Intelligent power knowledge question-answering method and device, electronic equipment and storage medium

Abstract

The invention provides an intelligent power knowledge question-answering method and device, electronic equipment and a storage medium. The method comprises the steps of determining a question input by a user; the question input by the user is preprocessed, question analysis and question reconstruction are conducted on the preprocessed question input by the user, and a question query statement is determined; inputting the question query statement into a pre-trained intelligent question and answer model for question query, and determining a first question answer; performing answer correction on the first question answer through a natural language processing model, and determining an answer set; wherein the answer set comprises a plurality of second question answers; and sorting the second question answers, determining a question and answer result according to a sorting result, and displaying the question and answer result to a question input user.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence technology, and in particular to a method, device, electronic device and storage medium for intelligent question-answering of electric power knowledge. Background Art

[0002] With the development of the power industry and the increasing complexity of power systems, higher requirements are being placed on the acquisition and query of power knowledge. Traditional methods of querying power knowledge primarily rely on manual retrieval and expert consultation, which is inefficient and prone to errors. In recent years, the development of artificial intelligence technology, particularly the emergence of large language models, has opened up new possibilities for intelligent question-and-answering of power knowledge. Large language models can process large amounts of text data, learn rich knowledge from it, and provide users with accurate question-and-answer services. However, existing large language models still have certain limitations when it comes to power expertise. The accuracy of feedback is low, and the response speed is slow, which cannot meet the power industry's requirements for accurate and rapid feedback. Summary of the Invention

[0003] In view of this, the purpose of this application is to propose an intelligent question-answering method, device, electronic device and storage medium for power knowledge.

[0004] Based on the above objectives, this application provides an intelligent question-answering method for power knowledge, including:

[0005] Identify user input issues;

[0006] Preprocessing the user input question, and performing question analysis and question reconstruction on the preprocessed user input question to determine a question query statement; wherein the question query statement is a structured query statement;

[0007] Input the question query statement into a pre-trained intelligent question-answering model to perform question query and determine the answer to the first question;

[0008] Correcting the answer to the first question using a natural language processing model to determine an answer set; wherein the answer set includes multiple answers to the second question;

[0009] The plurality of answers to the second question are sorted, a question-and-answer result is determined according to the sorting result, and the question-and-answer result is displayed to the user who inputs the question.

[0010] Optionally, the intelligent question-answering model is pre-trained by the following method, including:

[0011] Determine professional corpus in the power field;

[0012] Preprocessing the professional corpus in the electric power field, and extracting entities and entity association information from the preprocessed professional corpus using named entity recognition technology and relationship extraction technology to determine entities and entity issues; wherein the preprocessing includes: cleaning, deduplication and formatting;

[0013] Constructing a question-answer corpus based on the entity questions and entity problems;

[0014] Iteratively train the intelligent question answering model based on the question and answer corpus, and record the number of iterations and the loss function value of each iteration;

[0015] In response to determining that the number of iterations and the loss function value meet preset conditions, the training of the intelligent question and answer model is completed and the intelligent question and answer model is determined.

[0016] Optionally, the pre-processing of the user input question includes:

[0017] The user input questions are cleaned, deduplicated and formatted.

[0018] Optionally, the step of performing question parsing and question reconstruction on the pre-processed user input question to determine a question query statement includes:

[0019] By using named entity recognition technology and relationship extraction technology, the pre-processed user input question is parsed to determine the question entity and entity association problem in the user input question;

[0020] Reconstruct the problem entity and the entity-related problem to determine the problem query statement.

[0021] Optionally, inputting the question query statement into a pre-built intelligent question-answering model to perform question query and determine the first answer includes:

[0022] Constructing a question query path based on the question query statement and the power knowledge graph of the intelligent question-answering model;

[0023] Perform a question query according to the question query path to determine the first answer.

[0024] Optionally, the correcting the first answer using a natural language processing model to determine an answer set includes:

[0025] The first answer is corrected through a natural language processing model to determine multiple answers to the second question, and the answer set is determined based on the multiple answers to the second question.

[0026] Optionally, sorting the plurality of answers to the second question and determining the question-answer result according to the sorting result includes:

[0027] Calculating the relevance between any answer to the second question and the question input by the user to determine text relevance;

[0028] Performing spelling check, grammar check, and semantic analysis on any answer to the second question, and determining a spelling check result, a grammar check result, and a semantic analysis result;

[0029] The plurality of answers to the second question are sorted according to the text relevance, spelling check result, grammar check result and semantic analysis result of any answer to the second question, and the question-answering result is determined according to the sorting result.

[0030] Based on the same inventive concept, the present application also provides an intelligent question-answering device for electric power knowledge, including:

[0031] a question input module configured to determine a user input question;

[0032] A preprocessing module is configured to preprocess the user input question, and perform question analysis and question reconstruction on the preprocessed user input question to determine a question query statement; wherein the question query statement is a structured query statement;

[0033] A question query module is configured to input the question query statement into a pre-trained intelligent question-answering model to perform question query and determine an answer to the first question;

[0034] an answer correction module, configured to correct the answer to the first question using a natural language processing model to determine an answer set; wherein the answer set includes multiple answers to the second question;

[0035] The result generation module is configured to sort the multiple answers to the second question, determine the question and answer result according to the sorting result, and display the question and answer result to the user who input the question.

[0036] Based on the same inventive concept, an embodiment of the present application also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and runnable on the processor, wherein when the processor executes the program, the intelligent question-answering method as described in any one of the above items is implemented.

[0037] Based on the same inventive concept, an embodiment of the present application also provides a non-transitory computer-readable storage medium, which stores computer instructions, and the computer instructions are used to enable a computer to execute any of the above-mentioned intelligent question-answering methods.

[0038] From the above, it can be seen that the electric power knowledge intelligent question-answering method, device, electronic device and storage medium provided by the present application include: determining the user input question; preprocessing the user input question, and performing question analysis and question reconstruction on the preprocessed user input question to determine the question query statement; inputting the question query statement into a pre-trained intelligent question-answering model to perform question query and determine the answer to the first question; correcting the answer to the first question through a natural language processing model to determine an answer set; wherein the answer set includes multiple second question answers; sorting the multiple second question answers, determining the question and answer result according to the sorting result, and displaying the question and answer result to the question input user. The present application constructs an efficient query path by reconstructing the user input question into a question query statement after parsing it, and performing question query through the question query statement, thereby reducing unnecessary search space, improving query efficiency, and sorting the multiple second question answers to determine the question and answer result, effectively improving the accuracy of the answer. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] In order to more clearly illustrate the technical solutions in this application or related technologies, the following briefly introduces the drawings required for use in the embodiments or related technical descriptions. Obviously, the drawings described below are merely embodiments of this application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0040] Figure 1 A flowchart of the intelligent question-answering method for electric power knowledge according to an embodiment of the present application;

[0041] Figure 2 This is a schematic diagram of the training process of the intelligent question-answering model in the intelligent question-answering method for power knowledge in an embodiment of the present application;

[0042] Figure 3 A schematic diagram of an intelligent question-answering model for electric power knowledge according to an embodiment of the present application;

[0043] Figure 4 This is a schematic structural diagram of an electronic device according to an embodiment of the present application. DETAILED DESCRIPTION

[0044] In order to make the objectives, technical solutions and advantages of this application more clear, this application is further described in detail below in combination with specific embodiments and with reference to the accompanying drawings.

[0045] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in the embodiments of the present application should have the usual meanings understood by people with ordinary skills in the field to which this application belongs. The "first", "second" and similar words used in the embodiments of the present application do not indicate any order, quantity or importance, but are only used to distinguish different components. "Include" or "comprise" and similar words mean that the elements or objects appearing before the word cover the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0046] In order to facilitate understanding of the technical solutions of the present disclosure, some technical terms involved in the present disclosure are introduced below.

[0047] BERT, or Bidirectional Encoder Representations from Transformers, is a language representation model based on the Transformer architecture. It was proposed by the Google AI team in 2018. BERT has achieved significant improvements in natural language processing (NLP) and has become a benchmark model for many NLP tasks, such as text classification, question-answering systems, and named entity recognition.

[0048] Named Entity Recognition (NER) is a fundamental task in natural language processing (NLP). It aims to extract and classify specific entities, such as names of people, places, and organizations, from text. Its primary purpose is to group these entities into corresponding entity types, and NER plays a vital role in a variety of applications, including information extraction, intelligent question answering, and personalized recommendations.

[0049] Relation extraction (RE) is a key task in natural language processing (NLP). Its primary goal is to automatically identify and extract relationships between entities from text. This technology plays a vital role in applications such as knowledge graph construction, sentiment analysis, and question-answering systems. Relation extraction methods include rule-based, supervised, unsupervised, semi-supervised, and deep learning-based approaches.

[0050] In order to make the technical solution of the present disclosure clearer and easier to understand, the power knowledge intelligent question-answering method provided by the embodiment of the present disclosure is described in detail below with reference to the accompanying drawings.

[0051] As mentioned in the background technology section, with the development of the power industry, the complexity of the power system has increased, and higher requirements have been placed on the acquisition and query of power knowledge. Traditional power knowledge query methods mainly rely on manual retrieval and expert consultation, which are inefficient and prone to errors. In recent years, with the development of artificial intelligence technology, especially the emergence of large language models, new possibilities have been provided for intelligent question and answer of power knowledge. Large language models can process large amounts of text data and learn rich knowledge from them to provide users with accurate question and answer services. However, existing large language models still have certain limitations in terms of power professional knowledge. The accuracy of feedback questions is low and the response speed is slow, which cannot meet the power industry's requirements for accuracy and speed of feedback questions.

[0052] In view of this, an embodiment of the present application provides an intelligent question-answering method, device, electronic device and storage medium for electric power knowledge, including: determining a question input by a user; preprocessing the question input by the user, and performing question parsing and question reconstruction on the preprocessed question input by the user to determine a question query statement; wherein the question query statement is a structured query statement; inputting the question query statement into a pre-trained intelligent question-answering model to perform a question query and determine the answer to the first question; correcting the answer to the first question through a natural language processing model to determine an answer set; wherein the answer set includes multiple answers to the second question; sorting the multiple answers to the second question, determining a question-answering result based on the sorting result, and displaying the question-answering result to the user who input the question. The present application constructs an efficient query path by reconstructing the question input by the user to form a question query statement, and performing question query through the question query statement, thereby reducing unnecessary search space, improving query efficiency, and sorting the multiple answers to the second question to determine the question-answering result, effectively improving the accuracy of the answer.

[0053] like Figure 1 As shown, the electric power knowledge intelligent question-answering method includes:

[0054] Step S102: Determine the user input question;

[0055] Step S104: pre-processing the user input question, and performing question analysis and question reconstruction on the pre-processed user input question to determine a question query statement;

[0056] Step S106: input the question query statement into the pre-trained intelligent question-answering model to perform question query and determine the answer to the first question;

[0057] Step S108: Correct the first question answer through a natural language processing model to determine an answer set; wherein, the answer set includes multiple second question answers.

[0058] Step S110: Sort the multiple second question answers, determine the Q&A result according to the sorting result, and display the Q&A result to the user who input the question.

[0059] In step S102, "user input question" can be a question in the power field input through the client of a mobile device or the client of a PC device. It includes the subject that has the problem, the reason for a certain problem, and the handling method to solve the problem. For example, the user input question can be "What are the possible reasons and handling methods for the abnormal increase in the temperature of the transformer?". It includes the subject with the problem "transformer", the problem "abnormal increase in the temperature of the transformer", the "reason" for the abnormal increase in temperature, and the "handling method" for the abnormal increase in temperature.

[0060] In some optional embodiments, the specific scenario for the user input question can be: Xiao Zhang, a power engineer, found that the temperature of a transformer in a substation was abnormally high during the inspection process. Then he input the question "What are the possible reasons and handling methods for the abnormal increase in the temperature of the transformer?" through the handheld terminal on the system interface.

[0061] In step S104, the preprocessing of the user input question includes: cleaning, de-duplicating, and formatting the user input question. Among them, data cleaning includes word segmentation, stop word removal, and noise removal. Word segmentation is to split the sentence input by the user into independent words or phrases, which is a basic step in natural language processing (NLP) and helps with subsequent analysis and processing. Stop word removal is to remove meaningless words in the sentence, such as "de", "le", "zai", etc. These words do not help much in understanding the sentence meaning but increase the processing complexity. Noise removal includes removing punctuation marks, special characters, numbers and other factors that may interfere with understanding, as well as handling spelling mistakes, abbreviations, etc. De-duplication is to identify whether there are repeated words or characters in the question input by the user. If there are repeated words or characters in the question input by the user, further determine whether the meaning to be expressed by the question after removing the repeated words or characters is the same as before deletion. If not, the repeated words are deleted. Formatting processing includes reorganizing the question after cleaning and de-duplication. Among them, during the reorganization, the question is adaptively modified to make the user input question more accurate. For example, when the user input question is "What are the possible reasons and handling methods for the abnormal increase in the temperature of the transformer le de de?", after cleaning and de-duplicating this question, it becomes "What are the possible reasons and handling methods for the abnormal increase in the temperature of the transformer respectively?".

[0062] In some embodiments, the pre-processed user input question is subjected to question parsing and question reconstruction to determine the question query statement, specifically including: using named entity recognition technology and relationship extraction technology to perform question parsing on the pre-processed user input question to determine the question entity and entity-related question in the user input question; and performing question reconstruction on the question entity and the entity-related question to determine the question query statement.

[0063] In step S106, the intelligent question answering model is a question answering model that is pre-built and trained. Specifically, Figure 2 As shown, the intelligent question answering model is pre-trained through the following steps, including:

[0064] Step S202: Determine professional corpus in the electric power field;

[0065] Step S204: pre-processing the professional corpus in the electric power field, and extracting entities and entity association information from the pre-processed professional corpus using named entity recognition technology and relationship extraction technology to determine entities and entity issues; wherein the pre-processing includes: cleaning, deduplication, and formatting;

[0066] Step S206: construct a question-answer corpus based on the entity question and entity question;

[0067] Step S208: Iteratively train the intelligent question answering model based on the question and answer corpus, and record the number of iterations and the loss function value of each iteration;

[0068] Step S210: In response to determining that the number of iterations and the loss function value meet the preset conditions, the training of the intelligent question and answer model is completed, and the intelligent question and answer model is determined.

[0069] In some embodiments, a multi-source heterogeneous original dialogue data from various power documents, manuals, reports, standards, cases, forums, patents, etc. is collected and integrated to construct a power professional corpus. Professional corpus in the power field includes but is not limited to professional knowledge in sub-fields such as power system design, operation and maintenance, fault diagnosis, new energy technology, and smart grid. Entities can be the subject in the question and the problems existing in the subject, and entity questions are questions related to the entity. For example, when the question is "What are the possible causes and treatment methods for abnormal temperature increase of the transformer?", the entities are "transformer" and "abnormal temperature increase", and the entity questions are "possible causes" and "treatment methods".

[0070] In some embodiments, efficient training methods such as low-rank decomposition are used to train the intelligent question-answering model, reducing the training cost of large models while ensuring the performance and accuracy of the model.

[0071] In some embodiments, low-rank decomposition is used in training models to reduce training costs by reducing the number of parameters that need to be optimized, thereby reducing computational complexity and storage requirements. The process of applying low-rank decomposition in training models specifically includes: parameter initialization, parameter freezing, training the low-rank matrix, and combining parameters.

[0072] Parameter initialization involves initializing two or more low-rank matrices based on a certain weight matrix of the model. The ranks of these low-rank matrices are much smaller than the rank of the original weight matrix, so the number of their parameters is also greatly reduced.

[0073] Parameter freezing involves keeping the original weight matrix unchanged during training, i.e. freezing its parameters to prevent it from being updated during training for a new task. This is done to preserve the knowledge learned by the model during the pre-training phase.

[0074] Training low-rank matrices involves focusing on training these low-rank matrices and adjusting their parameters by optimizing the loss function for the target task. Because low-rank matrices have far fewer parameters than the original weight matrices, the computational cost of training is significantly reduced. Furthermore, as the number of parameters in the low-rank matrices decreases, the storage space required by the model also decreases. Furthermore, by freezing the original weight matrices, the model retains the knowledge learned during pre-training, which is crucial for quickly adapting to new tasks.

[0075] The combined parameters include: during the forward propagation process, the updated low-rank matrix is ​​combined with the original weight matrix to generate the weight matrix for the current task. This combination ensures that the model can adapt to the needs of new tasks while maintaining pre-training knowledge.

[0076] Therefore, low-rank decomposition achieves the goal of reducing training costs in training models by reducing the number of parameters that need to be optimized. By selecting an appropriate low-rank decomposition method and adjusting and optimizing it based on the actual situation of the model, the training efficiency and performance of the model are further improved.

[0077] Furthermore, low-rank decomposition methods include singular value decomposition (SVD) and nuclear norm minimization. Among them, singular value decomposition is a commonly used low-rank decomposition method that decomposes a matrix into the product of three matrices: a left singular matrix, a diagonal matrix (containing singular values), and a right singular matrix. By selecting the larger singular values ​​in the diagonal matrix (i.e., ignoring the smaller singular values), a high-rank matrix can be approximated as a low-rank matrix, thereby achieving compression. Nuclear norm minimization is another low-rank decomposition method that achieves compression by minimizing the nuclear norm of the matrix (i.e., the sum of the matrix singular values). This method is more flexible than SVD and can be applied to various types of matrices.

[0078] Step S106 inputs the question query statement into a pre-built intelligent question-answering model to perform question query and determine the first answer, including: constructing a question query path based on the question query statement and the power knowledge graph of the intelligent question-answering model; performing question query according to the question query path to determine the first answer.

[0079] In some embodiments, a question query path is constructed based on the question query statement and the power knowledge graph of the intelligent question-answering model, specifically including: analyzing the syntax and semantics of the question query statement, determining the index, determining the data access method, etc.

[0080] In some embodiments, for a power knowledge graph: Based on the information from the corpus building module, the power knowledge graph is constructed using technologies such as entity recognition and relationship extraction. Nodes in the graph represent entities in the power sector (e.g., equipment, technology, concepts, etc.), and edges represent relationships between entities (e.g., inclusion, association, causality, etc.), forming a structured power knowledge network.

[0081] Specifically, the power knowledge graph is a graph-based data structure composed of nodes (representing entities) and edges (representing relationships). In the power sector, these nodes can represent various power equipment (such as generators, transformers, switches), technologies (such as smart grid technology and power transmission technology), concepts (such as voltage, current, and resistance), and entities related to the power market (such as power companies and energy policies). Edges describe the various relationships between these entities, such as inclusion, association, and causality.

[0082] In some optional implementations, the construction of the power knowledge graph specifically includes: (1) Data collection: Collect data from various corpora in the power field, including power technical documents, equipment manuals, market reports, academic papers, etc. These data may contain structured data (such as tables in a database) and unstructured data (such as text descriptions, images, etc.). (2) Entity recognition: Use the entity recognition algorithm in natural language processing technology (NLP) to identify entities in the power field from the collected text data. For example, from the sentence "This substation adopts advanced smart grid technology", the two entities "substation" and "smart grid technology" can be identified. (3) Relationship extraction: After identifying the entities, further analyze the context information in the text to extract the relationship between the entities. For example, from the above sentence, the "adoption" relationship between "substation" and "smart grid technology" can be extracted. (4) Knowledge representation: The identified entities and relationships are represented in symbolic form and constructed into nodes and edges of the knowledge graph. For example, in the knowledge graph, “substation” and “smart grid technology” are two nodes, which are connected by an edge representing the “adoption” relationship.

[0083] For example, for a knowledge graph fragment about power transformers, the nodes include: transformer (an electrical equipment entity), cooling system (a component of the transformer), insulating oil (a medium used inside the transformer), and rated capacity (an important parameter of the transformer). The edges are: transformer-contains-cooling-system (indicating that the transformer contains the cooling system component), transformer-uses-insulating-oil (indicating that the transformer uses insulating oil as a medium), and transformer-has-rated-capacity (indicating that the transformer has a specific rated capacity parameter).

[0084] Step S108 corrects the first answer through a natural language processing model to determine an answer set, including: correcting the first answer through a natural language processing model to determine multiple answers to the second question, and determining the answer set based on the multiple answers to the second question.

[0085] Step S110 sorts the multiple answers to the second question and determines the question-and-answer result based on the sorting results, including: calculating the relevance between any answer to the second question and the question input by the user to determine the text relevance; performing spelling check, grammar check and semantic analysis on any answer to the second question to determine the spelling check result, grammar check result and semantic analysis result; sorting the multiple answers to the second question based on the text relevance, spelling check result, grammar check result and semantic analysis result of any answer to the second question, and determining the question-and-answer result based on the sorting results.

[0086] In some embodiments, the system displays the optimal answer to Xiao Zhang, which lists in detail several possible reasons for the abnormal increase in transformer temperature (such as overload operation, cooling system failure, insulation aging, etc.) and corresponding treatment methods (such as reducing the load, checking and repairing the cooling system, replacing insulation materials, etc.).

[0087] In some examples, power engineers conducted appropriate inspections and actions based on the system's responses, successfully resolving the issue of abnormally high transformer temperatures. The power engineers then submitted satisfaction ratings and feedback through the system interface, helping the system further optimize performance.

[0088] As can be seen from the above, the electric power knowledge intelligent question-answering method provided by the present application includes: determining the user input question; preprocessing the user input question, and performing question analysis and question reconstruction on the preprocessed user input question to determine the question query statement; inputting the question query statement into the pre-trained intelligent question-answering model to perform question query and determine the answer to the first question; correcting the answer to the first question through the natural language processing model to determine the answer set; wherein the answer set includes multiple second question answers; sorting the multiple second question answers, determining the question and answer result according to the sorting result, and displaying the question and answer result to the question input user. The present application constructs an efficient query path by reconstructing the user input question into a question query statement after parsing it, and performing question query through the question query statement, thereby reducing unnecessary search space, improving query efficiency, and sorting the multiple second question answers to determine the question and answer result, effectively improving the accuracy of the answer.

[0089] It should be noted that the method of the embodiment of the present application can be performed by a single device, such as a computer or server. The method of this embodiment can also be applied in a distributed scenario and performed by multiple devices working together. In such a distributed scenario, one of the multiple devices may only perform one or more steps of the method of the embodiment of the present application, and the multiple devices will interact with each other to complete the method.

[0090] It should be noted that the above description is limited to some embodiments of the present application. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims may be performed in an order different from that described in the above embodiments and still achieve the desired results. Furthermore, the processes depicted in the accompanying drawings do not necessarily require the specific order or sequential order shown to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0091] Based on the same inventive concept, corresponding to any of the above-mentioned embodiment methods, the present application also provides an intelligent question-and-answer device for electric power knowledge.

[0092] refer to Figure 3 The electric power knowledge intelligent question-answering device comprises:

[0093] The question input module 302 is configured to determine a user input question;

[0094] The preprocessing module 304 is configured to preprocess the user input question, and perform question analysis and question reconstruction on the preprocessed user input question to determine a question query statement;

[0095] The question query module 306 is configured to input the question query statement into a pre-trained intelligent question-answering model to perform a question query and determine an answer to the first question;

[0096] An answer correction module 308 is configured to correct the answer to the first question using a natural language processing model to determine an answer set; wherein the answer set includes multiple answers to the second question;

[0097] The result generation module 310 is configured to sort the multiple answers to the second question, determine the question-answering result according to the sorting result, and display the question-answering result to the user who input the question.

[0098] For the convenience of description, the above devices are described as being divided into various modules according to their functions. Of course, when implementing this application, the functions of each module can be implemented in the same or multiple software and / or hardware.

[0099] The device of the above embodiment is used to implement the corresponding intelligent question-answering method of power knowledge in any of the above embodiments, and has the beneficial effects of the corresponding method embodiment, which will not be repeated here.

[0100] Based on the same inventive concept, corresponding to any of the above-mentioned embodiments and methods, the present application also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and runnable on the processor, wherein when the processor executes the program, the intelligent question-and-answer method for electric power knowledge described in any of the above embodiments is implemented.

[0101] Figure 4 10 is a schematic diagram showing a more specific hardware structure of an electronic device provided in this embodiment. The device may include: a processor 1010, a memory 1020, an input / output interface 1030, a communication interface 1040, and a bus 1050. The processor 1010, the memory 1020, the input / output interface 1030, and the communication interface 1040 are communicatively connected to each other within the device via the bus 1050.

[0102] The processor 1010 can be implemented using a general-purpose CPU (Central Processing Unit), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of this specification.

[0103] The memory 1020 can be implemented in the form of ROM (Read Only Memory), RAM (Random Access Memory), static storage devices, dynamic storage devices, etc. The memory 1020 can store an operating system and other application programs. When the technical solutions provided in the embodiments of this specification are implemented through software or firmware, the relevant program code is stored in the memory 1020 and is called and executed by the processor 1010.

[0104] The input / output interface 1030 is used to connect input / output modules to implement information input and output. The input / output modules can be configured as components within the device (not shown in the figure) or can be externally connected to the device to provide corresponding functions. Input devices may include a keyboard, mouse, touch screen, microphone, various sensors, etc., and output devices may include a display, speaker, vibrator, indicator light, etc.

[0105] The communication interface 1040 is used to connect to a communication module (not shown) to enable communication between the device and other devices. The communication module can communicate via a wired method (such as USB, network cable, etc.) or a wireless method (such as mobile network, WiFi, Bluetooth, etc.).

[0106] The bus 1050 comprises a path for transmitting information between the various components of the device (eg, the processor 1010 , the memory 1020 , the input / output interface 1030 , and the communication interface 1040 ).

[0107] It should be noted that although the above device only shows the processor 1010, the memory 1020, the input / output interface 1030, the communication interface 1040, and the bus 1050, in a specific implementation, the device may also include other components necessary for normal operation. In addition, it will be understood by those skilled in the art that the above device may only include the components necessary to implement the embodiments of this specification, and does not necessarily include all the components shown in the figure.

[0108] The electronic device of the above embodiment is used to implement the corresponding intelligent question-answering method of power knowledge in any of the above embodiments, and has the beneficial effects of the corresponding method embodiment, which will not be repeated here.

[0109] Based on the same inventive concept, corresponding to any of the above-mentioned embodiment methods, the present application also provides a non-transitory computer-readable storage medium, which stores computer instructions, and the computer instructions are used to enable the computer to execute the intelligent question-and-answer method of electric power knowledge as described in any of the above embodiments.

[0110] The computer-readable media of this embodiment include permanent and non-permanent, removable and non-removable media that can be used to store information by any method or technology. The information can be computer-readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, read-only compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device.

[0111] The computer instructions stored in the storage medium of the above embodiment are used to enable the computer to execute the intelligent question-answering method for power knowledge as described in any of the above embodiments, and have the beneficial effects of the corresponding method embodiments, which will not be repeated here.

[0112] Based on the same inventive concept, corresponding to the intelligent power knowledge Q&A method described in any of the above embodiments, the present disclosure further provides a computer program product comprising computer program instructions. In some embodiments, the computer program instructions can be executed by one or more processors of a computer to cause the computer and / or the processor to perform the intelligent power knowledge Q&A method. For the execution entities corresponding to the steps in each embodiment of the intelligent power knowledge Q&A method, the processors executing the corresponding steps can belong to the corresponding execution entities.

[0113] The computer program product of the above embodiment is used to enable the computer and / or the processor to execute the intelligent question-answering method of power knowledge as described in any of the above embodiments, and has the beneficial effects of the corresponding method embodiments, which will not be repeated here.

[0114] Those skilled in the art should understand that the discussion of any of the above embodiments is merely illustrative and is not intended to imply that the scope of the present application (including the claims) is limited to these examples. Within the scope of the present application, the technical features in the above embodiments or different embodiments may be combined, the steps may be implemented in any order, and there are many other variations of the different aspects of the embodiments of the present application as described above, which are not provided in detail for the sake of simplicity.

[0115] In addition, for simplicity of description and discussion, and in order not to make the embodiment of the application difficult to understand, the known power supply / ground connection with integrated circuit (IC) chip and other components may or may not be shown in the accompanying drawings provided. In addition, the device can be shown in the form of a block diagram to avoid making the embodiment of the application difficult to understand, and this also takes into account the following fact, that is, the details of the embodiment of these block diagram devices are highly dependent on the platform to be implemented in the embodiment of the application (that is, these details should be fully within the scope of understanding of those skilled in the art). When specific details (for example, circuit) are set forth to describe exemplary embodiments of the application, it will be apparent to those skilled in the art that the embodiment of the application can be implemented without these specific details or when these specific details are changed. Therefore, these descriptions should be considered to be illustrative rather than restrictive.

[0116] Although the present invention has been described in conjunction with specific embodiments thereof, many alternatives, modifications, and variations of these embodiments will be apparent to those skilled in the art based on the foregoing description. For example, other memory architectures (e.g., dynamic RAM (DRAM)) may utilize the embodiments discussed.

[0117] The embodiments of the present application are intended to cover all such substitutions, modifications, and variations that fall within the broad scope of the appended claims. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the embodiments of the present application should be included in the scope of protection of this application.

Claims

1. An intelligent question-answering method for electric power knowledge, characterized in that: include: Identify user input issues; Preprocessing the user input question, and performing question analysis and question reconstruction on the preprocessed user input question to determine a question query statement; wherein the question query statement is a structured query statement; Inputting the question query statement into a pre-trained intelligent question-answering model to perform question query and determine the answer to the first question; Correcting the answer to the first question through a natural language processing model to determine an answer set; wherein the answer set includes multiple answers to the second question; The plurality of answers to the second question are sorted, a question-and-answer result is determined according to the sorting result, and the question-and-answer result is displayed to the user who inputs the question.

2. The method according to claim 1, characterized in that The intelligent question answering model is pre-trained by the following method, including: Determine professional corpus in the power sector; Preprocessing the professional corpus in the electric power field, and extracting entities and entity-related information from the preprocessed professional corpus in the electric power field through named entity recognition technology and relationship extraction technology to determine entities and entity problems; wherein the preprocessing includes: cleaning, deduplication and formatting; Constructing a question-answer corpus based on the entity questions and entity problems; Iteratively train the intelligent question answering model according to the question and answer corpus, and record the number of iterations and the loss function value of each iteration; In response to determining that the number of iterations and the loss function value meet preset conditions, the training of the intelligent question and answer model is completed and the intelligent question and answer model is determined.

3. The method according to claim 1, characterized in that The preprocessing of the user input question includes: The user input questions are cleaned, deduplicated and formatted.

4. The method according to claim 1, characterized in that: The step of parsing and reconstructing the preprocessed user input question to determine the question query statement includes: By using named entity recognition technology and relationship extraction technology, the pre-processed user input question is analyzed to determine the question entity and entity association problem in the user input question; Reconstruct the problem entity and the entity-related problem to determine the problem query statement.

5. The method according to claim 1, characterized in that The step of inputting the question query statement into a pre-built intelligent question-answering model to query the question and determine the first answer includes: Constructing a question query path according to the question query statement and the power knowledge graph of the intelligent question-answering model; A question query is performed according to the question query path to determine the first answer.

6. The method according to claim 1, characterized in that The step of correcting the answer to the first question by using a natural language processing model to determine an answer set includes: The answer to the first question is corrected through a natural language processing model to determine multiple answers to the second question, and the answer set is determined based on the multiple answers to the second question.

7. The method according to claim 1, characterized in that The step of sorting the plurality of answers to the second question and determining the question-answering result according to the sorting result includes: Calculate the relevance between any answer to the second question and the question input by the user to determine the text relevance; Performing spelling check, grammar check and semantic analysis on any answer to the second question, and determining a spelling check result, a grammar check result and a semantic analysis result; According to the text relevance, spelling check result, grammar check result and semantic analysis result of any answer to the second question, multiple answers to the second question are sorted, and the question-and-answer result is determined according to the sorting result.

8. An intelligent question-answering device for electric power knowledge, characterized in that: include: a question input module configured to determine a user input question; A preprocessing module is configured to preprocess the user input question, and perform question analysis and question reconstruction on the preprocessed user input question to determine a question query statement; wherein the question query statement is a structured query statement; A question query module is configured to input the question query statement into a pre-trained intelligent question-answering model to perform question query and determine an answer to the first question; an answer correction module, configured to correct the answer to the first question through a natural language processing model to determine an answer set; wherein the answer set includes multiple answers to the second question; The result generation module is configured to sort the multiple answers to the second question, determine the question and answer result according to the sorting result, and display the question and answer result to the user who inputs the question.

9. An electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable by the processor, wherein the processor implements the method according to any one of claims 1 to 7 when executing the computer program. 10 . A non-transitory computer-readable storage medium storing computer instructions for causing a computer to execute the method according to claim 1 .

Images