Industrial field knowledge question answering method and device, equipment, storage medium and product

Abstract

The invention discloses an industrial field knowledge question answering method and device, equipment, a storage medium and a product, and relates to the technical field of industrial manufacturing, and the method comprises the steps: carrying out the key information extraction of a question input by a user, and carrying out the matching of an extraction result with a thinking chain cue word template, and obtaining prompt information; the prompt information is input into the large model for answer prediction, and a prediction result is obtained; key prediction information in the prediction result is extracted, a meta-model set is called based on the key prediction information for prediction, a question and answer result output by the meta-model set is obtained, and the meta-model set comprises meta-models obtained by disassembling an industrial mechanism model and an optimization algorithm. The answer prediction is performed based on the large model, and then the optimization prediction is performed based on the meta-model set to obtain the question and answer result, so that compared with an existing mode of performing question and answer only depending on the large model, the method can meet the requirements of complex professional questions in the industrial field, realize refined question and answer processing, and ensure the accuracy and specialty of question and answer.

Description

Technical Field

[0001] This application relates to the technical field of industrial manufacturing, and particularly to a method, device, equipment, storage medium and product for knowledge Q&A in the industrial field. Background Art

[0002] In the past few decades, the production optimization and monitoring in the industrial manufacturing field can be summarized into two methods: Firstly, through symbolicism, rule systems and expert systems, solutions for specific industrial tasks have been achieved, and the production efficiency and sustainability have been improved by integrating various information technologies; with the emergence of big data and the improvement of computing power, after the manufacturing industry entered the next-generation technology and wide application stage of the artificial intelligence field (i.e., the AI2.0 stage), it has become more dependent on data-driven machine learning and deep learning methods, which can not only perform predictive maintenance of industrial processes to reduce production line failures, but also use relevant algorithms for product monitoring to ensure product quality. In November 2022, ChatGPT3.5 came out, which has triggered a series of research booms on large models in the fields of text generation and intelligent dialogue, and the large model technology is leading a new round of industrial revolution. In the future, AI large models will be integrated into every link of the industrial manufacturing field, from product R & D design and process, to quality control and operation, and then to organizational collaboration and operation management, accelerating the intelligent upgrading of industrial manufacturing enterprises. Although large models (such as GPT, etc.) have demonstrated powerful capabilities in natural language processing and understanding, and can deeply analyze and answer complex questions raised by users, there are certain limitations in relying solely on large models for Q&A processing, especially in professional fields, where the accuracy and professionalism of the answers are difficult to fully guarantee. Summary of the Invention

[0003] The main purpose of this application is to provide a method, device, equipment, storage medium and product for knowledge Q&A in the industrial field, aiming to solve the technical problem that existing large models have broad and insufficiently distinguishable answer content when facing industrial knowledge Q&A with a high degree of professionalism.

[0004] To achieve the above purpose, this application proposes a method for knowledge Q&A in the industrial field, and the method for knowledge Q&A in the industrial field includes:

[0005] Extract key information from the question input by the user, and match the extraction result with the thought chain prompt word template to obtain prompt information;

[0006] Input the prompt information into the large model for answer prediction to obtain a prediction result;

[0007] Extract the key prediction information from the prediction results, and based on the key prediction information, call the meta-model set for prediction to obtain the Q&A results output by the meta-model set, where the meta-model set includes meta-models obtained by disassembling the industrial mechanism model and the optimization algorithm.

[0008] Optionally, before the step of extracting the key prediction information from the prediction results, calling the meta-model set for prediction based on the key prediction information, and obtaining the Q&A results output by the meta-model set, further includes:

[0009] Obtain the industrial mechanism model and the industrial optimization algorithm;

[0010] Disassemble the industrial mechanism model and the industrial optimization algorithm to obtain meta-models;

[0011] Construct a meta-model set based on the meta-models.

[0012] Optionally, the step of constructing a meta-model set based on the meta-models includes:

[0013] Determine the partial order relationship between the meta-models;

[0014] Construct a directed acyclic graph based on the partial order relationship and the meta-models;

[0015] Determine the meta-model set according to the directed acyclic graph.

[0016] Optionally, the thought chain prompt word template includes: application background information, expected answer format, object and audience, few-shot prompt words;

[0017] After the step of constructing a meta-model set based on the meta-models, further includes:

[0018] Determine the cascade relationship between the meta-models in the meta-model set;

[0019] Construct a thought chain prompt word template according to the cascade relationship.

[0020] Optionally, the step of extracting the key prediction information from the prediction results, calling the meta-model set for prediction based on the key prediction information, and obtaining the Q&A results output by the meta-model set includes:

[0021] Extract key information from the prediction results to obtain key prediction information;

[0022] Determine the target meta-model in the meta-model set corresponding to the key prediction information;

[0023] Perform re-prediction on the prediction results based on the target meta-model to obtain the Q&A results.

[0024] Optionally, the interconnection relationships between the meta-models in the meta-model set are represented by the following transformation equations:

[0025]

[0026] Among them, DP(i) is used to represent the dynamic programming algorithm, which is a meta-model in the meta-model set. i is used to represent the state of DP(i). s represents a continuous series of sub-process index sequences that occur before the cascade industrial specific process index occurs. N is used to represent the number of process index sequences. f: N N ×I×X J ×y K →y, g: N N ×I → (N ∪ φ) J ,h: N N ×I → (N ∪ φ) K ,I is used to represent the state space, X is used to represent the input space, DP is used to represent the output space, y is used to represent any domain, J represents the number of input tokens, K represents the dimension of the output state space. If the token length for determining state i is less than J or the previous state for determining state i is less than K, then φ represents a placeholder. s φ and DP(φ) are not used in the function f.

[0027] In addition, to achieve the above object, the present application also proposes an industrial domain knowledge Q&A device, which includes:

[0028] An input module, which is used to extract key information from the question input by the user, and match the extraction result with the thinking chain prompt word template to obtain prompt information;

[0029] A large model prediction module, which is used to input the prompt information into the large model for answer prediction to obtain a prediction result;

[0030] A meta-model set prediction module, which is used to extract key prediction information from the prediction result, and based on the key prediction information, call the meta-model set for prediction to obtain the Q&A result output by the meta-model set. Among them, the meta-model set includes meta-models obtained by disassembling the industrial mechanism model and the optimization algorithm.

[0031] In addition, to achieve the above object, the present application also proposes an industrial domain knowledge Q&A device, which includes: a memory, a processor, and a computer program stored on the memory and executable on the processor. The computer program is configured to implement the steps of the industrial domain knowledge Q&A method as described above.

[0032] In addition, to achieve the above object, the present application also provides a storage medium, which is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, the steps of the industrial field knowledge Q&A method described above are implemented.

[0033] In addition, to achieve the above object, the present application also provides a computer program product, which includes a computer program. When the computer program is executed by a processor, the steps of the industrial field knowledge Q&A method described above are implemented.

[0034] The present application extracts key information from the questions input by users, matches the extraction results with the thought chain prompt word template to obtain prompt information, inputs the prompt information into a large model for answer prediction to obtain a prediction result, extracts the key prediction information in the prediction result, and calls a meta-model set for prediction based on the key prediction information to obtain the Q&A result output by the meta-model set. The meta-model set includes meta-models obtained by disassembling industrial mechanism models and optimization algorithms. Since the present application first performs answer prediction based on a large model and then optimizes the prediction result based on the meta-model set obtained by disassembling industrial mechanism models and optimization algorithms to obtain the Q&A result, compared with the existing method of simply relying on a large model for Q&A, the above method of the present application can meet the requirements of complex professional questions in the industrial field, achieve refined Q&A processing, and ensure the accuracy and professionalism of Q&A. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] The drawings here are incorporated into the specification and constitute a part of this specification, showing embodiments consistent with the present application and used together with the specification to explain the principles of the present application.

[0036] To more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, for those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0037] Figure 1 It is a schematic flowchart provided for the first embodiment of the industrial field knowledge Q&A method of the present application;

[0038] Figure 2 It is a schematic flowchart provided for the second embodiment of the industrial field knowledge Q&A method of the present application;

[0039] Figure 3 It is the overall schematic flowchart provided for the second embodiment of the industrial field knowledge Q&A method of the present application;

[0040] Figure 4It is a schematic diagram of the module structure of the industrial field knowledge Q&A device according to an embodiment of the present application;

[0041] Figure 5 It is a schematic diagram of the device structure of the hardware operating environment involved in the industrial field knowledge Q&A method according to an embodiment of the present application.

[0042] The implementation, functional features and advantages of the present application will be further described in conjunction with the embodiments with reference to the accompanying drawings. Specific embodiments

[0043] It should be understood that the specific embodiments described herein are only used to explain the technical solutions of the present application and are not used to limit the present application.

[0044] In order to better understand the technical solutions of the present application, the following will be described in detail in conjunction with the accompanying drawings of the specification and specific embodiments.

[0045] The main solution of the embodiment of the present application is: extracting key information from the question input by the user, matching the extraction result with the thought chain prompt word template to obtain prompt information; inputting the prompt information into the large model for answer prediction to obtain a prediction result; extracting the key prediction information in the prediction result, and based on the key prediction information, calling the meta-model set for prediction to obtain the Q&A result output by the meta-model set, where the meta-model set includes meta-models obtained by disassembling the industrial mechanism model and the optimization algorithm.

[0046] Compared with the existing method of relying solely on the large model for Q&A, the present application has significant advantages in the following aspects:

[0047] Modular design principle: disassembling a complex system into multiple independent meta-models, so that each meta-model is responsible for processing specific subtasks or functions. The meta-model set has good scalability and can dynamically update knowledge nodes and relationships to keep the knowledge base up-to-date and accurate.

[0048] By extracting the keywords of the question input by the user and matching the thought chain prompt word template, relevant knowledge nodes can be quickly located, improving the efficiency and accuracy of information retrieval. The large model can perform preliminary Q&A processing, and then use the graph database (i.e., the meta-model set) for refinement and verification to form a cascaded processing mechanism.

[0049] Combining the professional knowledge of the meta-model set and the powerful processing ability of the large model, using the thought chain prompt word template, it is possible to perform refined processing on the knowledge in a specific industrial field, systematically construct answers, ensure clear logic and well-organized answers, help users better understand complex questions, meet the complex and professional question needs in the industry, and provide accurate and in-depth professional knowledge services.

[0050] It should be noted that the execution subject of this embodiment can be a computing service device with data processing, network communication, and program running functions, such as a tablet computer, a personal computer, a mobile phone, etc., or an electronic device, an industrial field knowledge Q&A device, etc. that can implement the above functions. Hereinafter, taking the industrial field knowledge Q&A device as an example, this embodiment and the following embodiments will be described.

[0051] Based on this, an embodiment of the present application provides an industrial field knowledge Q&A method. Refer to Figure 1 , Figure 1 which is a schematic flowchart of the first embodiment of the industrial field knowledge Q&A method of the present application.

[0052] In this embodiment, the industrial field knowledge Q&A method includes steps S10 to S30:

[0053] Step S10, extract key information from the question input by the user, and match the extraction result with the thought chain prompt word template to obtain prompt information;

[0054] It should be noted that the extraction of key information from the question input by the user can be to extract the key information in the question input by the user by means of keyword extraction or feature extraction. The thought chain prompt word template can be a thought chain prompt word format template formed according to the cascade relationship between complex industrial meta-models in the meta-model set. The matching of the extraction result with the thought chain prompt word template to obtain prompt information can be to select the corresponding thought chain prompt word template according to the extraction result, and then splice it to obtain prompt information.

[0055] Step S20, input the prompt information into a large model for answer prediction to obtain a prediction result;

[0056] It should be noted that the large model can be an existing artificial intelligence large model. In this embodiment, first, the large model is guided by the prompt information spliced with the thought chain prompt word template to generate a logical answer, and the prediction result output by the large model is obtained.

[0057] Step S30, extract the key prediction information in the prediction result, and call the meta-model set for prediction based on the key prediction information to obtain the Q&A result output by the meta-model set, where the meta-model set includes meta-models obtained by disassembling industrial mechanism models and optimization algorithms.

[0058] It should be noted that the extraction of the key prediction information from the prediction result can be carried out by means of key information extraction and / or feature extraction, etc. to extract the key prediction information from the prediction result. The invocation of the meta-model set for prediction based on the key prediction information can be to invoke the corresponding meta-model in the meta-model set based on the key prediction information to obtain the Q&A result. The industrial mechanism model and the optimization algorithm can be an analysis model / algorithm based on the simulation and principle of equipment and products, used to describe the physical, chemical, mechanical and other laws in the industrial process. It can be expressed by means of mathematical equations, algorithms, logic, etc., and can be used to simulate, predict and optimize the industrial process.

[0059] In this embodiment, the key information of the question input by the user is extracted, and the extraction result is matched with the thought chain prompt word template to obtain the prompt information; the prompt information is input into the large model for answer prediction to obtain the prediction result; the key prediction information in the prediction result is extracted, and based on the key prediction information, the meta-model set is invoked for prediction to obtain the Q&A result output by the meta-model set, where the meta-model set includes the meta-models obtained by disassembling the industrial mechanism model and the optimization algorithm. Since this embodiment first performs answer prediction based on the large model, and then optimizes the prediction result based on the meta-model set obtained by disassembling the industrial mechanism model and the optimization algorithm to obtain the Q&A result, compared with the existing method of simply relying on the large model for Q&A, the above method of this embodiment can meet the needs of complex professional questions in the industrial field, realize refined Q&A processing, and ensure the accuracy and professionalism of the Q&A.

[0060] Based on the first embodiment of the present application, in the second embodiment of the present application, the same or similar content as that in the above-mentioned first embodiment can be referred to the above introduction and will not be repeated hereinafter. On this basis, please refer to Figure 2 , before the step S10, the following steps are further included:

[0061] Step S001: Obtain the industrial mechanism model and the industrial optimization algorithm;

[0062] Step S002: Disassemble the industrial mechanism model and the industrial optimization algorithm to obtain meta-models;

[0063] It should be noted that the disassembling of the industrial mechanism model and the industrial optimization algorithm to obtain meta-models can be to disassemble the complex industrial mechanism model and industrial optimization algorithm according to the reaction mechanism to form meta-models that cannot be further disassembled.

[0064] Step S003: Construct a meta-model set based on the meta-models.

[0065] It should be noted that constructing a meta-model set based on the meta-model can form a directed acyclic graph (DAG) in the state space based on the partial order relationship between the disassembled meta-models to obtain the meta-model set.

[0066] Further, step S003 may include: determining the partial order relationship between the meta-models;

[0067] Constructing a directed acyclic graph based on the partial order relationship and the meta-models;

[0068] Determining the meta-model set according to the directed acyclic graph.

[0069] It should be noted that there is a partial order relationship between different meta-models, that is, if under a certain specific cascaded industrial problem, sub-problem j should be solved before sub-problem i (for example, a certain index needs to call the meta-model M j (Θ j ) calculation is completed, but the key input parameters in M j (Θ j ) need to be calculated by calling the meta-model M i (Θ i ))), here it is defined that state j is before state i (denoted as j < i), that is, the output value of the dynamic programming algorithm DP(i) depends on DP(j). This partial order will form a directed acyclic graph (DAG) in the state space, and the directed acyclic graph is the meta-model set representing the cascaded relationship between the meta-models. Specifically, the complex model in the cascaded industrial background is disassembled into indivisible meta-models according to the reaction mechanism to form a complex industrial process model set, realizing the modular management of the system. Specifically, it can refer to the following formula:

[0070] M(Θ) = f M (M 1 (Θ 1 ), M 2 (Θ 2 ), …, M N (Θ L ))

[0071] MML = {M 1 (Θ 1 ), M 2 (Θ 2 ), …, M N (Θ L )}

[0072] Among them, M represents the complex mechanism model in this cascaded industrial background, that is, the industrial mechanism model, f M represents the functional relationship between different meta-models in the mechanism model, where MML represents the meta-model library, that is, the meta-model set, M 1 、M 2, …, M L represent all the indivisible meta - models in complex industrial scenarios, where Θ 1 , Θ 2 , …, Θ L represent all the relevant input parameter sets in the corresponding meta - models respectively, and L represents the number of meta - models.

[0073] Furthermore, in order to guide the large - model to output more accurate prediction results, the thought - chain prompt - word template includes: application background information, expected answer format, object and audience, and few - shot prompt words;

[0074] After the step of constructing the meta - model set based on the meta - models, it further includes:

[0075] Determine the cascade relationship between the meta - models in the meta - model set;

[0076] Construct a thought - chain prompt - word template according to the cascade relationship.

[0077] It should be noted that constructing the thought - chain prompt - word template according to the cascade relationship can be to construct a thought - chain prompt - word template including application background information, expected answer format, object and audience, and few - shot prompt words according to the cascade relationship between the meta - models. Among them, regarding the background information of technical applications, that is, the application background information: due to the high requirements for the accuracy of process correlation in cascade industries, the similarity of industrial link naming, and the specificity of application backgrounds, it is necessary to clarify the industrial scope of the question - and - answer in the prompt - word background, and it is necessary to go from the large - category process to a specific sub - process, and this granularity cannot be further divided. Taking the process of purifying and removing cobalt as an example, if the operator wants to know the cobalt - removing reaction rate in a certain reaction tank during the purification of cobalt, the background information needs to clarify two keywords: "hydrometallurgical zinc smelting" and "purification of cobalt removal". Among them, "hydrometallurgical zinc smelting" belongs to the large - category process of the large - category "zinc smelting", and "purification of cobalt removal" belongs to the smallest sub - process that cannot be further divided. Secondly, it is the specification of the answer expected from the large - model, that is, the expected answer format, which can clarify the information to be included in the answer, such as whether it belongs to professional operation and technical information or popular science information, etc. Then, it is necessary to clarify the object and audience of the large - model question - and - answer, which can be obtained through multiple rounds of question - and - answer with the user when the user inputs a question. Finally, it is the few - shot prompt words, which guide the example of the answer format, that is, the answer logic of the industrial thought - chain is guided in the form of few - shot prompt words, and the answer content is expected to be obtained.

[0078] To visualize how the large - model solves industrial problems under the guidance of the thought - chain, this step defines a transformation equation T (as shown in the following formula) to represent the interconnection relationship between the meta - models under cascade industrial sub - problems, that is, how to solve sub - problems based on the results of previous sub - problems:

[0079] DP(i) = Τ(s, i, {(j, DP(j)): j < i})

[0080] where s represents a continuous series of sub - process index sequences s that occur before the cascade industrial specific process index occurs 1 , s 2 , … s N , and each state i before and after only depends on a finite number of markers s in the input sequence and a finite number n of previous states. Therefore, the above formula can be represented by the following formula:

[0081] DP(i) = f(i, s g(n，i) , DP(h(n, i)))

[0082] = f(i, s g1(n，i) , …, s gj(n，i) , DP(h 1 (n, i)), …, DP(h K (n, i)))

[0083] where DP(i) is used to characterize the dynamic programming algorithm and is a meta - model in the meta - model set. i is used to characterize the state of DP(i). s represents a continuous series of sub - process index sequences that occur before the cascade industrial specific process index occurs. N is used to characterize the number of process index sequences. The functional relationships f, g, h determine the transformation equation T. f: N N × I × X J × y K → y, g: N N × I → (N ∪ φ) J , h: N N × I → (N ∪ φ) K , I is used to characterize the state space, X is used to characterize the input space, DP is used to characterize the output space, y is used to characterize any domain, J represents the number of input tokens, K represents the dimension of the output state space. If the token length determining state i is less than J or the previous states determining state i are less than K, then φ represents a placeholder. s φ and DP(φ) are not used in the function f. Finally, the solution aggregated by DP(i) is output by the large model.

[0084] In specific implementation, reference can be made to Figure 3 , Figure 3This is the overall process schematic diagram provided for the second embodiment of the industrial field knowledge Q&A method of this application. In this embodiment, first, the complex industrial mechanism model and optimization algorithm are disassembled into indivisible meta-models to form a meta-model set of the complex industrial process. Then, by decoupling the relationships between the meta-model modules, a thinking chain prompt word library for a specific process is established. In the third step, according to the cascade relationship between the complex industrial meta-models, an industrial thinking chain prompt word template is formed, which includes four parts: application background information, expected answer format, object and audience, and few-shot prompt words. For the input question, first extract the keywords in the user's question, match the thinking chain prompt word template, and guide the large model to generate a logical answer. Subsequently, according to the keywords output by the large model Q&A, call the small model for further prediction to achieve refined answer processing. This method not only improves the response speed of the system but also can be refined and verified at different levels to ensure the accuracy and professionalism of the answer. In addition, this method can effectively improve the development efficiency of optimal control algorithms for new processes, new optimization problems, etc. This embodiment has significant advantages in modular design, real-time update, refined processing, and clear logic, can meet the needs of complex professional problems in the industrial field, and provide accurate and in-depth professional knowledge services.

[0085] In one embodiment, by extracting the keyword information of the user input question, and splicing according to the extracted keyword information and the thinking chain prompt word template corresponding to the user input question (the corresponding thinking chain prompt word template can be matched according to process position, personnel identity, ion concentration, working condition, etc.), through "thinking chain prompt word + question", guide the large model to generate a logical answer. Specifically, the industrial field knowledge Q&A device first identifies the key concepts and core content in the question raised by the user, and then selects a suitable thinking chain prompt word template based on this content. These templates can include, for example, process, procedure, key indicators, working condition and other prompt words, so as to help construct a systematic and well-organized answer. Then, according to the keywords output by the large model Q&A, call the small model (i.e., the meta-model) for prediction to achieve more refined answer processing.

[0086] This embodiment constructs a thinking chain prompt word library for a specific industry, that is, a thinking chain prompt word template. By disassembling the complex cascaded industrial process into meta-models, different meta-models are used as nodes of the graph data, the association relationships between different nodes are constructed, and the answers obtained through the thinking chain prompt word engineering of the industrial process are used for real-time call of the model. In the actual application process, there are alternative solutions in the following operation details:

[0087] 1. Meta-model disassembly. In the actual deployment and application process, the model can be disassembled at different granularities according to its specific process requirements, rather than setting each disassembled model as a meta-model;

[0088] 2. Model library management. In this embodiment, the model library management adopted is to manage the meta-model set using the Neo4j graph database. There are also solutions that use other databases to construct relationship graphs.

[0089] 3. The prompt template in this embodiment includes four parts: technical background, answer specification, object orientation, and guiding answer format, which can be adjusted according to different industrial scenarios and granularities in actual operations.

[0090] 4. Guided prompt construction. In this embodiment, the prompt construction adopted is the chain-of-thought prompt. In actual operation processes, other prompt construction methods such as the tree-of-thought can be used according to different processes to improve the answer accuracy.

[0091] 5. This embodiment uses the few-shot chain-of-thought process in the prompt, and the number of prompt chains-of-thought can be adjusted according to the effect in actual operations.

[0092] This embodiment understands and analyzes the answers of the large model, extracts key information and keywords, and then calls the corresponding small model for further prediction and processing based on these keywords. This method can not only improve the response speed of the system, but also be refined and verified at different levels to ensure the accuracy and professionalism of the answers. At the same time, the prediction results of the small model can be fed back to the large model to further optimize and adjust the question-and-answer strategy, forming a dynamic and interactive optimization process, thereby continuously improving the overall performance and user experience of the question-and-answer system.

[0093] It should be noted that the above examples are only for understanding this application and do not constitute a limitation on the industrial domain knowledge question-and-answer method of this application. Based on this technical concept, more forms of simple transformations are within the protection scope of this application.

[0094] This application also provides an industrial domain knowledge question-and-answer device. Please refer to Figure 4 ., The industrial domain knowledge question-and-answer device includes:

[0095] An input module 10, configured to extract key information from the question input by the user, and match the extraction result with the chain-of-thought prompt template to obtain prompt information.

[0096] A large model prediction module 20, configured to input the prompt information into the large model for answer prediction to obtain a prediction result.

[0097] A meta-model set prediction module 30, configured to extract key prediction information from the prediction result, and call the meta-model set for prediction based on the key prediction information to obtain the question-and-answer result output by the meta-model set, where the meta-model set includes meta-models obtained by disassembling the industrial mechanism model and the optimization algorithm.

[0098] In this embodiment, key information is extracted from the questions input by users, and the extraction results are matched with the thought chain prompt word templates to obtain prompt information; the prompt information is input into a large model for answer prediction to obtain a prediction result; the key prediction information in the prediction result is extracted, and based on the key prediction information, a meta-model set is called for prediction to obtain the Q&A result output by the meta-model set, where the meta-model set includes meta-models obtained by disassembling industrial mechanism models and optimization algorithms. Since this embodiment first performs answer prediction based on a large model, and then optimizes the prediction result based on the meta-model set obtained by disassembling the industrial mechanism model and the optimization algorithm to obtain the Q&A result, compared with the existing method of simply relying on a large model for Q&A, the above method in this embodiment can meet the requirements of complex professional questions in the industrial field, achieve refined Q&A processing, and ensure the accuracy and professionalism of Q&A.

[0099] The industrial field knowledge Q&A device provided by this application adopts the industrial field knowledge Q&A method in the above embodiment, which can solve the technical problems that existing large models have broad and insufficiently distinguishable answers when facing industrial knowledge Q&A with a high degree of professionalism. Compared with the prior art, the beneficial effects of the industrial field knowledge Q&A device provided by this application are the same as those of the industrial field knowledge Q&A method provided by the above embodiment, and other technical features in the industrial field knowledge Q&A device are the same as those disclosed in the above embodiment method, which will not be elaborated here.

[0100] This application provides an industrial field knowledge Q&A device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein, the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the industrial field knowledge Q&A method in the first embodiment above.

[0101] The following refers to Figure 5 , which shows a schematic structural diagram of an industrial field knowledge Q&A device suitable for implementing the embodiments of this application. The industrial field knowledge Q&A device in the embodiments of this application may include, but is not limited to, mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Descriptions), PMPs (Portable Media Players), in-vehicle terminals (such as in-vehicle navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. Figure 5The industrial field knowledge Q&A device shown is merely an example and should not impose any limitations on the functions and scope of use of the embodiments of this application.

[0102] As Figure 5 shown, the industrial field knowledge Q&A device may include a processing device 1001 (such as a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes according to the program stored in the read-only memory (ROM: Read Only Memory) 1002 or the program loaded from the storage device 1003 into the random access memory (RAM: Random Access Memory) 1004. In the RAM 1004, various programs and data required for the operation of the industrial field knowledge Q&A device are also stored. The processing device 1001, the ROM 1002, and the RAM 1004 are connected to each other through a bus 1005. The input / output (I / O) interface 1006 is also connected to the bus. Generally, the following systems can be connected to the I / O interface 1006: an input device 1007 including, for example, a touch screen, a touchpad, a keyboard, a mouse, an image sensor, a microphone, an accelerometer, a gyroscope, etc.; an output device 1008 including, for example, a liquid crystal display (LCD: Liquid Crystal Display), a speaker, a vibrator, etc.; a storage device 1003 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 1009. The communication device 1009 can allow the industrial field knowledge Q&A device to communicate with other devices wirelessly or wiredly to exchange data. Although the figure shows an industrial field knowledge Q&A device with various systems, it should be understood that it is not required to implement or have all the systems shown. More or fewer systems can be alternatively implemented or had.

[0103] In particular, according to the embodiments disclosed in this application, the process described above with reference to the flowchart can be implemented as a computer software program. For example, the embodiments disclosed in this application include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program contains program codes for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from the network through the communication device, or installed from the storage device 1003, or installed from the ROM 1002. When the computer program is executed by the processing device 1001, the above functions defined in the method of the embodiments disclosed in this application are executed.

[0104] The industrial field knowledge Q&A device provided by this application adopts the industrial field knowledge Q&A method in the above embodiment, and can solve the technical problem that existing large models have broad and insufficiently distinguishable answers when facing industrial knowledge Q&A with a relatively high degree of professionalism. Compared with the prior art, the beneficial effects of the industrial field knowledge Q&A device provided by this application are the same as those of the industrial field knowledge Q&A method provided by the above embodiment, and other technical features in this industrial field knowledge Q&A device are the same as those disclosed in the previous embodiment method, which will not be elaborated here.

[0105] It should be understood that each part disclosed in this application can be implemented by hardware, software, firmware or a combination thereof. In the description of the above embodiments, specific features, structures, materials or characteristics can be combined in a suitable manner in any one or more embodiments or examples.

[0106] The above is only the specific implementation manner of this application, but the protection scope of this application is not limited thereto. Any person skilled in the art within the technical scope disclosed in this application can easily think of changes or substitutions, which should all be covered within the protection scope of this application. Therefore, the protection scope of this application should be subject to the protection scope of the claimed rights.

[0107] This application provides a computer-readable storage medium with computer-readable program instructions (i.e., computer programs) stored thereon, and the computer-readable program instructions are used to execute the industrial field knowledge Q&A method in the above embodiment.

[0108] The computer-readable storage medium provided by this application can be, for example, a USB flash drive, but is not limited to electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or components, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: electrical connections with one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM) or flash memory, optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the above. In this embodiment, the computer-readable storage medium can be any tangible medium that contains or stores a program, and this program can be used by or in combination with an instruction execution system, device, or component. The program code contained on the computer-readable storage medium can be transmitted using any appropriate medium, including but not limited to: wires, optical cables, RF (radio frequency), etc., or any suitable combination of the above.

[0109] The above computer-readable storage medium can be included in an industrial field knowledge Q&A device; it can also exist independently without being assembled into an industrial field knowledge Q&A device.

[0110] The above computer-readable storage medium carries one or more programs. When the above one or more programs are executed by an industrial field knowledge Q&A device, the industrial field knowledge Q&A device is caused to execute the above industrial field knowledge Q&A method.

[0111] Computer program code for performing the operations of this application can be written in one or more programming languages or combinations thereof. The above programming languages include object-oriented programming languages - such as Java, Smalltalk, C++; they also include conventional procedural programming languages - such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, executed as an independent software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer can be connected to the user's computer through any type of network - including a local area network (LAN) or a wide area network (WAN) - or can be connected to an external computer (for example, by using an Internet service provider to connect through the Internet).

[0112] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architectures, functions, and operations of systems, methods, and computer program products according to various embodiments of the present application. In this regard, each block in the flowchart or block diagram may represent a module, a segment of a program, or a part of code that contains one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than that marked in the accompanying drawings. For example, two consecutive blocks shown may actually be executed substantially in parallel, and they may sometimes be executed in the reverse order, depending on the functions involved. It should also be noted that each block in the block diagram and / or flowchart, as well as combinations of blocks in the block diagram and / or flowchart, can be implemented by a dedicated hardware-based system that performs the specified functions or operations, or can be implemented by a combination of dedicated hardware and computer instructions.

[0113] The modules described in the embodiments of the present application can be implemented in software or in hardware. In some cases, the name of the module does not constitute a limitation on the unit itself.

[0114] The readable storage medium provided by the present application is a computer-readable storage medium that stores computer-readable program instructions (i.e., computer programs) for executing the above-mentioned industrial field knowledge answering method, and can solve the technical problem that existing large models have broad and insufficiently distinguishable answers when facing industrial knowledge answering with a high degree of professionalism. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided by the present application are the same as those of the industrial field knowledge answering method provided by the above embodiments, and will not be elaborated here.

[0115] The present application also provides a computer program product, including a computer program, and when the computer program is executed by a processor, it implements the steps of the industrial field knowledge answering method as described above.

[0116] The computer program product provided by the present application can solve the technical problem that existing large models have broad and insufficiently distinguishable answers when facing industrial knowledge answering with a high degree of professionalism. Compared with the prior art, the beneficial effects of the computer program product provided by the present application are the same as those of the industrial field knowledge answering method provided by the above embodiments, and will not be elaborated here.

[0117] The above are only some embodiments of the present application, and thus do not limit the patent scope of the present application. Any equivalent structural transformation made under the technical concept of the present application by using the content of the specification and drawings of the present application, or any direct / indirect application in other related technical fields is included in the patent protection scope of the present application.

Claims

1. An industrial knowledge question answering method, characterized in that: The industrial domain knowledge question answering method comprises the following steps: Extract key information from the question input by the user, and match the extracted result with the thinking chain prompt word template to obtain prompt information; Input the prompt information into the large model to predict the answer and obtain the prediction result; Extract key prediction information from the prediction result, call the metamodel set for prediction based on the key prediction information, and obtain the question and answer result output by the metamodel set, wherein the metamodel set includes the metamodel obtained by disassembling the industrial mechanism model and the optimization algorithm.

2. The industrial domain knowledge question answering method according to claim 1, characterized in that: Before the step of extracting key prediction information from the prediction result, calling a meta-model set to perform prediction based on the key prediction information, and obtaining a question-answering result output by the meta-model set, the step further includes: Obtain industrial mechanism models and industrial optimization algorithms; Decomposing the industrial mechanism model and the industrial optimization algorithm to obtain a meta-model; A metamodel set is constructed based on the metamodel.

3. The industrial domain knowledge question answering method according to claim 2, characterized in that: The step of constructing a metamodel set based on the metamodel comprises: determining a partial order relationship between the meta-models; Constructing a directed acyclic graph based on the partial order relationship and the metamodel; A meta-model set is determined according to the directed acyclic graph.

4. The industrial domain knowledge question answering method according to claim 3, characterized in that: The thought chain prompt word template includes: application background information, expected answer format, object and audience, and few-shot prompt words; After the step of constructing a metamodel set based on the metamodel, the method further includes: Determine the cascade relationship between each metamodel in the metamodel set; A thinking chain prompt word template is constructed according to the cascading relationship.

5. The industrial domain knowledge question answering method according to any one of claims 1 to 4, characterized in that: The step of extracting key prediction information from the prediction result, calling a meta-model set to perform prediction based on the key prediction information, and obtaining a question-answering result output by the meta-model set includes: Extracting key information from the prediction results to obtain key prediction information; Determine a target metamodel in the metamodel set corresponding to the key prediction information; The prediction result is re-predicted based on the target meta-model to obtain a question-answering result.

6. The industrial domain knowledge question answering method according to any one of claims 1 to 4, characterized in that: The interconnection relationship between the metamodels in the metamodel set is expressed by the following transformation equation: DP(i)=f(i,s g(n，i) ,DP(h(n,i))) =f(i,s g1(n，i) ,…,s gj(n，i) ,DP(h1(n,i)),…,DP(h K (n,i))) Among them, DP(i) is used to characterize the dynamic programming algorithm and is a metamodel in the metamodel set. i is used to characterize the state of DP(i). s represents a series of sub-process indicator sequences that occur before the occurrence of the cascade industry specific process indicator. N is used to characterize the number of process indicator sequences. f:N N ×I×X J ×y K →y,g:N N ×I→(N∪φ) J , h:N N ×I→(N∪φ) K , I is used to represent the state space, X is used to represent the input space, DP is used to represent the output space, y is used to represent any domain, J represents the number of input tokens, K represents the dimension of the output state space, if the length of the token that determines state i is less than J or the previous state that determines state i is less than K, then φ represents a placeholder, s φ and DP(φ) are not used in function f.

7. An industrial knowledge question-answering device, characterized in that: The industrial field knowledge question answering device comprises: The input module is used to extract key information from the questions input by the user, and match the extracted results with the thinking chain prompt word template to obtain prompt information; A large model prediction module is used to input the prompt information into the large model to predict the answer and obtain the prediction result; The meta-model set prediction module is used to extract key prediction information from the prediction results, call the meta-model set for prediction based on the key prediction information, and obtain the question-answering results output by the meta-model set, wherein the meta-model set includes the meta-model obtained by disassembling the industrial mechanism model and the optimization algorithm.

8. An industrial knowledge question-answering device, characterized in that: The device comprises: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program is configured to implement the steps of the industrial domain knowledge question answering method according to any one of claims 1 to 6.

9. A storage medium, characterized in that: The storage medium is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, the steps of the industrial field knowledge question and answer method according to any one of claims 1 to 6 are implemented.

10. A computer program product, characterized in that The computer program product comprises a computer program, and when the computer program is executed by a processor, the steps of the industrial domain knowledge question answering method according to any one of claims 1 to 6 are implemented.

Images