CN117827923A - Query demand processing method and device, computer equipment and storage medium - Google Patents
Query demand processing method and device, computer equipment and storage medium Download PDFInfo
- Publication number
- CN117827923A CN117827923A CN202410006852.5A CN202410006852A CN117827923A CN 117827923 A CN117827923 A CN 117827923A CN 202410006852 A CN202410006852 A CN 202410006852A CN 117827923 A CN117827923 A CN 117827923A
- Authority
- CN
- China
- Prior art keywords
- information
- target
- code
- preset
- query requirement
- 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
- 238000003860 storage Methods 0.000 title claims abstract description 24
- 238000003672 processing method Methods 0.000 title description 4
- 238000000034 method Methods 0.000 claims abstract description 73
- 238000012545 processing Methods 0.000 claims abstract description 24
- 238000000605 extraction Methods 0.000 claims description 31
- 238000011156 evaluation Methods 0.000 claims description 25
- 230000015654 memory Effects 0.000 claims description 24
- 230000005856 abnormality Effects 0.000 claims description 23
- 230000002159 abnormal effect Effects 0.000 claims description 20
- 238000004458 analytical method Methods 0.000 claims description 12
- 239000000284 extract Substances 0.000 claims description 6
- 238000004891 communication Methods 0.000 claims description 4
- 230000008569 process Effects 0.000 abstract description 24
- 230000002950 deficient Effects 0.000 abstract description 2
- 230000006870 function Effects 0.000 description 10
- 238000010586 diagram Methods 0.000 description 8
- 208000025174 PANDAS Diseases 0.000 description 5
- 208000021155 Paediatric autoimmune neuropsychiatric disorders associated with streptococcal infection Diseases 0.000 description 5
- 240000004718 Panda Species 0.000 description 5
- 235000016496 Panda oleosa Nutrition 0.000 description 5
- 230000011218 segmentation Effects 0.000 description 5
- 230000009286 beneficial effect Effects 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 3
- 238000007405 data analysis Methods 0.000 description 3
- 238000012216 screening Methods 0.000 description 3
- 230000008901 benefit Effects 0.000 description 2
- 238000009826 distribution Methods 0.000 description 2
- 230000008030 elimination Effects 0.000 description 2
- 238000003379 elimination reaction Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 230000008439 repair process Effects 0.000 description 2
- 239000007787 solid Substances 0.000 description 2
- 238000013459 approach Methods 0.000 description 1
- 238000004140 cleaning Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 125000004122 cyclic group Chemical group 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000010295 mobile communication Methods 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 230000008929 regeneration Effects 0.000 description 1
- 238000011069 regeneration method Methods 0.000 description 1
- 238000009877 rendering Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
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
- G06F16/245—Query processing
- G06F16/2457—Query processing with adaptation to user needs
-
- 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
- G06F16/22—Indexing; Data structures therefor; Storage structures
- G06F16/2228—Indexing structures
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Software Systems (AREA)
- Computational Linguistics (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention provides a method, a device, a computer device and a storage medium for processing query requirements, wherein the method comprises the following steps: acquiring a target query requirement currently input by a user, and extracting key information of the target query requirement; inquiring related target table structure information from a preset information base by utilizing key information, wherein the preset information base comprises table structure information corresponding to a plurality of data tables; generating a first code text based on the target query requirement and the target table structure information; in a preset operation environment, executing the first code text to obtain a first code execution result, and generating feedback content corresponding to the target query requirement by utilizing the target query requirement and the first code execution result. The method solves the problems of the prior art that the query process of the table structure information is deficient and the code generation passing rate is low.
Description
Technical Field
The present invention relates to the field of data processing, and in particular, to a method and apparatus for processing a query requirement, a computer device, and a storage medium.
Background
In recent years, large language models have made significant research progress in the field of question-answering systems. These systems rely mainly on sentence embedding, vector library retrieval, and large language model rendering functions, and their knowledge sources are mainly derived from non-relational data, such as PDF, word, TXT. These systems perform well when dealing with text knowledge in a particular area. However, statistical task processing methods require a deeper understanding of the data structure and processing language by the operator, limiting their application in certain scenarios. In addition, existing query demand processing methods typically analyze in a passive manner when processing relational data. Such a passive approach may result in a system consuming significant computing resources, reducing efficiency, and may not take full advantage of the relational data.
Therefore, to solve these problems, it is necessary to further optimize the data processing flow and improve the efficiency and accuracy of the system, solve the problem of the deficiency of the large language model to the table structure information, and design some pre-operations of code generation to improve the code passing rate of the large predictive model generation.
Disclosure of Invention
In view of this, the embodiments of the present invention provide a method, an apparatus, a computer device, and a storage medium for processing a query requirement, so as to solve the problems of the prior art that the query process of table structure information is deficient and the code generation passing rate is low.
In a first aspect, an embodiment of the present invention provides a method for processing a query requirement, where the method includes:
acquiring a target query requirement currently input by a user, and extracting key information of the target query requirement;
inquiring related target table structure information from a preset information base by utilizing the key information, wherein the preset information base comprises table structure information corresponding to a plurality of data tables;
generating a first code text based on the target query requirement and the target table structure information;
and in a preset operation environment, executing the first code text to obtain a first code execution result, and generating feedback content corresponding to the target query requirement by utilizing the target query requirement and the first code execution result.
Further, the extracting the key information of the target query requirement includes:
acquiring a first constraint condition and a key information extraction rule;
configuring a preset model by utilizing the first constraint condition and the key information extraction rule, and taking the configured preset model as a first model;
and inputting the target query requirement to the first model so that the first model obtains requirement information meeting the first constraint condition from the target query requirement, and extracting the key information from the requirement information according to the key information extraction rule.
Further, before the key information is used for inquiring the associated target table structure information from a preset information base, the method comprises the following steps:
obtaining a plurality of data tables, and performing de-duplication on each column of data in each data table to obtain an effective value;
acquiring the column names corresponding to the effective values and the number of the effective values;
generating table structure information based on the column names, the effective values and the effective value quantity, and storing the table structure information in the preset information base.
Further, the querying the related target table structure information from the preset information base by using the key information includes:
inquiring a target column name and/or a target effective value matched with the key information from the preset information base, and taking table structure information corresponding to the target column name and/or the target effective value as candidate table structure information;
determining the similarity between the key information and each candidate list structure information;
and arranging the candidate list structure information in a descending order according to the similarity to obtain a list structure information sequence, and selecting a preset number of candidate list structure information from the list structure information sequence as the target list structure information.
Further, the generating the first code text based on the target query requirement and the target table structure information includes:
acquiring task description information, task output rules and code generation rules;
configuring a preset model by utilizing the task description information, the task output rule and the code generation rule, and taking the configured preset model as a second model;
and inputting the target query requirement and the target table structure information into the second model, so that the second model extracts target content matched with the task description information from the target query requirement and the target table structure information, outputs an output result corresponding to the target content according to the task output rule, and generates a first code text corresponding to the output result according to the code generation rule.
Further, in the preset operation environment, after the first code text is executed to obtain a first code execution result, the method further includes:
acquiring first evaluation information corresponding to the first code execution result;
determining whether the first code text has an execution abnormality according to the first evaluation information;
If the first code text has the execution abnormality, determining abnormal content of the first code text, and inputting the abnormal content and the first code text into the second model based on the abnormal content, so that the second model generates a second code text based on the abnormal content and the first code text;
executing the second code text in a preset operation environment to obtain a second code execution result;
acquiring second evaluation information corresponding to the second code execution result, and determining whether the second code text has an execution abnormality according to the second evaluation information;
and if the second code text does not have the execution abnormality, generating feedback content corresponding to the target query requirement by using the target query requirement and the second code execution result.
Further, the generating feedback content corresponding to the target query requirement by using the target query requirement and the first code execution result includes:
acquiring a feedback rule;
and inputting the target query requirement, the first code execution result and the feedback rule into a preset model so that the preset model analyzes the target query requirement and the first code execution result to obtain an analysis result, and generating the feedback content according to the feedback rule by the analysis result.
In a second aspect, an embodiment of the present invention provides a device for processing a query requirement, where the device includes:
the acquisition module is used for acquiring a target query requirement currently input by a user and extracting key information of the target query requirement;
the query module is used for querying related target table structure information from a preset information base by utilizing the key information, wherein the preset information base comprises table structure information corresponding to a plurality of data tables;
the generation module is used for generating a first code text based on the target query requirement and the target table structure information;
and the execution module is used for executing the first code text in a preset operation environment to obtain a first code execution result, and generating feedback content corresponding to the target query requirement by utilizing the target query requirement and the first code execution result.
In a third aspect, an embodiment of the present invention provides a computer apparatus, including: the memory and the processor are in communication connection, the memory stores computer instructions, and the processor executes the computer instructions to perform the method of the first aspect or any implementation manner corresponding to the first aspect.
In a fourth aspect, an embodiment of the present invention provides a computer readable storage medium having stored thereon computer instructions for causing a computer to perform the method of the first aspect or any of its corresponding embodiments.
The embodiment of the application has the following beneficial effects:
according to the method provided by the embodiment of the application, the corresponding query codes can be automatically generated by extracting the key information of the target query requirement currently input by the user, so that the workload of manually writing the query codes is reduced, and the efficiency is improved; by utilizing the table structure information corresponding to a plurality of data tables in the preset information base, the process can realize the query of cross-table and cross-data, and the relevance and the integration of the information are enhanced; the table structure information in the preset information base can provide accurate table structure information for the generated first code text, so that the query accuracy is ensured; the feedback content is generated according to the target query requirement and the query result, and more visual and clearer result display is provided.
The method provided by the embodiment of the application can more accurately understand the real requirements of the user by extracting the key information of the target query requirements, so that answers or services which more accord with the user expectations are provided; through a preset model and a key information extraction rule, the system can rapidly acquire the requirement information meeting the constraint condition from the target query requirement, so that the query efficiency is improved; by configuring a preset model, the system can adapt to different query requirements and constraint conditions, and the flexibility and adaptability of the system are improved; by the key information extraction rule, key information can be accurately extracted from the requirement information, and the accuracy of information extraction is improved.
The method provided by the embodiment of the application can remove redundant and repeated information in the data through duplication elimination, so that the data is clearer and standard; the column structure and the data distribution of each data table can be defined by acquiring the column names and the number of the effective values corresponding to the effective values, so that a foundation is provided for subsequent data analysis and processing; the unified management of the data can be realized by storing the table structure information in a preset information base; the storage mode of the preset information base enables inquiring and updating of the table structure information to be more convenient, and quick access to the information can be achieved.
According to the method provided by the embodiment of the application, the list name and/or the target effective value of the target matched with the key information are inquired from the preset information base, so that the table structure information which is most matched with the key information can be accurately screened, and the screening accuracy and efficiency are improved; the similarity between the key information and each candidate list structure information is determined, and the candidate list structure information is arranged in a descending order according to the similarity, so that the list structure information which is most similar to the key information can be found out quickly, and a basis is provided for subsequent decisions; selecting a preset number of candidate table structure information in the table structure information sequence as target table structure information, which is helpful for making decisions in a plurality of similar candidate schemes and optimizing the selection process; through sorting and selection based on similarity, the decision making process is more scientific and objective, and subjectivity and randomness are reduced.
According to the method provided by the embodiment of the application, the task requirements can be accurately understood and related tasks can be automatically completed by acquiring the task description information, the task output rule and the code generation rule; the task description information, the task output rule and the code generation rule are utilized to configure a preset model, so that the accurate extraction of target content matched with the task description information is realized; outputting an output result corresponding to the target content according to the task output rule, and ensuring the accuracy and consistency of the output result; generating a first code text corresponding to the output result according to the code generation rule, which is helpful for quickly generating the code text meeting the requirements and reducing coding errors; by configuring the preset model, the system can adapt to different task descriptions, output rules and code generation rules, and the flexibility and adaptability of the code generation process are improved.
The method provided by the embodiment of the application can ensure the effectiveness and the safety of the codes by checking whether the first code text has the execution abnormality, and is beneficial to the optimization of the subsequent code text; if the first code text has the execution abnormality, the second model can generate a second code text according to the abnormal content and the first code text, and the process can realize the cyclic optimization of the code text until the code text is stable; by comparing the execution result of the second code text with the second evaluation information, it is possible to further confirm whether the second code text is executed correctly, which helps to ensure that the generated code is accurate; if the second code text has no execution exception, feedback content may be generated using the target query requirement and the second code execution result. The whole query process for the user requirement is realized, the final query result is directly presented to the user in a more refined mode, and the high efficiency and accuracy of the whole process are ensured.
The method provided by the embodiment of the application can analyze and evaluate the code execution result according to the preset rule by acquiring the feedback rule, is beneficial to comprehensively checking the correctness and the effectiveness of the code execution result by the system, and timely discovers and processes potential problems; by inputting the target query requirement, the first code execution result and the feedback rule into the preset model, the code execution can be analyzed from multiple dimensions, and the comprehensive understanding of the performance, stability and safety of the code execution is facilitated; the preset model can find potential performance bottlenecks, errors or potential safety hazards through analysis of target query requirements and first code execution results.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings that are needed in the description of the embodiments or the prior art will be briefly described, and it is obvious that the drawings in the description below are some embodiments of the present invention, and other drawings can be obtained according to the drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flow diagram of a method of processing query requirements according to some embodiments of the invention;
FIG. 2 is a schematic diagram of configuring a first model according to some embodiments of the invention;
FIG. 3 is a schematic diagram of building a preset information library according to some embodiments of the present invention;
FIG. 4 is a schematic diagram of an example of configuring a second model according to some embodiments of the invention;
FIG. 5 is a schematic diagram of check code text according to some embodiments of the invention;
FIG. 6 is a flow chart of another method of processing query requirements according to some embodiments of the invention;
FIG. 7 is a block diagram of an apparatus for processing query requirements according to some embodiments of the invention;
fig. 8 is a schematic diagram of a hardware structure of a computer device according to an embodiment of the present invention.
Detailed Description
For the purpose of making the objects, technical solutions and advantages of the embodiments of the present invention more apparent, the technical solutions of 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, and it is apparent that the described embodiments are some embodiments of the present invention, but not all embodiments of the present invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
According to embodiments of the present invention, there is provided a method, apparatus, computer device and storage medium for processing a query requirement, it should be noted that the steps illustrated in the flowchart of the drawings may be performed in a computer system such as a set of computer executable instructions, and that although a logical order is illustrated in the flowchart, in some cases, the steps illustrated or described may be performed in an order different from that illustrated herein.
In this embodiment, a method for processing a query requirement is provided, and fig. 1 is a flowchart of a method for processing a query requirement according to an embodiment of the present invention, as shown in fig. 1, where the flowchart includes the following steps:
step S11, obtaining a target query requirement currently input by a user, and extracting key information of the target query requirement.
In the embodiment of the application, in a conventional statistical question-answering system, the requirements of users generally require association, statistics and calculation of complex field, and conventional search methods are to use a model to vectorize the whole sentence, and search from an established vector database, which results in serious influence on the searched result by the granularity of the sentence used in vectorization.
In the embodiment of the application, the key information extraction is performed on the demands of the user in a prompt mode based on the intelligent question-answering system, and the specific extraction process comprises the following steps: acquiring a first constraint condition and a key information extraction rule; configuring a preset model by using the first constraint condition and the key information extraction rule, and taking the configured preset model as a first model; and inputting the target query requirement into the first model so that the first model obtains the requirement information meeting the first constraint condition from the target query requirement, and extracting the key information from the requirement information according to the key information extraction rule.
In the embodiment of the application, the key information of the target query requirement is extracted, which comprises the following steps A1-A3:
and A1, acquiring a first constraint condition and a key information extraction rule.
In this embodiment of the present application, the first constraint condition may constrain a domain in which content is generated, and the key information extraction rule may include a task description and a sample, where the key information extraction rule is an information extraction rule to be followed when extracting multiple pieces of key information, and the task description and the sample are used to guide an output manner and content of a preset model.
And A2, configuring a preset model by using the first constraint condition and the key information extraction rule, and taking the configured preset model as a first model.
As an example, as shown in fig. 2, the first constraint is obtained as: brand vehicle field; in the key information extraction rule, task description: extracting keywords from the query requirements according to the given query requirements; sample example: the sample requirement is to inquire the vehicle model with the lowest price of the Y system in the X field, and the sample output is the X field, the Y system, the price and the vehicle model. And configuring the preset model by using the constraint conditions and the key information extraction rules, so that the configured preset model can be obtained, and the preset model is used as a first model.
And step A3, inputting the target query requirement into the first model so that the first model obtains the requirement information meeting the first constraint condition from the target query requirement, and extracting the key information from the requirement information according to the key information extraction rule.
As one example, the target query requirement is: inquiring vehicles with P functions in the Z brand of vehicles, inquiring how many vehicles are in the G area, and inputting the target requirement into the configured first model, so that the first model obtains requirement information meeting a first constraint condition from the requirements of inquiring the vehicle types with P functions in the Z brand of vehicles and inquiring how many vehicles are in the G area, namely inquiring the vehicles with P functions in the Z brand of vehicles, and extracting key information from the requirement information according to a key information extraction rule is as follows: z brand, P function, vehicle model.
The method provided by the embodiment of the application can more accurately understand the real requirements of the user by extracting the key information of the target query requirements, so that answers or services which more accord with the user expectations are provided; through a preset model and a key information extraction rule, the system can rapidly acquire the requirement information meeting the constraint condition from the target query requirement, so that the query efficiency is improved; by configuring a preset model, the system can adapt to different query requirements and constraint conditions, and the flexibility and adaptability of the system are improved; by the key information extraction rule, key information can be accurately extracted from the requirement information, and the accuracy of information extraction is improved.
Step S12, inquiring related target table structure information from a preset information base by utilizing the key information, wherein the preset information base comprises table structure information corresponding to a plurality of data tables.
In the embodiment of the present application, the preset information base is a vector database storing table structure information. In this preset information base, each table structure information is represented as a vector, which can be generated by the embedding model BGE and the vector database FAISS; the table structure information is regarded as a data table and is stored vectorized using an embedded model and a vector database. Vectorization refers to converting a data table or other data type into a vector form for efficient similarity matching and retrieval. In the vectorization process, a deduplication process is performed on each column first to reduce the data throughput. And secondly, unfolding segmentation by taking the columns as the reference, and selecting an equal segmentation strategy to ensure the equalization and the accuracy of the data. And finally, adding related column names and corresponding effective value numbers to the segmented effective values to construct a preset information base.
In the embodiment of the application, before the related target table structure information is queried from the preset information base by utilizing the key information, the corresponding preset information base is required to be generated, and the table structure description related to the corresponding requirement is automatically generated according to the input statement of the user, so that more reliable codes are generated in an auxiliary mode.
The specific process of constructing the preset information base can be understood as a vectorization process, as shown in fig. 3, performing de-duplication and segmentation on each column of data in each data table, adding a relevant column name and a corresponding number of valid values to the segmented valid values, performing vectorization on each column, constructing a plurality of corresponding table structure information based on the column name, the valid values and the number of valid values, and storing the plurality of table structure information in the preset information base.
In the embodiment of the application, before the related target table structure information is queried from the preset information base by utilizing the key information, the method specifically comprises the following steps of B1-B3:
and B1, acquiring a plurality of data tables, and performing de-duplication on each column of data in each data table to obtain an effective value.
In this embodiment of the present application, before performing deduplication on each column of data in each data table, table structure information needs to be segmented in the data table with columns as references, and after segmentation, each column of data in each data table is obtained, and then data deduplication, that is, data cleaning or data preprocessing is performed. The purpose of this process is to improve the data quality and remove duplicate, invalid or erroneous data, thereby making the data analysis more accurate and reliable. The specific process of deduplication may include: reading a plurality of data tables, which can be accomplished using file reading functions in various programming languages or database query languages (e.g., SQL); duplicate values may be removed for each column in the data table using a deduplication function or method; and obtaining the effective value of each column of data immediately after the deduplication processing is completed. For example, for each column in the data table, a data deduplication operation may be performed in Python using a drop_complexes function in a set or pandas library, and in SQL, a data deduplication operation may be performed using a DISTINCT key.
And step B2, acquiring the column names corresponding to the effective values and the number of the effective values.
In the embodiment of the application, after each column of data in each data table is de-duplicated, each column of data is traversed, and the column name and the corresponding number of each effective value are recorded, and the information can be stored through a data structure.
As an example, assume a data table named "data" containing three columns of data: "A", "B" and "C". After the weight is removed, the following effective value information can be obtained: valid value of column a: [ 'a1', 'a2', 'a3' ]; valid values for column B: [ 'b1', 'b2' ]; valid value of column C: the column names corresponding to the valid values may include: the valid values of columns and their number may include: the effective value of the column "A" is [ "a1", "a2", "a3" ], and the number of effective values is 3; valid values for column B: the effective value number is 2, [ 'b1', 'b2' ]; valid value of column C: the number of effective values is 4, [ ' c1', ' c2', ' c3', ' c4', '.
And B3, generating table structure information based on the column names, the effective values and the effective value quantity, and storing the table structure information in a preset information base.
In the embodiments of the present application,
may include: determining the header of the table according to the column names; filling the effective value of each column in the table according to the effective value and the number; adding other information related to the table structure according to the need, such as data types, constraint conditions and the like; representing each table structure information as a vector; and storing the generated table structure information in a vectorized form in a data table, wherein the data table is stored in a preset database.
The method provided by the embodiment of the application can remove redundant and repeated information in the data through duplication elimination, so that the data is clearer and standard; the column structure and the data distribution of each data table can be defined by acquiring the column names and the number of the effective values corresponding to the effective values, so that a foundation is provided for subsequent data analysis and processing; the unified management of the data can be realized by storing the table structure information in a preset information base; the storage mode of the preset information base enables inquiring and updating of the table structure information to be more convenient, and quick access to the information can be achieved.
In the embodiment of the application, the key information is utilized to query the related target table structure information from the preset information base, and the method specifically comprises the following steps C1-C3:
and C1, inquiring a target column name and/or a target effective value matched with the key information from a preset information base, and taking the table structure information corresponding to the target column name and/or the target effective value as candidate table structure information.
In the embodiment of the application, the key information extracted from the preset information base according to the query requirement provided by the user is queried. The key information may be viewed as keywords extracted according to the query requirements provided by the user, for finding column names and/or valid values that match the keywords. The result of the query is a series of candidate table structure information containing table structure information corresponding to the target column name and/or target valid value.
And C2, determining the similarity between the key information and each candidate list structure information.
In the embodiment of the present application, the purpose of this step is to evaluate the similarity of each candidate table structure information and the key information. The similarity may be calculated based on various factors, such as character string similarity, semantic similarity, or structural similarity, and by calculating the similarity, the matching degree of each candidate table structural information and the key information may be obtained.
As one example, if the similarity between the key information and each candidate table structure information is determined by means of text comparison, a string similarity algorithm is used to calculate the similarity between the two texts. The string similarity algorithm may include: cosine similarity, jaccard similarity, edit distance, etc., text typically needs to be preprocessed, such as to remove stop words, to perform word segmentation, to transform cases, etc., before string similarity calculation is performed, to eliminate some unimportant information and make the comparison more accurate.
And C3, arranging the candidate list structure information in a descending order according to the similarity to obtain a list structure information sequence, and selecting a preset number of candidate list structure information from the list structure information sequence as target list structure information.
In the embodiment of the application, after the similarity between each candidate list structure information and the key information is calculated, the candidate list structure information is arranged into a list structure information sequence according to a similarity descending mode. The purpose of this is to facilitate the selection of candidate table structure information with the highest similarity. The specific descending order method can be selected according to the data type and application scenario. For example, for numerical data, the numerical data is arranged in descending order directly according to the numerical value of the similarity; for text data, the similarity is normalized, and then the similarity is arranged in descending order. After the list structure information sequences arranged in a descending order are obtained, a certain amount of candidate list structure information is selected as target list structure information, and the amount is a preset value and can be adjusted according to actual requirements.
According to the method provided by the embodiment of the application, the list name and/or the target effective value of the target matched with the key information are inquired from the preset information base, so that the table structure information which is most matched with the key information can be accurately screened, and the screening accuracy and efficiency are improved; the similarity between the key information and each candidate list structure information is determined, and the candidate list structure information is arranged in a descending order according to the similarity, so that the list structure information which is most similar to the key information can be found out quickly, and a basis is provided for subsequent decisions; selecting a preset number of candidate table structure information in the table structure information sequence as target table structure information, which is helpful for making decisions in a plurality of similar candidate schemes and optimizing the selection process; through sorting and selection based on similarity, the decision making process is more scientific and objective, and subjectivity and randomness are reduced.
Step S13, generating a first code text based on the target query requirement and the target table structure information.
In the embodiment of the present application, step S13 specifically includes the following steps D1 to D3:
and D1, acquiring task description information, task output rules and code generation rules.
In the embodiments of the present application, task description information refers to a specific description or illustration of a task provided to a model. This information is used to instruct the model on how to perform a particular task or answer a particular question. Task output rules refer to the output format or specification that a model follows when performing a particular task, defining how the model generates and presents the results of the task to ensure accuracy and consistency of the output. Code generation rules define the basic syntax and structure of the code, such as variable declarations, function definitions, control structures, etc., which ensure that the generated code complies with the syntax specifications of a particular programming language, and may include in particular: grammar rules, programming formats, and functional implementation.
And D2, configuring a preset model by using the task description information, the task output rule and the code generation rule, and taking the configured preset model as a second model.
In this embodiment of the present application, the preset model may be a large language model, and the task description information, the task output rule, and the code generation rule configure to instruct the large language model to sequentially complete a plurality of task outputs, and the specific task may include: demand analysis, table structure analysis, construction of SQL statements and writing of Python codes. By employing a progressive task ordering strategy, it is ensured that the output of each step provides the necessary information and direction for the next step, thereby ensuring high quality of the final code.
As an example, as shown in fig. 4, the task description information, the task output rule, and the code generation rule may be specifically:
defining roles of the preset model: the generation object of the pandas python code;
task description information: available tools are python's pandas; the user inputs the address, table structure information and the requirement to be realized of the excel or csv file; the table information contains all information for solving the requirement, including list names, list descriptions, real list values, the number of corresponding list values types and descriptions of the whole table;
task output rules: column_info_list associated with the input sentence, specifically includes: column name, valid value, and column description (meaning of column, data characteristics, wherein the data characteristics include data type, data format, unit); a corresponding sql code; the output format of the executable python command line is json, and key is column_in_list, short, sql, code.
Code generation rules: screening column list to be processed; when the list is subjected to numerical processing such as sorting, numerical comparison, maximum value taking and the like, the form of valid values corresponding to columnnem in columm_intolist is observed first, and then, whether additional processing is needed, such as invalid value removal or unit removal, is judged. Note that reference is made to the list of examples given, which are truly existing examples, but it is possible that not all samples are given. It is not suggested to directly use the contents in the input sentence, and the contents in the table structure given by multiple references are further corrected for the user's needs. And finally, the result print is sent out.
And configuring a preset model by utilizing the specific task description information, the task output rule and the code generation rule to obtain the configured preset model, wherein the preset model is the second model.
And D3, inputting the target query requirement and the target table structure information into a second model, so that the second model extracts target content matched with the task description information from the target query requirement and the target table structure information, outputs an output result corresponding to the target content according to a task output rule, and generates a first code text corresponding to the output result according to a code generation rule.
As one example, assume that the target query requirement of the user is: product information having a vending number greater than 100 and a vending price less than 20 is queried. The target table structure information includes: the four columns of "Product ID", "Product Name", "Sales Quantity" and "Sales Price". The task output rules of the second model include: the queried Product information meeting the conditions includes "Product ID", "Product Name", "Sales Quantity" and "Sales Price". The code generation rule includes: generating a corresponding SQL query statement according to the target query requirement and the table structure information; executing SQL query sentences by using a pandas library of Python, and extracting a query result; and formatting the query result according to the task output rule, and generating a corresponding code text.
Inputting the target query requirement and the target table structure information into a second model, and extracting target contents by the second model: the second model generates corresponding SQL query sentences according to the target content, and executes the query sentences and extracts results by using a pandas library of Python; and formatting the query result into a JSON format according to the task output rule, and generating a corresponding first code text.
According to the method provided by the embodiment of the application, the task requirements can be accurately understood and related tasks can be automatically completed by acquiring the task description information, the task output rule and the code generation rule; the task description information, the task output rule and the code generation rule are utilized to configure a preset model, so that the accurate extraction of target content matched with the task description information is realized; outputting an output result corresponding to the target content according to the task output rule, and ensuring the accuracy and consistency of the output result; generating a first code text corresponding to the output result according to the code generation rule, which is helpful for quickly generating the code text meeting the requirements and reducing coding errors; by configuring the preset model, the system can adapt to different task descriptions, output rules and code generation rules, and the flexibility and adaptability of the code generation process are improved.
Step S14, executing the first code text in a preset operation environment to obtain a first code execution result, and generating feedback content corresponding to the target query requirement by using the target query requirement and the first code execution result.
In the embodiment of the present application, after executing the first code text in a preset operation environment to obtain a first code execution result, the method includes the following steps: acquiring first evaluation information corresponding to a first code execution result; determining whether the first code text has an execution abnormality according to the first evaluation information; if the first code text has the execution abnormality, determining abnormal content of the first code text, and inputting the abnormal content and the first code text into a second model based on the abnormal content, so that the second model generates a second code text based on the abnormal content and the first code text; executing the second code text in a preset operation environment to obtain a second code execution result; acquiring second evaluation information corresponding to a second code execution result, and determining whether the second code text has an execution abnormality according to the second evaluation information; and if the second code text does not have the execution abnormality, generating feedback content corresponding to the target query requirement by utilizing the target query requirement and the second code execution result.
In this embodiment, as shown in fig. 5, the code generated by the second model in the code execution stage is transmitted to the Python shell for statistics and calculation. Firstly, executing the generated code by using exec, transmitting the file to be processed in a local variable mode, and outputting the execution result of the code in a printing mode to an instantiated String IO. In addition, in order to increase the pass rate of the code, a loop generation mechanism is added. And after each time of executing the generated code, performing simple result evaluation to judge whether error reporting or abnormality occurs in the executing process. If yes, the error report content and the generated code are input into the model regeneration code as a history record. If not, the exit loop proceeds to the next stage.
As an example, assuming that the currently acquired first evaluation information is "code execution failure and database connection failure", it may be determined that the first code text has an execution exception, the exception content is the database connection failure, the exception content and the first code text are input into the second model, the second model may analyze according to the exception content and the first code text, generate a corresponding repair code or exception handling code, generate a second code text according to the repair code or the exception handling code, execute the second code text in a preset operation environment to obtain a second code execution result and second evaluation information, and if the second evaluation information is "query success and the return result is a list containing two product information", determine that the second code text does not have an execution exception, generate feedback content corresponding to the target query requirement by using the target query requirement and the second code execution result, and send the feedback content to the user side.
In the embodiment of the present application, generating feedback content corresponding to the target query requirement by using the target query requirement and the first code execution result includes the following steps: acquiring a feedback rule; inputting the target query requirement, the first code execution result and the feedback rule into a preset model, so that the preset model analyzes the target query requirement and the first code execution result to obtain an analysis result, and generating feedback content according to the feedback rule by the analysis result.
In the embodiment of the application, the preset model is responsible for information summarization and feedback content correction. The preset model inputs the query requirement of the user, the first code execution result and the answer requirement into the preset model in a text form, so that the preset model generates content meeting the standard, and the generated feedback content is displayed to the user in a text form, so that the question and answer of one-time statistical task is completed.
According to the method provided by the embodiment of the application, the first code text is executed in the preset operation environment, so that the first code execution result is obtained. Monitoring the code running effect is realized; by acquiring the feedback rule, the code execution result can be analyzed and evaluated according to the preset rule, so that the system is facilitated to comprehensively check the correctness and the effectiveness of the code execution result, and potential problems can be found and processed in time; by inputting the target query requirement, the first code execution result and the feedback rule into the preset model, the code execution can be analyzed from multiple dimensions, and the comprehensive understanding of the performance, stability and safety of the code execution is facilitated; the preset model can find potential performance bottlenecks, errors or potential safety hazards through analysis of target query requirements and first code execution results.
Fig. 6 is a flowchart of another method for processing a query requirement according to an embodiment of the present invention, as shown in fig. 6, in which steps 1-5 are a process of building a preset information base, and steps 6-14 are a process of generating feedback content corresponding to a target query requirement, and the method includes:
step 1, obtaining a plurality of data tables;
step 2, de-duplicating each column of data in each data table;
step 3, obtaining a column name corresponding to the effective value and an effective value effective amount;
step 4, generating table structure information;
step 5, storing the table structure information in a preset information base;
step 6, acquiring the query requirement currently input by the user;
step 7, extracting key information of the query requirement;
step 8, inquiring the table structure information associated with the key information;
step 9, generating a first code text based on the query requirement and the table structure information;
step 10, executing a first code text to obtain a first code execution result;
step 11, evaluating the execution result of the first code to obtain first evaluation information, if the first evaluation information is that the execution abnormality exists, determining abnormal content, executing step 12, and if the execution abnormality does not exist in the first code text, executing step 14;
Step 12, generating a second code text according to the abnormal content and the first code text;
step 13, executing the second code text to obtain a second code execution result;
step 14, evaluating the execution result of the second code to obtain second evaluation information, if the second evaluation information is that the execution abnormality exists, determining abnormal content, returning to step 12, and if the execution abnormality does not exist in the second code text, executing step 14;
and 15, generating feedback content corresponding to the target query requirement.
The present embodiment provides a device for processing a query requirement, as shown in fig. 7, including:
the acquiring module 71 is configured to acquire a target query requirement currently input by a user, and extract key information of the target query requirement;
a query module 72, configured to query the related target table structure information from a preset information base by using the key information, where the preset information base includes table structure information corresponding to a plurality of data tables;
a generating module 73, configured to generate a first code text based on the target query requirement and the target table structure information;
the execution module 74 is configured to execute the first code text in a preset operation environment to obtain a first code execution result, and generate feedback content corresponding to the target query requirement by using the target query requirement and the first code execution result.
In an embodiment of the present application, the apparatus further includes: the storage module is used for acquiring a plurality of data tables, and performing de-duplication on each column of data in each data table to obtain an effective value; acquiring the column names corresponding to the effective values and the number of the effective values; generating table structure information based on the column names, the effective values and the effective value quantity, and storing the table structure information in a preset information base.
In an embodiment of the present application, the apparatus further includes: the verification module is used for acquiring first evaluation information corresponding to the first code execution result; determining whether the first code text has an execution abnormality according to the first evaluation information; if the first code text has the execution abnormality, determining abnormal content of the first code text, and inputting the abnormal content and the first code text into a second model based on the abnormal content, so that the second model generates a second code text based on the abnormal content and the first code text; executing the second code text in a preset operation environment to obtain a second code execution result; acquiring second evaluation information corresponding to a second code execution result, and determining whether the second code text has an execution abnormality according to the second evaluation information; and if the second code text does not have the execution abnormality, generating feedback content corresponding to the target query requirement by utilizing the target query requirement and the second code execution result.
In the embodiment of the present application, the obtaining module 71 is configured to obtain a first constraint condition and a key information extraction rule; configuring a preset model by using the first constraint condition and the key information extraction rule, and taking the configured preset model as a first model; and inputting the target query requirement into the first model so that the first model obtains the requirement information meeting the first constraint condition from the target query requirement, and extracting the key information from the requirement information according to the key information extraction rule.
In the embodiment of the present application, the query module 72 is configured to query, from a preset information base, a target column name and/or a target effective value that are matched with the key information, and use table structure information corresponding to the target column name and/or the target effective value as candidate table structure information; determining the similarity between the key information and each candidate list structure information; and arranging the candidate table structure information in a descending order according to the similarity to obtain a table structure information sequence, and selecting a preset number of candidate table structure information from the table structure information sequence as target table structure information.
In the embodiment of the present application, the generating module 73 is configured to obtain task description information, a task output rule, and a code generating rule; configuring a preset model by using task description information, a task output rule and a code generation rule, and taking the configured preset model as a second model; and inputting the target query requirement and the target table structure information into a second model, so that the second model extracts target content matched with the task description information from the target query requirement and the target table structure information, outputs an output result corresponding to the target content according to a task output rule, and generates a first code text corresponding to the output result according to a code generation rule.
In the embodiment of the present application, the execution module 74 is configured to obtain a feedback rule; inputting the target query requirement, the first code execution result and the feedback rule into a preset model, so that the preset model analyzes the target query requirement and the first code execution result to obtain an analysis result, and generating feedback content according to the feedback rule by the analysis result.
Referring to fig. 8, fig. 8 is a schematic structural diagram of a computer device according to an alternative embodiment of the present invention, as shown in fig. 8, the computer device includes: one or more processors 10, memory 20, and interfaces for connecting the various components, including high-speed interfaces and low-speed interfaces. The various components are communicatively coupled to each other using different buses and may be mounted on a common motherboard or in other manners as desired. The processor may process instructions executing within the computer device, including instructions stored in or on memory to display graphical information of the GUI on an external input/output device, such as a display device coupled to the interface. In some alternative embodiments, multiple processors and/or multiple buses may be used, if desired, along with multiple memories and multiple memories. Also, multiple computer devices may be connected, each providing a portion of the necessary operations (e.g., as a server array, a set of blade servers, or a multiprocessor system).
The processor 10 may be a central processor, a network processor, or a combination thereof. The processor 10 may further include a hardware chip, among others. The hardware chip may be an application specific integrated circuit, a programmable logic device, or a combination thereof. The programmable logic device may be a complex programmable logic device, a field programmable gate array, a general-purpose array logic, or any combination thereof.
Wherein the memory 20 stores instructions executable by the at least one processor 10 to cause the at least one processor 10 to perform a method for implementing the embodiments described above.
The memory 20 may include a storage program area that may store an operating system, at least one application program required for functions, and a storage data area; the storage data area may store data created from the use of the computer device of the presentation of a sort of applet landing page, and the like. In addition, the memory 20 may include high-speed random access memory, and may also include non-transitory memory, such as at least one magnetic disk storage device, flash memory device, or other non-transitory solid-state storage device. In some alternative embodiments, memory 20 may optionally include memory located remotely from processor 10, which may be connected to the computer device via a network. Examples of such networks include, but are not limited to, the internet, intranets, local area networks, mobile communication networks, and combinations thereof.
Memory 20 may include volatile memory, such as random access memory; the memory may also include non-volatile memory, such as flash memory, hard disk, or solid state disk; the memory 20 may also comprise a combination of the above types of memories.
The computer device also includes a communication interface 30 for the computer device to communicate with other devices or communication networks.
The embodiments of the present invention also provide a computer readable storage medium, and the method according to the embodiments of the present invention described above may be implemented in hardware, firmware, or as a computer code which may be recorded on a storage medium, or as original stored in a remote storage medium or a non-transitory machine readable storage medium downloaded through a network and to be stored in a local storage medium, so that the method described herein may be stored on such software process on a storage medium using a general purpose computer, a special purpose processor, or programmable or special purpose hardware. The storage medium can be a magnetic disk, an optical disk, a read-only memory, a random access memory, a flash memory, a hard disk, a solid state disk or the like; further, the storage medium may also comprise a combination of memories of the kind described above. It will be appreciated that a computer, processor, microprocessor controller or programmable hardware includes a storage element that can store or receive software or computer code that, when accessed and executed by the computer, processor or hardware, implements the methods illustrated by the above embodiments.
Although embodiments of the present invention have been described in connection with the accompanying drawings, various modifications and variations may be made by those skilled in the art without departing from the spirit and scope of the invention, and such modifications and variations fall within the scope of the invention as defined by the appended claims.
Claims (10)
1. A method for processing a query requirement, comprising:
acquiring a target query requirement currently input by a user, and extracting key information of the target query requirement;
inquiring related target table structure information from a preset information base by utilizing the key information, wherein the preset information base comprises table structure information corresponding to a plurality of data tables;
generating a first code text based on the target query requirement and the target table structure information;
and in a preset operation environment, executing the first code text to obtain a first code execution result, and generating feedback content corresponding to the target query requirement by utilizing the target query requirement and the first code execution result.
2. The method of claim 1, wherein the extracting key information of the target query requirement comprises:
acquiring a first constraint condition and a key information extraction rule;
Configuring a preset model by utilizing the first constraint condition and the key information extraction rule, and taking the configured preset model as a first model;
and inputting the target query requirement to the first model so that the first model obtains requirement information meeting the first constraint condition from the target query requirement, and extracting the key information from the requirement information according to the key information extraction rule.
3. The method of claim 1, comprising, prior to querying the associated target table structure information from a preset information base using the key information:
obtaining a plurality of data tables, and performing de-duplication on each column of data in each data table to obtain an effective value;
acquiring the column names corresponding to the effective values and the number of the effective values;
generating table structure information based on the column names, the effective values and the effective value quantity, and storing the table structure information in the preset information base.
4. A method according to claim 3, wherein querying the associated target table structure information from a preset information base using the key information comprises:
Inquiring a target column name and/or a target effective value matched with the key information from the preset information base, and taking table structure information corresponding to the target column name and/or the target effective value as candidate table structure information;
determining the similarity between the key information and each candidate list structure information;
and arranging the candidate list structure information in a descending order according to the similarity to obtain a list structure information sequence, and selecting a preset number of candidate list structure information from the list structure information sequence as the target list structure information.
5. The method of claim 1, wherein the generating a first code text based on the target query requirement and the target table structure information comprises:
acquiring task description information, task output rules and code generation rules;
configuring a preset model by utilizing the task description information, the task output rule and the code generation rule, and taking the configured preset model as a second model;
and inputting the target query requirement and the target table structure information into the second model, so that the second model extracts target content matched with the task description information from the target query requirement and the target table structure information, outputs an output result corresponding to the target content according to the task output rule, and generates a first code text corresponding to the output result according to the code generation rule.
6. The method of claim 5, wherein after executing the first code text in a predetermined execution environment to obtain a first code execution result, the method further comprises:
acquiring first evaluation information corresponding to the first code execution result;
determining whether the first code text has an execution abnormality according to the first evaluation information;
if the first code text has the execution abnormality, determining abnormal content of the first code text, and inputting the abnormal content and the first code text into the second model based on the abnormal content, so that the second model generates a second code text based on the abnormal content and the first code text;
executing the second code text in a preset operation environment to obtain a second code execution result;
acquiring second evaluation information corresponding to the second code execution result, and determining whether the second code text has an execution abnormality according to the second evaluation information;
and if the second code text does not have the execution abnormality, generating feedback content corresponding to the target query requirement by using the target query requirement and the second code execution result.
7. The method of claim 1, wherein generating feedback content corresponding to the target query requirement using the target query requirement and the first code execution result comprises:
acquiring a feedback rule;
and inputting the target query requirement, the first code execution result and the feedback rule into a preset model so that the preset model analyzes the target query requirement and the first code execution result to obtain an analysis result, and generating the feedback content according to the feedback rule by the analysis result.
8. An apparatus for processing a query requirement, the apparatus comprising:
the acquisition module is used for acquiring a target query requirement currently input by a user and extracting key information of the target query requirement;
the query module is used for querying related target table structure information from a preset information base by utilizing the key information, wherein the preset information base comprises table structure information corresponding to a plurality of data tables;
the generation module is used for generating a first code text based on the target query requirement and the target table structure information;
and the execution module is used for executing the first code text in a preset operation environment to obtain a first code execution result, and generating feedback content corresponding to the target query requirement by utilizing the target query requirement and the first code execution result.
9. A computer device, comprising:
a memory and a processor in communication with each other, the memory having stored therein computer instructions which, upon execution, cause the processor to perform the method of any of claims 1 to 7.
10. A computer readable storage medium having stored thereon computer instructions for causing a computer to perform the method of any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410006852.5A CN117827923A (en) | 2024-01-02 | 2024-01-02 | Query demand processing method and device, computer equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410006852.5A CN117827923A (en) | 2024-01-02 | 2024-01-02 | Query demand processing method and device, computer equipment and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117827923A true CN117827923A (en) | 2024-04-05 |
Family
ID=90523889
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410006852.5A Pending CN117827923A (en) | 2024-01-02 | 2024-01-02 | Query demand processing method and device, computer equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117827923A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118069819A (en) * | 2024-04-22 | 2024-05-24 | 浙江大学 | A method and device for processing data visualization tasks based on large models |
| CN118643243A (en) * | 2024-08-15 | 2024-09-13 | 深圳市智慧城市科技发展集团有限公司 | Browser control method, browser control device and readable storage medium |
-
2024
- 2024-01-02 CN CN202410006852.5A patent/CN117827923A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118069819A (en) * | 2024-04-22 | 2024-05-24 | 浙江大学 | A method and device for processing data visualization tasks based on large models |
| CN118643243A (en) * | 2024-08-15 | 2024-09-13 | 深圳市智慧城市科技发展集团有限公司 | Browser control method, browser control device and readable storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111522816B (en) | Data processing method, device, terminal and medium based on database engine | |
| CN108256074B (en) | Verification processing method and device, electronic equipment and storage medium | |
| US11243923B2 (en) | Computing the need for standardization of a set of values | |
| CN117827923A (en) | Query demand processing method and device, computer equipment and storage medium | |
| CN110580308B (en) | Information auditing method and device, electronic equipment and storage medium | |
| CN108427720A (en) | System log sorting technique | |
| WO2022237253A1 (en) | Test case generation method, apparatus and device | |
| CN113760891B (en) | Data table generation method, device, equipment and storage medium | |
| CN112181490B (en) | Method, device, equipment and medium for identifying function category in function point evaluation method | |
| CN110109678B (en) | Code audit rule base generation method, device, equipment and medium | |
| US11334603B2 (en) | Efficiently finding potential duplicate values in data | |
| CN112445775A (en) | Fault analysis method, device, equipment and storage medium of photoetching machine | |
| CN113627168B (en) | Method, device, medium and equipment for checking component packaging conflict | |
| CN110555212A (en) | Document verification method and device based on natural language processing and electronic equipment | |
| CN115454445A (en) | Code checking method and device, computer readable storage medium and terminal | |
| Howard | Phonetic spelling algorithm implementations for R | |
| US11797578B2 (en) | Technologies for unsupervised data classification with topological methods | |
| CN116719950A (en) | Intelligent question-answering method and system based on knowledge graph sub-graph retrieval | |
| CN115470034A (en) | Log analysis method, device and storage medium | |
| CN115034187A (en) | Annotating method, annotating device, computer equipment and storage medium | |
| Zhi et al. | LLM-powered Zero-shot Online Log Parsing | |
| CN113064597A (en) | Redundant code identification method, device and equipment | |
| CN112667666A (en) | SQL operation time prediction method and system based on N-gram | |
| CN116484230B (en) | Method for identifying abnormal business data and training method of AI digital person | |
| CN119202133A (en) | Shared dimension optimization method, device, equipment and storage medium for data warehouse model |
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 |