CN107704521A - A kind of question and answer processing server, client and implementation method - Google Patents
A kind of question and answer processing server, client and implementation method Download PDFInfo
- Publication number
- CN107704521A CN107704521A CN201710801967.3A CN201710801967A CN107704521A CN 107704521 A CN107704521 A CN 107704521A CN 201710801967 A CN201710801967 A CN 201710801967A CN 107704521 A CN107704521 A CN 107704521A
- Authority
- CN
- China
- Prior art keywords
- question
- answer
- vector
- sentence
- language material
- 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
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
- G06F16/35—Clustering; Classification
- G06F16/355—Class or cluster creation or modification
-
- 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
Abstract
The invention discloses a kind of question and answer processing server, client and implementation method, the server includes:Term vector training unit and sentence vector generation dispensing unit, the term vector training unit, to go out the language material of natural language according to different question and answer scene configurations, and train to obtain term vector model by the language material, the sentence vector generation dispensing unit, according to the term vector model, the sentence in the language material to be converted to the vector in vector space and obtains a vector model, and provides the configuration interface of the self-defined weight in this vector model.The language material of present invention training term vector can allow user to replace, the adjustment of weight can be carried out for the keyword in sentence vector, so as to train different term vector models according to different session operational scenarios configurations, so that the degree of accuracy of model is higher, so as to improve the matching accuracy rate of question and answer processing server.In addition, the present invention can also meet the vocabulary processing of user's particular/special requirement, the quiz answers more met are matched.
Description
Technical field
The present invention relates to computer software fields, natural language processing field, more particularly to a kind of question and answer processing server,
Client and implementation method.
Background technology
Question answering system refers to such a scene:For a computer system, when user inputs a natural language
During the problem of (Natural Language), the question answering system can be found in default question and answer storehouse and the meaning asked questions
(being related to natural language processing) compares the answer of fitting and return.
At this stage in general way be all by question sentence by natural language model conversation for high-dimensional vector space vector (i.e.
Sentence vector), then it is compared with the question sentence in system and finally provides answer.But for the no applied field of question answering system
Scape, the algorithm of generation sentence vector often have differences, it is impossible to dynamically adapt to a variety of scenes so that final effect is past
Past is not especially good.
The content of the invention
The technical problem to be solved in the present invention is to provide a kind of question and answer processing server, by existing conventional method
On the basis of, the links for training sentence vector are become into configurable variable, so as to realize that the system all has to different scenes
There is preferable effect.
Present invention also offers client, for there is the user of specific demand that it can be allowed to have enough flexibility ratios to configure
The characteristic of whole system, so as to better meet the particular/special requirement of user.
Solve above-mentioned technical problem, the invention provides a kind of question and answer processing server, including:Term vector training unit and
Sentence vector generation dispensing unit,
The term vector training unit, to go out the language material of natural language according to different question and answer scene configurations, and pass through institute
Predicate material trains to obtain term vector model,
The sentence vector generation dispensing unit, according to the term vector model, the sentence in the language material to be changed
A vector model is obtained for the vector in vector space, and the configuration interface of the self-defined weight in this vector model is provided.
Further, the method for the language material for going out natural language according to different question and answer scene configurations is:
Training for different question and answer scenes obtains corresponding term vector model,
Pass through upload/importing and the language material of the natural language of the question and answer scene matching.
Further, self-defined weight configuration interface carries out assigning power manually by word attribute in sentence.
Further, server also includes:Question and answer storehouse pretreatment unit, to deposit according to different question and answer scene question and answer
Answer.
Further, server also includes:Matching unit, to match question and answer difference question and answer field based on sentence vector space
The answer of scape question and answer.
Further, server also to:One API service interface is provided.
The invention provides a kind of client, including:
Go out the language material interface of natural language according to different question and answer scene configurations,
And, there is provided the self-defined weight configuration interface in sentence vector model,
Train to obtain term vector model by the language material, further according to the term vector model, by the sentence in the language material
The vector that son is converted in vector space obtains a vector model,
Pass through web server in the client, there is provided the interface of above-mentioned calling altogether.
Further, question and answer result is obtained by accessing API server.
Further, client also includes:Manual configuration interface, to provide matching somebody with somebody for the specific vocabulary in a vector model
Put weights interface.
The invention provides a kind of implementation method of question and answer processing, comprise the following steps:
Go out the language material of natural language according to different question and answer scene configurations, and train by the language material to obtain term vector mould
Type,
According to the term vector model, the sentence of the language material is converted to the vector in vector space and obtains sentence vector mould
Type, and the configuration interface of the self-defined weight in this vector model is provided.
Beneficial effects of the present invention:
1) the question and answer processing server in the present invention, can be according to different question and answer scenes due to including term vector training unit
The language material of natural language is configured, and trains by the language material to obtain term vector model, can be according to different session operational scenarios
Different term vector models is trained in configuration so that the degree of accuracy of model is higher., can root due to generating dispensing unit including sentence vector
According to the term vector model, the sentence in the language material is converted to the vector in vector space and obtains a vector model, and is carried
For the self-defined weight configuration interface in this vector model, the weight in sentence vector generating algorithm can adjust so that entirely ask
The matching accuracy rate for answering system is higher.So as to the question and answer processing server in the present invention, the links for training sentence vector are become
Into configurable variable, so as to realize that the system has the pretty good effect of comparison to different scenes.
2) further, due to providing the self-defined weight configuration interface in providing sentence vector model in client, from
And for there is the user of specific demand that it can be allowed to have enough flexibility ratios to configure the characteristic of whole system, can be more preferable
Meet the particular/special requirement of user.For example it can be assigned manually by the part of speech of word, the frequency of occurrences even some special words
Power, so as to reach best effect and Consumer's Experience.
3) present invention trains question and answer data pattern file, then is provided out servicing in the form of API server, can answer
For several scenes, the calling of different clients.
Brief description of the drawings
Fig. 1 is the question and answer processing server schematic diagram in one embodiment of the invention;
Fig. 2 is the client terminal structure schematic diagram in one embodiment of the invention;
Fig. 3 is the implementation method schematic flow sheet in one embodiment of the invention;
Fig. 4 is the question and answer processing server structural representation in one embodiment of the present invention;
Fig. 5 is the concrete operations schematic diagram on the question and answer processing server in Fig. 4.
Embodiment
The principle of the disclosure is described referring now to some example embodiments.It is appreciated that these embodiments are merely for saying
It is bright and help it will be understood by those skilled in the art that with the purpose of the embodiment disclosure and describe, rather than suggest the model to the disclosure
Any restrictions enclosed.Content of this disclosure described here can in a manner of described below outside various modes implement.
As described herein, term " comprising " and its various variants are construed as open-ended term, it means that " bag
Include but be not limited to ".Term "based" is construed as " being based at least partially on ".Term " one embodiment " it is understood that
For " at least one embodiment ".Term " another embodiment " is construed as " at least one other embodiment ".
Refer to Fig. 1 is the question and answer processing server schematic diagram in one embodiment of the invention, and one kind in the present embodiment is asked
Processing server is answered, including:Term vector training unit 1 and sentence vector generation dispensing unit 2, the term vector training unit 1, are used
To go out the language material of natural language according to different question and answer scene configurations, and train by the language material to obtain term vector model, it is described
Sentence vector generation dispensing unit 2, according to the term vector model, the sentence in the language material is converted in vector space
Vector obtain a vector model, and provide the self-defined weight in this vector model configuration interface.Specifically, in term vector
The language material of a large amount of natural languages is needed in training unit 1, calculates term vector model.The problem of natural language understanding, will be converted into machine
The problem of device learns, each vocabulary is shown as in natural language processing (NLP, Natural Language Processing)
One very long vector.This vectorial dimension is vocabulary size, wherein most elements are 0, the value of only one dimension
For 1, this dimension just represents current word.The training of term vector may include but be not limited to, word2vec algorithms or GLOVE
Algorithm.The specific implementation of those algorithms has all had corresponding open source projects such as gensim to provide.Serviced as question and answer processing
Device only needs to call the interface of these projects.Further, the term vector mould trained in the term vector training unit 1
Type is in fact one and have recorded all words to the file of the point of a hyperspace, be exactly per a line in file word, to
Amount } form.The main purpose of the term vector model is that the vocabulary in natural language is mapped into a hyperspace.Specifically
Ground, the process entirely mapped are that a line started with target word is found in term vector model file, then the vector of the row is
The vector that the word is mapped to.Vocabulary can to a certain extent express between corresponding vector express the meaning between vocabulary in space
Contact.
In certain embodiments, using CBOW (the CBoW MODEL C ontinuous Bag-of- in word2vec algorithms
Words Model) calculate term vector model.
In certain embodiments, term vector model is calculated using the Skip-gram in word2vec algorithms.Skip-gram
The essence of model is that the cosine between the input vector and target word that calculate input word output vector is similar
Degree, and carry out softmax normalization.
In certain embodiments, term vector model is calculated using GLOVE algorithms.
As preferred in the present embodiment, the method for the language material for going out natural language according to different question and answer scene configurations is further
For:Training for different question and answer scenes obtains corresponding term vector model, passes through upload/importing and the question and answer scene matching
Natural language language material.In practice in use, the training in the term vector space for different scenes, if using more conforming to work as
The language material of preceding scene, effect have very big lifting.Upload/the importing includes but is not limited to, and user uploads corpus
Content imports language material bag.For example using the demand of the question and answer processing server in the present embodiment it is to be based on question answering system
The rigorous knowledge type question answering system of linguistic organization, if the effect of the term vector model generated using common chat language material is just very
Difference, and use chat effect as language close to the rigorous text of review, encyclopaedia etc will be more preferable.So by uploading/leading
Enter the language material text of specialty, the accuracy that matching is answered can be lifted.Because the language material of the training of term vector is to be matched somebody with somebody
Put, so user can more conform to the language material of its scene to reach more preferable effect such as table 1 below by upload/importing.
Table 1
Scene | Language material | Mode |
Public safety | Wal-Mart's encyclopaedia language material | Language material uploads |
Shopping guide's scene | Daily shopping language material | Language material API |
Virtual reality scenario | VR language materials | Language material uploads |
SUV automobile scenes | Automobile chooses language material | Language material imports |
Bit coin scene | Block chain language material | Language material uploads |
Lunar exploration scene | Aero-Space language material | Language material imports |
Sentence vector generation dispensing unit 2 in the present embodiment, the generation knot based on the term vector training unit 1
Fruit, the sentence in the language material is converted to the vector in vector space and obtains a vector model, and this vector model is provided
In self-defined weight configuration interface.Specifically, the generation result based on term vector, for the sentence of a natural language, and
A vector being converted into vector space.Those skilled in the art can understand, a kind of enforceable method be for
Each word in sentence, calculates its term vector, and the sentence vector of referred to as whole sentence is being weighted by certain mode.
Concrete implementation mode in certain embodiments is:
If the sentence s of natural language can be divided into n word, word1, word2 ... wordn, by described by participle
Term vector model in term vector training unit 1,
Those words can correspond into n vector:V1, v2 ... vn, can generate a user in sentence vector generation module can
With the weight function W (word) of self-defining, then final sentence vector is equal to vi*W (wordi) i=1 to n is added.This reality
Interface can be configured by providing the self-defined weight in this vector model by applying in example, frequency can occur by the part of speech of word
It is the W (word) that rate even some special words, which carry out assigning power manually, so as to reach best effect.
As preferred in the present embodiment, self-defined weight configuration interface, which passes through word attribute in sentence, to carry out assigning power manually.
Refer to Fig. 2 is the client terminal structure schematic diagram in one embodiment of the invention, a kind of client in the present embodiment,
Including:Go out the language material interface 5 of natural language according to different question and answer scene configurations, and, there is provided it is self-defined in sentence vector model
Weight configures interface 6, trains to obtain term vector model by the language material, further according to the term vector model, by the language material
In the vector that is converted in vector space of sentence obtain a vector model, pass through web server in the client, there is provided altogether
The interface of above-mentioned calling.Due to including term vector training unit in the server, nature can be gone out according to different question and answer scene configurations
The language material of language, and train by the language material to obtain term vector model, can be according to different session operational scenarios configuration training not
Same term vector model so that the degree of accuracy of model is higher.Further, since it is single also to include sentence vector generation configuration in the server
Member, the sentence in the language material can be converted to the vector in vector space and obtain sentence vector mould according to the term vector model
Type, and the configuration interface of the self-defined weight in this vector model is provided, the weight in sentence vector generating algorithm can adjust so that
The matching accuracy rate of whole question answering system is higher.So as to the question and answer processing server in the present invention, each of sentence vector will be trained
Link becomes configurable variable, so as to realize that the system has the pretty good effect of comparison to different scenes.Lead in client
Too short link may have access to above-mentioned question and answer processing server.
The client can be the client such as PC, Android, iPhone, WP, iPad, Mac.HTML5 bis- can also be based on
The access window of secondary exploitation.User can go out the language material of natural language by the client according to different question and answer scene configurations
Simultaneously weight adjustment can be carried out by providing the configuration interface of the self-defined weight in this vector model.
As preferred in the present embodiment, question and answer result is obtained by accessing API server.By calling API to connect
Mouthful, obtain answer/reply of question and answer problem.
As preferred in the present embodiment, client also includes:Manual configuration interface, to provide in a vector model
The configuration weights interface of specific vocabulary.
Refer to Fig. 3 is the implementation method schematic flow sheet in one embodiment of the invention, the implementation method in the present embodiment
Including:
Step S100 goes out the language material of natural language according to different question and answer scene configurations, and trains to obtain word by the language material
Vector model,
According to the term vector model, the vector sentence of the language material is converted in vector space obtains step S101
Sentence vector model, and the configuration interface of the self-defined weight in this vector model is provided
It refer to Fig. 4 and Fig. 5, the question and answer processing server in the present embodiment, it includes:Term vector training unit 1 and sentence
Vector generation dispensing unit 2, the term vector training unit 1, to go out the language of natural language according to different question and answer scene configurations
Material, and train to obtain term vector model by the language material, the sentence vector generation dispensing unit 2, to according to institute's predicate to
Model is measured, the sentence in the language material is converted to the vector in vector space and obtains a vector model, and this vector is provided
Self-defined weight configuration interface in model.As preferred in the present embodiment, question and answer processing server also includes:Question and answer storehouse is pre-
Processing unit 3, to deposit the answer according to different question and answer scene question and answer.As preferred in the present embodiment, question and answer processing clothes
Business device also includes:Matching unit 4, to match the answer of question and answer difference question and answer scene question and answer based on sentence vector space.Pass through
Term vector training unit 1, sentence vector generation dispensing unit 2 and question and answer storehouse pretreatment unit 3, train question and answer data model text
Part, it is provided out servicing providing an API (calling interface) service interface, the chat softwares such as wechat, qq can be based on.
It should be appreciated that each several part of the present invention can be realized with hardware, software, firmware or combinations thereof.Above-mentioned
In embodiment, software that multiple steps or method can be performed in memory and by suitable instruction execution system with storage
Or firmware is realized.If, and in another embodiment, can be with well known in the art for example, realized with hardware
Any one of row technology or their combination are realized:With the logic gates for realizing logic function to data-signal
Discrete logic, have suitable combinational logic gate circuit application specific integrated circuit, programmable gate array (PGA), scene
Programmable gate array (FPGA) etc..
In the description of this specification, reference term " one embodiment ", " some embodiments ", " example ", " specifically show
The description of example " or " some examples " etc. means specific features, structure, material or the spy for combining the embodiment or example description
Point is contained at least one embodiment or example of the present invention.In this manual, to the schematic representation of above-mentioned term not
Necessarily refer to identical embodiment or example.Moreover, specific features, structure, material or the feature of description can be any
One or more embodiments or example in combine in an appropriate manner.
In general, the various embodiments of the disclosure can be with hardware or special circuit, software, logic or its any combination
Implement.Some aspects can be implemented with hardware, and some other aspect can be with firmware or software implementation, and the firmware or software can
With by controller, microprocessor or other computing devices.Although the various aspects of the disclosure be shown and described as block diagram,
Flow chart is represented using some other drawing, but it is understood that frame described herein, equipment, system, techniques or methods can
With in a non limiting manner with hardware, software, firmware, special circuit or logic, common hardware or controller or other calculating
Equipment or some combinations are implemented.
Although in addition, operation is described with particular order, this is understood not to require this generic operation with shown suitable
Sequence is performed or performed with generic sequence, or requires that all shown operations are performed to realize expected result.In some feelings
Under shape, multitask or parallel processing can be favourable.Similarly, begged for although the details of some specific implementations is superincumbent
By comprising but these are not necessarily to be construed as any restrictions to the scope of the present disclosure, but the description of feature is only pin in
To specific embodiment.Some features described in some embodiments of separation can also be held in combination in single embodiment
OK.Mutually oppose, the various features described in single embodiment can also in various embodiments be implemented separately or to appoint
The mode of what suitable sub-portfolio is implemented.
Claims (10)
- A kind of 1. question and answer processing server, it is characterised in that including:Term vector training unit and sentence vector generation dispensing unit,The term vector training unit, to go out the language material of natural language according to different question and answer scene configurations, and pass through institute's predicate Material training obtains term vector model,Sentence vector generation dispensing unit, to according to the term vector model, by the sentence in the language material be converted to Vector in quantity space obtains a vector model, and provides the configuration interface of the self-defined weight in this vector model.
- 2. question and answer processing server according to claim 1, it is characterised in that go out nature according to different question and answer scene configurations The method of the language material of language is further:Training for different question and answer scenes obtains corresponding term vector model,Pass through upload/importing and the language material of the natural language of the question and answer scene matching.
- 3. question and answer processing server according to claim 1, it is characterised in that self-defined weight configuration interface passes through sentence Middle word attribute carries out assigning power manually.
- 4. question and answer processing server according to claim 1, it is characterised in that also include:Question and answer storehouse pretreatment unit, use To deposit the answer according to different question and answer scene question and answer.
- 5. the question and answer processing server according to claim 1 or 4, it is characterised in that also include:Matching unit, to base The answer of question and answer difference question and answer scene question and answer is matched in sentence vector space.
- 6. question and answer processing server according to claim 1, it is characterised in that also to:One API service interface is provided.
- A kind of 7. client, it is characterised in that including:Go out the language material interface of natural language according to different question and answer scene configurations,And, there is provided the self-defined weight configuration interface in sentence vector model,Train to obtain term vector model by the language material, further according to the term vector model, the sentence in the language material is turned The vector being changed in vector space obtains a vector model,Pass through web server in the client, there is provided the interface of above-mentioned calling altogether.
- 8. client according to claim 7, it is characterised in that obtain question and answer result by accessing API server.
- 9. client according to claim 7, it is characterised in that also include:Manual configuration interface, to provide sentence vector The configuration weights interface of specific vocabulary in model.
- 10. a kind of implementation method of question and answer processing, it is characterised in that comprise the following steps:Go out the language material of natural language according to different question and answer scene configurations, and train by the language material to obtain term vector model,According to the term vector model, the sentence of the language material is converted to the vector in vector space and obtains a vector model, And provide the configuration interface of the self-defined weight in this vector model.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710801967.3A CN107704521A (en) | 2017-09-07 | 2017-09-07 | A kind of question and answer processing server, client and implementation method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710801967.3A CN107704521A (en) | 2017-09-07 | 2017-09-07 | A kind of question and answer processing server, client and implementation method |
Publications (1)
Publication Number | Publication Date |
---|---|
CN107704521A true CN107704521A (en) | 2018-02-16 |
Family
ID=61172228
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710801967.3A Pending CN107704521A (en) | 2017-09-07 | 2017-09-07 | A kind of question and answer processing server, client and implementation method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107704521A (en) |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109460448A (en) * | 2018-08-31 | 2019-03-12 | 厦门快商通信息技术有限公司 | It is a kind of can autonomous configuration FAQ service framework |
CN109598658A (en) * | 2018-11-30 | 2019-04-09 | 暨南大学 | A kind of newborn service system and method for entering a school |
CN111858838A (en) * | 2019-04-04 | 2020-10-30 | 拉扎斯网络科技(上海)有限公司 | Menu calibration method and device, electronic equipment and nonvolatile storage medium |
CN111881263A (en) * | 2020-08-12 | 2020-11-03 | 福州大学 | Service recommendation online optimization method for intelligent home scene |
CN112732877A (en) * | 2019-10-14 | 2021-04-30 | 阿里巴巴集团控股有限公司 | Data processing method, device and system |
US11704497B2 (en) | 2020-09-09 | 2023-07-18 | International Business Machines Corporation | Generating and using a sentence model for answer generation |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104731895A (en) * | 2015-03-18 | 2015-06-24 | 北京京东尚科信息技术有限公司 | Auto-answer method and device |
US20160034573A1 (en) * | 2012-10-09 | 2016-02-04 | Amazon Technologies, Inc. | Analyzing user searches of verbal media content |
CN105653519A (en) * | 2015-12-30 | 2016-06-08 | 贺惠新 | Mining method of field specific word |
CN105955965A (en) * | 2016-06-21 | 2016-09-21 | 上海智臻智能网络科技股份有限公司 | Question information processing 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 |
CN106897263A (en) * | 2016-12-29 | 2017-06-27 | 北京光年无限科技有限公司 | Robot dialogue exchange method and device based on deep learning |
-
2017
- 2017-09-07 CN CN201710801967.3A patent/CN107704521A/en active Pending
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20160034573A1 (en) * | 2012-10-09 | 2016-02-04 | Amazon Technologies, Inc. | Analyzing user searches of verbal media content |
CN104731895A (en) * | 2015-03-18 | 2015-06-24 | 北京京东尚科信息技术有限公司 | Auto-answer method and device |
CN105653519A (en) * | 2015-12-30 | 2016-06-08 | 贺惠新 | Mining method of field specific word |
CN105955965A (en) * | 2016-06-21 | 2016-09-21 | 上海智臻智能网络科技股份有限公司 | Question information processing 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 |
CN106897263A (en) * | 2016-12-29 | 2017-06-27 | 北京光年无限科技有限公司 | Robot dialogue exchange method and device based on deep learning |
Cited By (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109460448A (en) * | 2018-08-31 | 2019-03-12 | 厦门快商通信息技术有限公司 | It is a kind of can autonomous configuration FAQ service framework |
CN109598658A (en) * | 2018-11-30 | 2019-04-09 | 暨南大学 | A kind of newborn service system and method for entering a school |
CN109598658B (en) * | 2018-11-30 | 2023-05-02 | 暨南大学 | New-use entrance service system and method |
CN111858838A (en) * | 2019-04-04 | 2020-10-30 | 拉扎斯网络科技(上海)有限公司 | Menu calibration method and device, electronic equipment and nonvolatile storage medium |
CN112732877A (en) * | 2019-10-14 | 2021-04-30 | 阿里巴巴集团控股有限公司 | Data processing method, device and system |
CN111881263A (en) * | 2020-08-12 | 2020-11-03 | 福州大学 | Service recommendation online optimization method for intelligent home scene |
US11704497B2 (en) | 2020-09-09 | 2023-07-18 | International Business Machines Corporation | Generating and using a sentence model for answer generation |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107704521A (en) | A kind of question and answer processing server, client and implementation method | |
CN108319599B (en) | Man-machine conversation method and device | |
KR102055355B1 (en) | Batch normalization layers | |
CN105912500B (en) | Machine learning model generation method and device | |
CN109961780A (en) | A kind of man-machine interaction method, device, server and storage medium | |
CN110347863B (en) | Speaking recommendation method and device and storage medium | |
CN107741976A (en) | Intelligent response method, apparatus, medium and electronic equipment | |
KR101666930B1 (en) | Target speaker adaptive voice conversion method using deep learning model and voice conversion device implementing the same | |
WO2019019649A1 (en) | Method and apparatus for generating investment portfolio product, storage medium and computer device | |
CN105893535A (en) | Intelligent question and answer method, knowledge base optimizing method and device and intelligent knowledge base | |
US11011176B1 (en) | Voice synthesis for virtual agents | |
CN112818025A (en) | Test question generation method, device and system, computer storage medium and program product | |
CN107396144A (en) | A kind of barrage distribution method and device | |
US20230022004A1 (en) | Dynamic vocabulary customization in automated voice systems | |
CN107111761A (en) | Technology for providing machine language translation of the user images seizure feedback to be improved | |
CN110019838A (en) | Intelligent Answer System and intelligent terminal | |
US20230396567A1 (en) | Chat bot utilizing metaphors to both relay and obtain information | |
CN109961141A (en) | Method and apparatus for generating quantization neural network | |
EP3573050A1 (en) | Computing platform and method for modifying voice data | |
US10937411B2 (en) | Methods, apparatus, and articles of manufacture to generate voices for artificial speech based on an identifier represented by frequency dependent bits | |
CN114023313B (en) | Training of speech processing model, speech processing method, apparatus, device and medium | |
US20230005466A1 (en) | Speech synthesis method, and electronic device | |
US20210192139A1 (en) | Language processing device, language processing system and language processing method | |
CN110019730A (en) | Automatic interaction system and intelligent terminal | |
Rogozhnikov et al. | A review of tests for exponentiality |
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 | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20180216 |
|
RJ01 | Rejection of invention patent application after publication |