It includes:
Natural language text to be resolved is subjected to word cutting, obtains corresponding multiple word cutting segments;
Concept tagging is carried out to each word cutting segment, obtains at least one concepts tab, wherein the concepts tab is used for
Map the abstract concept of word cutting segment;
Permutation and combination is carried out at least one described concepts tab, obtains multiple concepts tab sequences, wherein different
Between concepts tab sequence, the word cutting boundary that the concepts tab in each concepts tab sequence is covered is not overlapped;
For each concepts tab sequence, it is intended to knowledge network in conjunction with the entity pre-established and carries out intention derivation, obtain
The intention and slot position of the natural language text, wherein the entity is intended to include multiple and different business scenarios in knowledge network
Entity be intended to set, and different business scene entity be intended to set between by entity it is interrelated.
Certainly, a kind of storage medium comprising computer executable instructions provided in the embodiment of the present invention calculates
The method operation that machine executable instruction is not limited to the described above, can also be performed application provided in any embodiment of that present invention
Relevant operation in the text playback method of terminal.
The computer storage medium of the embodiment of the present invention, can be using any of one or more computer-readable media
Combination.Computer-readable medium can be computer-readable signal media or computer readable storage medium.It is computer-readable
Storage medium for example may be-but not limited to-the system of electricity, magnetic, optical, electromagnetic, infrared ray or semiconductor, device or
Device, or any above combination.The more specific example (non exhaustive list) of computer readable storage medium includes: tool
There are electrical connection, the portable computer diskette, hard disk, random access memory (RAM), read-only memory of one or more conducting wires
(ROM), erasable programmable read only memory (EPROM or flash memory), optical fiber, portable compact disc read-only memory (CD-
ROM), light storage device, magnetic memory device or above-mentioned any appropriate combination.In this document, computer-readable storage
Medium can be any tangible medium for including or store program, which can be commanded execution system, device or device
Using or it is in connection.
Computer-readable signal media may include in a base band or as carrier wave a part propagate data-signal,
Wherein carry computer-readable program code.The data-signal of this propagation can take various forms, including but unlimited
In electromagnetic signal, optical signal or above-mentioned any appropriate combination.Computer-readable signal media can also be that computer can
Any computer-readable medium other than storage medium is read, which can send, propagates or transmit and be used for
By the use of instruction execution system, device or device or program in connection.
The program code for including on computer-readable medium can transmit with any suitable medium, including --- but it is unlimited
In wireless, electric wire, optical cable, RF etc. or above-mentioned any appropriate combination.
The computer for executing operation of the present invention can be write with one or more programming languages or combinations thereof
Program code, described program design language include object oriented program language-such as Java, Smalltalk, C++,
It further include conventional procedural programming language-such as " C " language or similar programming language.Program code can be with
It fully executes, partly execute on the user computer on the user computer, being executed as an independent software package, portion
Divide and partially executes or executed on a remote computer or server completely on the remote computer on the user computer.?
Be related in the situation of remote computer, remote computer can pass through the network of any kind --- including local area network (LAN) or
Wide area network (WAN)-be connected to subscriber computer, or, it may be connected to outer computer (such as mentioned using Internet service
It is connected for quotient by internet).Note that the above is only a better embodiment of the present invention and the applied technical principle.This field
Technical staff is able to carry out for a person skilled in the art it will be appreciated that the invention is not limited to the specific embodiments described herein
Various apparent variations are readjusted and are substituted without departing from protection scope of the present invention.Therefore, although by implementing above
Example is described in further detail the present invention, but the present invention is not limited to the above embodiments only, is not departing from the present invention
It can also include more other equivalent embodiments in the case where design, and the scope of the present invention is by scope of the appended claims
It determines.
Claims (12)
1. a kind of natural language analytic method, which is characterized in that the described method includes:
Natural language text to be resolved is subjected to word cutting, obtains corresponding multiple word cutting segments;
Concept tagging is carried out to each word cutting segment, obtains at least one concepts tab, wherein the concepts tab is for mapping
The abstract concept of word cutting segment;
Permutation and combination is carried out at least one described concepts tab, obtains multiple concepts tab sequences, wherein in different concepts
Between sequence label, the word cutting boundary that the concepts tab in each concepts tab sequence is covered is not overlapped;
For each concepts tab sequence, it is intended to knowledge network in conjunction with the entity pre-established and carries out intention derivation, obtains described
The intention and slot position of natural language text, wherein the entity is intended to the reality in knowledge network including multiple and different business scenarios
Body is intended to set, and interrelated by entity between the entity intention set of different business scene.
2. being obtained at least the method according to claim 1, wherein carrying out concept tagging to each word cutting segment
One concepts tab, comprising:
By the combination of adjacent word cutting segment in each word cutting segment and the multiple word cutting segment, with the knowledge pre-established
Vocabulary is matched;
If there are matched concepts tab in the knowledge vocabulary, using the matched concepts tab as it is described at least one
Concepts tab.
3. the method according to claim 1, wherein described be directed to each concepts tab sequence, in conjunction with building in advance
Vertical entity is intended to knowledge network and carries out intention derivation, obtains the intention and slot position of the natural language text, comprising:
It is suitable according to expression identical with the natural language text for multiple concepts tabs in each concepts tab sequence
Sequence is intended to knowledge network with the entity respectively and is matched successively by single concepts tab and adjacent multiple concepts tabs;
If being matched to intention, the intention is obtained in the entity and is intended to corresponding entity in knowledge network, by the entity pair
It the concepts tab answered and not yet carries out matched concepts tab and merges into new concepts tab sequence, and the concepts tab new to this
Sequence executes above-mentioned matching operation, until all financial resourcess concept tag match in current concepts sequence label finishes or matches not
Until intention;
By being all intended to of obtaining of matching in above-mentioned matching process, according to concepts tab corresponding when being matched to intention current
Coverage rate in concepts tab sequence determines final intention, and using remaining concepts tab as slot position.
4. the method according to claim 1, wherein being directed to each concepts tab sequence, in conjunction with pre-establishing
Entity be intended to knowledge network carry out intention derivation, before obtaining the intention and slot position of the natural language text, the method
Further include:
To each concepts tab sequence, arranged according to concepts tab coverage rate, concepts tab number and concepts tab score value
Sequence chooses the N number of concepts tab sequence for coming front, wherein concepts tab covers from multiple concepts tab sequences after sequence
The priority of lid rate, concepts tab number and concepts tab score value successively reduces, and N is positive integer;
Correspondingly, being directed to each concepts tab sequence, it is intended to knowledge network in conjunction with the entity pre-established and carries out intention derivation, obtain
To the intention and slot position of the natural language text, comprising:
To each of N number of concepts tab sequence concepts tab sequence, it is intended to knowledge knowledge network in conjunction with the entity pre-established
Network carries out intention derivation, obtains the intention and slot position of the natural language text.
5. method according to any one of claims 1-4, which is characterized in that natural language text progress to be resolved
Before word cutting, the method also includes:
The natural language text is pre-processed, wherein the pretreatment, which includes at least, removes punctuation mark, capital and small letter turn
It changes.
6. a kind of natural language resolver, which is characterized in that described device includes:
Word cutting module obtains corresponding multiple word cutting segments for natural language text to be resolved to be carried out word cutting;
Concept tagging module obtains at least one concepts tab, wherein institute for carrying out concept tagging to each word cutting segment
Concepts tab is stated for mapping the abstract concept of word cutting segment;
Composite module, for obtaining multiple concepts tab sequences at least one described concepts tab progress permutation and combination,
In, between different concepts tab sequences, the word cutting boundary that the concepts tab in each concepts tab sequence is covered is not overlapped;
It is intended to derivation module, for being directed to each concepts tab sequence, is intended to knowledge network in conjunction with the entity pre-established and carries out
It is intended to derive, obtains the intention and slot position of the natural language text, wherein the entity is intended in knowledge network include multiple
The entity of different business scene is intended to set, and interrelated by entity between the entity intention set of different business scene.
7. device according to claim 6, which is characterized in that the concept tagging module is specifically used for:
By the combination of adjacent word cutting segment in each word cutting segment and the multiple word cutting segment, with the knowledge pre-established
Vocabulary is matched;
If there are matched concepts tab in the knowledge vocabulary, using the matched concepts tab as it is described at least one
Concepts tab.
8. according to the method any in claim 6, which is characterized in that the intention derivation module is specifically used for:
It is suitable according to expression identical with the natural language text for multiple concepts tabs in each concepts tab sequence
Sequence is intended to knowledge network with the entity respectively and is matched successively by single concepts tab and adjacent multiple concepts tabs;
If being matched to intention, the intention is obtained in the entity and is intended to corresponding entity in knowledge network, by the entity pair
It the concepts tab answered and not yet carries out matched concepts tab and merges into new concepts tab sequence, and the concepts tab new to this
Sequence executes above-mentioned matching operation, until all financial resourcess concept tag match in current concepts sequence label finishes or matches not
Until intention;
By being all intended to of obtaining of matching in above-mentioned matching process, according to concepts tab corresponding when being matched to intention current
Coverage rate in concepts tab sequence determines final intention, and using remaining concepts tab as slot position.
9. device according to claim 6, which is characterized in that described device further include:
Sequence obtains module, for each concepts tab sequence, according to concepts tab coverage rate, concepts tab number and general
It reads label score value to be ranked up, from multiple concepts tab sequences after sequence, chooses the N number of concepts tab sequence for coming front
Column, wherein the priority of concepts tab coverage rate, concepts tab number and concepts tab score value successively reduces, and N is positive integer;
Correspondingly, the intention derivation module is used for:
To each of N number of concepts tab sequence concepts tab sequence, it is intended to knowledge knowledge network in conjunction with the entity pre-established
Network carries out intention derivation, obtains the intention and slot position of the natural language text.
10. according to the device any in claim 6-9, which is characterized in that described device further include:
Preprocessing module, for being pre-processed to the natural language text to be resolved, wherein the pretreatment is at least wrapped
It includes and removes punctuation mark, capital and small letter conversion.
11. a kind of server characterized by comprising
One or more processors;
Memory, for storing one or more programs,
When one or more of programs are executed by one or more of processors, so that one or more of processors are real
Now such as natural language analytic method as claimed in any one of claims 1 to 5.
12. a kind of computer readable storage medium, is stored thereon with computer program, which is characterized in that the program is by processor
Such as natural language analytic method as claimed in any one of claims 1 to 5 is realized when execution.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810713935.2A CN108959257B (en) | 2018-06-29 | 2018-06-29 | Natural language parsing method, device, server and storage medium |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810713935.2A CN108959257B (en) | 2018-06-29 | 2018-06-29 | Natural language parsing method, device, server and storage medium |
Publications (2)
Publication Number | Publication Date |
---|---|
CN108959257A true CN108959257A (en) | 2018-12-07 |
CN108959257B CN108959257B (en) | 2022-11-22 |
Family
ID=64485018
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810713935.2A Active CN108959257B (en) | 2018-06-29 | 2018-06-29 | Natural language parsing method, device, server and storage medium |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108959257B (en) |
Cited By (12)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109815481A (en) * | 2018-12-17 | 2019-05-28 | 北京百度网讯科技有限公司 | Method, apparatus, equipment and the computer storage medium of event extraction are carried out to text |
CN110188278A (en) * | 2019-05-31 | 2019-08-30 | 三角兽(北京)科技有限公司 | Application program recommended method, device, electronic equipment and computer readable storage medium |
CN110209446A (en) * | 2019-04-23 | 2019-09-06 | 华为技术有限公司 | The configuration method and device of slot position are combined in a kind of interactive system |
CN110633476A (en) * | 2019-09-27 | 2019-12-31 | 北京百度网讯科技有限公司 | Method and device for acquiring knowledge annotation information |
CN111859984A (en) * | 2020-07-23 | 2020-10-30 | 中国平安人寿保险股份有限公司 | Intention mining method, device, equipment and storage medium |
CN112036186A (en) * | 2019-06-04 | 2020-12-04 | 腾讯科技(深圳)有限公司 | Corpus labeling method and device, computer storage medium and electronic equipment |
CN112558488A (en) * | 2019-09-10 | 2021-03-26 | 珠海格力电器股份有限公司 | Method and device for controlling intelligent equipment |
CN112580298A (en) * | 2019-09-29 | 2021-03-30 | 大众问问(北京)信息科技有限公司 | Method, device and equipment for acquiring marked data |
JP2021081712A (en) * | 2019-11-14 | 2021-05-27 | バイドゥ オンライン ネットワーク テクノロジー (ベイジン) カンパニー リミテッド | Method, device, electronic apparatus, and computer readable storage media for voice interaction |
CN113626564A (en) * | 2021-10-09 | 2021-11-09 | 腾讯科技(深圳)有限公司 | Concept label generation method and device, electronic equipment and storage medium |
CN115840806A (en) * | 2022-12-16 | 2023-03-24 | 深圳市人马互动科技有限公司 | Method and related device for acquiring plot information based on natural language interaction |
WO2023098077A1 (en) * | 2021-12-01 | 2023-06-08 | 中国电信股份有限公司 | Network intent processing method and related device |
Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101395696A (en) * | 2006-03-03 | 2009-03-25 | 亚马逊技术股份有限公司 | Collaborative structured tagging for item encyclopedias |
CN102236677A (en) * | 2010-04-28 | 2011-11-09 | 北京大学深圳研究生院 | Question answering system-based information matching method and system |
CN103942204A (en) * | 2013-01-18 | 2014-07-23 | 佳能株式会社 | Method and device for mining intentions |
CN104850554A (en) * | 2014-02-14 | 2015-08-19 | 北京搜狗科技发展有限公司 | Searching method and system |
CN107193865A (en) * | 2017-04-06 | 2017-09-22 | 上海奔影网络科技有限公司 | Natural language is intended to understanding method and device in man-machine interaction |
CN107301227A (en) * | 2017-06-21 | 2017-10-27 | 北京百度网讯科技有限公司 | Search information analysis method and device based on artificial intelligence |
CN107818781A (en) * | 2017-09-11 | 2018-03-20 | 远光软件股份有限公司 | Intelligent interactive method, equipment and storage medium |
CN108038183A (en) * | 2017-12-08 | 2018-05-15 | 北京百度网讯科技有限公司 | Architectural entities recording method, device, server and storage medium |
-
2018
- 2018-06-29 CN CN201810713935.2A patent/CN108959257B/en active Active
Patent Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101395696A (en) * | 2006-03-03 | 2009-03-25 | 亚马逊技术股份有限公司 | Collaborative structured tagging for item encyclopedias |
CN102236677A (en) * | 2010-04-28 | 2011-11-09 | 北京大学深圳研究生院 | Question answering system-based information matching method and system |
CN103942204A (en) * | 2013-01-18 | 2014-07-23 | 佳能株式会社 | Method and device for mining intentions |
CN104850554A (en) * | 2014-02-14 | 2015-08-19 | 北京搜狗科技发展有限公司 | Searching method and system |
CN107193865A (en) * | 2017-04-06 | 2017-09-22 | 上海奔影网络科技有限公司 | Natural language is intended to understanding method and device in man-machine interaction |
CN107301227A (en) * | 2017-06-21 | 2017-10-27 | 北京百度网讯科技有限公司 | Search information analysis method and device based on artificial intelligence |
CN107818781A (en) * | 2017-09-11 | 2018-03-20 | 远光软件股份有限公司 | Intelligent interactive method, equipment and storage medium |
CN108038183A (en) * | 2017-12-08 | 2018-05-15 | 北京百度网讯科技有限公司 | Architectural entities recording method, device, server and storage medium |
Non-Patent Citations (6)
Title |
---|
YI WU: "LEARNING CONCEPT TEMPLATES FROM WEB IMAGES TO QUERY PERSONAL IMAGE", 《2007 IEEE INTERNATIONAL CONFERENCE ON MULTIMEDIA AND EXPO》 * |
ZHENGHUA LI, JIAYUAN CHAO: "Coupled POS Tagging on Heterogeneous Annotations", 《IEEE/ACM TRANSACTIONS ON AUDIO, SPEECH, AND LANGUAGE PROCESSING》 * |
丁建辉: "知识图谱补全算法综述", 《研究与开发》 * |
刘秀磊: "本体匹配中基于词义组合的词法分析算法", 《电子学报》 * |
官赛萍: "面向知识图谱的知识推理研究进展", 《软件学报》 * |
崔建青: "查询意图识别的关键技术研究", 《中国优秀硕士学位学位论文全文数据库》 * |
Cited By (21)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109815481A (en) * | 2018-12-17 | 2019-05-28 | 北京百度网讯科技有限公司 | Method, apparatus, equipment and the computer storage medium of event extraction are carried out to text |
CN110209446A (en) * | 2019-04-23 | 2019-09-06 | 华为技术有限公司 | The configuration method and device of slot position are combined in a kind of interactive system |
WO2020216134A1 (en) * | 2019-04-23 | 2020-10-29 | 华为技术有限公司 | Configuration method and device for combination slots in human-machine dialogue system |
CN110188278A (en) * | 2019-05-31 | 2019-08-30 | 三角兽(北京)科技有限公司 | Application program recommended method, device, electronic equipment and computer readable storage medium |
CN112036186A (en) * | 2019-06-04 | 2020-12-04 | 腾讯科技(深圳)有限公司 | Corpus labeling method and device, computer storage medium and electronic equipment |
CN112558488B (en) * | 2019-09-10 | 2022-02-15 | 珠海格力电器股份有限公司 | Method and device for controlling intelligent equipment |
CN112558488A (en) * | 2019-09-10 | 2021-03-26 | 珠海格力电器股份有限公司 | Method and device for controlling intelligent equipment |
CN110633476A (en) * | 2019-09-27 | 2019-12-31 | 北京百度网讯科技有限公司 | Method and device for acquiring knowledge annotation information |
CN110633476B (en) * | 2019-09-27 | 2024-04-05 | 北京百度网讯科技有限公司 | Method and device for acquiring knowledge annotation information |
CN112580298A (en) * | 2019-09-29 | 2021-03-30 | 大众问问(北京)信息科技有限公司 | Method, device and equipment for acquiring marked data |
CN112580298B (en) * | 2019-09-29 | 2024-05-07 | 大众问问(北京)信息科技有限公司 | Method, device and equipment for acquiring annotation data |
JP2021081712A (en) * | 2019-11-14 | 2021-05-27 | バイドゥ オンライン ネットワーク テクノロジー (ベイジン) カンパニー リミテッド | Method, device, electronic apparatus, and computer readable storage media for voice interaction |
JP7300435B2 (en) | 2019-11-14 | 2023-06-29 | バイドゥ オンライン ネットワーク テクノロジー(ペキン) カンパニー リミテッド | Methods, apparatus, electronics, and computer-readable storage media for voice interaction |
US11830482B2 (en) | 2019-11-14 | 2023-11-28 | Baidu Online Network Technology (Beijing) Co., Ltd | Method and apparatus for speech interaction, and computer storage medium |
CN111859984A (en) * | 2020-07-23 | 2020-10-30 | 中国平安人寿保险股份有限公司 | Intention mining method, device, equipment and storage medium |
CN111859984B (en) * | 2020-07-23 | 2023-02-14 | 中国平安人寿保险股份有限公司 | Intention mining method, device, equipment and storage medium |
CN113626564A (en) * | 2021-10-09 | 2021-11-09 | 腾讯科技(深圳)有限公司 | Concept label generation method and device, electronic equipment and storage medium |
CN113626564B (en) * | 2021-10-09 | 2021-12-17 | 腾讯科技(深圳)有限公司 | Concept label generation method and device, electronic equipment and storage medium |
WO2023098077A1 (en) * | 2021-12-01 | 2023-06-08 | 中国电信股份有限公司 | Network intent processing method and related device |
CN115840806B (en) * | 2022-12-16 | 2023-09-05 | 深圳市人马互动科技有限公司 | Method and related device for acquiring scenario information based on natural language interaction |
CN115840806A (en) * | 2022-12-16 | 2023-03-24 | 深圳市人马互动科技有限公司 | Method and related device for acquiring plot information based on natural language interaction |
Also Published As
Publication number | Publication date |
---|---|
CN108959257B (en) | 2022-11-22 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108959257A (en) | A kind of natural language analytic method, device, server and storage medium | |
CN107220235B (en) | Speech recognition error correction method and device based on artificial intelligence and storage medium | |
CN110069608B (en) | Voice interaction method, device, equipment and computer storage medium | |
US10796244B2 (en) | Method and apparatus for labeling training samples | |
US12008336B2 (en) | Multimodal translation method, apparatus, electronic device and computer-readable storage medium | |
CN109657054A (en) | Abstraction generating method, device, server and storage medium | |
CN111858843B (en) | Text classification method and device | |
CN109299320A (en) | A kind of information interacting method, device, computer equipment and storage medium | |
CN111832449B (en) | Engineering drawing display method and related device | |
CN111738791B (en) | Text processing method, device, equipment and storage medium | |
CN109408834A (en) | Auxiliary machinery interpretation method, device, equipment and storage medium | |
CN112820269A (en) | Text-to-speech method, device, electronic equipment and storage medium | |
CN110263218A (en) | Video presentation document creation method, device, equipment and medium | |
CN109710951A (en) | Supplementary translation method, apparatus, equipment and storage medium based on translation history | |
CN114757176A (en) | Method for obtaining target intention recognition model and intention recognition method | |
CN109145164A (en) | Data processing method, device, equipment and medium | |
CN109815481A (en) | Method, apparatus, equipment and the computer storage medium of event extraction are carried out to text | |
CN111368066A (en) | Method, device and computer readable storage medium for acquiring dialogue abstract | |
CN109597881A (en) | Matching degree determines method, apparatus, equipment and medium | |
CN109359276A (en) | A kind of anti-fake code generating method, device, equipment and storage medium | |
CN108985421A (en) | The generation method and recognition methods of encoded information | |
CN111814496A (en) | Text processing method, device, equipment and storage medium | |
CN110516125A (en) | Method, device and equipment for identifying abnormal character string and readable storage medium | |
CN113360683A (en) | Method for training cross-modal retrieval model and cross-modal retrieval method and device | |
CN110489740A (en) | Semantic analytic method and Related product |
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 |