CN108846125A - Talk with generation method, device, terminal and computer readable storage medium - Google Patents
Talk with generation method, device, terminal and computer readable storage medium Download PDFInfo
- Publication number
- CN108846125A CN108846125A CN201810694671.0A CN201810694671A CN108846125A CN 108846125 A CN108846125 A CN 108846125A CN 201810694671 A CN201810694671 A CN 201810694671A CN 108846125 A CN108846125 A CN 108846125A
- Authority
- CN
- China
- Prior art keywords
- key
- value pair
- value
- information
- answer
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Landscapes
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The embodiment of the present invention proposes that a kind of dialogue generation method, device, terminal and computer readable storage medium, method include：Query semantics information is extracted from query statement；Multiple key-value pairs relevant to query semantics information are searched in key value library；Obtain auxiliary information relevant to query semantics information；Auxiliary information is weighted with each key-value pair found, obtains replying semantic information；Semantic information will be replied and be converted to answer sentence.The embodiment of the present invention realizes by also introducing auxiliary information except according to the relevant key-value pair of query semantics information searching and generates answer sentence in such a way that much information combines, so that the answer sentence ultimately generated is more accurate.
Description
Technical field
The present invention relates to field of artificial intelligence more particularly to a kind of dialogue generation method, device, terminal and computers
Readable storage medium storing program for executing.
Background technique
In the prior art, the method for talking with generation includes two kinds：Retrieval type and spatial term.Wherein, retrieval type
Method be by user input sentence retrieved in corpus, find some relevant replies.But by corpus
Limitation, if user can not be replied without relevant sentence in corresponding corpus.Although and spatial term can be by omnipotent
The mode of sentence realize to can not answer statement answer, but it is this reply will affect user experience.In order to avoid using ten thousand
Energy sentence, existing spatial term generate answer frequently with the mode for extracting keyword and using the mode of maximum mutual information
Sentence, but the mode of extracting keyword is more difficult and it is difficult to ensure that the sentence generated can reply information above, and use
The mode of maximum mutual information, which screens answer sentence, can consume the more time.
Disclosed above- mentioned information are only used for reinforcing the understanding to background of the invention in the background technology, therefore it may be wrapped
Containing the information for not being formed as the prior art that those of ordinary skill in the art are known.
Summary of the invention
The embodiment of the present invention provides a kind of dialogue generation method, device, terminal and computer readable storage medium, to solve
One or more technical problem in the prior art.
In a first aspect, the embodiment of the invention provides a kind of dialogue generation methods, including：
Query semantics information is extracted from query statement；
Multiple key-value pairs relevant to the query semantics information are searched in key value library；
Obtain auxiliary information relevant to the query semantics information；
The auxiliary information is weighted with each key-value pair found, obtains replying semantic information；
The answer semantic information is converted into answer sentence.
With reference to first aspect, the embodiment of the present invention is searched in key value library in the first implementation of first aspect
Multiple key-value pairs relevant to the query semantics information, including：
Knowledge semantic information is extracted from the query semantics information；
Knowledge corpus relevant to the knowledge semantic information is obtained from knowledge base；
The knowledge semantic information and the knowledge corpus are formed into the key-value pair.
The first implementation with reference to first aspect, second implementation of the embodiment of the present invention in first aspect
In, the knowledge semantic information and the knowledge corpus are formed into the key-value pair, including：
The knowledge semantic information and the knowledge corpus are formed into the key-value pair by neural network.
With reference to first aspect, the embodiment of the present invention further includes in the third implementation of first aspect：
The answer semantic information and the query semantics information are formed into question and answer key-value pair；
The existing key-value pair stored in the question and answer key-value pair and the key value library is matched, if the key value library
The question and answer key-value pair is not present in the existing key-value pair, then the question and answer key-value pair is stored in the key value library.
The third implementation with reference to first aspect, four kind implementation of the embodiment of the present invention in first aspect
In, the existing key-value pair stored in the question and answer key-value pair and the key value library is matched, if the key value library is described
The question and answer key-value pair is not present in existing key-value pair, then the question and answer key-value pair is stored in the key value library, specific steps
Including：
The question and answer key-value pair is kept in the short-term memory area of the key value library；
The question and answer key-value pair is assessed by assessment models, if assessing the threshold value of the question and answer key-value pair beyond institute
The threshold value of existing key-value pair is stated, then the question and answer key-value pair is deposited into the long-term memory area of the key value library.
With reference to first aspect, the embodiment of the present invention further includes in the 5th kind of implementation of first aspect：
The language expression mode for replying sentence is adjusted by stochastic variable.
Second aspect, the embodiment of the invention provides a kind of dialogue generating means, including：
Extraction module, for extracting query semantics information from query statement；
Module is obtained, for searching multiple key-value pairs relevant to the query semantics information in key value library；
Supplementary information module, for obtaining auxiliary information relevant to the query semantics information；
Processing module obtains replying language for the auxiliary information to be weighted with each key-value pair found
Adopted information；
Conversion module, for the answer semantic information to be converted to answer sentence.
In a possible design, the acquisition module includes：
Extracting sub-module, for extracting knowledge semantic information from the query semantics information；
Acquisition submodule, for obtaining knowledge corpus relevant to the knowledge semantic information from knowledge base；
Key-value pair submodule, for the knowledge semantic information and the knowledge corpus to be built into the key-value pair.
In a possible design, further include：
Matching module, for the answer semantic information and the query semantics information to be formed question and answer key-value pair；
Enlargement module, for the existing key-value pair stored in the question and answer key-value pair and the key value library to be matched,
It, then will be described in question and answer key-value pair deposit when the question and answer key-value pair is not present in the existing key-value pair of the key value library
In key value library.
In a possible design, the enlargement module includes：
Short memory submodule, for keeping in the question and answer key-value pair in the short-term memory area of the key value library；
Long memory submodule, for being assessed by assessment models the question and answer key-value pair, if assessing the question and answer
The threshold value of key-value pair exceeds the threshold value of the existing key-value pair, then the question and answer key-value pair is deposited into the long-term of the key value library
Remember in area.
The third aspect, the embodiment of the invention provides a kind of dialogues to generate terminal, including：
The function that dialogue generates terminal can also execute corresponding software realization by hardware realization by hardware.
The hardware or software include one or more modules corresponding with above-mentioned function.
In a possible design, talking with includes processor and memory, the storage in the structure of the terminal of generation
Device is used to store the terminal for supporting dialogue to generate and executes the program for talking with generation method in above-mentioned first aspect, the processor quilt
It is configured for executing the program stored in the memory.Talk with generate terminal can also include communication interface, for pair
Words generate terminal and other equipment or communication.
Fourth aspect, the embodiment of the invention provides a kind of computer readable storage mediums, for storing dialogue generation
Computer software instructions used in terminal comprising for execute in above-mentioned first aspect talk with generation method be dialogue generation
Program involved in terminal.
A technical solution in above-mentioned technical proposal has the following advantages that or beneficial effect：By being believed according to query semantics
Breath is searched except relevant key-value pair, and auxiliary information is also introduced, and is realized and is generated answer in such a way that much information combines
Sentence, so that the answer sentence ultimately generated is more accurate.
Above-mentioned general introduction is merely to illustrate that the purpose of book, it is not intended to be limited in any way.Except foregoing description
Schematical aspect, except embodiment and feature, by reference to attached drawing and the following detailed description, the present invention is further
Aspect, embodiment and feature, which will be, to be readily apparent that.
Detailed description of the invention
In the accompanying drawings, unless specified otherwise herein, otherwise indicate the same or similar through the identical appended drawing reference of multiple attached drawings
Component or element.What these attached drawings were not necessarily to scale.It should be understood that these attached drawings depict only according to the present invention
Disclosed some embodiments, and should not serve to limit the scope of the present invention.
Fig. 1 is the flow chart for the dialogue generation method that embodiment of the present invention provides.
Fig. 2 is the flow diagram for the dialogue generation method that embodiment of the present invention provides.
Fig. 3 is the schematic diagram for the more new knowledge base that embodiment of the present invention provides.
Fig. 4 is the flow diagram of dialogue generation method knowledge base and key value library that embodiment of the present invention provides.
Fig. 5 is the structural schematic diagram for the dialogue generating means that embodiment of the present invention provides.
Fig. 6 is that the dialogue that embodiment of the present invention provides generates terminal structure schematic diagram.
Specific embodiment
Hereinafter, certain exemplary embodiments are simply just described.As one skilled in the art will recognize that
Like that, without departing from the spirit or scope of the present invention, described embodiment can be modified by various different modes.
Therefore, attached drawing and description are considered essentially illustrative rather than restrictive.
The embodiment of the invention provides a kind of dialogue generation methods, as shown in Figure 1 and Figure 2, including step：
S100：Query semantics information is extracted from query statement.
S200：Multiple key-value pairs relevant to query semantics information are searched in key value library (K-V, Key-Value).
S300：Obtain auxiliary information relevant to query semantics information.
S400：Auxiliary information is weighted with each key-value pair found, obtains replying semantic information.
S500：Semantic information will be replied and be converted to answer sentence.
It should be noted that although describing operating process in the present embodiment with particular order, this is not required that
Or hint must execute these operations in this particular order.Additionally or alternatively, it is convenient to omit certain steps, it will be more
A step is merged into a step and is executed, and/or a step is decomposed into execution of multiple steps.For example, step S200 can be first
It is executed in step S300 or step S300 can also be executed prior to step S200 or step S200 and step S300 is performed simultaneously, no
By which kind of operation will not the realization result to the present embodiment method have an impact.
In one embodiment, the query statement in step S100 can be understood as the language of the speech form of user's input
The sentence of sentence or textual form.Answer sentence in step S500 can be understood as the sentence or text shape of speech form
The sentence of formula.Query statement is converted into query semantics information by encoder.Semantic information conversion will be replied by decoder
To reply sentence.
It not only include the clause ingredient in every-day language in the query semantics information of user, it is also possible to include
About the inquiry of knowledge information, thus it is related to the clause ingredient of query semantics information and dialogue in addition to being found from key value library
Key-value pair except, building key-value pair associated with knowledge information can also be searched.Knowledge information is stored with so introducing
Knowledge base.
In one embodiment, multiple key-value pairs relevant to query semantics information are searched in key value library, including：
Knowledge semantic information is extracted from query semantics information.
Knowledge corpus relevant to knowledge semantic information is obtained from knowledge base.
Knowledge semantic information and knowledge corpus are formed into key-value pair.
Know for example, user's input " height of Yao Ming and age are how many ", Yao Ming's height therein and age data belong to
Know semantic information, retrieval can be exported about Yao Ming's height and the data at age as knowledge corpus in instruction library.
Since the knowledge corpus instruction found in knowledge base and knowledge semantic information are to matched data, and the two is not
Key-value pair is formed, therefore knowledge semantic information and knowledge corpus can be formed by key-value pair by neural network.
In one embodiment, as shown in figure 3, the data in knowledge base can pass through server update (server
Update knowledge memory (Knowledge Memory) is formed) to improve the data in knowledge base.
In one embodiment, knowledge base can be the subdata base in key value library, and knowledge base is also possible to and key value library
Database arranged side by side.As shown in figure 4, when knowledge semantic information and knowledge corpus will be formed key assignments by neural network by knowledge base
To rear, key-value pair is moved in key value library, in order to which key value library pair key-value pair relevant to query semantics information screens
And output.Wherein, MLP (Multi-Layer Perceptron, multilayer neural network) or RNN can be used in neural network
(Recurrent Neural Network, Recognition with Recurrent Neural Network).The type of neural network structure model can carry out as needed
Selection, however it is not limited to the neural network limited in the present embodiment, it should be understood that as long as can be realized above-mentioned technical proposal, can adopt
With any neural network structure model in the prior art.
In one embodiment, it is searched in key value library (K-V, Key-Value) relevant to query semantics information multiple
Key-value pair, including：By " being worth, Value " is extracted and shape V list for related key-value pair.Each " value " in Vlist is added
Power.
In one embodiment, in order to comprehensively finding out key assignments relevant to query semantics information from key value library
It is right, softmax function can be used and searched.
In one embodiment, auxiliary information relevant to query semantics information is obtained, it can be understood as inquire with user
Essence in sentence inquires the not high information of content correlation degree.Auxiliary information can be the information such as the sound of user or the tone.
For example, sound is sex, sound is any information such as adult or child, speech intonation, word speed.Or by knowing
Know the information related with certain keyword of the essence inquiry content in user's INQUIRE statement or word that map is associated.Pass through auxiliary
Information can make the answer semantic information generated by weighting more accurate perfect.
For example, the key-value pair associated with query semantics information found by key value library includes：<You are that male is female？,
Male>With<You are boy？, I is girl>Two groups of key-value pairs, and in auxiliary information include user inquiry when sound be male
Sound, the information when being weighted about " male " will be strengthened, so that the answer semantic information ultimately produced is quasi-
Exactness is higher.
In one embodiment, as shown in Fig. 2, can also include step after obtaining replying semantic information：
Semantic information will be replied and query semantics information forms question and answer key-value pair.
The existing key-value pair stored in question and answer key-value pair and key value library is matched, if in the existing key-value pair of key value library
There is no the question and answer key-value pairs, then question and answer key-value pair are stored in key value library, to supplement the corpus in key value library (key-value pair).It is logical
Crossing the above method may be implemented continuous updating to key value library, and achieving the purpose that enrich constantly extends key value library, enhance key value library
Learning ability.
In a specific embodiment, the existing key-value pair progress that will be stored in question and answer key-value pair and key value library
Match, if question and answer key-value pair is not present in the existing key-value pair of key value library, question and answer key-value pair is stored in key value library, specific steps
Including：
Question and answer key-value pair is kept in first in the short-term memory area of key value library.Short-term memory area can use for storage is current
The region of family dialogue corpus.
Question and answer key-value pair is assessed by assessment models, if the threshold value of assessment question and answer key-value pair exceeds existing key-value pair
Threshold value, then question and answer key-value pair is deposited into the long-term memory area of key value library.It long-term memory area can be useful to be stored with
Talk with the region of corpus or preferred dialogue corpus in family.
In order to increase the expressed in abundance degree of the answer sentence ultimately generated, avoids clause excessively single, language can replied
Stochastic variable is introduced in adopted information, the language expression mode for replying sentence is adjusted by stochastic variable, it then again will be adjusted
Sentence is replied to decode and export.
The embodiment of the invention provides a kind of dialogue generating means, as shown in figure 5, device includes：
Extraction module 10, for extracting query semantics information from query statement；
Module 20 is obtained, for searching multiple key-value pairs relevant to query semantics information in key value library；
Supplementary information module 30, for obtaining auxiliary information relevant to query semantics information；
Processing module 40 obtains replying semantic information for auxiliary information to be weighted with each key-value pair found；
Conversion module 50 is converted to answer sentence for that will reply semantic information.
In one embodiment, obtaining module 20 includes：
Extracting sub-module, for extracting knowledge semantic information from query semantics information；
Acquisition submodule, for obtaining knowledge corpus relevant to knowledge semantic information from knowledge base；
Key-value pair submodule, for knowledge semantic information and knowledge corpus to be built into key-value pair.
In one embodiment, dialogue generating means further include：
Matching module forms question and answer key-value pair for that will reply semantic information and query semantics information；
Enlargement module works as key value library for matching the existing key-value pair stored in question and answer key-value pair and key value library
Existing key-value pair in be not present question and answer key-value pair, then will question and answer key-value pair be stored in key value library in.
In a specific embodiment, enlargement module includes：
Short memory submodule, in the short-term memory area for question and answer key-value pair to be kept in key value library；
Long memory submodule, for being assessed by assessment models question and answer key-value pair, if assessment question and answer key-value pair
Threshold value exceeds the threshold value of existing key-value pair, then question and answer key-value pair is deposited into the long-term memory area of key value library.
The embodiment of the invention provides a kind of dialogues to generate terminal, as shown in fig. 6, including：
Memory 910 and processor 920 are stored with the computer journey that can be run on processor 920 in memory 910
Sequence.Processor 920 realizes the dialogue generation method in above-described embodiment when executing computer program.Memory 910 and processor
920 quantity can be one or more.
Communication interface 930 is communicated for memory 910 and processor 920 with outside.
Memory 910 may include high speed RAM memory, it is also possible to further include nonvolatile memory (non-
Volatile memory), a for example, at least magnetic disk storage.
If memory 910, processor 920 and the independent realization of communication interface 930, memory 910, processor 920
And communication interface 930 can be connected with each other by bus and complete mutual communication.Bus can be Industry Standard Architecture
Structure (ISA, Industry Standard Architecture) bus, external equipment interconnection (PCI, Peripheral
Component) bus or extended industry-standard architecture (EISA, Extended Industry Standard
Component) bus etc..Bus can be divided into address bus, data/address bus, control bus etc..For convenient for indicating, in Fig. 6 only
It is indicated with a thick line, it is not intended that an only bus or a type of bus.
Optionally, in specific implementation, if memory 910, processor 920 and communication interface 930 are integrated in one piece
On chip, then memory 910, processor 920 and communication interface 930 can complete mutual communication by internal interface.
The embodiment of the invention provides a kind of computer readable storage mediums, are stored with computer program, the program quilt
Processor execute when realize embodiment one include it is any as described in dialogue generation method.
The embodiment of the present invention is by except according to the relevant key-value pair of query semantics information searching, also introducing auxiliary letter
Breath realizes and generates answer sentence in such a way that much information combines, so that the answer sentence ultimately generated is more accurate.And
And since the auxiliary information can be any information, it can according to need and be adjusted.The key value library of the embodiment of the present invention
In key-value pair since itself is there are mapping relations, can guarantee the answer sentence generated and inquiry language to a certain extent
Sentence is associated.Due to introducing knowledge base in the embodiment of the present invention, knowledge language related with knowledge can be accurately obtained
Adopted information, and key-value pair can be generated at any time by neural network and arrived in key value library, the expansion to key value library is realized, is made
The answer sentence that must be generated is more complete.And construction of knowledge base is simple, and Input knowledge semantic information is inquired, and reduces
Calculating time loss improves the process that entire dialogue generates.
In the description of this specification, reference term " one embodiment ", " some embodiments ", " example ", " specifically show
The description of example " or " some examples " etc. means specific features, structure, material or spy described in conjunction with this embodiment or example
Point is included at least one embodiment or example of the invention.Moreover, particular features, structures, materials, or characteristics described
It may be combined in any suitable manner in any one or more of the embodiments or examples.In addition, without conflicting with each other, this
The technical staff in field can be by the spy of different embodiments or examples described in this specification and different embodiments or examples
Sign is combined.
In addition, term " first ", " second " are used for descriptive purposes only and cannot be understood as indicating or suggesting relative importance
Or implicitly indicate the quantity of indicated technical characteristic." first " is defined as a result, the feature of " second " can be expressed or hidden
It include at least one this feature containing ground.In the description of the present invention, the meaning of " plurality " is two or more, unless otherwise
Clear specific restriction.
Any process described otherwise above or method description are construed as in flow chart or herein, and expression includes
It is one or more for realizing specific logical function or process the step of executable instruction code module, segment or portion
Point, and the range of the preferred embodiment of the present invention includes other realization, wherein can not press shown or discussed suitable
Sequence, including according to related function by it is basic simultaneously in the way of or in the opposite order, to execute function, this should be of the invention
Embodiment person of ordinary skill in the field understood.
Expression or logic and/or step described otherwise above herein in flow charts, for example, being considered use
In the order list for the executable instruction for realizing logic function, may be embodied in any computer-readable medium, for
Instruction executing device, device or equipment (such as computer based device, including the device of processor or other can be held from instruction
Luggage sets, device or equipment instruction fetch and the device executed instruction) it uses, or combine these instruction executing devices, device or set
It is standby and use.For the purpose of this specification, " computer-readable medium ", which can be, any may include, stores, communicates, propagates or pass
Defeated program is for instruction executing device, device or equipment or the dress used in conjunction with these instruction executing devices, device or equipment
It sets.The more specific example (non-exhaustive list) of computer-readable medium includes following：Electricity with one or more wiring
Interconnecting piece (electronic device), portable computer diskette box (magnetic device), random access memory (RAM), read-only memory
(ROM), erasable edit read-only storage (EPROM or flash memory), fiber device and portable read-only memory
(CDROM).In addition, computer-readable medium can even is that the paper that can print described program on it or other suitable Jie
Matter, because can then be edited, be interpreted or when necessary with other for example by carrying out optical scanner to paper or other media
Suitable method is handled electronically to obtain described program, is then stored in computer storage.
It should be appreciated that each section of the invention can be realized with hardware, software, firmware or their combination.Above-mentioned
In embodiment, software that multiple steps or method can be executed in memory and by suitable instruction executing device with storage
Or firmware is realized.It, and in another embodiment, can be under well known in the art for example, if realized with hardware
Any one of column technology or their combination are realized：With for realizing the logic gates of logic function to data-signal
Discrete logic, with suitable combinational logic gate circuit specific integrated circuit, programmable gate array (PGA), scene
Programmable gate array (FPGA) etc..
Those skilled in the art are understood that realize all or part of step that above-described embodiment method carries
It suddenly is that relevant hardware can be instructed to complete by program, the program can store in a kind of computer-readable storage medium
In matter, which when being executed, includes the steps that one or a combination set of embodiment of the method.
It, can also be in addition, each functional unit in each embodiment of the present invention can integrate in a processing module
It is that each unit physically exists alone, can also be integrated in two or more units in a module.Above-mentioned integrated mould
Block both can take the form of hardware realization, can also be realized in the form of software function module.The integrated module is such as
Fruit is realized and when sold or used as an independent product in the form of software function module, also can store in a computer
In readable storage medium storing program for executing.The storage medium can be read-only memory, disk or CD etc..
The above description is merely a specific embodiment, but scope of protection of the present invention is not limited thereto, any
Those familiar with the art in the technical scope disclosed by the present invention, can readily occur in its various change or replacement,
These should be covered by the protection scope of the present invention.Therefore, protection scope of the present invention should be with the guarantor of the claim
It protects subject to range.
Claims (12)
1. a kind of dialogue generation method, which is characterized in that including：
Query semantics information is extracted from query statement；
Multiple key-value pairs relevant to the query semantics information are searched in key value library；
Obtain auxiliary information relevant to the query semantics information；
The auxiliary information is weighted with each key-value pair found, obtains replying semantic information；
The answer semantic information is converted into answer sentence.
2. the method as described in claim 1, which is characterized in that searched in key value library relevant to the query semantics information
Multiple key-value pairs, including：
Knowledge semantic information is extracted from the query semantics information；
Knowledge corpus relevant to the knowledge semantic information is obtained from knowledge base；
The knowledge semantic information and the knowledge corpus are formed into the key-value pair.
3. method according to claim 2, which is characterized in that the knowledge semantic information and the knowledge corpus are formed institute
Key-value pair is stated, including：
The knowledge semantic information and the knowledge corpus are formed into the key-value pair by neural network.
4. the method as described in claim 1, which is characterized in that further include：
The answer semantic information and the query semantics information are formed into question and answer key-value pair；
The existing key-value pair stored in the question and answer key-value pair and the key value library is matched, if the key value library is described
The question and answer key-value pair is not present in existing key-value pair, then the question and answer key-value pair is stored in the key value library.
5. method as claimed in claim 4, which is characterized in that will be stored in the question and answer key-value pair and the key value library
There is key-value pair to be matched, it, will be described if the question and answer key-value pair is not present in the existing key-value pair of the key value library
Question and answer key-value pair is stored in the key value library, and specific steps include：
The question and answer key-value pair is kept in the short-term memory area of the key value library；
The question and answer key-value pair is assessed by assessment models, if assess the threshold value of the question and answer key-value pair beyond it is described
There is the threshold value of key-value pair, then the question and answer key-value pair is deposited into the long-term memory area of the key value library.
6. the method as described in claim 1, which is characterized in that further include：
The language expression mode for replying sentence is adjusted by stochastic variable.
7. a kind of dialogue generating means, which is characterized in that including：
Extraction module, for extracting query semantics information from query statement；
Module is obtained, for searching multiple key-value pairs relevant to the query semantics information in key value library；
Supplementary information module, for obtaining auxiliary information relevant to the query semantics information；
Processing module obtains replying semantic letter for the auxiliary information to be weighted with each key-value pair found
Breath；
Conversion module, for the answer semantic information to be converted to answer sentence.
8. device as claimed in claim 7, which is characterized in that the acquisition module includes：
Extracting sub-module, for extracting knowledge semantic information from the query semantics information；
Acquisition submodule, for obtaining knowledge corpus relevant to the knowledge semantic information from knowledge base；
Key-value pair submodule, for the knowledge semantic information and the knowledge corpus to be built into the key-value pair.
9. device as claimed in claim 7, which is characterized in that further include：
Matching module, for the answer semantic information and the query semantics information to be formed question and answer key-value pair；
Enlargement module works as institute for matching the existing key-value pair stored in the question and answer key-value pair and the key value library
It states and the question and answer key-value pair is not present in the existing key-value pair of key value library, then the question and answer key-value pair is stored in the key assignments
In library.
10. device as claimed in claim 9, which is characterized in that the enlargement module includes：
Short memory submodule, for keeping in the question and answer key-value pair in the short-term memory area of the key value library；
Long memory submodule, for being assessed by assessment models the question and answer key-value pair, if assessing the question and answer key assignments
Pair threshold value exceed the existing key-value pair threshold value, then the question and answer key-value pair is deposited into the long-term memory of the key value library
Qu Zhong.
11. a kind of dialogue generates terminal, which is characterized in that including：
One or more processors；
Storage device, 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
Realize such as method described in any one of claims 1 to 6.
12. a kind of computer readable storage medium, is stored with computer program, which is characterized in that the program is held by processor
Such as method described in any one of claims 1 to 6 is realized when row.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810694671.0A CN108846125A (en) | 2018-06-29 | 2018-06-29 | Talk with generation method, device, terminal and computer readable storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810694671.0A CN108846125A (en) | 2018-06-29 | 2018-06-29 | Talk with generation method, device, terminal and computer readable storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108846125A true CN108846125A (en) | 2018-11-20 |
Family
ID=64200736
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810694671.0A Pending CN108846125A (en) | 2018-06-29 | 2018-06-29 | Talk with generation method, device, terminal and computer readable storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108846125A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110276071A (en) * | 2019-05-24 | 2019-09-24 | 众安在线财产保险股份有限公司 | A kind of text matching technique, device, computer equipment and storage medium |
| CN111401922A (en) * | 2020-03-09 | 2020-07-10 | 联想(北京)有限公司 | Question and answer information processing method and device and computer equipment |
| CN111563029A (en) * | 2020-03-13 | 2020-08-21 | 深圳市奥拓电子股份有限公司 | Testing method, system, storage medium and computer equipment for conversation robot |
| CN113194346A (en) * | 2019-11-29 | 2021-07-30 | 广东海信电子有限公司 | Display device |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104239459A (en) * | 2014-09-02 | 2014-12-24 | 百度在线网络技术（北京）有限公司 | Voice search method, voice search device and voice search system |
| CN106919655A (en) * | 2017-01-24 | 2017-07-04 | 网易（杭州）网络有限公司 | A kind of answer provides method and apparatus |
| WO2018097907A1 (en) * | 2016-10-20 | 2018-05-31 | Google Llc | Answer to question neural networks |
| CN108108426A (en) * | 2017-12-15 | 2018-06-01 | 杭州网蛙科技有限公司 | Understanding method, device and the electronic equipment that natural language is putd question to |
| CN108197167A (en) * | 2017-12-18 | 2018-06-22 | 深圳前海微众银行股份有限公司 | Human-computer dialogue processing method, equipment and readable storage medium storing program for executing |
-
2018
- 2018-06-29 CN CN201810694671.0A patent/CN108846125A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104239459A (en) * | 2014-09-02 | 2014-12-24 | 百度在线网络技术（北京）有限公司 | Voice search method, voice search device and voice search system |
| WO2018097907A1 (en) * | 2016-10-20 | 2018-05-31 | Google Llc | Answer to question neural networks |
| CN106919655A (en) * | 2017-01-24 | 2017-07-04 | 网易（杭州）网络有限公司 | A kind of answer provides method and apparatus |
| CN108108426A (en) * | 2017-12-15 | 2018-06-01 | 杭州网蛙科技有限公司 | Understanding method, device and the electronic equipment that natural language is putd question to |
| CN108197167A (en) * | 2017-12-18 | 2018-06-22 | 深圳前海微众银行股份有限公司 | Human-computer dialogue processing method, equipment and readable storage medium storing program for executing |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110276071A (en) * | 2019-05-24 | 2019-09-24 | 众安在线财产保险股份有限公司 | A kind of text matching technique, device, computer equipment and storage medium |
| CN110276071B (en) * | 2019-05-24 | 2023-10-13 | 众安在线财产保险股份有限公司 | Text matching method and device, computer equipment and storage medium |
| CN113194346A (en) * | 2019-11-29 | 2021-07-30 | 广东海信电子有限公司 | Display device |
| CN111401922A (en) * | 2020-03-09 | 2020-07-10 | 联想(北京)有限公司 | Question and answer information processing method and device and computer equipment |
| CN111563029A (en) * | 2020-03-13 | 2020-08-21 | 深圳市奥拓电子股份有限公司 | Testing method, system, storage medium and computer equipment for conversation robot |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111191016B (en) | Multi-round dialogue processing method and device and computing equipment | |
| CN110377716B (en) | Interaction method and device for conversation and computer readable storage medium | |
| CN110782870A (en) | Speech synthesis method, speech synthesis device, electronic equipment and storage medium | |
| Sun et al. | Deep and shallow features fusion based on deep convolutional neural network for speech emotion recognition | |
| Yu et al. | Sequential labeling using deep-structured conditional random fields | |
| CN113239700A (en) | Text semantic matching device, system, method and storage medium for improving BERT | |
| CN112182252B (en) | Intelligent medication question-answering method and device based on medicine knowledge graph | |
| CN108846125A (en) | Talk with generation method, device, terminal and computer readable storage medium | |
| CN105094315A (en) | Method and apparatus for smart man-machine chat based on artificial intelligence | |
| CN113239169B (en) | Answer generation method, device, equipment and storage medium based on artificial intelligence | |
| CN109325242B (en) | Method, device and equipment for judging whether sentences are aligned based on word pairs and translation | |
| CN112214585B (en) | Reply message generation method, system, computer device and storage medium | |
| CN108959556A (en) | Entity answering method, device and terminal neural network based | |
| CN110853626A (en) | Bidirectional attention neural network-based dialogue understanding method, device and equipment | |
| CN113987179A (en) | Knowledge enhancement and backtracking loss-based conversational emotion recognition network model, construction method, electronic device and storage medium | |
| US11783179B2 (en) | System and method for domain- and language-independent definition extraction using deep neural networks | |
| CN110197279A (en) | Transformation model training method, device, equipment and storage medium | |
| CN107832439A (en) | Method, system and the terminal device of more wheel state trackings | |
| CN110990555A (en) | End-to-end retrieval type dialogue method and system and computer equipment | |
| CN112183106B (en) | Semantic understanding method and device based on phoneme association and deep learning | |
| US20230094730A1 (en) | Model training method and method for human-machine interaction | |
| CN113761843B (en) | Voice editing method, electronic device and computer readable storage medium | |
| CN113761124B (en) | Training method of text coding model, information retrieval method and equipment | |
| CN113326383B (en) | Short text entity linking method, device, computing equipment and storage medium | |
| CN116955579B (en) | Chat reply generation method and device based on keyword knowledge retrieval |
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 |