CN118349569A - Relationship-aware production data query method, device, equipment, medium and product - Google Patents

Abstract

The invention provides a relational awareness production data query method, a relational awareness production data query device, relational awareness production data query equipment, a relational awareness production data query medium and a relational awareness production data query product, wherein the relational awareness production data query method comprises the following steps: acquiring a statement to be queried, and converting the statement to be queried into query SQL; inquiring in a current production database based on the query SQL to obtain an inquiry result, wherein the inquiry result comprises a subgraph of a production data relation diagram; the production data relation graph is dynamically updated based on the current production database, nodes in the production data relation graph correspond to production elements, and edges in the production data relation graph correspond to relations among the production elements. The invention does not need a manager to inquire each production element independently, and can improve the efficiency of obtaining the information of the auxiliary production decision.

Description

Relationship-aware production data query method, device, equipment, medium and product

Technical Field

The present invention relates to the field of artificial intelligence technology, and in particular to a relationship-aware production data query method, device, equipment, medium and product.

Background technique

In large-scale industrial production scenarios, there are many production factors, and there are correlations between production factors. Production management does not make decisions based solely on the data of a single production factor, but rather must make decisions based on multiple interrelated production factors.

In the prior art, when searching in the production database, only the data of a single production factor can be queried, and the query results cannot reflect the relationship between the production factors. The management personnel can only manually determine which related generation factors a certain production factor has, and query each production factor separately to obtain information for assisting production decision-making. This may not only miss related production factors, but also be inefficient.

Summary of the invention

The present invention provides a relationship-aware production data query method, device, equipment, medium and product to solve the defect of low efficiency in querying and obtaining information for assisting production decision-making in the prior art, thereby improving the efficiency of querying and obtaining information for assisting production decision-making.

The present invention provides a relationship-aware production data query method, comprising:

Obtaining a query statement, and converting the query statement into a query SQL;

Based on the query SQL, a query is performed in the current production database to obtain a query result, wherein the query result includes a subgraph of the production data relationship graph;

The production data relationship graph is dynamically updated based on the current production database, the nodes in the production data relationship graph correspond to production factors, and the edges in the production data relationship graph correspond to the relationships between production factors.

According to a relationship-aware production data query method provided by the present invention, converting the query statement to be queried into query SQL includes:

Inputting the query statement to be queried into a trained neural network model, and obtaining the query SQL output by the neural network model;

The neural network model includes a feature extraction layer and a conversion layer, the feature extraction layer is used to extract the intermediate semantic representation of the statement to be queried, and the conversion layer is used to convert the intermediate semantic representation into the query SQL, wherein the neural network model is trained based on multiple groups of training data, and each group of training data includes a sample statement to be queried and a query SQL label corresponding to the sample statement to be queried.

According to a relationship-aware production data query method provided by the present invention, the training process of the neural network model includes:

Selecting target training data from the training data, inputting sample query statements in the target training data set into the feature extraction layer, and obtaining the sample intermediate semantic representation output by the feature extraction layer;

Updating the feature extraction layer based on the sample intermediate semantic representation;

Repeating the step of selecting target training data from the training data until the parameters of the feature extraction layer converge, thereby obtaining the pre-trained feature extraction layer;

Inputting the sample query sentence into the neural network model, wherein the neural network model includes the pre-trained feature extraction layer;

Obtaining a sample query SQL output by the neural network model;

Determine a target training loss based on the sample query SQL and the query SQL label corresponding to the sample query statement;

The conversion layer and the feature extraction layer are updated based on the target training loss.

According to a relationship-aware production data query method provided by the present invention, the updating of the feature extraction layer based on the intermediate semantic representation of the sample includes:

Inputting the sample intermediate semantic representation into a reconstruction model to obtain a reconstructed text output by the reconstruction model;

The feature extraction layer is updated based on the reconstructed text and the sample query sentence.

According to a relationship-aware production data query method provided by the present invention, querying in the current production database based on the query SQL to obtain the query result includes:

Performing a query in the production database based on the query SQL to obtain at least one target production factor;

The subgraph of the production data relationship graph including the node corresponding to the target production factor is used as the query result corresponding to the query SQL.

According to a relationship-aware production data query method provided by the present invention, the query is performed in the current production database based on the query SQL, and after obtaining the query result, the method includes:

The subgraphs in the query result are sorted according to the degree of association between the subgraphs in the query result and the query statement.

The present invention also provides a relationship-aware production data query device, comprising:

An SQL conversion module is used to obtain a query statement and convert the query statement into a query SQL;

A query module, used to query the current production database based on the query SQL to obtain a query result, wherein the query result includes a subgraph of the production data relationship graph;

The production data relationship graph is dynamically updated based on the current production database, the nodes in the production data relationship graph correspond to production factors, and the edges in the production data relationship graph correspond to the relationships between production factors.

The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the relationship-aware production data query method described above is implemented.

The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the relationship-aware production data query method as described in any one of the above is implemented.

The present invention also provides a computer program product, comprising a computer program, wherein when the computer program is executed by a processor, the computer program implements any of the relationship-aware production data query methods described above.

The relationship-aware production data query method, device, equipment, medium and program product provided by the present invention convert the query statement into a query SQL, and after querying the production database based on the query SQL, output the query result in the form of a subgraph of the production data relationship graph. The production data relationship graph is dynamically updated based on the production database, and the nodes therein correspond to production factors, and the edges correspond to the relationships between production factors. That is to say, the query result can directly reflect the relationship between the various production factors, and there is no need for management personnel to query each production factor separately, which can improve the efficiency of obtaining information to assist production decisions.

BRIEF DESCRIPTION OF THE DRAWINGS

In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

FIG1 is a flow chart of a method for querying relationship-aware production data provided by the present invention;

FIG2 is a second flow chart of a relationship-aware production data query method provided by the present invention;

FIG3 is a schematic diagram of the structure of a relationship-aware production data query device provided by the present invention;

FIG. 4 is a schematic diagram of the structure of an electronic device provided by the present invention.

Detailed ways

In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

In today's industrial production and enterprise operations, the efficiency and accuracy of data management directly affect the quality of decision-making and the optimization of production processes. With the rapid development of information technology, especially the advancement of natural language processing (NLP), database technology and artificial intelligence (AI), enterprises are increasingly demanding intelligent data management solutions. However, although the development of these technologies has brought new opportunities for production data management, existing solutions still face many challenges in processing complex queries, achieving efficient database operations and maintenance, and providing intuitive user interaction.

With the development of cloud computing and big data technology, the scale and complexity of databases continue to increase, making database operation and management more difficult. Although the emergence of automated tools and platforms has simplified database management tasks to a certain extent, efficient management and query of large-scale and complex databases in a dynamically changing production environment still requires a high degree of professional knowledge and technical support. The popularization of intelligent production has greatly accelerated the generation and collection of production data, and enterprises have more urgent needs to analyze and utilize this data in real time to improve production efficiency and product quality. However, existing data management systems often lack sufficient flexibility and intelligence, and it is difficult to meet the needs of users to directly query and analyze production data through natural language. Existing data queries still rely on users to analyze production data to obtain related production factors and query them separately.

In summary, although the progress of database technology and artificial intelligence has brought new opportunities for production data management, new technical solutions are still needed to achieve efficient, intuitive and intelligent management of complex production data. It is in this context that the present invention is proposed to solve the problem that the existing technology requires managers to query each production factor separately, and to provide a more efficient, flexible and user-friendly solution for data management in intelligent production environments.

The following describes the relationship-aware production data query method provided by the present invention in conjunction with FIG1 . As shown in FIG1 , the method includes the steps of:

S110, obtaining a query statement to be queried, and converting the query statement to be queried into a query SQL;

S120, querying the current production database based on the query SQL to obtain a query result, where the query result includes a subgraph of the production data relationship graph;

Among them, the production data relationship graph is dynamically updated based on the current production database, the nodes in the production data relationship graph correspond to the production factors, and the edges in the production data relationship graph correspond to the relationships between the production factors.

The method provided by the present invention converts the query statement into a query SQL, performs a query in the production database based on the query SQL, and then outputs the query result in the form of a subgraph of a production data relationship graph. The production data relationship graph is dynamically updated based on the production database, wherein the nodes correspond to production factors and the edges correspond to the relationships between the production factors. That is to say, the query result can directly reflect the relationship between the various production factors, and there is no need for management personnel to query each production factor individually, thereby improving the efficiency of obtaining information to assist production decisions.

As shown in FIG2 , the query statement is a text input by the administrator through the front end that reflects the information that the administrator wants to query. The conversion of the query statement into the query SQL can be achieved by using a natural language processing model. Specifically, the conversion of the query statement into the query SQL includes:

Input the query statement to be queried into the trained neural network model, and obtain the query SQL output by the neural network model;

Among them, the neural network model includes a feature extraction layer and a conversion layer, as shown in Figure 2, the feature extraction layer is used to perform natural language parsing and extract the intermediate semantic representation of the query statement, and the conversion layer is used to convert the intermediate semantic representation into query SQL. Among them, the neural network model is trained based on multiple groups of training data, and each group of training data includes sample query statements and query SQL labels corresponding to the sample query statements.

The conversion layer can be implemented based on the existing RAT-SQL algorithm, and the parameters of the feature extraction layer and the conversion layer can be determined through training. For storage efficiency considerations, the data in the production database will not be stored in the form of natural language expressions. To query in the production database, it is necessary to use SQL to query in order to match and obtain accurate results. The present invention can improve query accuracy by using a neural network model to directly convert the query statement into a query SQL suitable for database query.

Specifically, the training process of the neural network model includes:

Select target training data from the training data, input the sample query statements in the target training data set into the feature extraction layer, and obtain the sample intermediate semantic representation output by the feature extraction layer;

Update the feature extraction layer based on the intermediate semantic representation of the sample;

Repeat the step of selecting target training data from the training data until the parameters of the feature extraction layer converge, thereby obtaining a pre-trained feature extraction layer;

Inputting the sample query sentence into the neural network model, wherein the neural network model includes a pre-trained feature extraction layer;

Get the sample query SQL output by the neural network model;

Determine the target training loss based on the query SQL labels corresponding to the sample query SQL and the sample query statements;

Update the parameters of the conversion layer and feature extraction layer based on the target training loss.

In the present invention, the training of the neural network model is divided into two stages. The first stage is to train the feature extraction layer so that the feature extraction layer can extract semantic representations that can reflect the semantics of the input sentence. Then, the conversion layer and the feature extraction layer are trained so that the pre-trained feature extraction layer and the conversion layer can be more adapted, and the conversion layer can convert the input semantic representation into accurate SQL.

Update the feature extraction layer based on the intermediate semantic representation of the sample, including:

Input the intermediate semantic representation of the sample into the reconstruction model to obtain the reconstructed text output by the reconstruction model;

Update the feature extraction layer based on the reconstructed text and sample query sentences.

Since the intermediate semantic representation of the sample does not have corresponding label data, it is impossible to determine the loss based on the difference between the data output by the feature extraction layer and the label data and then update the feature extraction layer. The method provided by the present invention uses a sentence reconstruction model to measure whether the intermediate semantic representation of the sample can accurately reflect the semantics.

Specifically, the reconstruction model may be a transformer model. The reconstruction model outputs a reconstructed text based on the intermediate semantic representation of the input sample, and updates the feature extraction layer based on the reconstructed text and the sample query sentence, including:

Update the feature extraction layer and reconstruction model based on the reconstructed text and sample query sentences.

The closer the output reconstructed text is to the sample query sentence, the smaller the training loss is, and vice versa.

The reconstruction model may use a text generation model that has been trained using a general database as an initial state before training.

The target training loss reflects the difference between the query SQL label corresponding to the sample query SQL and the sample query statement. The greater the difference, the greater the target training loss.

Update the parameters of the conversion layer and feature extraction layer based on the target training loss, including:

The conversion layer is updated based on the target training loss, and the feature extraction layer is fine-tuned.

That is to say, during the second stage of training the neural network model, the feature extraction layer is only fine-tuned, and the conversion layer is updated according to the normal training loss.

After obtaining the query SQL, query the current production database based on the query SQL to obtain the query results, including:

Based on the query SQL, query is performed in the production database to obtain at least one target production factor;

The subgraph of the production data relationship graph including the nodes corresponding to the target production factors is used as the query result corresponding to the query SQL.

Based on the query SQL, a query is performed in the production database to obtain one or more data in the production database. Based on the mapping relationship between these data and production factors, at least one target production factor corresponding to the query SQL can be obtained. For example, the query statement entered by the user is "find the task schedule of production line A today". After converting it into SQL for query, the name of the corresponding production product of production line A today can be obtained. The name of the production product is the target production factor. Based on the target production factor, the subgraph including the target production factor is searched in the production data relationship diagram. The production of a product is associated with multiple other factors, such as parts, suppliers, technicians, production standards, production raw materials, etc. The relationship between these production factors is represented in advance in the production data relationship diagram. Therefore, the subgraph obtained by the query can intuitively reflect which production factors need to be arranged accordingly if the task schedule of production line A today needs to be modified.

In order to make the query results conform to the actual production situation, the method provided by the present invention updates the production data relationship diagram in real time, that is, the production data relationship diagram is dynamically updated based on the current production database. The process of updating the production data relationship diagram can be expressed as: R _t+1 =f(R _t ,ΔS), where R _t+1 and R _t represent the production data relationship diagram at time t+1 and t respectively, ΔS represents the change of the database model, and f is the relationship diagram update function. The present invention uses a dynamic relationship diagram construction and adjustment mechanism, which enables the system to adapt to the changes of the database model in real time, thereby improving the flexibility and adaptability of the data management system.

As shown in FIG2 , after querying based on the query SQL, the query result is fed back to the user front end, and then the user's query result feedback can be received, and the neural network model can be updated based on the feedback.

Furthermore, after obtaining the query result, since the query result may include multiple subgraphs, in the method provided by the present invention, each subgraph in the query result is sorted according to the degree of association between each subgraph in the query result and the query statement.

Specifically, sorting each subgraph in the query result according to the association procedure between each subgraph in the query result and the query statement includes:

Search the historical query results for the target subgraphs corresponding to the subgraphs in the current query results. The similarity between the target subgraph corresponding to the subgraph and the subgraph is greater than the similarity between other target subgraphs and the subgraph.

The historical query statements corresponding to each target subgraph are obtained, and each subgraph is sorted based on the semantic similarity between the historical query statements corresponding to the target subgraph and the current query statement.

For example, there are three subgraphs ABC in this query result, and the corresponding target subgraphs are abc. The target subgraphs abc are obtained based on the three historical query statements A, B, and C. Then, based on the semantic similarity between A, B, and C and the query statements of this query, the three subgraphs ABC in this query result are sorted. The higher the corresponding semantic similarity, the higher the ranking.

The method provided by the present invention provides a more intuitive and flexible user interface through a context-aware natural language understanding module and an interactive query feedback loop, allowing users to easily query and analyze production data in natural language form, thereby greatly improving the user experience.

The following is a description of the relationship-aware production data query device provided by the present invention. The relationship-aware production data query device described below and the relationship-aware production data query method described above can be referred to each other. As shown in FIG3 , the relationship-aware production data query device provided by the present invention includes:

SQL conversion module 310, used to obtain the query statement to be queried and convert the query statement to be queried SQL;

A query module 320, configured to query the current production database based on the query SQL to obtain a query result, wherein the query result includes a subgraph of the production data relationship graph;

Among them, the production data relationship graph is dynamically updated based on the current production database, the nodes in the production data relationship graph correspond to the production factors, and the edges in the production data relationship graph correspond to the relationships between the production factors.

FIG4 illustrates a schematic diagram of the physical structure of an electronic device. As shown in FIG4 , the electronic device may include: a processor 410, a communication interface 420, a memory 430 and a communication bus 440, wherein the processor 410, the communication interface 420 and the memory 430 communicate with each other through the communication bus 440. The processor 410 may call the logic instructions in the memory 430 to execute the relationship-aware production data query method, which includes: obtaining a statement to be queried, converting the statement to be queried into a query SQL; querying in the current production database based on the query SQL to obtain a query result, and the query result includes a subgraph of a production data relationship graph; wherein the production data relationship graph is dynamically updated based on the current production database, the nodes in the production data relationship graph correspond to production factors, and the edges in the production data relationship graph correspond to the relationship between production factors.

In addition, the logic instructions in the above-mentioned memory 430 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.

On the other hand, the present invention also provides a computer program product, which includes a computer program. The computer program can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the relationship-aware production data query method provided by the above methods, which includes: obtaining a statement to be queried, and converting the statement to be queried into a query SQL; querying the current production database based on the query SQL to obtain a query result, and the query result includes a subgraph of the production data relationship graph; wherein the production data relationship graph is dynamically updated based on the current production database, the nodes in the production data relationship graph correspond to production factors, and the edges in the production data relationship graph correspond to the relationship between production factors.

On the other hand, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to execute the relationship-aware production data query method provided by the above-mentioned methods, the method comprising: obtaining a statement to be queried, converting the statement to be queried into a query SQL; querying in the current production database based on the query SQL to obtain a query result, the query result comprising a subgraph of a production data relationship graph; wherein the production data relationship graph is dynamically updated based on the current production database, the nodes in the production data relationship graph correspond to production factors, and the edges in the production data relationship graph correspond to the relationships between production factors.

The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Those of ordinary skill in the art may understand and implement it without creative work.

Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM/RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims (10)

1. A relationship-aware production data query method, characterized by comprising: Obtaining a query statement, and converting the query statement into a query SQL; Based on the query SQL, a query is performed in the current production database to obtain a query result, wherein the query result includes a subgraph of the production data relationship graph; The production data relationship graph is dynamically updated based on the current production database, the nodes in the production data relationship graph correspond to production factors, and the edges in the production data relationship graph correspond to the relationships between production factors. 2. The relationship-aware production data query method according to claim 1, characterized in that the step of converting the query statement to be queried into query SQL comprises: Inputting the query statement to be queried into a trained neural network model, and obtaining the query SQL output by the neural network model; The neural network model includes a feature extraction layer and a conversion layer, the feature extraction layer is used to extract the intermediate semantic representation of the statement to be queried, and the conversion layer is used to convert the intermediate semantic representation into the query SQL, wherein the neural network model is trained based on multiple groups of training data, and each group of training data includes a sample statement to be queried and a query SQL label corresponding to the sample statement to be queried. 3. The relationship-aware production data query method according to claim 2, wherein the training process of the neural network model comprises: Selecting target training data from the training data, inputting sample query statements in the target training data set into the feature extraction layer, and obtaining the sample intermediate semantic representation output by the feature extraction layer; Updating the feature extraction layer based on the sample intermediate semantic representation; Repeating the step of selecting target training data from the training data until the parameters of the feature extraction layer converge, thereby obtaining the pre-trained feature extraction layer; Inputting the sample query sentence into the neural network model, wherein the neural network model includes the pre-trained feature extraction layer; Obtaining a sample query SQL output by the neural network model; Determine a target training loss based on the sample query SQL and the query SQL label corresponding to the sample query statement; The conversion layer and the feature extraction layer are updated based on the target training loss. 4. The relationship-aware production data query method according to claim 3, characterized in that the updating of the feature extraction layer based on the sample intermediate semantic representation comprises: Inputting the sample intermediate semantic representation into a reconstruction model to obtain a reconstructed text output by the reconstruction model; The feature extraction layer is updated based on the reconstructed text and the sample query sentence. 5. The relationship-aware production data query method according to claim 1, characterized in that the querying in the current production database based on the query SQL to obtain the query result comprises: Performing a query in the production database based on the query SQL to obtain at least one target production factor; The subgraph of the production data relationship graph including the node corresponding to the target production factor is used as the query result corresponding to the query SQL. 6. The relationship-aware production data query method according to claim 5, characterized in that after querying the current production database based on the query SQL and obtaining the query result, the method comprises: The subgraphs in the query result are sorted according to the degree of association between the subgraphs in the query result and the query statement. 7. A relationship-aware production data query device, comprising: An SQL conversion module is used to obtain a query statement and convert the query statement into a query SQL; A query module, used to query the current production database based on the query SQL to obtain a query result, wherein the query result includes a subgraph of the production data relationship graph; The production data relationship graph is dynamically updated based on the current production database, the nodes in the production data relationship graph correspond to production factors, and the edges in the production data relationship graph correspond to the relationships between production factors. 8. An electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the relationship-aware production data query method as described in any one of claims 1 to 6 is implemented. 9. A non-transitory computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the relationship-aware production data query method as described in any one of claims 1 to 6 is implemented. 10. A computer program product, comprising a computer program, characterized in that when the computer program is executed by a processor, the relationship-aware production data query method as described in any one of claims 1 to 6 is implemented.