CN116108058A - Automatic generation method of commit query language based on transition conversion system for code warehouse - Google Patents
Automatic generation method of commit query language based on transition conversion system for code warehouse Download PDFInfo
- Publication number
- CN116108058A CN116108058A CN202310390767.9A CN202310390767A CN116108058A CN 116108058 A CN116108058 A CN 116108058A CN 202310390767 A CN202310390767 A CN 202310390767A CN 116108058 A CN116108058 A CN 116108058A
- Authority
- CN
- China
- Prior art keywords
- query language
- language
- commit
- commit query
- ast
- 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
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/24—Querying
- G06F16/242—Query formulation
- G06F16/2433—Query languages
- G06F16/2448—Query languages for particular applications; for extensibility, e.g. user defined types
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/28—Databases characterised by their database models, e.g. relational or object models
- G06F16/283—Multi-dimensional databases or data warehouses, e.g. MOLAP or ROLAP
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F8/00—Arrangements for software engineering
- G06F8/40—Transformation of program code
- G06F8/41—Compilation
- G06F8/42—Syntactic analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F8/00—Arrangements for software engineering
- G06F8/40—Transformation of program code
- G06F8/51—Source to source
-
- 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
Abstract
The invention relates to the field of computers, and discloses a method for automatically generating a commit query language based on a transition conversion system for a code warehouse. The abstract grammar tree AST is used as the intermediate representation to complete the conversion generation from the natural language to the field specific language commit query language, thereby saving learning time and energy, effectively helping programmers complete search query operation and better completing development tasks.
Description
Technical Field
The invention relates to the technical field of computers, in particular to a method for automatically generating a commit query language based on a transition conversion system aiming at a code warehouse.
Background
With the rapid development of the internet and the continuous improvement of the network technology level in the current society, the market demand for programming is also increasing, and when a programmer develops a piece of software, the programmer usually applies to more existing code modules, and after some changes and combinations are performed on the code modules, the code modules meeting the requirements are completed. Therefore, the code warehouse is generated due to the potential, and programmers can better call the existing code segments in the warehouse through the code warehouse, so that the self programming efficiency and the code utilization rate are greatly improved. In addition, the code warehouse can well manage codes submitted by programmers, effectively store the codes, and has great progress significance for development of code programming and software engineering.
When a programmer utilizes a code warehouse to program and develop, common problems exist: when searching for a commit required by a query, a query matching the self-desired commit record is entered, the results returned by the code repository to the programmer may be somewhat different from their desired targets, there may be some irrelevant junk information, and it is difficult to obtain the desired commit record. In order to acquire the commit record with higher matching degree, a programmer needs to input the matched commit query language according to the grammar of the commit query language, however, the grammar of the commit query language is complex, and the query language in the special field can be well applied only by corresponding learning, so that the application efficiency of the code warehouse is greatly influenced by the programmer, and unnecessary waste in time, energy and the like is caused. Therefore, it is important for current code warehouse development to propose a method to implement the generation of query languages from natural language to commit.
In the document CN 112988785A, an SQL conversion method and system based on language model encoding and multitask decoding are disclosed, which are based on conversion from natural language to SQL query language realized by a transducer network, a pre-training language model and a multitask decoder, and the conversion method and system have slower speed and lower efficiency when in use.
Disclosure of Invention
In order to solve the problems, the invention discloses a method for automatically generating a commit query language based on a transition conversion system aiming at a code warehouse.
The abstract grammar tree AST is used as intermediate representation of natural language and commit query language, a transition conversion system is used as the basis, the conversion generation from the natural language to the abstract grammar tree AST is realized under the guidance of abstract grammar description language ASDL constructed for the commit query language, the conversion generation from the abstract grammar tree AST to the commit query language is realized under the guidance of the commit query language grammar, thereby completing the automatic generation of the commit query language, and when a programmer searches for query, the corresponding natural language can be input, so that the commit result with higher matching degree with an expected target can be directly obtained, the time and energy for learning the commit query language grammar and the detailed use method are saved, and the searching, managing and use efficiency of a code warehouse is improved.
The technical scheme of the invention is as follows: a method for automatically generating a commit query language based on a transition conversion system for a code warehouse comprises the following steps:
step 1: firstly, constructing a corresponding abstract grammar description language ASDL for the domain specific language commit query language according to the grammar of the domain specific language commit query language, and using the abstract grammar description language ASDL as a grammar guide for converting natural language into an abstract grammar tree AST in a subsequent conversion system;
step 2: taking a JSON format as a data storage format, taking a natural language as a question, taking a commit query language as an answer, and constructing a data set required by training for training a subsequent model;
step 3: constructing a transition conversion system suitable for the commit query language by taking an abstract syntax description language ASDL matched with the commit query language as a guide, converting an input natural language into a corresponding abstract syntax tree AST through a series of tree construction operations by utilizing a first part of the conversion system, and storing the abstract syntax tree AST;
step 4: on the basis of a conversion system, a neural encoder-decoder network with enhanced cyclic connection is used for parameterizing a parser P for judging and calculating the probability of a series of operations generated by an AST abstract syntax tree, and the encoder and the decoder are both standard two-way long-short-term memory networks, namely LSTM networks;
step 5: obtaining a saved abstract syntax tree AST, applying a transition conversion system suitable for a commit query language, and converting the abstract syntax tree AST into the commit query language according to a conversion function suitable for the commit query language grammar to complete automatic generation of the commit query language;
step 6: and 2) training the model through the data set constructed in the step 2), comparing answers output by the model with answers of the commit query language in the data set, and obtaining corresponding evaluating results of the commit query language performance.
In step 1, according to the description of the grammar in the commit query language grammar definition file CommitQuery.g4 in the code warehouse source code, two basic components of type and constructor are set in the abstract grammar description language ASDL which is matched for the construction, one composite type is defined by the constructor set under the type, the constructor uses the field to specify the language construction of the specific type, and the Conditions constructor under the composite type cond_expr represents the query statement expression, and has two fields: op and value, an AST is made up of multiple constructors, each node on the tree corresponds to a type field in the constructors, except the root node, which represents the root constructor, according to the cardinality of the field, the node can hold one or more constructors as its value.
In step 2, a JSON format is used as a data storage format, a natural language is used as a question, a commit query language is used as an answer, and a data set required for training is constructed, for example: { "query": "search was submitted my since the last worship" "" commit ":" after (last week) committed-by-me "" "id":1}, each JSON data additionally contains an id number, and after construction is complete, is used for training of subsequent models.
In step 3, using an abstract syntax description language ASDL matched with the commit query language as a guide, applying a transition conversion system suitable for the commit query language, and traversing from top to bottom from left to right to construct a corresponding abstract syntax tree AST by a series of tree construction operations from an initial derivation with a single root node, wherein each time step may trigger a new extension derivation to continuously perfect the current AST, and the derivation actions are divided into three: APPLYCONSTR [ c ], REDUCE and GENTOKEN [ v ], when there is no extension on the derivative, a new front field is derived, that is, the inputted natural language is converted into the corresponding abstract grammar tree AST and stored.
Wherein, in step 4, on the basis of the conversion system, a neural encoder-decoder network with enhanced cyclic connection is used to parameterize the parser P for determining the probability of calculating the series of operations generated by the AST abstract syntax tree, the encoder and the decoder are both a standard two-way long and short-term memory network, i.e. LSTM network, and the encoder encodes n token input words x, x 1.
Step 5 firstly obtains a saved abstract syntax tree AST, applies a transition conversion system suitable for a commit query language, and converts the abstract syntax tree AST into the commit query language according to a conversion function defined in query. Py and suitable for the commit query language grammar, wherein the basic grammar of the commit query language is "" criterion "" criterion "", EOF, criterion is a query condition, the content of the criterion includes after, before, committer, author, path, message, and the abstract syntax tree AST is converted into the commit query language to complete automatic generation of the commit query language.
And 6) training the model by the data set constructed in the step 2), and comparing the answers output by the model with the answers of the commit query language in the data set to obtain a corresponding commit query language performance evaluation result.
The invention has the advantages that: 1. the invention constructs a corresponding abstract syntax description language ASDL for the code warehouse source code according to the CommitQuery.g4 as a guide, uses the abstract syntax description language ASDL as a grammar guide in a subsequent conversion system, converts an input natural language into a corresponding abstract syntax tree AST through a series of tree construction functions under the guidance of the abstract syntax description language ASDL, applies another part of a transition conversion system after the abstract syntax tree AST is generated, constructs a corresponding syntax conversion function according to the grammar of the commit query language, converts the abstract syntax tree AST into the commit query language according to the conversion function, and completes the automatic generation of the commit query language.
2. The invention can save the time and energy of a programmer for learning the grammar of the commit query language and detailed using method when using the query function of the code warehouse by realizing the automatic generation of the commit query language, and improve the searching, managing and using efficiency of the code warehouse.
3. The invention mainly completes the conversion generation of the natural language to the field specific language commit query language by taking the abstract syntax tree AST as the intermediate representation, saves the time and energy for a programmer to learn the commit query language grammar and the using method aiming at the query function in the code warehouse, effectively helps the programmer to complete the search query operation and better completes the development task.
Drawings
FIG. 1 is a flow diagram illustration of the present invention;
fig. 2 is an input-output operation of the embodiment.
Description of the embodiments
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.
As shown in fig. 1, the method for automatically generating the commit query language based on the transition conversion system for the code warehouse comprises the following steps:
step 1: firstly, constructing a corresponding abstract syntax description language ASDL for the field specific language commit query language according to the grammar of the field specific language commit query language, which is used as a syntax guide for converting natural language into an abstract syntax tree AST in a subsequent conversion system, in the step 1, constructing a matched abstract syntax description language ASDL for the field specific language commit query language according to the description of the grammar in a commit query language grammar definition file CommitQuery.g4 in a code warehouse source code, wherein the ASDL comprises two basic components of types and constructors, one composite type is defined by an constructor set under the type, the constructor uses fields of the constructor to specify the language construction of the specific type, and the Condition constructor under the composite type cond_expr represents a query statement expression and comprises two fields: op and value, an AST is composed of a plurality of constructors, each node on the tree corresponds to a type field in constructors of the constructors, the root node is except for the root node, the root node represents the root constructor, according to the cardinality of the field, the node can store one or more constructors as its value;
step 2: the JSON format is used as a data storage format, the natural language is used as a question, the commit query language is used as an answer, a data set required by training is constructed and used for training a subsequent model, the JSON format is used as the data storage format in step 2, the natural language is used as the question, the commit query language is used as the answer, and the data set required by training is constructed, for example: { "query": "search from my post-worship" "commit" "" after (last week) committed-by-me "" "id":1}, each JSON data additionally contains an id number, and after construction is complete, is used for training of subsequent models;
step 3: constructing a transition conversion system suitable for the commit query language by taking the abstract syntax description language ASDL matched with the commit query language as a guide, converting an input natural language into a corresponding abstract syntax tree AST by utilizing a first part of the conversion system through a series of tree construction operations and storing, applying the transition conversion system suitable for the commit query language by taking the abstract syntax description language ASDL matched with the commit query language as a guide, and constructing the corresponding abstract syntax tree AST by traversing from left to right from one initial derivative with a single root node through a series of tree construction operations, wherein each time step possibly triggers new extension derivation to continuously perfect the current AST, and the derivative action is divided into three steps: APPLYCONSTR [ c ], REDUCE and GENTOKEN [ v ], when the new front domain front field is derived without extension on the derivative, representing that the input natural language is converted into a corresponding abstract grammar tree AST and is stored;
step 4: parameterizing the parser P by using a neural encoder-decoder network with enhanced cyclic connectivity for determining the probability of a series of operations generated by computing an AST abstract syntax tree, the encoder and decoder being a standard two-way long and short term memory network, i.e., an LSTM network, and parameterizing the parser P by using a neural encoder-decoder network with enhanced cyclic connectivity for determining the probability of a series of operations generated by computing an AST abstract syntax tree, the encoder and decoder being a standard two-way long and short term memory network, i.e., an LSTM network, the encoder encoding n token input utterances x, x 1..n into vector representations h 1..n;
step 5: obtaining a stored abstract syntax tree AST, applying a transition conversion system suitable for a commit query language, and according to a conversion function defined in query.py and suitable for the commit query language grammar, wherein the basic grammar of the commit query language is ' criterion (' +criterion) ' criterion EOF, criterion is a query condition, the content of the criterion is after, before, committer, author, path, message, converting the abstract syntax tree AST into the commit query language, and completing automatic generation of the commit query language;
step 6: and 2) training the model through the data set constructed in the step 2), comparing answers output by the model with answers of the commit query language in the data set, and obtaining corresponding evaluating results of the commit query language performance.
In step 3, using an abstract syntax description language ASDL matched with the commit query language as a guide, applying a transition conversion system suitable for the commit query language, and traversing from top to bottom from left to right to construct a corresponding abstract syntax tree AST by a series of tree construction operations from an initial derivation with a single root node, wherein each time step may trigger a new extension derivation to continuously perfect the current AST, and the derivation actions are divided into three: APPLYCONSTR [ c ], REDUCE and GENTOKEN [ v ], when there is no extension on the derivative, a new front field is derived, that is, the inputted natural language is converted into the corresponding abstract grammar tree AST and stored.
The automatic generation method of the commit query language based on the transition conversion system for the code warehouse is experimentally verified as follows: approximately two thousand data sets are manually constructed, and the data format is as follows: { "query": search is submitted from my after the last worship "" commit "" "after (last week) committed-by-me" "" id ":1}, the problem, the query language and the data sequence number are contained, the problem, the query language and the data sequence number are divided into a training set and a testing set respectively, the model is trained, the performance and the accuracy of the model are evaluated through an evaluation system, and the result proves that the automatic generation method of the query language based on the conversion system has higher accuracy and practicability, and the average accuracy value can reach 80%.
In terms of user interface, fig. 2 is an input-output operation of an embodiment, from which it can be seen that the required content of the required query described in natural language is input in a text box, and the automatically generated commit query language is output in the output text box: after (last week) committed-by-me verifies the feasibility and accuracy of the method.
It will be appreciated by persons skilled in the art that the embodiments of the invention described above and shown in the drawings are by way of example only and not limitation, and that the objects of the invention have been fully and effectively achieved. The functional and structural principles of the present invention have been shown and described in the examples and embodiments of the invention may be modified or practiced without departing from the principles described.
Claims (7)
1. The automatic generation method of the commit query language based on the transition conversion system for the code warehouse is characterized by comprising the following steps:
step 1: firstly, constructing a corresponding abstract grammar description language ASDL for the domain specific language commit query language according to the grammar of the domain specific language commit query language, and using the abstract grammar description language ASDL as a grammar guide for converting natural language into an abstract grammar tree AST in a subsequent conversion system;
step 2: taking a JSON format as a data storage format, taking a natural language as a question, taking a commit query language as an answer, and constructing a data set required by training for training a subsequent model;
step 3: constructing a transition conversion system suitable for the commit query language by taking an abstract syntax description language ASDL matched with the commit query language as a guide, converting an input natural language into a corresponding abstract syntax tree AST through a series of tree construction operations by utilizing a first part of the conversion system, and storing the abstract syntax tree AST;
step 4: on the basis of a conversion system, a neural encoder-decoder network with enhanced cyclic connection is used for parameterizing a parser P for judging and calculating the probability of a series of operations generated by an AST abstract syntax tree, and the encoder and the decoder are both standard two-way long-short-term memory networks, namely LSTM networks;
step 5: obtaining a saved abstract syntax tree AST, applying a transition conversion system suitable for a commit query language, and converting the abstract syntax tree AST into the commit query language according to a conversion function suitable for the commit query language grammar to complete automatic generation of the commit query language;
step 6: and (3) comparing the output answers with the answers of the query language through training of the model by the data set constructed in the step (2) to obtain corresponding assessment results of the performance of the query language of the commit.
2. The method for automatically generating a commit query language for a code warehouse based on a transition transformation system of claim 1, wherein: in the step 1, according to the description of the grammar in the commit query language grammar definition file CommitQuery.g4 in the code warehouse source code, two basic components of types and constructors are formed in the abstract grammar description language ASDL which is matched for the construction, one composite type is defined by an constructor set under the type, the constructor uses the field to specify the language construction of the specific type, and the Conditions constructor under the composite type cond_expr represents the query statement expression and has two fields: op and value, an AST is made up of multiple constructors, each node on the tree corresponds to a type field in the constructors, except the root node, which represents the root constructor, according to the cardinality of the field, the node can hold one or more constructors as its value.
3. The method for automatically generating a commit query language for a code warehouse based on a transition transformation system of claim 1, wherein: in the step 2, a data set required by training is constructed by taking a JSON format as a data storage format, taking a natural language as a question and a commit query language as an answer, and each JSON data additionally comprises an id sequence number and is used for training a subsequent model after the construction is completed.
4. The method for automatically generating a commit query language for a code warehouse based on a transition transformation system of claim 1, wherein: in the step 3, using an abstract syntax description language ASDL matched with the commit query language as a guide, applying a transition conversion system suitable for the commit query language, and traversing from top to bottom from left to right by a series of tree construction operations from an initial derivation with a single root node to construct a corresponding abstract syntax tree AST, wherein each time step may trigger a new extension derivation to continuously perfect the current AST, and the derivation actions are divided into three: APPLYCONSTR [ c ], REDUCE and GENTOKEN [ v ], when there is no extension on the derivative, a new front field is derived, that is, the inputted natural language is converted into the corresponding abstract grammar tree AST and stored.
5. The method for automatically generating a commit query language for a code warehouse based on a transition transformation system of claim 1, wherein: in said step 4, on the basis of the conversion system, a neural encoder-decoder network with enhanced cyclic connection is used to parameterize the parser P for determining the probability of calculating the series of operations generated by the AST abstract syntax tree, both the encoder and the decoder being a standard two-way long and short-term memory network, i.e. LSTM network, the encoder encoding n token input utterances x, x 1.
6. The method for automatically generating a commit query language for a code warehouse based on a transition transformation system of claim 1, wherein: step 5 is to obtain the abstract syntax tree AST stored first, and apply the transition conversion system suitable for the commit query language.
7. The method for automatically generating a commit query language for a code warehouse based on a transition transformation system of claim 1, wherein: and step 6, training the model by the data set constructed in the step 2, and comparing the answers output by the model with the answers of the commit query language in the data set to obtain a corresponding commit query language performance evaluation result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310390767.9A CN116108058A (en) | 2023-04-13 | 2023-04-13 | Automatic generation method of commit query language based on transition conversion system for code warehouse |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310390767.9A CN116108058A (en) | 2023-04-13 | 2023-04-13 | Automatic generation method of commit query language based on transition conversion system for code warehouse |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116108058A true CN116108058A (en) | 2023-05-12 |
Family
ID=86261960
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310390767.9A Pending CN116108058A (en) | 2023-04-13 | 2023-04-13 | Automatic generation method of commit query language based on transition conversion system for code warehouse |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116108058A (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112069199A (en) * | 2020-08-20 | 2020-12-11 | 浙江大学 | Multi-round natural language SQL conversion method based on intermediate syntax tree |
| CN114282497A (en) * | 2021-12-24 | 2022-04-05 | 思必驰科技股份有限公司 | Method and system for converting text into SQL |
| CN114880341A (en) * | 2022-04-26 | 2022-08-09 | 同济大学 | Text conversion encoder, text conversion SQL query analysis method and system |
| CN115062052A (en) * | 2022-05-10 | 2022-09-16 | 电子科技大学 | Method for translating from natural language problem to SQL based on map |
-
2023
- 2023-04-13 CN CN202310390767.9A patent/CN116108058A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112069199A (en) * | 2020-08-20 | 2020-12-11 | 浙江大学 | Multi-round natural language SQL conversion method based on intermediate syntax tree |
| CN114282497A (en) * | 2021-12-24 | 2022-04-05 | 思必驰科技股份有限公司 | Method and system for converting text into SQL |
| CN114880341A (en) * | 2022-04-26 | 2022-08-09 | 同济大学 | Text conversion encoder, text conversion SQL query analysis method and system |
| CN115062052A (en) * | 2022-05-10 | 2022-09-16 | 电子科技大学 | Method for translating from natural language problem to SQL based on map |
Non-Patent Citations (1)
| Title |
|---|
| QQ_40631568: "文章翻译|TRANX:A Transition-based Neural Abstract Syntax Parser for Semantic Parsing and Code Generation", Retrieved from the Internet <URL:blog.csdn.net/qq_40631568/article/details/125902823> * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110188167B (en) | End-to-end dialogue method and system integrating external knowledge | |
| Fu et al. | A survey on complex question answering over knowledge base: Recent advances and challenges | |
| Shi et al. | Incsql: Training incremental text-to-sql parsers with non-deterministic oracles | |
| CN109933602B (en) | Method and device for converting natural language and structured query language | |
| KR102027141B1 (en) | A program coding system based on artificial intelligence through voice recognition and a method thereof | |
| CN111930906A (en) | Knowledge graph question-answering method and device based on semantic block | |
| CN116820429B (en) | Training method and device of code processing model, electronic equipment and storage medium | |
| CN110673840A (en) | Automatic code generation method and system based on tag graph embedding technology | |
| CN112364125B (en) | Text information extraction system and method combining reading course learning mechanism | |
| US20220129450A1 (en) | System and method for transferable natural language interface | |
| CN105354180A (en) | Method and system for realizing open semantic interaction service | |
| CN113268561B (en) | Problem generation method based on multi-task joint training | |
| CN110084323A (en) | End-to-end semanteme resolution system and training method | |
| CN115238045B (en) | Method, system and storage medium for extracting generation type event argument | |
| WO2022146546A1 (en) | Multi-lingual code generation with zero-shot inference | |
| CN113779062A (en) | SQL statement generation method and device, storage medium and electronic equipment | |
| CN113326367B (en) | Task type dialogue method and system based on end-to-end text generation | |
| CN110909174B (en) | Knowledge graph-based method for improving entity link in simple question answering | |
| CN117112767A (en) | Question and answer result generation method, commercial query big model training method and device | |
| CN116974554A (en) | Code data processing method, apparatus, computer device and storage medium | |
| CN116108058A (en) | Automatic generation method of commit query language based on transition conversion system for code warehouse | |
| CN115470232A (en) | Model training and data query method and device, electronic equipment and storage medium | |
| CN114881010A (en) | Chinese grammar error correction method based on Transformer and multitask learning | |
| CN114968325A (en) | Code annotation generation method and device, processor and electronic equipment | |
| CN114036268A (en) | Task type multi-turn dialogue method and system based on intention gate |
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 |