CN109657047A - A kind of voice automatic question-answering method and system based on crawler technology and machine learning - Google Patents
A kind of voice automatic question-answering method and system based on crawler technology and machine learning Download PDFInfo
- Publication number
- CN109657047A CN109657047A CN201811619723.4A CN201811619723A CN109657047A CN 109657047 A CN109657047 A CN 109657047A CN 201811619723 A CN201811619723 A CN 201811619723A CN 109657047 A CN109657047 A CN 109657047A
- Authority
- CN
- China
- Prior art keywords
- answer
- final result
- weight
- obtains
- result
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/22—Procedures used during a speech recognition process, e.g. man-machine dialogue
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/22—Procedures used during a speech recognition process, e.g. man-machine dialogue
- G10L2015/225—Feedback of the input speech
Landscapes
- Engineering & Computer Science (AREA)
- Computational Linguistics (AREA)
- Health & Medical Sciences (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Human Computer Interaction (AREA)
- Physics & Mathematics (AREA)
- Acoustics & Sound (AREA)
- Multimedia (AREA)
- Machine Translation (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention discloses a kind of voice automatic question-answering method and system based on crawler technology and machine learning, belongs to speech recognition, machine learning and web crawlers technical field.The present invention parses the enquirement voice of acquisition, obtains key to the issue word set;Based on tree algorithm is promoted, retrieval is filtered to key to the issue word set, then obtain final result set from the result after filtering retrieval;Final result is selected in final result set, and the result after filtering retrieval is handled according to final result, it is after filtering is retrieved after processing as a result, for selecting next time, for corresponding method there are also corresponding system, system includes speech recognition module, processing module and selection processing module.The present invention is used for voice automatic question answering.
Description
Technical field
A kind of voice automatic question-answering method and system based on crawler technology and machine learning is used for voice automatic question answering,
Belong to speech recognition, machine learning and web crawlers technical field.
Background technique
Microsoft's speech recognition module (Microsoft Speech SDK): Microsoft Speech SDK is that Microsoft provides
Software development kit, the Speech API (SAPI) provided mainly include two broad aspects:
1.API for Text-to-Speech
2.API for Speech Recognition
Wherein APl for Text-to-Speech, is exactly the interface of Microsoft's tts engine, and by it, we can be easy to
Powerful text voice program is established on ground, and the word function of reading aloud of Kingsoft Powerword has just used this and write APl, and at present almost
All texts, which read aloud tool all, to be developed with this SDK.It is exactly as API for Speech Recognition and TTS
Corresponding speech recognition, voice technology are a kind of exciting technologies, but due to current speech recognition technology accuracy and
Recognition speed is not satisfactory, also not up to widely applied requirement.
The interdependent syntactic analysis in Stamford (Stanford CcreNLP): Stanford parser be by Stanford University from
The open source parser of right Language Processing group development is that a JAVA based on probability statistics syntactic analysis is realized.Analysis
Device currently provides 5 Chinese syntax.
Web crawlers: (be otherwise known as web crawlers webpage spider, and network robot is more frequent among the community FOAF
Referred to as webpage follower), be it is a kind of according to certain rules, automatically grab the program or script of web message.
Machine learning: it is the core of artificial intelligence, is the fundamental way for making computer have intelligence, and application spreads people
The every field of work intelligence, it is mainly using conclusion, comprehensive rather than deduction.
Promote tree algorithm: (Boosting) method that promoted is that one kind is widely used and very effective statistical learning method.
It is based on such a thought: for a complex task, the judgement of multiple experts being carried out comprehensive income appropriate and is gone out
Judgement, it is good individually to judge than any one expert.
Existing automatic answering system is typically all to input key to the issue word by user, and answering system is being inscribed according to keyword
Correspondence problem is retrieved in library, and answer is then showed into user.
It is that can be putd question to by voice using speech recognition technology, user there are also a kind of mode.But for problem
Answer is still using manual entry.
The response answer of above two mode all relies on manual entry, and a part is system manager, and there are also one
Divide from user's active typing, have the following deficiencies:
One, the answer in exam pool typing and update be manually entered dependent on personnel, the accuracy of answer needs relevant people
Member has relevant professional, if typing mistake, can not find accurate answer, typing is also very time-consuming, also will cause people
The waste etc. of power resource;
Two, answer is limited and fixed, can not replacement problem at any time newest answer, cause the problem that user experience is bad.
Summary of the invention
Aiming at the problem that the studies above, the intelligence based on speech recognition and machine learning that the purpose of the present invention is to provide a kind of
Energy answer method and system, the answer for solving intelligent response in the prior art have relied on manual entry, and typing information has
Limit, causes that the accuracy of response message is low, updates not in time and professional requirement and high problem to typing personnel.
In order to achieve the above object, the present invention adopts the following technical scheme:
A kind of intelligent response method based on speech recognition and machine learning, which is characterized in that following steps:
Step 1 parses the enquirement voice of acquisition, obtains key to the issue word set;
Step 2 is based on promoting tree algorithm, is filtered retrieval, then the knot after filtering retrieval to key to the issue word set
Final result set is obtained in fruit;
Step 3 selects final result in final result set, and according to final result to the result after filtering retrieval into
Row processing, it is after filtering is retrieved in replacement step 2 after processing as a result, for selecting next time.
Further, the specific steps of the step 1 are as follows:
Step 1.1 identifies enquirement voice, obtains problem;
Step 1.2 decomposes problem using the interdependent syntactic analysis in Stamford, obtains set of keywords.
Further, the specific steps of the step 2 are as follows:
Step 2.1 is filtered parsing to the keyword in set of keywords based on syntax library, excludes subject and the tone helps
Word obtains template problem;
Step 2.2 carries out interpretation extension to each keyword in template problem;
Step 2.3, the answer of the template problem after extending is interpreted in search in exam pool, obtains existing basic answer and answer
Corresponding weight, obtained existing basic answer and the corresponding weight of answer are an existing boosted tree;
Step 2.4 in internet is retrieved the template problem after extension by the way of web crawlers, is interconnected
Basic answer and the corresponding weight of answer are netted, according to obtained result and tree algorithm is promoted, establishes boosted tree;
The boosted tree of step 2.5, merging step 2.3 and step 2.4 obtains new boosted tree, the right to choose from new boosted tree
It is worth the basic answer of ranking top N, obtains final result set.
Further, the specific steps of the step 3 are as follows:
Step 3.1, user select final result in final result set;
Step 3.2, record final result, in new boosted tree corresponding to increase problem after the weight of the answer, simultaneously
Trimming increases the new boosted tree after weight, removes the too low answer of part weight, removes to replace after the too low answer of weight and repair
The new boosted tree obtained before cutting, for selecting next time.
A kind of intelligent response system based on speech recognition and machine learning characterized by comprising
Speech recognition module: parsing the enquirement voice of acquisition, obtains key to the issue word set;
Processing module: based on tree algorithm is promoted, retrieval is filtered to key to the issue word set, then after filtering retrieval
As a result it obtains final result set in or receives to feed back to be replaced the result after filtering retrieval;
Selection processing module: selecting final result in final result set, and according to final result to filtering retrieval after
Result handled, and the result after feedback processing is to processing module.
Further, the implementation of the speech recognition module includes the following steps:
To puing question to voice to identify, problem is obtained;
Problem is decomposed using the interdependent syntactic analysis in Stamford, obtains set of keywords.
Further, the implementation of the processing module includes the following steps:
Parsing is filtered to the keyword in set of keywords based on syntax library, subject and auxiliary words of mood is excluded, obtains
Template problem;
Interpretation extension is carried out to each keyword in template problem;
The answer of the template problem after extending is interpreted in search in exam pool, obtains existing basic answer and the corresponding power of answer
Value, obtained existing basic answer and the corresponding weight of answer are an existing boosted tree;
The template problem after extension is retrieved in internet by the way of web crawlers, Internet basic is obtained and answers
Case and the corresponding weight of answer according to obtained result and promote tree algorithm, establish boosted tree;
Merge above two boosted tree and obtain new boosted tree, the basis of weight ranking top N is selected from new boosted tree
Answer obtains final result set.
Further, the implementation of the selection processing module includes the following steps:
User selects final result in final result set;
Final result is recorded, the weight of the answer in new boosted tree corresponding to increase problem is trimmed after increasing weight
New boosted tree removes the too low answer of part weight, and after removing the too low answer of weight, result is fed back to processing module.
The present invention compared with the existing technology, its advantages are shown in:
One, the present invention is grasped in real time using web crawlers technology puts question to the newest answer of voice on the internet, eliminates
The burden of manual entry;Meanwhile by the way of machine learning, by machine learning (promoted tree algorithm) be user's selection with
While meeting the answer of regard, the use of user can be also utilized, the accuracy answered a question is improved;
Two, exam pool of the invention can update the newest answer in internet from internet and machine learning gained at any time,
It is also to promote selection accuracy next time for lower trained AI and (improve corresponding answer that user, which selects the movement of answer, simultaneously
Probability of occurrence);
Three, the present invention trims the new boosted tree of merging, moreover it is possible to remove certain irrelevant answers on internet,
Reduce redundant data;
Four, the present invention can be reduced human cost by the way of web crawlers technology and machine learning.
Detailed description of the invention
Fig. 1 is flow diagram of the invention.
Specific embodiment
Below in conjunction with the drawings and the specific embodiments, the invention will be further described.
A kind of intelligent response method based on speech recognition and machine learning, following steps:
Step 1 parses the enquirement voice of acquisition, obtains key to the issue word set;Specific steps are as follows:
Step 1.1 identifies enquirement voice, obtains problem;
Step 1.2 decomposes problem using the interdependent syntactic analysis in Stamford, obtains set of keywords.
Step 2 is based on promoting tree algorithm, is filtered retrieval, then the knot after filtering retrieval to key to the issue word set
Final result set is obtained in fruit;Specific steps are as follows:
Step 2.1 is filtered parsing to the keyword in set of keywords based on syntax library, excludes subject and the tone helps
Word obtains template problem;
Step 2.2 carries out interpretation extension to each keyword in template problem;
Step 2.3, the answer of the template problem after extending is interpreted in search in exam pool, obtains existing basic answer and answer
Corresponding weight, obtained existing basic answer and the corresponding weight of answer are an existing boosted tree;
Step 2.4 in internet is retrieved the template problem after extension by the way of web crawlers, is interconnected
Basic answer and the corresponding weight of answer are netted, according to obtained result and tree algorithm is promoted, establishes boosted tree;
The boosted tree of step 2.5, merging step 2.3 and step 2.4 obtains the new boosted tree (knot as after filtering retrieval
Fruit), the basic answer of weight ranking top N is selected from new boosted tree, obtains final result set.
Step 3 selects final result in final result set, and according to final result to the result after filtering retrieval into
Row processing, it is after filtering is retrieved in replacement step 2 after processing as a result, for selecting next time.Specific steps are as follows:
Step 3.1, user select final result in final result set;
Step 3.2, record final result, in new boosted tree corresponding to increase problem after the weight of the answer, simultaneously
Trimming increases the new boosted tree after weight, removes the too low answer of part weight, removes to replace after the too low answer of weight and repair
The new boosted tree obtained before cutting, for selecting next time.
A kind of intelligent response system based on speech recognition and machine learning, comprising:
Speech recognition module: parsing the enquirement voice of acquisition, obtains key to the issue word set;Implementation includes
Following steps:
To puing question to voice to identify, problem is obtained;
Problem is decomposed using the interdependent syntactic analysis in Stamford, obtains set of keywords.
Processing module: based on tree algorithm is promoted, retrieval is filtered to key to the issue word set, then after filtering retrieval
As a result it obtains final result set in or receives to feed back to be replaced the result after filtering retrieval;Implementation includes following step
It is rapid:
Parsing is filtered to the keyword in set of keywords based on syntax library, subject and auxiliary words of mood is excluded, obtains
Template problem;
Interpretation extension is carried out to each keyword in template problem;
The answer of the template problem after extending is interpreted in search in exam pool, obtains existing basic answer and the corresponding power of answer
Value, obtained existing basic answer and the corresponding weight of answer are an existing boosted tree;
The template problem after extension is retrieved in internet by the way of web crawlers, Internet basic is obtained and answers
Case and the corresponding weight of answer according to obtained result and promote tree algorithm, establish boosted tree;
Merge above two boosted tree and obtain new boosted tree, the basis of weight ranking top N is selected from new boosted tree
Answer obtains final result set.
Selection processing module: selecting final result in final result set, and according to final result to filtering retrieval after
Result handled, and the result after feedback processing is to processing module.Implementation includes the following steps:
User selects final result in final result set;
Final result is recorded, the weight of the answer in new boosted tree corresponding to increase problem is trimmed after increasing weight
New boosted tree removes the too low answer of part weight, and after removing the too low answer of weight, result is fed back to processing module.
Embodiment
Voice is identified using Microsoft's speech recognition module (Microsoft Speech SDK), obtains problem, example
Such as: " what I should eat today ";Problem is decomposed using Stanford CoreNLP (the interdependent syntactic analysis in Stamford),
Obtain set of keywords be combined into I, today, should, eat and what.
Syntax library selects Baidu's natural language processing API, can do morphological analysis, identifies part of speech, based on syntax library to pass
Keyword in key word set is filtered parsing, i.e., to me, today, should, eat and what is parsed, exclude subject and language
Gas auxiliary word etc., obtains what template problem eats today;Each keyword of template problem is extended, as time word can carry out it is a variety of
It interprets, such as today is Saturday, December 23 and Winter Solstice, can be obtained what today eats, what Saturday eats, and ten
What February 23 ate, what Winter Solstice eats;Answer is searched in exam pool, obtains existing basic answer (leaf node), such as
It eats apple 23 today, eats vegetables 4 today, eat meat 35 today, 23,4 and 35 be the weight of existing basic answer, is obtained
Existing basis answer and the corresponding weight of answer are a boosted trees, exam pool be it is self-built, for saving problem and corresponding
Boosted tree, so what is be retrieved is boosted tree;
In internet, (such as Baidu's question and answer, know) retrieve problem by the way of web crawlers, obtain mutually
The basic answer of networking, and weight is calculated according to Internet basic answer frequency, based on the Internet basic for promoting tree algorithm, obtaining
Answer and the corresponding weight of answer establish boosted tree, such as Saturday eats 1, gruel, eats 7, rice cake December 23, the winter
To eating 13, dumpling;Merge two boosted trees, the basic answer of weight ranking top N is selected from new boosted tree, obtains final
Answer set, such as I eat apple today, I eats meat today, I eats rice cake December 23, I eats dumpling in Winter Solstice.
User selects final result and is recorded in final result set, and increases this in boosted tree corresponding to problem and answer
The weight of case, such as user have selected me to eat dumpling in Winter Solstice;The problem of by " what I should eat today " corresponding boosted tree
In the weight of the answer increase, trim new boosted tree after increasing weight, removing training AI after the too low answer of part weight is
User selects to prepare next time, and training A1, is that the new boosted tree after trimming is replaced with new boosted tree.
The above is only the representative embodiment in the numerous concrete application ranges of the present invention, to protection scope of the present invention not structure
At any restrictions.It is all using transformation or equivalence replacement and the technical solution that is formed, all fall within rights protection scope of the present invention it
It is interior.
Claims (8)
1. a kind of intelligent response method based on speech recognition and machine learning, which is characterized in that following steps:
Step 1 parses the enquirement voice of acquisition, obtains key to the issue word set;
Step 2 is based on promoting tree algorithm, is filtered retrieval to key to the issue word set, then from the result after filtering retrieval
Obtain final result set;
Step 3 selects final result in final result set, and according to final result to the result after filtering retrieval at
Reason, it is after filtering is retrieved in replacement step 2 after processing as a result, for selecting next time.
2. a kind of intelligent response method based on speech recognition and machine learning according to claim 1, which is characterized in that
The specific steps of the step 1 are as follows:
Step 1.1 identifies enquirement voice, obtains problem;
Step 1.2 decomposes problem using the interdependent syntactic analysis in Stamford, obtains set of keywords.
3. a kind of intelligent response method based on speech recognition and machine learning according to claim 1 or 2, feature exist
In the specific steps of the step 2 are as follows:
Step 2.1 is filtered parsing to the keyword in set of keywords based on syntax library, excludes subject and auxiliary words of mood,
Obtain template problem;
Step 2.2 carries out interpretation extension to each keyword in template problem;
Step 2.3, the answer of the template problem after extending is interpreted in search in exam pool, obtains existing basic answer and answer is corresponding
Weight, obtained existing basic answer and the corresponding weight of answer are an existing boosted tree;
Step 2.4 in internet is retrieved the template problem after extension by the way of web crawlers, obtains the Internet-based
Plinth answer and the corresponding weight of answer according to obtained result and promote tree algorithm, establish boosted tree;
The boosted tree of step 2.5, merging step 2.3 and step 2.4 obtains new boosted tree, and weight row is selected from new boosted tree
The basic answer of name top N, obtains final result set.
4. a kind of intelligent response method based on speech recognition and machine learning according to claim 3, which is characterized in that
The specific steps of the step 3 are as follows:
Step 3.1, user select final result in final result set;
Step 3.2, record final result are trimmed simultaneously in new boosted tree corresponding to increase problem after the weight of the answer
New boosted tree after increasing weight, removes the too low answer of part weight, removes after the too low answer of weight before replacement trimming
Obtained new boosted tree, for selecting next time.
5. a kind of intelligent response system based on speech recognition and machine learning characterized by comprising
Speech recognition module: parsing the enquirement voice of acquisition, obtains key to the issue word set;
Processing module: based on tree algorithm is promoted, retrieval, then the result after filtering retrieval are filtered to key to the issue word set
In obtain final result set or receive feedback to filtering retrieval after result be replaced;
Selection processing module: final result is selected in final result set, and according to final result to the knot after filtering retrieval
Fruit is handled, and the result after feedback processing is to processing module.
6. a kind of intelligent response method based on speech recognition and machine learning according to claim 1, which is characterized in that
The implementation of the speech recognition module includes the following steps:
To puing question to voice to identify, problem is obtained;
Problem is decomposed using the interdependent syntactic analysis in Stamford, obtains set of keywords.
7. a kind of intelligent response method based on speech recognition and machine learning according to claim 5 or 6, feature exist
In the implementation of the processing module includes the following steps:
Parsing is filtered to the keyword in set of keywords based on syntax library, subject and auxiliary words of mood is excluded, obtains template
Problem;
Interpretation extension is carried out to each keyword in template problem;
The answer of the template problem after extending is interpreted in search in exam pool, obtains existing basic answer and the corresponding weight of answer,
Obtained existing basic answer and the corresponding weight of answer are an existing boosted tree;
The template problem after extension is retrieved in internet by the way of web crawlers, obtain Internet basic answer and
The corresponding weight of answer according to obtained result and promotes tree algorithm, establishes boosted tree;
Merge above two boosted tree and obtain new boosted tree, the basic answer of weight ranking top N is selected from new boosted tree,
Obtain final result set.
8. a kind of intelligent response method based on speech recognition and machine learning according to claim 7, which is characterized in that
The implementation of the selection processing module includes the following steps:
User selects final result in final result set;
Record final result, the weight of the answer in new boosted tree corresponding to increase problem, increase trim after weight it is new
Boosted tree removes the too low answer of part weight, and after removing the too low answer of weight, result is fed back to processing module.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811619723.4A CN109657047B (en) | 2018-12-27 | 2018-12-27 | Voice automatic question-answering method and system based on crawler technology and machine learning |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811619723.4A CN109657047B (en) | 2018-12-27 | 2018-12-27 | Voice automatic question-answering method and system based on crawler technology and machine learning |
Publications (2)
Publication Number | Publication Date |
---|---|
CN109657047A true CN109657047A (en) | 2019-04-19 |
CN109657047B CN109657047B (en) | 2020-09-29 |
Family
ID=66117547
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201811619723.4A Active CN109657047B (en) | 2018-12-27 | 2018-12-27 | Voice automatic question-answering method and system based on crawler technology and machine learning |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109657047B (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111782794A (en) * | 2020-05-29 | 2020-10-16 | 北京沃东天骏信息技术有限公司 | Question-answer response method and device |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107220380A (en) * | 2017-06-27 | 2017-09-29 | 北京百度网讯科技有限公司 | Question and answer based on artificial intelligence recommend method, device and computer equipment |
CN108595696A (en) * | 2018-05-09 | 2018-09-28 | 长沙学院 | A kind of human-computer interaction intelligent answering method and system based on cloud platform |
CN108717433A (en) * | 2018-05-14 | 2018-10-30 | 南京邮电大学 | A kind of construction of knowledge base method and device of programming-oriented field question answering system |
CN108932349A (en) * | 2018-08-17 | 2018-12-04 | 齐鲁工业大学 | Medical automatic question-answering method and device, storage medium, electronic equipment |
-
2018
- 2018-12-27 CN CN201811619723.4A patent/CN109657047B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107220380A (en) * | 2017-06-27 | 2017-09-29 | 北京百度网讯科技有限公司 | Question and answer based on artificial intelligence recommend method, device and computer equipment |
CN108595696A (en) * | 2018-05-09 | 2018-09-28 | 长沙学院 | A kind of human-computer interaction intelligent answering method and system based on cloud platform |
CN108717433A (en) * | 2018-05-14 | 2018-10-30 | 南京邮电大学 | A kind of construction of knowledge base method and device of programming-oriented field question answering system |
CN108932349A (en) * | 2018-08-17 | 2018-12-04 | 齐鲁工业大学 | Medical automatic question-answering method and device, storage medium, electronic equipment |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111782794A (en) * | 2020-05-29 | 2020-10-16 | 北京沃东天骏信息技术有限公司 | Question-answer response method and device |
Also Published As
Publication number | Publication date |
---|---|
CN109657047B (en) | 2020-09-29 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106528845B (en) | Retrieval error correction method and device based on artificial intelligence | |
CN106708893B (en) | Search query word error correction method and device | |
CN100458795C (en) | Intelligent word input method and input method system and updating method thereof | |
CN106649768A (en) | Deep question answering-based questions and answers clarifying method and device | |
US9613317B2 (en) | Justifying passage machine learning for question and answer systems | |
CN103365925B (en) | Obtain polyphone phonetic, method based on phonetic retrieval and related device thereof | |
EP1800294B1 (en) | System and methods for improving accuracy of speech recognition | |
CN107818164A (en) | A kind of intelligent answer method and its system | |
CN109241258A (en) | A kind of deep learning intelligent Answer System using tax field | |
US20050131677A1 (en) | Dialog driven personal information manager | |
CN104679910A (en) | Intelligent answering method and system | |
CN106649742A (en) | Database maintenance method and device | |
CN110287482B (en) | Semi-automatic participle corpus labeling training device | |
CN1351744A (en) | Recognition engines with complementary language models | |
WO2020010834A1 (en) | Faq question and answer library generalization method, apparatus, and device | |
CN108304445A (en) | A kind of text snippet generation method and device | |
CN109740053A (en) | Sensitive word screen method and device based on NLP technology | |
CN110362824A (en) | A kind of method, apparatus of automatic error-correcting, terminal device and storage medium | |
CN110110039A (en) | A kind of method that the visualization of intelligent sound robot is built | |
CN113361266A (en) | Text error correction method, electronic device and storage medium | |
CN116521893A (en) | Control method and control device of intelligent dialogue system and electronic equipment | |
CN117331730A (en) | Fault diagnosis method, device, electronic equipment and storage medium | |
KR20210062934A (en) | Text document cluster and topic generation apparatus and method thereof | |
CN112182019B (en) | Semantic parsing search method for power grid statistics professional index feature extraction | |
CN109657047A (en) | A kind of voice automatic question-answering method and system based on crawler technology and machine learning |
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 |