CN110532358A - A kind of template automatic generation method towards knowledge base question and answer - Google Patents
A kind of template automatic generation method towards knowledge base question and answer Download PDFInfo
- Publication number
- CN110532358A CN110532358A CN201910604477.3A CN201910604477A CN110532358A CN 110532358 A CN110532358 A CN 110532358A CN 201910604477 A CN201910604477 A CN 201910604477A CN 110532358 A CN110532358 A CN 110532358A
- Authority
- CN
- China
- Prior art keywords
- template
- answer
- knowledge base
- relationship
- label
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/33—Querying
- G06F16/332—Query formulation
- G06F16/3329—Natural language query formulation or dialogue systems
-
- 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
- G06F16/36—Creation of semantic tools, e.g. ontology or thesauri
- G06F16/374—Thesaurus
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Artificial Intelligence (AREA)
- Human Computer Interaction (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
A kind of template automatic generation method towards knowledge base question and answer provided by the invention, including relationship dictionary building process, and according to problem answers to the process for automatically generating question template and query template.Wherein, relationship dictionary building process is on the basis of a large amount of corpus, and natural language phrase is corresponded to the relationship dictionary of knowledge base relation by building.According to problem answers query graph is obtained first according to problem answers pair to the process for automatically generating question template and query template from knowledge base, natural language problem is aligned with query graph, on the basis of alignment, automatically generates question template and query template.The present invention does not depend on artificial constructed template, the template for being able to solve knowledge base question and answer automatically generates problem, template generation process is automatic, efficient, it solves the problems, such as that the template generation of traditional artificial template construction method is at high cost, template number is limited, is convenient for a series of development of subsequent uses (such as natural language knowledge base question and answer).
Description
Technical field
The present invention relates to technical field of information retrieval is related to, it is related to knowledge base question and answer technology, is to be related to more specifically
A kind of preparation work of the knowledge base answering method based on template, according to known problem answers to automatically generating towards knowledge base
The method of question and answer template.
Background technique
With the fast development of computer networking technology, information retrieval is inseparable with the work of people, life.In information
Searching field, with the variation of user demand, traditional search engines also expose some shortcomings.Firstly, traditional search engines
The webpage for retrieving return is excessive, and user is difficult to filter out the information oneself really wanted.Secondly, Search Requirement can only pass through logic
Combined keyword is expressed, and the expression of user query intention is limited.And knowledge base question and answer technology allows user to use nature
Language question sentence is putd question to, and accurate problem answers are searched from knowledge base.Therefore, knowledge base question and answer become natural language processing
Prior development direction in field has important researching value and meaning.
Knowledge base question and answer based on template are a kind of classical ways, mainly total by the conclusion to known problem answers pair
Knot, therefrom summarizes general mode, constructs template.The problem of proposing for user, is matched, structure using existing template
The subgraph mode for producing answer inquiry, then retrieves answer in knowledge base.Having for knowledge base question and answer based on template is higher
Accuracy rate, the problem of for template can be matched to, can usually answer correct.Obviously, the knowledge base question and answer side based on template
Template building is a top priority in method, and template construction method determines the accuracy rate of knowledge base question and answer.Existing method is adopted more
Template is constructed by artificial means, and known problem answers pair are observed by expert, summarize general template.But in face of complexity
Natural language problem and problem quantity growing day by day, traditional artificial template's construction method expose template generation at
The problems such as this height, template number be limited, limit to the development of knowledge base answering method.
Summary of the invention
To solve the above problems, the invention discloses a kind of template automatic generation methods towards knowledge base question and answer.Firstly,
A dictionary about natural language phrase and knowledge base relationship map is collected by remote measure of supervision；Secondly, being asked according to training
Topic answer from knowledge base to constructing query graph；Then, the problem after syntax parsing is aligned with query graph by relationship dictionary；
Finally, extracting question template and query template automatically, template library is added.
In order to achieve the above object, the invention provides the following technical scheme:
A kind of template automatic generation method towards knowledge base question and answer, comprising the following steps:
Step 1, relationship dictionary creation
Several semi-structured informations are obtained from the corpus of text with mark；Sentence in given corpus, if the sentence
Contain and contain only and refers to m there are two entity1And m2, and two entities refer between phrase p be no more than four words, while
Two entities refer to corresponding entity e in knowledge base1And e2Between there are relationship r, then obtain the mapping f of phrase p Yu relationship r；With
Confidence level of the number that the number that mapping f occurs in corpus occurs in corpus divided by phrase p as the mapping；From corpus
Obtain all mappings and its confidence level constituent relation dictionary；
Step 2, query graph constructs
According to problem answers pair, search includes the minimum connected graph of problematic entities and answer entity in knowledge base, will even
CVT node in logical figure replaces with variable node, and answer entity is also replaced with variable node, obtains query graph；
Step 3, natural language problem is aligned with query graph
The mapping between natural language problem and query graph is found, entity alignment is carried out according to entity link result first,
Entity in problem is referred to and is aligned with entity node in query graph；Then all entities refer in removal problem, in reservation problem
Other words；According to relationship in other words in problem and query graph, relationship dictionary is inquired, if relationship and query graph in entry
Middle relationship is completely the same, and phrase is the subsequence of problem word in entry, then it is assumed that deposits between the phrase and relationship in problem
It is mapping, the confidence level of mapping is the confidence level of the entry, and word in problem is aligned with relationship in query graph；
Step 4, question template and query template automatically generate
Interdependent syntactic analysis and part-of-speech tagging are done to problem, obtain syntax dependency tree, is extracted in syntax dependency tree comprising real
The minimum subtree that body refers to and relationship refers to replaces original word to generate question template using part of speech label；
On the basis of the query graph that step 2 constructs, specific relationship and entity are replaced with the alignment label in step 3, it is raw
At query template；
Template library is added in question template together with query template.
Further, the confidence calculations formula of f is mapped in the step 1 are as follows:
Wherein, count (f) be map f occur number, count (p) be phrase corpus appearance number.
Further, mapping and its confidence level cooperation are an entry in the relationship dictionary.
Further, in the step 2, if problem corresponds to multiple answers, the query graph that each answer generates is different, only
Retain F1It is worth highest query graph.
Further, in the step 2, for giving problem q and answer set A={ a1, a2..., an, F1Value calculates public
Formula are as follows:
Wherein, TP=| A*∩ A |, FP=| A*- A |, FN=| A-A*|, A*It is answered for what the query graph was obtained from knowledge base
Case set.
Further, confidence level is selected most if there are multiple mappings to meet the requirements in relationship dictionary in the step 3
High mapping, for word in problem to be aligned with relationship in query graph.
Further, problem part-of-speech tagging result is abstracted in the step 4, similar part of speech label is abstracted as
One label.
Further, part of speech label NNS, NNP, NNPS are abstracted as a NN label in the step 4, by part of speech label
RBR, RBS are abstracted as a RB label, part of speech label WDT, WP, WP $, WRB are abstracted as a WB label, by part of speech label
JJR, JJS are abstracted as a JJ label, and part of speech label PRP $ is abstracted as a PRP label, by part of speech label VBD, VBG,
VBN, VBP, VBZ are abstracted as a VB label.
Compared with prior art, the invention has the advantages that and the utility model has the advantages that
The present invention does not need artificial constructed knowledge base question and answer template, but on the basis of the relationship dictionary constructed in advance, from
Dynamic to carry out template generation: on the basis of a large amount of corpus, natural language phrase is corresponded to the relative of knowledge base relation by building
Allusion quotation；According to problem answers pair, search includes the minimum connected graph building query graph of problematic entities and answer entity in knowledge base；
On the basis of relationship dictionary, by the entity in natural language problem refer to relationship refer to respectively in query graph entity and pass
System's alignment；According to alignment as a result, doing interdependent syntactic analysis and part-of-speech tagging to problem, original word is replaced to generate problem using part of speech
Template replaces specific relationship and entity name in query graph with alignment label, generates query template.
Therefore, the present invention does not depend on artificial constructed template, and the template for being able to solve knowledge base question and answer automatically generates problem, mould
Plate generating process is automatic, efficiently, and the template generation for solving traditional artificial template construction method is at high cost, template number is limited
Problem is convenient for a series of development of subsequent uses (such as natural language knowledge base question and answer).
Detailed description of the invention
Fig. 1 is a kind of template automatic generation method flow chart towards knowledge base question and answer provided by the invention.
Specific embodiment
Technical solution provided by the invention is described in detail below with reference to specific embodiment, it should be understood that following specific
Embodiment is only illustrative of the invention and is not intended to limit the scope of the invention.
A kind of template automatic generation method towards knowledge base question and answer provided by the invention, including relationship dictionary creation mistake
Journey, and according to problem answers to the process for automatically generating question template and query template.Wherein, relationship dictionary building process is
On the basis of a large amount of corpus, natural language phrase is corresponded to the relationship dictionary of knowledge base relation by building.According to problem answers
To the process for automatically generating question template and query template first according to problem answers pair, query graph is obtained from knowledge base, it will
Natural language problem is aligned with query graph.On the basis of alignment, question template and query template are automatically generated.Specifically,
Process of the present invention is as shown in Figure 1, comprise the following steps:
One, relationship dictionary creation
Step 1, relationship dictionary creation.
Natural language problem often with the phenomenon for having one justice of more words, and the triple relation name in knowledge base be it is unique,
The present invention is realized by building relationship dictionary from natural language phrase to the mapping of knowledge base relation.This example is with wikipedia
(Wikipedia) text information with mark in (can also select other corpus as the corpus source of dictionary creation as needed
Source), the sentence in corpus is given, if the sentence contains and contains only there are two entity, while the entity of two entities refers to
Between phrase be no more than four words, shaped like m1 p m2, ρ is the phrase of of length no more than 4 words, m1And m2It is knowledge respectively
Entity e in library1And e2Refer to, while the e in knowledge base1And e2Between there are relationship r (one jump or two hop relationship), then
To the mapping f of phrase p and relationship r.
Corpus is handled according to the above method, is collected into 495,358 mappings altogether.It wherein include several identical mappings, system
The number that each mapping occurs is counted, count (f) is denoted as, the number that statistics phrase occurs in corpus is denoted as count (p), root
The confidence level of each mapping is calculated according to following formula:
It will map and its confidence level combination is used as an entry, component relationship dictionary, finally obtained relationship dictionary shares
96,338 entries.
Two, according to problem answers to automatically generating question template and query template
Step 2, query graph constructs.
Current invention assumes that entity link work has been completed, the entity in acquisition problem is referred to and its in knowledge base
Corresponding entity, the set that these entities are constituted are denoted as Eq.Given problem q and answer set A={ a1, a2..., an, in A
Each answer ai, search includes a in knowledge baseiWith entity E whole in problemqMinimum connected graph.It will be in the connected graph
CVT (Compound Value Types, compound Value Types) node replace with variable node CVT replaces answer entity node
It is changed to variable node x obtains query graph.
During query graph building, need to carry out aforesaid operations to answer entity each in answer set A, if one
Problem has multiple answers, then can obtain multiple queries figure.To the multiple queries figure of a problem answers pair, by calculating its F1
Value, it is final only to retain F1It is worth highest query graph.F1It is as follows to be worth calculation formula:
Wherein, A*For the answer set that the query graph is inquired in knowledge base, TP=| A*∩ A |, FP=| A*- A |,
FN=| A-A*|。
Step 3, natural language problem is aligned with query graph.
After obtaining the query graph of problem, need to find the mapping between natural language problem and query graph, as rear
After the mapping between question template and query template.
Firstly, doing entity alignment according to entity link result, entity in problem is referred into entity corresponding with query graph
Node is aligned, and the entity in label problem refers to that with the entity of query graph be entity.Then, all entities mention in removal problem
And other words in reservation problem.For the relationship r in query graphq, relationship dictionary is inquired, if relationship and r in entryqComplete one
It causes, and phrase is the subsequence that problem retains word in entry, then it is assumed that phrase p and relationship r in problemqBetween exist
Mapping, the confidence level of mapping are the confidence level of the entry, and word in problem is aligned with relationship in query graph.Finally, if closing
There are multiple mappings to meet the requirements in copula allusion quotation, then retains the highest mapping of confidence level, is used for word in problem and query graph
Middle relationship alignment.Phrase p is exactly relationship rqReferring in problem.Label relationship refers to p and relationship rqFor pred label.
Step 4, question template and query template automatically generate.
In terms of question template generation, this example using Stamford natural language processing kit (https: //
Nlp.stanford.edu/software/ interdependent syntactic analysis and part-of-speech tagging) are done to problem, obtain syntax dependency tree.It extracts
It is wherein referred to comprising entity and relationship refers to the minimum subtree for marking word, it is problematic instead of original morphology to reuse part of speech label
Template.The parsing result of Stamford part-of-speech tagging tool includes 36 kinds of part of speech labels, since meticulous part of speech label will affect
Wherein similar part of speech label is abstracted as one by the generalization ability of template, the present invention.The NNS that former part of speech label includes is (common
Noun, plural form), NNP (proper noun), NNPS (proper noun, plural form) indicate in a template is all noun, institute
They to be abstracted as to NN (common noun).Part of speech label RBR, RBS are abstracted as a RB label.By part of speech label WDT,
WP, WP $, WRB are abstracted as a WB label.Part of speech label J JR, JJS are abstracted as a JJ label.Part of speech label PRP $ is taken out
As for a PRP label.Part of speech label VBD, VBG, VBN, VBP, VBZ are abstracted as a VB label.
In terms of query template generation, based on query graph, with the alignment label in step 3 replace specific relationship and
Entity name forms query template.Template library is finally added in question template together with query template.
The present invention during the experiment, uses the English Wikipedia in October, 2017 to construct relationship dictionary as corpus.
Problem is done using Stamford natural language processing kit (https: //nlp.stanford.edu/software/) interdependent
Syntactic analysis and part-of-speech tagging.WebQuestionsSP training set is used, to set, to use as known problem answers
2015 edition data collection of Freebase is as knowledge base.The template side of automatically generating proposed by the present invention towards knowledge base question and answer
The problem of method, the relationship dictionary of construction shares 96,338 entries, automatically generates and query template share 1,286.
The technical means disclosed in the embodiments of the present invention is not limited only to technological means disclosed in above embodiment, further includes
Technical solution consisting of any combination of the above technical features.It should be pointed out that for those skilled in the art
For, various improvements and modifications may be made without departing from the principle of the present invention, these improvements and modifications are also considered as
Protection scope of the present invention.
Claims (8)
1. a kind of template automatic generation method towards knowledge base question and answer, which comprises the following steps:
Step 1, relationship dictionary creation
Several semi-structured informations are obtained from the corpus of text with mark；Sentence in given corpus, if the sentence contains
And it contains only and refers to m there are two entity1And m2, and two entities refer between phrase p be no more than four words, while in knowledge
Two entities refer to corresponding entity e in library1And e2Between there are relationship r, then obtain the mapping f of phrase p Yu relationship r；With mapping
Confidence level of the number that the number that f occurs in corpus occurs in corpus divided by phrase p as the mapping；It is obtained from corpus
All mappings and its confidence level constituent relation dictionary；
Step 2, query graph constructs
According to problem answers pair, search includes the minimum connected graph of problematic entities and answer entity in knowledge base, by connected graph
In CVT node replace with variable node, answer entity is also replaced with into variable node, obtains query graph；
Step 3, natural language problem is aligned with query graph
The mapping between natural language problem and query graph is found, entity alignment is carried out according to entity link result first, will be asked
Entity is referred in topic is aligned with entity node in query graph；Then all entities refer in removal problem, other in reservation problem
Word；According to relationship in other words in problem and query graph, relationship dictionary is inquired, if closing in relationship and query graph in entry
It is completely the same, and phrase is the subsequence of problem word in entry, then it is assumed that exist between the phrase and relationship in problem and reflect
It penetrates, the confidence level of mapping is the confidence level of the entry, and word in problem is aligned with relationship in query graph；
Step 4, question template and query template automatically generate
Interdependent syntactic analysis and part-of-speech tagging are done to problem, obtain syntax dependency tree, extracts and is mentioned in syntax dependency tree comprising entity
And the minimum subtree referred to relationship, replace original word to generate question template using part of speech label；
On the basis of the query graph that step 2 constructs, specific relationship and entity are replaced with the alignment label in step 3, generation is looked into
Ask template；
Template library is added in question template together with query template.
2. the template automatic generation method according to claim 1 towards knowledge base question and answer, which is characterized in that the step
The confidence calculations formula of f is mapped in 1 are as follows:
Wherein, count (f) be map f occur number, count (p) be phrase corpus appearance number.
3. the template automatic generation method according to claim 1 towards knowledge base question and answer, which is characterized in that the relationship
Mapping and its confidence level cooperation are an entry in dictionary.
4. the template automatic generation method according to claim 1 towards knowledge base question and answer, which is characterized in that the step
In 2, if problem corresponds to multiple answers, the query graph that each answer generates is different, only retains F1It is worth highest query graph.
5. the template automatic generation method according to claim 4 towards knowledge base question and answer, which is characterized in that the step
In 2, for giving problem q and answer set A={ a1, a2..., an, F1It is worth calculation formula are as follows:
Wherein, TP=| A*∩ A |, FP=| A*- A |, FN=| A-A*|, A*The answer set obtained from knowledge base for the query graph
It closes.
6. the template automatic generation method according to claim 1 towards knowledge base question and answer, which is characterized in that the step
If there are multiple mappings to meet the requirements in relationship dictionary in 3, the highest mapping of confidence level is selected, is used for word in problem
It is aligned with relationship in query graph.
7. the template automatic generation method according to claim 1 towards knowledge base question and answer, which is characterized in that the step
Problem part-of-speech tagging result is abstracted in 4, similar part of speech label is abstracted as a label.
8. the template automatic generation method according to claim 7 towards knowledge base question and answer, which is characterized in that the step
Part of speech label NNS, NNP, NNPS are abstracted as a NN label in 4, part of speech label RBR, RBS are abstracted as a RB label,
Part of speech label WDT, WP, WP $, WRB are abstracted as a WB label, part of speech label J JR, JJS are abstracted as a JJ label, it will
Part of speech label PRP $ is abstracted as a PRP label, and part of speech label VBD, VBG, VBN, VBP, VBZ are abstracted as a VB label.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910604477.3A CN110532358B (en) | 2019-07-05 | 2019-07-05 | Knowledge base question-answering oriented template automatic generation method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910604477.3A CN110532358B (en) | 2019-07-05 | 2019-07-05 | Knowledge base question-answering oriented template automatic generation method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110532358A true CN110532358A (en) | 2019-12-03 |
| CN110532358B CN110532358B (en) | 2023-08-22 |
Family
ID=68659486
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910604477.3A Active CN110532358B (en) | 2019-07-05 | 2019-07-05 | Knowledge base question-answering oriented template automatic generation method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110532358B (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111339269A (en) * | 2020-02-20 | 2020-06-26 | 来康科技有限责任公司 | Knowledge graph question-answer training and application service system with automatically generated template |
| CN111770236A (en) * | 2020-02-13 | 2020-10-13 | 北京沃东天骏信息技术有限公司 | Conversation processing method, device, system, server and storage medium |
| CN112632237A (en) * | 2020-12-07 | 2021-04-09 | 厦门渊亭信息科技有限公司 | Knowledge graph-based question-answer template automatic generation method and device |
| CN112800177A (en) * | 2020-12-31 | 2021-05-14 | 北京智源人工智能研究院 | FAQ knowledge base automatic generation method and device based on complex data types |
| CN114090620A (en) * | 2022-01-19 | 2022-02-25 | 支付宝(杭州)信息技术有限公司 | Query request processing method and device |
| CN114444512A (en) * | 2022-01-24 | 2022-05-06 | 中科合肥智慧农业协同创新研究院 | Ontology knowledge base-based automatic labeling method for natural language field data set |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20040167875A1 (en) * | 2003-02-20 | 2004-08-26 | Eriks Sneiders | Information processing method and system |
| CN105528349A (en) * | 2014-09-29 | 2016-04-27 | 华为技术有限公司 | Method and apparatus for analyzing question based on knowledge base |
| CN105701253A (en) * | 2016-03-04 | 2016-06-22 | 南京大学 | Chinese natural language interrogative sentence semantization knowledge base automatic question-answering method |
| CN106484675A (en) * | 2016-09-29 | 2017-03-08 | 北京理工大学 | Fusion distributed semantic and the character relation abstracting method of sentence justice feature |
-
2019
- 2019-07-05 CN CN201910604477.3A patent/CN110532358B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20040167875A1 (en) * | 2003-02-20 | 2004-08-26 | Eriks Sneiders | Information processing method and system |
| CN105528349A (en) * | 2014-09-29 | 2016-04-27 | 华为技术有限公司 | Method and apparatus for analyzing question based on knowledge base |
| CN105701253A (en) * | 2016-03-04 | 2016-06-22 | 南京大学 | Chinese natural language interrogative sentence semantization knowledge base automatic question-answering method |
| CN106484675A (en) * | 2016-09-29 | 2017-03-08 | 北京理工大学 | Fusion distributed semantic and the character relation abstracting method of sentence justice feature |
Non-Patent Citations (2)
| Title |
|---|
| DEPENG HU 等: "ACQA_ONTO: AN ONTOLOGY APPROACH FOR", 《IET INTERNATIONAL CONFERENCE ON INFORMATION SCIENCE AND CONTROL ENGINEERING 2012 (ICISCE 2012)》 * |
| 汪卫明 等: "基于语义模板的医学问答自动生成", 《武汉大学学报(理学版)》 * |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111770236A (en) * | 2020-02-13 | 2020-10-13 | 北京沃东天骏信息技术有限公司 | Conversation processing method, device, system, server and storage medium |
| CN111339269A (en) * | 2020-02-20 | 2020-06-26 | 来康科技有限责任公司 | Knowledge graph question-answer training and application service system with automatically generated template |
| CN111339269B (en) * | 2020-02-20 | 2023-09-26 | 来康科技有限责任公司 | Knowledge graph question-answering training and application service system capable of automatically generating templates |
| CN112632237A (en) * | 2020-12-07 | 2021-04-09 | 厦门渊亭信息科技有限公司 | Knowledge graph-based question-answer template automatic generation method and device |
| CN112800177A (en) * | 2020-12-31 | 2021-05-14 | 北京智源人工智能研究院 | FAQ knowledge base automatic generation method and device based on complex data types |
| CN114090620A (en) * | 2022-01-19 | 2022-02-25 | 支付宝(杭州)信息技术有限公司 | Query request processing method and device |
| CN114444512A (en) * | 2022-01-24 | 2022-05-06 | 中科合肥智慧农业协同创新研究院 | Ontology knowledge base-based automatic labeling method for natural language field data set |
| CN114444512B (en) * | 2022-01-24 | 2024-04-09 | 中科合肥智慧农业协同创新研究院 | Automatic labeling method for natural language field data set based on ontology knowledge base |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110532358B (en) | 2023-08-22 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110532358A (en) | A kind of template automatic generation method towards knowledge base question and answer | |
| CN107291687B (en) | Chinese unsupervised open type entity relation extraction method based on dependency semantics | |
| CN104361127B (en) | The multilingual quick constructive method of question and answer interface based on domain body and template logic | |
| CN104216913B (en) | Question answering method, system and computer-readable medium | |
| WO2018000272A1 (en) | Corpus generation device and method | |
| JP2017511922A (en) | Method, system, and storage medium for realizing smart question answer | |
| WO2020010834A1 (en) | Faq question and answer library generalization method, apparatus, and device | |
| CN109062904B (en) | Logic predicate extraction method and device | |
| Gupta et al. | A novel approach towards building a portable nlidb system using the computational paninian grammar framework | |
| CN107656921B (en) | Short text dependency analysis method based on deep learning | |
| CN108665141B (en) | Method for automatically extracting emergency response process model from emergency plan | |
| CN109783806A (en) | A kind of text matching technique using semantic analytic structure | |
| CN114218472A (en) | Intelligent search system based on knowledge graph | |
| CN110175585A (en) | It is a kind of letter answer correct system and method automatically | |
| CN114625748A (en) | SQL query statement generation method and device, electronic equipment and readable storage medium | |
| CN114528312A (en) | Method and device for generating structured query language statement | |
| Gaillard et al. | Tuuurbine: a generic CBR engine over RDFS | |
| CN113779062A (en) | SQL statement generation method and device, storage medium and electronic equipment | |
| CN112507089A (en) | Intelligent question-answering engine based on knowledge graph and implementation method thereof | |
| US11487795B2 (en) | Template-based automatic software bug question and answer method | |
| CN112395425A (en) | Data processing method and device, computer equipment and readable storage medium | |
| CN117251455A (en) | Intelligent report generation method and system based on large model | |
| CN111831624A (en) | Data table creating method and device, computer equipment and storage medium | |
| CN113157875A (en) | Knowledge graph question-answering system, method and device | |
| WO2012133941A1 (en) | Method for matching elements in schemas of databases using bayesian network |
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 |