CN108304561A - A kind of semantic understanding method, equipment and robot based on finite data - Google Patents
A kind of semantic understanding method, equipment and robot based on finite data Download PDFInfo
- Publication number
- CN108304561A CN108304561A CN201810128775.5A CN201810128775A CN108304561A CN 108304561 A CN108304561 A CN 108304561A CN 201810128775 A CN201810128775 A CN 201810128775A CN 108304561 A CN108304561 A CN 108304561A
- Authority
- CN
- China
- Prior art keywords
- user
- keyword
- dialogue state
- word
- user speech
- 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
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/33—Querying
- G06F16/3331—Query processing
- G06F16/334—Query execution
- G06F16/3344—Query execution using natural language analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/33—Querying
- G06F16/332—Query formulation
- G06F16/3329—Natural language query formulation or dialogue systems
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/237—Lexical tools
- G06F40/247—Thesauruses; Synonyms
-
- 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)
- Physics & Mathematics (AREA)
- Computational Linguistics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Audiology, Speech & Language Pathology (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Mathematical Physics (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- Human Computer Interaction (AREA)
- Machine Translation (AREA)
Abstract
The invention discloses a kind of semantic understanding method, equipment and robot based on finite data, belongs to semantic understanding technical field.Semantic understanding method disclosed herein based on finite data includes：It from the keyword inquired in the keywords database pre-established in current collected user speech, determines that the dialogue state of current collected user speech marks according to the keyword inquired, determines and cache the dialogue state and mark the discourse context to be formed；The problem of being putd question to according to cached discourse context, user and individual subscriber dictionary, analysis of key word, extract with it is asked in user speech the problem of in the corresponding relevant conditional code of keyword of the usual word of user；According to the determining dialogue state label of dialogue state circulation and its discourse context, the usual word of user of memory and extraction and analysis and the problems in the user speech of acquisition of formation, final semanteme is determined in conjunction with the Corpus Analysis pre-established, positions the answer asked a question in user speech.
Description
Technical field
The present invention relates to semantic understanding technical fields, and in particular to a kind of semantic understanding method based on finite data is set
Standby and robot.
Background technology
Currently, the semantic understanding technology based on big data is just in full swing.Especially about convolutional neural networks CNN
With the deep learning research of Recognition with Recurrent Neural Network RNN, chat robots are pushed to a new high.
But current all kinds of chat robots, for example, little Du robots of Baidu, the siri of Apple Inc., Microsoft small ice
Deng needing to be further improved.It is smoothly horizontal that some robots can't be fully achieved human-computer interaction, give an irrelevant answer or do not know
The embarrassment of institute's cloud is often troubling.It would therefore be highly desirable to a chat robots with more hommization, to reach the expection of user
Effect.
Invention content
Provided herein is a kind of semantic understanding method, equipment and robot based on finite data, can solve existing machine
The not smooth problem low with accuracy of conference process.
The semantic understanding method based on finite data that disclosed herein is a kind of, includes at least：
From the keyword inquired in the keywords database pre-established in current collected user speech, dialogue state is utilized
Circulation mechanism is determining and caches the corresponding dialogue state label of inquired keyword, wherein the dialogue state mark cached
Note is used to form discourse context；
The discourse context to be formed, collected user is marked to carry the dialogue state cached with extraction mechanism using memory
The individual subscriber dictionary and keyword of the problem of asking and history accumulation, carry out logic of language comparison, determine user in user speech
The corresponding keyword of usual word；
Using language material location mechanism according to the dialogue state label cached and its discourse context formed, identified use
The corresponding keyword of the usual word in family and the problems in the user speech of acquisition determine most in conjunction with the Corpus Analysis pre-established
Whole semanteme positions the final result asked a question in user speech.
Optionally, the above method further includes：
When being talked for the first time with any user, the individual subscriber dictionary of the user, and the user accumulated from history are automatically created
The user usual word consistent with keywords semantics is extracted in dialogue to store, in the usual word of the new user of appearance, update
The individual subscriber dictionary.
Optionally, in the above method, the keyword that the basis inquires determines pair of current collected user speech
Speech phase marks, and caches identified dialogue state and mark to form discourse context, including：
It being encoded according to the keyword inquired, coding obtains the conditional code for being used to indicate the dialogue state label,
All conditional codes encoded by the interim banked cache of conditional code sequence, all conditional codes cached constitute discourse context.
Optionally, described that current collected user speech is inquired from the keywords database pre-established in the above method
In keyword before, this method further includes：
Preset wake-up word is collected, into normal operating condition.
Optionally, the above method further includes：
Keyword is extracted from the user speech of acquisition in advance, establishes keywords database.
Optionally, the above method further includes：
When the answer that there is no problem in the corpus, learn new language material, to update corpus content.
The semantic understanding equipment based on finite data that there is disclosed herein a kind of, includes at least：
Dialogue state circulation module, from the pass inquired in the keywords database pre-established in current collected user speech
Keyword, and determine that the dialogue state of current collected user speech marks according to the keyword inquired, and cache and determine
Dialogue state mark to form discourse context；
The dialogue state cached is marked the discourse context to be formed, collected user to put question to by memory and extraction module
The problem of and history accumulation individual subscriber dictionary and keyword, carry out logic of language comparison, determine collected user speech
The corresponding keyword of the middle usual word of user；
Language material locating module, according to dialogue state circulate module determine dialogue state mark and its formed to language
The problems in the user speech in border, the corresponding keyword of the usual word of user and acquisition, in conjunction with the Corpus Analysis pre-established
It determines final semanteme, positions the answer asked a question in user speech.
Optionally, in above equipment, the memory and extraction module when being talked for the first time with any user, automatically create this
The individual subscriber dictionary of user, and the user usual word consistent with keywords semantics is extracted from the user session that history is accumulated
It is stored, and in the usual word of the new user of appearance, updates the individual subscriber dictionary.
Optionally, in above equipment, dialogue state circulation module determines current collected according to the keyword inquired
The dialogue state of user speech marks, and caches identified dialogue state and mark to form discourse context, including：
The dialogue state circulation module, is encoded according to the keyword inquired, and coding obtains being used to indicate described
The conditional code of dialogue state label, all conditional codes encoded by the interim banked cache of conditional code sequence, the institute cached
Stateful code constitutes discourse context.
Optionally, above equipment further includes：
It wakes up and sleep block, into normal operating condition, is collecting suspend mode word when collecting preset wake-up word
When, into dormant state.
Optionally, above equipment further includes：
Keyword is arranged and extraction module, extracts keyword from the user speech of acquisition in advance, establishes keywords database.
Optionally, above equipment further includes：
Study module when the answer that there is no problem in corpus, learns new language material, to update corpus content.
There is disclosed herein a kind of robot, including memory, processor and it is stored on the memory and can be in institute
State the computer program run on processor, wherein the processor is realized as described above when executing the computer program
All processing of semantic understanding method based on finite data.
There is disclosed herein a kind of robots, include at least the semantic understanding equipment based on finite data as described above.
Robot conversation procedure can be made more smooth and accurate using technical scheme.
Description of the drawings
Fig. 1 is the semantic understanding device structure schematic diagram based on finite data in the embodiment of the present invention；
Fig. 2 is the operation principle schematic diagram of dialogue state circulation mechanism in structure shown in Fig. 1；
Fig. 3 is an interactive specific example process schematic for the embodiment of the present invention；
Fig. 4 is the session operational scenarios tree exemplary plot of extraction during human-computer dialogue shown in Fig. 3.
Specific implementation mode
To make the objectives, technical solutions, and advantages of the present invention clearer, below in conjunction with specific implementation mode pair
Technical solution of the present invention is described in further detail.It should be noted that in the absence of conflict, embodiments herein and
Feature in embodiment can be arbitrarily combined with each other.
Embodiment 1
Present inventor find it is existing have human-computer interaction technology (such as chat robots) be based on big data,
Its user oriented is uncertain, using CNN and RNN theories for solving the problems, such as that complicated semantic understanding still is apparent not enough at present,
The fluency of application effect shortage language, accuracy, flexibility.Based on this discovery, present inventor proposes, one kind is based on having
The semantic understanding equipment of data is limited, includes mainly dialogue state circulation module, memory and extraction module and language material locating module.
Dialogue state circulation module, from the pass inquired in the keywords database pre-established in current collected user speech
Keyword determines the corresponding dialogue state label of keyword that simultaneously caching query arrives, wherein the dialogue state label cached is used for
Form discourse context；
In the present embodiment, dialogue state circulation module can be encoded, coding is used according to the keyword inquired
It is stateful in the conditional code for indicating the dialogue state label, then by the institute that the interim banked cache of conditional code sequence encodes
Code, all conditional codes cached may make up discourse context.
Memory and extraction module, using memory with extraction mechanism marked according to the dialogue state cached to be formed to language
Border, collected user put question to the problem of and history accumulation individual subscriber dictionary, association context, problem, and with keyword into
Row logic of language compares, and determines the corresponding keyword of the usual word of user in the problem of user of acquisition puts question to.Language material locating module,
The dialogue state label determined according to dialogue state circulation module and its discourse context formed, the corresponding key of the usual word of user
Word and the problems in the user speech of acquisition determine final semanteme in conjunction with the Corpus Analysis pre-established, position user
The final result asked a question in voice.Specifically, language material locating module can combine the corpus comprehensive analysis pre-established,
And by repeatedly asking in reply user, to understand customer problem, that is, the real semanteme of user is understood, finally position and carried in user speech
The final result of problem.
In addition, on the basis of above-mentioned module, the semantic understanding equipment based on finite data can also be following a kind of or several
Kind module：It wakes up and sleep block, into normal operating condition, is collecting not mainly when collecting preset wake-up word
When dormancy word, into dormant state.
Keyword is arranged and extraction module, extracts keyword from the user speech of acquisition in advance, establishes keywords database.
Study module when the answer that there is no problem in corpus, learns new language material, to update corpus content.
The specific example for the semantic understanding equipment based on finite data that the present embodiment provides a kind of below, can be placed in dialogue
In robot, including above-mentioned all modules, that is, include：It wakes up and sleep block, dialogue state circulation module, memory and extraction mould
Block, keyword setting and extraction module, language material locating module and study module, as shown in Figure 1.
1, it wakes up and sleep block, into normal operating condition, is collecting suspend mode when collecting preset wake-up word
When word, into dormant state.This module is regarded as conventional modules, and the present embodiment does not do special limit to the specific implementation of the module
System.
I.e. using wake-up, dormancy mechanism, arbitrarily chips in avoid robot, only waken up when needing human-computer dialogue, otherwise
Suspend mode.For example, it is to wake up word that " hello for robot ", which can be arranged, robot enters normal operating condition, carries out dialogue and prepares；It can
Can be for suspend mode word with setting " goodbye ", robot responds goodbye, so that it may to enter dormant state, no longer chip in.Specifically, waking up
Word and suspend mode word can be that system default is arranged, and can also be independently arranged by user.
2, it is (i.e. current to inquire current collected voice from the keywords database pre-established for dialogue state circulation module
The problem of people is asked) in keyword, the context for the problem of people is asked is determined according to the keyword inquired, and cache and determine
Context (in the present embodiment adoption status code come indicate current session state circulate label, i.e., one group of conditional code is joined together
It can indicate that the context of its current session, and this group of conditional code can be buffered in conditional code sequence temporary library) for memory
Use is associated with extraction module；
It should be noted that semanteme of the identical language expressed by different language environments is different, i.e., the context usually said
It is different semantic multifarious.The application constructs a kind of dialogue state circulation mechanism thus, and dialogue state circulation module uses should
Mechanism realizes its function.The essence that dialogue state circulation mechanism is constituted is the language expressed by the record by human-computer dialogue process
Border specifically can complete the accurate recording of dialog procedure by a set of encoding mechanism.The present embodiment state circulate mechanism,
There are one states (being also believed to precondition), referred to as " preceding state " before robot speaks every time；Robot has one after finishing words
A state, referred to as " rear state ", or it is " migration state ", this migration state is indicated with coding, is exactly state circulation label (the present embodiment
In be equivalent to conditional code).In this way, migration state can also be used as condition, it is formed dialog procedure, is constantly accumulated as context.Its
Principle is as shown in Figure 2.
In the present embodiment, the rule of conditional code coding is：Different dialogue scene is interrelated to may be constructed a scene tree,
It is layered several, conditional code coding is carried out according to scene level, can define dormant state, general state, singlet and polymorphic, in rear state only
Singlet has general state, singlet or polymorphic in preceding state.Dormant state is robot standby mode, it cannot be said that words.General state is in preceding state
" condition " arbitrary situation.Singlet refers to unique Dialog Token, or the case where unique " condition ".Polymorphic refers to that multiple dialogues are
The case where premise, or to say that in preceding state, " condition " has multiple.Cryptoprinciple has three, first, dialogue answer is oriented to principle, second is that field
Scape is layered principle, third, accumulation principle.As it can be seen that the present embodiment uses dialogue state circulation module, human-computer dialogue is combined
User speaks front and back state in journey and robot speaks front and back state, keeps interactive transition more natural, improves people
The fluency of machine dialogue, also, the front and back state spoken with reference to user so that it is more accurate that robot understands user semantic,
Dialogue replies also more accurate.
3, memory and extraction module, extract for the dialog history of each user and store individual subscriber dictionary respectively, and
The problem of bonding state code, user put question to and individual subscriber dictionary, analysis of key word, extraction and people institute from individual subscriber dictionary
The consistent keyword of the semanteme of the usual word of user (personal touch for extracting user session), update user in the problem of asking
People's dictionary.
Wherein, the memory and extraction mechanism that above-mentioned memory is used with extraction module：Mainly consider the individual character and language of people
Reason and good sense decorrelation, therefore individual subscriber dictionary (comprising personal synonym), i.e. user are established respectively to the people or user talked with
In people's dictionary, include the correspondence of keyword and the usual word of user, i.e., it is corresponding with same keyword there are one or multiple use
The usual word in family, and the usual word of these users is consistent with the semanteme of the keyword.In this way, helping to be more accurately located semanteme.
Establish memory by conditional code temporary library, personal dictionary, then in conjunction with conditional code, people is asked the problem of and personal dictionary, analyze
Keyword.The mechanism used by above-mentioned memory and the extraction module can be seen that using this mechanism, it is possible to distinguish go out different use
The personalized difference of family word improves the accuracy that robot replies in human-computer dialogue to be more accurately located semanteme.
Specifically, memory and extraction module can automatically create the user when this equipment and any user are talked for the first time
Individual subscriber dictionary, and the usual word progress of consistent with the keywords semantics user of extraction from the user session that history is accumulated
Storage, and in the usual word of the new user of appearance, update the individual subscriber dictionary.
4, keyword setting and extraction module pre-set keywords database, and extract key from the user speech of acquisition
Word, to establish keywords database；This module is regarded as conventional modules, and the present embodiment does not do especially the specific implementation of the module
Limitation.
The keyword that above-mentioned module uses is arranged and the mechanism of extraction is：Sentence answer is positioned in view of current keyword search
A kind of method is generally used, but how keyword is arranged, and can have many methods, with morpheme setting, vocabulary setting or word
Group setting, effect are all not quite similar.Therefore the method for taking mixing to be arranged in the present embodiment, according to theme and most question-answer sentences, into
Setting, including common synonym are mixed after row analysis.
5, language material locating module, according to dialogue state circulation conditional code, memory and extraction and analysis personal word, to
Personal touch, the problem of context and people are asked etc. of family dialogue, and analyzed in many ways in conjunction with the corpus pre-established, with
And user is repeatedly asked in reply, and to determine the final semanteme for the problem of user is carried, then quick location answer.
Specifically, language material locating module is realized using a kind of language material location mechanism.I.e. when people proposes problem, often language
Justice not can determine that, by inquiry " conditional code sequence temporary library ", " keywords database ", " individual subscriber dictionary " and " corpus ",
It is asked in reply again, target is gradually reduced, it is final to determine " correct option ".
6, study module：If the answer or robot that there is no problem in corpus are said when not knowing, robot can be given
Say the word " prepare study ", allows the language material that robot learning is new, finally corpus is given to increase new content.This module is regarded as normal
Scale block, the present embodiment are not particularly limited the specific implementation of the module.
By taking a specific people and robot session operational scenarios as an example, the process for carrying out language material positioning is as shown in Figure 3.It is right from this
It is as shown in Figure 4 to talk about the scene tree established in the process.Scene tree according to Fig.4, can carry out conditional code coding.
It is noted that the present embodiment also provides a kind of robot, each mould described in embodiment 1 is included at least
Block.
Embodiment 2
The semantic understanding method based on finite data that the present embodiment provides a kind of, main includes following operation：
From the keyword inquired in the keywords database pre-established in current collected user speech, dialogue state is utilized
Circulation mechanism is determining and caches the corresponding dialogue state label of inquired keyword, wherein the dialogue state mark cached
Note is used to indicate the discourse context of user and robot；
Using memory and extraction module mechanism, the problem of dialogue state cached label, user are putd question to and user
People's dictionary carries out logic of language comparison with the keyword inquired, determines that the usual word of user corresponds in the problem of user puts question to
Keyword；
According to determining dialogue state label and its discourse context formed, keyword corresponding with the usual word of user and
The problems in user speech of acquisition determines final semanteme in conjunction with the Corpus Analysis pre-established, positions in user speech
The answer asked a question.
Wherein, it determines that the dialogue state of current collected user speech marks according to the keyword inquired, and caches
Identified dialogue state marks the specific operation process to form discourse context to can be found in following operation：
It being encoded according to the keyword inquired, coding obtains the conditional code for being used to indicate the dialogue state label,
All conditional codes encoded by the interim banked cache of conditional code sequence, all conditional codes cached constitute discourse context.
In addition, according to use habit, before above method operation, preset wake-up word can be first acquired, and acquiring
Enter normal operating condition to after waking up word, carries out subsequent operation.
Based on method described above, can keyword be extracted from the user speech of acquisition in advance, establish and update key
Dictionary.
When the answer that there is no problem in corpus, new language material can also be learnt, to update corpus content.
Involved individual subscriber dictionary can be user session accumulate according to history to establish among the above, i.e., with times
When one user talks for the first time, the individual subscriber dictionary of the user can be automatically created, and from the user session that history is accumulated
The extraction user usual word consistent with keywords semantics stores, and in the usual word of the new user of appearance, described in update
Individual subscriber dictionary.
Method provided in this embodiment can be realized based on the equipment of above-described embodiment 1, therefore the other details of the above method
The corresponding contents of above-described embodiment 1 are can be found in, details are not described herein.
It is also noted that the application also provides a kind of robot, it may include memory, processor and be stored in storage
On device and the computer program that can run on a processor, wherein embodiment 2 may be implemented in the computer program that processor executes
Described in the methodical processing of institute.
One of ordinary skill in the art will appreciate that all or part of step in the above method can be instructed by program
Related hardware is completed, and described program can be stored in computer readable storage medium, such as read-only memory, disk or CD
Deng.Optionally, all or part of step of above-described embodiment can also be realized using one or more integrated circuits.Accordingly
Ground, the form that hardware may be used in each module/unit in above-described embodiment are realized, the shape of software function module can also be used
Formula is realized.The application is not limited to the combination of the hardware and software of any particular form.
The above, only preferred embodiments of the invention, are not intended to limit the scope of the present invention.It is all this
Within the spirit and principle of invention, any modification, equivalent substitution, improvement and etc. done should be included in the protection model of the present invention
Within enclosing.
Claims (14)
1. a kind of semantic understanding method based on finite data, includes at least：
From the keyword inquired in the keywords database pre-established in current collected user speech, circulated using dialogue state
Mechanism is determining and caches the corresponding dialogue state label of inquired keyword, wherein the dialogue state label cached is used
In formation discourse context；
The discourse context to be formed, collected user is marked to put question to the dialogue state cached with extraction mechanism using memory
The individual subscriber dictionary and keyword of problem and history accumulation, carry out logic of language comparison, determine that user is usual in user speech
The corresponding keyword of word；
It is marked according to the dialogue state cached using language material location mechanism and its discourse context of formation, identified user is used to
The corresponding keyword of word and the problems in the user speech of acquisition determine finally in conjunction with the Corpus Analysis pre-established
Semanteme positions the final result asked a question in user speech.
2. the method as described in claim 1, which is characterized in that the method further includes：
When being talked for the first time with any user, the individual subscriber dictionary of the user, and the user session accumulated from history are automatically created
The middle extraction user usual word consistent with keywords semantics stores, in the usual word of the new user of appearance, described in update
Individual subscriber dictionary.
3. the method as described in claim 1, which is characterized in that described determining using dialogue state circulation mechanism and cache and looked into
The corresponding dialogue state label of keyword ask, including：
It is encoded according to the keyword inquired, coding obtains the conditional code for being used to indicate the dialogue state label, passes through
All conditional codes that the interim banked cache of conditional code sequence encodes, all conditional codes cached constitute discourse context.
4. method as claimed in claim 1,2 or 3, which is characterized in that described inquired from the keywords database pre-established is worked as
Before keyword in preceding collected user speech, this method further includes：
Preset wake-up word is collected, into normal operating condition.
5. method as claimed in claim 4, which is characterized in that the method further includes：
Keyword is extracted from the user speech of acquisition in advance, updates keywords database.
6. method as claimed in claim 4, which is characterized in that the method further includes：
When the answer that there is no problem in the corpus, learn new language material, to update corpus content.
7. a kind of semantic understanding equipment based on finite data, includes at least：
Dialogue state circulation module, from the key inquired in the keywords database pre-established in current collected user speech
Word, and determine and cache the corresponding dialogue state label of inquired keyword, wherein the dialogue state label cached is used
In formation discourse context；
Memory and extraction module, by the dialogue state cached mark the discourse context to be formed, collected user put question to ask
The individual subscriber dictionary and keyword of topic and history accumulation, carry out logic of language comparison, determine and used in collected user speech
The corresponding keyword of the usual word in family；
Language material locating module, the dialogue state label for the module determination that circulated according to dialogue state and its discourse context, the use of formation
The corresponding keyword of the usual word in family and the problems in the user speech of acquisition determine most in conjunction with the Corpus Analysis pre-established
Whole semanteme positions the answer asked a question in user speech.
8. equipment as claimed in claim 7, which is characterized in that
The memory and extraction module when being talked for the first time with any user, automatically create the individual subscriber dictionary of the user, and from
The extraction user usual word consistent with keywords semantics stores in the user session of history accumulation, and is occurring newly
When the usual word of user, the individual subscriber dictionary is updated.
9. equipment as claimed in claim 7, which is characterized in that the dialogue state circulation module is determined and cached and inquired
The corresponding dialogue state label of keyword arrived, including：
The dialogue state circulation module, is encoded, coding obtains being used to indicate the dialogue according to the keyword inquired
The conditional code of status indication, all conditional codes encoded by the interim banked cache of conditional code sequence, all shapes cached
State code constitutes discourse context.
10. the equipment as described in claim 7,8 or 9, which is characterized in that the equipment further includes：
Wake-up and sleep block, when collecting preset wake-up word, into normal operating condition, when collecting suspend mode word,
Into dormant state.
11. equipment as claimed in claim 10, which is characterized in that the equipment further includes：
Keyword is arranged and extraction module, extracts keyword from the user speech of acquisition in advance, establishes update keywords database.
12. equipment as claimed in claim 10, which is characterized in that the equipment further includes：
Study module when the answer that there is no problem in corpus, learns new language material, to update corpus content.
13. a kind of robot, including memory, processor and it is stored on the memory and can runs on the processor
Computer program, which is characterized in that the processor is realized when executing the computer program as any in claim 1-6
The processing of method described in.
14. a kind of robot, which is characterized in that include at least the equipment described in any one of claim 7-12.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810128775.5A CN108304561B (en) | 2018-02-08 | 2018-02-08 | A kind of semantic understanding method, equipment and robot based on finite data |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810128775.5A CN108304561B (en) | 2018-02-08 | 2018-02-08 | A kind of semantic understanding method, equipment and robot based on finite data |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108304561A true CN108304561A (en) | 2018-07-20 |
| CN108304561B CN108304561B (en) | 2019-03-29 |
Family
ID=62864952
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810128775.5A Active CN108304561B (en) | 2018-02-08 | 2018-02-08 | A kind of semantic understanding method, equipment and robot based on finite data |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108304561B (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110265013A (en) * | 2019-06-20 | 2019-09-20 | 平安科技（深圳）有限公司 | The recognition methods of voice and device, computer equipment, storage medium |
| CN111831801A (en) * | 2020-05-27 | 2020-10-27 | 北京市农林科学院 | Man-machine conversation method and system |
| CN112365892A (en) * | 2020-11-10 | 2021-02-12 | 杭州大搜车汽车服务有限公司 | Man-machine interaction method, device, electronic device and storage medium |
| CN117034953A (en) * | 2023-10-07 | 2023-11-10 | 湖南东良数智科技有限公司 | System for utilizing personal copybook library and intelligent session thereof |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104219388A (en) * | 2014-08-28 | 2014-12-17 | 小米科技有限责任公司 | Voice control method and device |
| CN104347074A (en) * | 2013-07-31 | 2015-02-11 | 通用汽车环球科技运作有限责任公司 | Systems and methods for managing dialog context in speech systems |
| CN106128453A (en) * | 2016-08-30 | 2016-11-16 | 深圳市容大数字技术有限公司 | The Intelligent Recognition voice auto-answer method of a kind of robot and robot |
| CN106528522A (en) * | 2016-08-26 | 2017-03-22 | 南京威卡尔软件有限公司 | Scenarized semantic comprehension and dialogue generation method and system |
| CN106777018A (en) * | 2016-12-08 | 2017-05-31 | 竹间智能科技（上海）有限公司 | To the optimization method and device of read statement in a kind of intelligent chat robots |
| CN107102988A (en) * | 2017-04-27 | 2017-08-29 | 长沙军鸽软件有限公司 | A kind of method that session is actively initiated based on personal exclusive corpus |
| CN107493353A (en) * | 2017-10-11 | 2017-12-19 | 宁波感微知著机器人科技有限公司 | A kind of intelligent robot cloud computing method based on contextual information |
| CN107644642A (en) * | 2017-09-20 | 2018-01-30 | 广东欧珀移动通信有限公司 | Method for recognizing semantics, device, storage medium and electronic equipment |
-
2018
- 2018-02-08 CN CN201810128775.5A patent/CN108304561B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104347074A (en) * | 2013-07-31 | 2015-02-11 | 通用汽车环球科技运作有限责任公司 | Systems and methods for managing dialog context in speech systems |
| CN104219388A (en) * | 2014-08-28 | 2014-12-17 | 小米科技有限责任公司 | Voice control method and device |
| CN106528522A (en) * | 2016-08-26 | 2017-03-22 | 南京威卡尔软件有限公司 | Scenarized semantic comprehension and dialogue generation method and system |
| CN106128453A (en) * | 2016-08-30 | 2016-11-16 | 深圳市容大数字技术有限公司 | The Intelligent Recognition voice auto-answer method of a kind of robot and robot |
| CN106777018A (en) * | 2016-12-08 | 2017-05-31 | 竹间智能科技（上海）有限公司 | To the optimization method and device of read statement in a kind of intelligent chat robots |
| CN107102988A (en) * | 2017-04-27 | 2017-08-29 | 长沙军鸽软件有限公司 | A kind of method that session is actively initiated based on personal exclusive corpus |
| CN107644642A (en) * | 2017-09-20 | 2018-01-30 | 广东欧珀移动通信有限公司 | Method for recognizing semantics, device, storage medium and electronic equipment |
| CN107493353A (en) * | 2017-10-11 | 2017-12-19 | 宁波感微知著机器人科技有限公司 | A kind of intelligent robot cloud computing method based on contextual information |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110265013A (en) * | 2019-06-20 | 2019-09-20 | 平安科技（深圳）有限公司 | The recognition methods of voice and device, computer equipment, storage medium |
| CN111831801A (en) * | 2020-05-27 | 2020-10-27 | 北京市农林科学院 | Man-machine conversation method and system |
| CN112365892A (en) * | 2020-11-10 | 2021-02-12 | 杭州大搜车汽车服务有限公司 | Man-machine interaction method, device, electronic device and storage medium |
| CN117034953A (en) * | 2023-10-07 | 2023-11-10 | 湖南东良数智科技有限公司 | System for utilizing personal copybook library and intelligent session thereof |
| CN117034953B (en) * | 2023-10-07 | 2023-12-19 | 湖南东良数智科技有限公司 | System for utilizing personal copybook library and intelligent session thereof |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108304561B (en) | 2019-03-29 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108920622B (en) | Training method, training device and recognition device for intention recognition | |
| US10540965B2 (en) | Semantic re-ranking of NLU results in conversational dialogue applications | |
| CN108304561B (en) | A kind of semantic understanding method, equipment and robot based on finite data | |
| US20210319051A1 (en) | Conversation oriented machine-user interaction | |
| CN105095182B (en) | A kind of return information recommendation method and device | |
| CN108681574B (en) | Text abstract-based non-fact question-answer selection method and system | |
| CN107562863A (en) | Chat robots reply automatic generation method and system | |
| CN110222182B (en) | Statement classification method and related equipment | |
| CN111177355B (en) | Man-machine conversation interaction method and device based on search data and electronic equipment | |
| CN108304372A (en) | Entity extraction method and apparatus, computer equipment and storage medium | |
| CN110111780A (en) | Data processing method and server | |
| CN110334347A (en) | Information processing method, relevant device and storage medium based on natural language recognition | |
| CN110059169B (en) | Intelligent robot chat context implementation method and system based on corpus labeling | |
| US20230350929A1 (en) | Method and system for generating intent responses through virtual agents | |
| US20150302056A1 (en) | Method, system, and storage medium for information search | |
| CN112151015B (en) | Keyword detection method, keyword detection device, electronic equipment and storage medium | |
| KR20070102267A (en) | Dialog management system, and method of managing dialog using example-based dialog modeling technique | |
| CN112163425A (en) | Text entity relation extraction method based on multi-feature information enhancement | |
| US11636272B2 (en) | Hybrid natural language understanding | |
| CN113468894B (en) | Dialogue interaction method and device, electronic equipment and computer readable storage medium | |
| CN113361266A (en) | Text error correction method, electronic device and storage medium | |
| CN109271459A (en) | Chat robots and its implementation based on Lucene and grammer networks | |
| CN112131885A (en) | Semantic recognition method and device, electronic equipment and storage medium | |
| CN114625855A (en) | Method, apparatus, device and medium for generating dialogue information | |
| CN108595609A (en) | Generation method, system, medium and equipment are replied by robot based on personage IP |
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 |