Virtual expert creation method and device, equipment and medium

Abstract

The invention provides a virtual expert creation method and device, equipment and a medium, and relates to the technical field of artificial intelligence, and the method specifically comprises the steps: receiving an expert knowledge base uploaded by an artificial expert, and carrying out the preprocessing of the expert knowledge base; performing text splitting and extraction on the preprocessed expert knowledge base on the basis of a large language model, and constructing an initial knowledge graph on the basis of extracted knowledge points and semantic relationships; receiving a modification operation instruction of the artificial expert on the initial knowledge graph through a man-machine interaction interface of a large language model; and in response to the operation instruction, dynamically adjusting and optimizing the initial knowledge graph, and generating a virtual expert with a professional domain knowledge graph. The method is used for replacing a traditional rule-based virtual expert system, and is more intelligent and efficient.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to a method, device, equipment and medium for creating a virtual expert. Background Art

[0002] Currently, the construction of virtual expert systems has been widely implemented in many fields. However, before the integration of large language model technology, the current technology generally faces several challenges and limitations in the implementation process. The following is the current status of virtual expert system construction:

[0003] Traditional rule-based virtual expert systems mainly rely on pre-set rules and logic to parse and make decisions based on inputs through a specific rule base. Although this approach is highly efficient in dealing with structured and clear problems, the formulation and maintenance of rules are particularly complex and difficult in the face of complex, ambiguous and dynamically changing environments. In addition, current systems generally lack efficient interaction mechanisms, which limits the system's full use of the intuition and experience of artificial experts. It is difficult for experts to easily input new knowledge or ideas into the system, which restricts the system's flexibility and self-optimization capabilities.

[0004] Therefore, there is an urgent need to provide a new virtual expert system construction solution to improve the above problems. Summary of the invention

[0005] The purpose of the present invention is to provide a virtual expert creation method, device, medium and equipment to replace the traditional rule-based virtual expert system, which is more intelligent and efficient.

[0006] In a first aspect, an embodiment of the present invention provides a method for creating a virtual expert, comprising: receiving an expert knowledge base uploaded by an artificial expert, and preprocessing the expert knowledge base; performing text segmentation and extraction on the preprocessed expert knowledge base based on a large language model, and constructing an initial knowledge graph based on the extracted knowledge points and semantic relationships; receiving modification operation instructions of the artificial expert on the initial knowledge graph through a human-computer interaction interface of the large language model; and dynamically adjusting and optimizing the initial knowledge graph in response to the operation instructions to generate a virtual expert with a professional field knowledge graph.

[0007] In a possible implementation, the preprocessed knowledge base is subjected to text segmentation and extraction based on a large language model, and an initial knowledge graph is constructed based on the extracted knowledge points and semantic relationships, including: gradually segmenting the chapters, paragraphs, and terms in the knowledge base based on the large language model; performing deep semantic analysis on the segmented text to extract key knowledge points; and creating a preliminary knowledge graph based on the extracted knowledge points and the relationships between the knowledge points.

[0008] In a possible implementation, after generating a virtual expert with a professional domain knowledge graph, it also includes: obtaining a natural language query instruction input by a user; performing intent recognition on the natural language query instruction, and based on the intent recognition result, querying from the database corresponding to the virtual expert to generate a query result.

[0009] In a possible implementation, preprocessing the expert knowledge base includes: performing text formatting and denoising on the documents in the expert knowledge base.

[0010] In a possible implementation, it also includes: receiving the modification operation of the artificial expert on the knowledge graph in the virtual expert through the human-computer interaction interface of the large language model; in response to the modification operation, adding a new information unit to the knowledge graph of the virtual expert, and linking the newly added information unit with the knowledge points in the existing knowledge graph.

[0011] In a possible implementation, after generating a virtual expert with a professional field knowledge graph, it also includes: generating a virtual character of the artificial expert based on the portrait information authorized by the artificial expert; and displaying the facial expressions, body movements and language output content of the virtual character in a display interface.

[0012] In a possible implementation, the expert knowledge base includes professional materials in at least two subject areas.

[0013] In a second aspect, an embodiment of the present invention further provides a virtual expert creation device, comprising: a receiving unit, configured to receive an expert knowledge base uploaded by a human expert;

[0014] A preprocessing unit, used for preprocessing the expert knowledge base;

[0015] A processing unit is used to perform text segmentation and extraction on the preprocessed expert knowledge base based on a large language model, and to construct an initial knowledge graph based on the extracted knowledge points and semantic relationships;

[0016] The receiving unit is further used to receive the modification operation instruction of the artificial expert on the initial knowledge graph through the human-computer interaction interface of the large language model;

[0017] The processing unit is also used to dynamically adjust and optimize the initial knowledge graph in response to the operation instruction, and generate a virtual expert with a professional field knowledge graph.

[0018] In a third aspect, an embodiment of the present invention further provides an electronic device, comprising a memory and a processor, wherein the memory stores a program that can be run on the processor, and when the program is executed by the processor, the electronic device implements a method in any possible embodiment of the first aspect.

[0019] In a fourth aspect, an embodiment of the present invention further provides a computer-readable storage medium, wherein the computer-readable storage medium includes a program, and when the program runs on an electronic device, the electronic device executes a method in any possible implementation of the first aspect.

[0020] In a fifth aspect, an embodiment of the present invention further provides a computer program product, which, when executed on an electronic device, enables the electronic device to execute a method of any possible implementation of the first aspect.

[0021] The present invention provides a virtual expert creation method, device, medium and equipment. The beneficial effects of the technical solution are mainly reflected in the following aspects:

[0022] 1. Intelligence and efficiency: By introducing a large language model to deeply process the expert knowledge base, this technical solution can automatically build and optimize the knowledge graph, thereby creating a virtual expert with professional domain knowledge. This not only greatly reduces the need for manual intervention and improves processing efficiency, but also significantly improves the intelligence level of the virtual expert.

[0023] 2. Human-computer interaction and flexibility: This technical solution provides a human-computer interaction interface, allowing human experts to dynamically modify and optimize the generated initial knowledge graph. This design not only increases the flexibility of the system, enabling virtual experts to better meet actual needs, but also further improves the accuracy and reliability of virtual experts through the participation of human experts.

[0024] 3. Multi-domain applicability: This technical solution supports processing expert knowledge bases containing at least two subject areas, which means that it can be applied to multiple professional fields to create virtual experts with different expertise. This cross-domain capability makes this technical solution widely applicable and of practical application value.

[0025] 4. User-friendliness: This technical solution supports natural language queries, can identify user intent and generate corresponding query results. This makes the interaction between users and virtual experts more natural and convenient, improving the user experience. In addition, by generating a virtual character of an artificial expert and displaying its facial expressions, body movements and language output content in the display interface, the user's interactive experience with the virtual expert is further enhanced. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0027] Figure 1 A schematic diagram of a virtual expert creation method provided by an embodiment of the present invention;

[0028] Figure 2 A schematic diagram of another medical virtual expert creation method provided by an embodiment of the present invention;

[0029] Figure 3 A schematic diagram of the structure of a virtual expert creation device provided by an embodiment of the present invention;

[0030] Figure 4 A schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0031] The embodiments of the present invention relate to artificial intelligence (AI) and machine learning (ML) technologies, and implement solutions based on deep learning networks and machine learning in artificial intelligence.

[0032] With the research and advancement of artificial intelligence technology, artificial intelligence has been studied and applied in many fields, such as common smart homes, smart customer service, virtual assistants, smart speakers, smart marketing, unmanned driving, autonomous driving, robots, smart medical care, etc. I believe that with the development of technology, artificial intelligence will be applied in more fields and play an increasingly important role.

[0033] Machine learning is a multi-disciplinary subject that involves probability theory, statistics, approximation theory, convex analysis, algorithm complexity theory and other disciplines. It specializes in studying how computers simulate or implement human learning behavior to acquire new knowledge or skills and reorganize existing knowledge structures to continuously improve their performance. Machine learning is the core of artificial intelligence and the fundamental way to make computers intelligent. Its applications are spread across all areas of artificial intelligence. Machine learning and deep learning usually include artificial neural networks, belief networks, reinforcement learning, transfer learning, inductive learning and other technologies.

[0034] In the description of the embodiments of the present invention, the terms used in the following embodiments are only for the purpose of describing specific embodiments, and are not intended to be used as limitations of the present invention. As used in the specification of the present invention and the appended claims, the singular expressions "a", "said", "above", "the" and "this" are intended to also include expressions such as "one or more", unless there is a clear contrary indication in the context. It should also be understood that in the following embodiments of the present invention, "at least one", "one or more" refer to one or more (including two). The term "and / or" is used to describe the association relationship of associated objects, indicating that three relationships may exist; for example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone, where A and B can be singular or plural. The character " / " generally indicates that the objects associated before and after are in an "or" relationship.

[0035] References to "one embodiment" or "some embodiments" etc. described in this specification mean that one or more embodiments of the present invention include specific features, structures or characteristics described in conjunction with the embodiment. Therefore, the statements "in one embodiment", "in some embodiments", "in some other embodiments", "in some other embodiments", etc. that appear in different places in this specification do not necessarily refer to the same embodiment, but mean "one or more but not all embodiments", unless otherwise specifically emphasized in other ways. The terms "including", "comprising", "having" and their variations all mean "including but not limited to", unless otherwise specifically emphasized in other ways. The term "connection" includes direct connection and indirect connection, unless otherwise specified. "First" and "second" are used for descriptive purposes only and cannot be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated.

[0036] In the embodiments of the present invention, the words "exemplarily" or "for example" are used to indicate examples, illustrations or explanations. Any embodiment or implementation scheme described as "exemplarily" or "for example" in the embodiments of the present invention should not be interpreted as being more preferred or more advantageous than other embodiments or implementation schemes. Specifically, the use of words such as "exemplarily" or "for example" is intended to present related concepts in a specific way.

[0037] The embodiment of the present invention provides a method for creating a virtual expert. Figure 1 As shown, the process of the method can be executed by an electronic device, and the method includes the following steps:

[0038] S101, receiving an expert knowledge base uploaded by a human expert, and preprocessing the expert knowledge base;

[0039] S102, performing text segmentation and extraction on the preprocessed expert knowledge base based on the large language model, and constructing an initial knowledge graph based on the extracted knowledge points and semantic relationships;

[0040] S103, receiving the modification operation instruction of the artificial expert on the initial knowledge graph through the human-computer interaction interface of the large language model;

[0041] S104, in response to the operation instruction, dynamically adjust and optimize the initial knowledge graph to generate a virtual expert with a professional field knowledge graph.

[0042] In addition, the expert knowledge base is preprocessed in the above S101, specifically including: text formatting and denoising the documents in the expert knowledge base. Preprocessing is an important prerequisite for text analysis, which involves a series of steps such as text cleaning, word segmentation, and part-of-speech tagging. The expert knowledge base usually includes professional materials in the field of artificial experts themselves, such as academic papers, technical reports, white papers, etc. This embodiment can remove noise information in the text, such as punctuation marks, special characters, etc., by preprocessing the knowledge base, making the text data purer and convenient for subsequent analysis.

[0043] In another possible embodiment, the above S101 performs text segmentation and extraction on the preprocessed knowledge base based on the large language model, and constructs an initial knowledge graph based on the extracted knowledge points and semantic relationships, including: gradually splitting each chapter, paragraph and term in the knowledge base based on the large language model; performing deep semantic analysis on the split text to extract key knowledge points; and creating a preliminary knowledge graph based on the extracted knowledge points and the relationship between the knowledge points. In other words, through the text segmentation function of the large language model, the knowledge base can be split into independent sentences or phrases, providing a basis for subsequent semantic analysis. Semantic relationships are not just simple "include" or "included", they may involve multiple types such as cause and effect, parallel, and transition. By accurately identifying these semantic relationships, the logical connection between each knowledge point in the text can be more accurately understood, thereby constructing a more complete and accurate knowledge graph. At the same time, the pre-analysis function of the large language model can identify keywords and phrases in the text, extract key information in the text, and provide data support for the subsequent knowledge graph construction. The large language model of this embodiment can process information at different levels, from basic concepts to advanced theories, covering the breadth and depth of professional materials, and fully reflecting the knowledge structure in the expert field.

[0044] After extracting key information, the initial knowledge graph is constructed based on the extracted knowledge points and semantic relationships. The knowledge graph is a graph structure data model used to represent the relationship between entities. In the process of building a knowledge graph, it is necessary to first define entities and relationships, and then use the extracted knowledge points and semantic relationships to associate entities and relationships to form a graph structure. Through the construction of the knowledge graph, we can more clearly understand the content in the knowledge base, discover the laws and trends therein, and provide strong support for subsequent research and application. For example, when processing an ancient historical document, the historical figures, events, places and other entities in the document, as well as the relationships between them, can be extracted through a large language model. Then, these entities and relationships can be associated to form a knowledge graph containing rich information. Through this knowledge graph, we can have a deeper understanding of the historical events and relationships between people in the document, discover the laws and trends therein, and provide new perspectives and methods for historical research.

[0045] In the above S103 of this embodiment, the artificial expert can also view, edit and annotate the initial knowledge graph through the human-computer interaction interface of the large language model. The artificial expert can participate in the construction and correction process of the knowledge graph in real time to ensure the accuracy and authority of the knowledge graph. Moreover, the expert's annotations and modifications can provide timely feedback and adjust the model to improve the reliability of the final knowledge graph. In addition, the artificial expert can also modify the knowledge graph in the virtual expert through the human-computer interaction interface of the large language model; for example, the artificial expert adds a new information unit to the original knowledge graph and links it with the existing knowledge points to form a more complete and detailed knowledge network.

[0046] In a possible embodiment, after generating a virtual expert with a professional domain knowledge graph, the process further includes: obtaining a natural language query instruction input by a user; performing intent recognition on the natural language query instruction, and based on the intent recognition result, querying from a database corresponding to the virtual expert to generate a query result. In addition, a virtual character of the artificial expert can be generated based on the portrait information authorized by the artificial expert; and the facial expressions, body movements, and language output content of the virtual character can be displayed in a display interface, which can enhance the interactive fun.

[0047] After generating the query results, the system will further analyze user feedback. Users can express their satisfaction or dissatisfaction with the query results through simple gestures, language or other interactive methods. The system will record these feedbacks and use machine learning algorithms to learn and optimize, so as to provide more accurate and user-friendly answers in future queries.

[0048] At the same time, in order to further enhance the user experience, the system also introduces an intelligent recommendation function. Based on the user's historical query records, feedback data, and the context information of the current query, the system will intelligently recommend relevant professional field knowledge, case studies, or the latest developments to the user. These recommendations can not only help users gain a deeper understanding of a certain professional field, but also stimulate their desire to explore and promote the dissemination and sharing of knowledge.

[0049] In addition to basic query and recommendation functions, the system also has advanced learning and communication capabilities. Users can have in-depth discussions and exchanges with virtual experts through natural language, and raise their own opinions and questions. Virtual experts will give targeted answers and suggestions based on their rich professional knowledge and learned user habits. This interactive learning and communication method can not only help users quickly acquire knowledge, but also enhance the connection and trust between users and the system.

[0050] In addition, the system also supports multi-user online communication at the same time. Users can communicate and cooperate with peers or experts from all over the world by creating or joining different discussion groups. This cross-regional and cross-domain communication method provides users with a broader learning and communication platform, which helps promote knowledge sharing and innovation. In short, by generating virtual experts with professional knowledge graphs and combining advanced technologies such as natural language processing and machine learning, it provides users with an efficient, convenient and intelligent learning and communication platform. This platform can not only help users quickly acquire knowledge and solve problems, but also stimulate users' desire for exploration and innovative spirit, and promote the dissemination and development of knowledge.

[0051] Combined with the following Figure 2 The flowchart shown introduces in detail the method for constructing a virtual expert for medical diagnosis, including the specific implementation of each step and the technical support behind it.

[0052] First of all, the uploading and preprocessing of professional materials are the cornerstones of building the system. Medical experts upload a large number of medical research papers, clinical trial reports, case analysis, professional medical instruction manuals and other medical documents to the system. These documents are a treasure trove of medical knowledge, but they have different formats and complex contents, which require unified processing by the system. The system uses advanced text processing technology to remove noise in the documents, such as advertisements and irrelevant information, to ensure the purity of the text content. At the same time, the system will also format and structure the documents to make them easier for subsequent text analysis and knowledge extraction.

[0053] Next, the extraction of medical knowledge and graph construction based on a large language model is the core of the system. The system uses advanced natural language processing technology and deep learning models to split and extract text from pre-processed medical documents. Through the training of deep learning models, the system can accurately identify medical knowledge points in documents, such as disease symptoms, diagnostic methods, treatment plans, etc. Then, the system uses deep semantic parsing technology to extract these key medical information and construct a preliminary medical knowledge graph. This graph displays medical knowledge points in the form of a graph, which is convenient for medical experts to view and understand.

[0054] In the interactive optimization stage of the medical knowledge graph, the system provides an intuitive and easy-to-use interface for medical experts through interactive tools based on large models. Medical experts can view the preliminary medical knowledge graph through this interface, edit and annotate it, and then the system determines whether it is necessary to reuse the large language model for analysis. The system receives and processes expert feedback in real time and dynamically adjusts the knowledge graph based on the expert's opinions. This interactive optimization method not only improves the accuracy and completeness of the knowledge graph, but also enables medical experts to participate in the construction process of the knowledge graph, enhancing their sense of participation and belonging.

[0055] Finally, the construction and deployment of the virtual expert system is a reflection of the value of the system. After multiple iterations of optimization, the medical knowledge graph has formed a dynamic and timely updated diagnosis and treatment knowledge system. This knowledge system not only contains a wealth of medical knowledge, but also can be updated in real time according to the latest medical research and practical experience. Based on this knowledge system, the system can be built into a virtual expert system capable of intelligent feedback and reasoning. This system can simulate the way of thinking of medical experts and provide intelligent assistance and decision support for the doctor's diagnosis and treatment process. Doctors can obtain more accurate diagnostic suggestions and more reasonable treatment plans through interaction with the virtual expert system. At the same time, this system can also be deployed in the internal system of medical institutions or as a cloud service for use by more doctors and researchers.

[0056] In summary, the construction method of medical diagnosis virtual expert system is a complex and delicate process. It requires close cooperation between medical experts, technicians and computer systems to achieve the aggregation, extraction and intelligent application of medical knowledge. The emergence of this system not only improves medical efficiency and accuracy, but also injects new vitality into the development of the medical field.

[0057] In some embodiments of the present invention, the present invention discloses a virtual expert creation device, such as Figure 3 As shown, the device is used to implement the methods recorded in the above method embodiments, and includes: a receiving unit 301, a pre-processing unit 302 and a processing unit 303.

[0058] A receiving unit 301 is used to receive an expert knowledge base uploaded by a human expert;

[0059] A preprocessing unit 302, used for preprocessing the expert knowledge base;

[0060] The processing unit 303 is used to perform text segmentation and extraction on the preprocessed expert knowledge base based on the large language model, and construct an initial knowledge graph based on the extracted knowledge points and semantic relationships;

[0061] The receiving unit 301 is further used to receive the modification operation instruction of the artificial expert on the initial knowledge graph through the human-computer interaction interface of the large language model;

[0062] The processing unit 303 is also used to dynamically adjust and optimize the initial knowledge graph in response to the operation instruction, and generate a virtual expert with a professional field knowledge graph.

[0063] All relevant contents of each step involved in the above method embodiment can be referred to the functional description of the corresponding functional module, and will not be repeated here.

[0064] The above device may be executed by a chip or a chip module. The various modules / units included in the various devices and products described in the above embodiments may be software modules / units or hardware modules / units, or may be partially software modules / units and partially hardware modules / units.

[0065] Next, an electronic device provided by an embodiment of the present application is introduced. Figure 4 , Figure 4 This is a schematic diagram of a structure of an electronic device provided in an embodiment of the present application. The electronic device 400 can be specifically a mobile phone, a tablet, a laptop computer, a smart wearable device, etc., which is not limited here. Figure 1 Functions of the corresponding method embodiments. Specifically, the electronic device 400 includes: a receiver 401, a transmitter 402, a processor 403 and a memory 404 (wherein the number of processors 403 in the electronic device 400 can be one or more), wherein the processor 403 may include an application processor 4031 and a communication processor 4032. In some embodiments of the present application, the receiver 401, the transmitter 402, the processor 403 and the memory 404 may be connected via a bus or other means.

[0066] The memory 404 may include a read-only memory and a random access memory, and provides instructions and data to the processor 403. A portion of the memory 404 may also include a non-volatile random access memory (NVRAM). The memory 404 stores processor and operation instructions, executable modules or data structures, or subsets thereof, or extended sets thereof, wherein the operation instructions may include various operation instructions for implementing various operations.

[0067] The processor 403 controls the operation of the electronic device. In a specific application, the various components of the electronic device are coupled together through a bus system, wherein the bus system includes not only a data bus but also a power bus, a control bus, and a status signal bus, etc. However, for the sake of clarity, various buses are referred to as bus systems in the figure.

[0068] The method disclosed in the above embodiment of the present application can be applied to the processor 403, or implemented by the processor 403. The processor 403 can be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method can be completed by an integrated logic circuit of hardware or software instructions in the processor 403. The above processor 403 can be a general-purpose processor, a digital signal processor (digital signal processing, DSP), a microprocessor or a microcontroller, and a vision processor (vision processing unit, VPU), a tensor processor (tensorprocessing unit, TPU) and other processors suitable for AI computing, and can further include an application specific integrated circuit (application specific integrated circuit, ASIC), a field programmable gate array (field-programmable gate array, FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components. The processor 403 can implement or execute the disclosed methods, steps and logic block diagrams in the embodiments of the present application. The general processor can be a microprocessor or the processor can also be any conventional processor, etc. The steps of the method disclosed in the embodiment of the present application can be directly embodied as being executed by a hardware decoding processor, or being executed by a combination of hardware and software modules in a decoding processor. The software module can be located in a mature storage medium in the art such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory, or an electrically erasable programmable memory, a register, etc. The storage medium is located in the memory 404, and the processor 403 reads the information in the memory 404, and completes the steps in the above method embodiment in combination with its hardware.

[0069] The receiver 401 can be used to receive input digital or character information and generate signal input related to the relevant settings and function control of the electronic device. The transmitter 402 can be used to output digital or character information through the first interface; the transmitter 402 can also be used to send instructions to the disk group through the first interface to modify the data in the disk group; the transmitter 402 can also include a display device such as a display screen.

[0070] The embodiment of the present application also provides a computer program product, which, when executed on a computer, enables the computer to execute the steps executed by the aforementioned device, or enables the computer to execute the steps executed by the aforementioned device.

[0071] A computer-readable storage medium is also provided in an embodiment of the present application. The computer-readable storage medium stores a program for performing signal processing. When the program is run on a computer, the computer executes the steps executed by the aforementioned device, or the computer executes the steps executed by the aforementioned device.

[0072] The execution device, training device or electronic device provided in the embodiments of the present application may specifically be a chip, and the chip includes: a processing unit and a communication unit, wherein the processing unit may be, for example, a processor, and the communication unit may be, for example, an input / output interface, a pin or a circuit, etc. The processing unit may execute the computer execution instructions stored in the storage unit so that the chip in the execution device executes the data processing method described in the above embodiment, or so that the chip in the training device executes the data processing method described in the above embodiment. Optionally, the storage unit is a storage unit in the chip, such as a register, a cache, etc. The storage unit may also be a storage unit located outside the chip in the wireless access device, such as a read-only memory (ROM) or other types of static storage devices that can store static information and instructions, a random access memory (RAM), etc.

[0073] For details, please refer to Figure 4 , Figure 4 A schematic diagram of the structure of a chip provided in an embodiment of the present application, the chip can be expressed as a neural network processor NPU400, NPU 400 is mounted on the host CPU (Host CPU) as a coprocessor, and the host CPU assigns tasks. The core part of the NPU is the operation circuit 403, which is controlled by the controller 404 to extract matrix data from the memory and perform multiplication operations.

[0074] NPU 400 can achieve Figure 4The described embodiments provide a method for processing conversation data.

[0075] More specifically, in some implementations, the operation circuit 403 in the NPU 400 includes a plurality of processing units (Process Engine, PE) therein. In some implementations, the operation circuit 403 is a two-dimensional systolic array. The operation circuit 403 can also be a one-dimensional systolic array or other electronic circuits capable of performing mathematical operations such as multiplication and addition. In some implementations, the operation circuit 403 is a general-purpose matrix processor.

[0076] For example, assume there is an input matrix A, a weight matrix B, and an output matrix C. The operation circuit takes the corresponding data of matrix B from the weight memory 402 and caches it on each PE in the operation circuit. The operation circuit takes the matrix A data from the input memory 401 and performs matrix operation with matrix B, and the partial result or final result of the matrix is ​​stored in the accumulator 408.

[0077] The unified memory 406 is used to store input data and output data. The weight data is directly transferred to the weight memory 402 through the direct memory access controller (DMAC) 405. The input data is also transferred to the unified memory 406 through the DMAC.

[0078] BIU stands for Bus Interface Unit, i.e., bus interface unit 410 , which is used for interaction between AXI bus, DMAC and instruction fetch buffer (IFB) 409 .

[0079] The bus interface unit 410 (BIU) is used for the instruction fetch memory 409 to obtain instructions from the external memory, and is also used for the storage unit access controller 404 to obtain the original data of the input matrix A or the weight matrix B from the external memory.

[0080] DMAC is mainly used to transfer input data in the external memory DDR to the unified memory 406 or to transfer weight data to the weight memory 402 or to transfer input data to the input memory 401.

[0081] The vector calculation unit 407 includes multiple operation processing units, and further processes the output of the operation circuit 403 when necessary, such as vector multiplication, vector addition, exponential operation, logarithmic operation, size comparison, etc. It is mainly used for non-convolutional / fully connected layer network calculations in neural networks, such as Batch Normalization, pixel-level summation, upsampling of feature planes, etc.

[0082] In some implementations, the vector calculation unit 407 can store the processed output vector to the unified memory 406. For example, the vector calculation unit 407 can apply a linear function; or a nonlinear function to the output of the operation circuit 403, such as linear interpolation of the feature plane extracted by the convolution layer, and then, for example, a vector of accumulated values ​​to generate an activation value. In some implementations, the vector calculation unit 407 generates a normalized value, a pixel-level summed value, or both. In some implementations, the processed output vector can be used as an activation input to the operation circuit 403, for example, for use in a subsequent layer in a neural network.

[0083] An instruction fetch buffer 409 connected to the controller 404 is used to store instructions used by the controller 404;

[0084] Unified memory 406, input memory 401, weight memory 402 and instruction fetch memory 409 are all On-Chip memories. External memories are private to the NPU hardware architecture.

[0085] The processor mentioned in any of the above places may be a general-purpose central processing unit, a microprocessor, an ASIC, or one or more integrated circuits for controlling the execution of the above programs.

[0086] It should also be noted that the device embodiments described above are merely schematic, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed over multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. In addition, in the drawings of the device embodiments provided by the present application, the connection relationship between the modules indicates that there is a communication connection between them, which may be specifically implemented as one or more communication buses or signal lines.

[0087] Through the description of the above implementation mode, the technicians in the field can clearly understand that the present application can be implemented by means of software plus necessary general hardware, and of course, it can also be implemented by special hardware including special integrated circuits, special CPUs, special memories, special components, etc. In general, all functions completed by computer programs can be easily implemented by corresponding hardware, and the specific hardware structure used to implement the same function can also be various, such as analog circuits, digital circuits or special circuits. However, for the present application, software program implementation is a better implementation mode in more cases. Based on such an understanding, the technical solution of the present application is essentially or the part that contributes to the prior art can be embodied in the form of a software product, which is stored in a readable storage medium, such as a computer floppy disk, a U disk, a mobile hard disk, a ROM, a RAM, a disk or an optical disk, etc., including a number of instructions to enable a computer device (which can be a personal computer, a training device, or a network device, etc.) to execute the methods described in each embodiment of the present application.

[0088] In the above embodiments, all or part of the embodiments may be implemented by software, hardware, firmware or any combination thereof. When implemented by software, all or part of the embodiments may be implemented in the form of a computer program product.

[0089] The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions may be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions may be transmitted from a website site, a computer, a training device, or a data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) mode to another website site, computer, training device, or data center. The computer-readable storage medium may be any available medium that a computer can store or a data storage device such as a training device, a data center, etc. that includes one or more available media integrations. The available medium may be a magnetic medium, (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium (e.g., a solid-state drive (SSD)), etc.

Claims

1. A method for creating a virtual expert, characterized in that: include: Receiving an expert knowledge base uploaded by a human expert and preprocessing the expert knowledge base; Based on the large language model, the preprocessed expert knowledge base is split and extracted, and the initial knowledge graph is constructed based on the extracted knowledge points and semantic relationships; Receiving modification operation instructions of the artificial expert on the initial knowledge graph through the human-computer interaction interface of the large language model; In response to the operation instruction, the initial knowledge graph is dynamically adjusted and optimized to generate a virtual expert with a professional field knowledge graph.

2. The method according to claim 1, characterized in that Based on the large language model, the preprocessed knowledge base is split and extracted, and the initial knowledge graph is constructed based on the extracted knowledge points and semantic relationships, including: Based on the large language model, each chapter, paragraph and term in the knowledge base is gradually split; Perform deep semantic analysis on the split text to extract key knowledge points; Create a preliminary knowledge graph based on the extracted knowledge points and the relationships between them.

3. The method according to claim 1, characterized in that After generating a virtual expert with a professional domain knowledge graph, it also includes: Obtaining natural language query instructions input by the user; The intention of the natural language query instruction is recognized, and based on the intention recognition result, a query is performed from the database corresponding to the virtual expert to generate a query result.

4. The method according to claim 1, characterized in that: Preprocessing the expert knowledge base includes: The documents in the expert knowledge base are subjected to text formatting and denoising processing.

5. The method according to claim 1, characterized in that Also includes: Receiving the modification operation of the artificial expert on the knowledge graph in the virtual expert through the human-computer interaction interface of the large language model; In response to the modification operation, a new information unit is added to the knowledge graph of the virtual expert, and the new information unit is linked to the knowledge points in the existing knowledge graph.

6. The method according to claim 1, characterized in that After generating a virtual expert with a professional domain knowledge graph, it also includes: generating a virtual character of the artificial expert according to the portrait information authorized by the artificial expert; The facial expressions, body movements and language output content of the virtual character are displayed in the display interface.

7. The method according to claim 1, characterized in that The expert knowledge base includes professional materials in at least two subject areas.

8. A virtual expert creation device, characterized in that: include: A receiving unit, used for receiving an expert knowledge base uploaded by a human expert; A preprocessing unit, used for preprocessing the expert knowledge base; A processing unit is used to perform text segmentation and extraction on the preprocessed expert knowledge base based on a large language model, and to construct an initial knowledge graph based on the extracted knowledge points and semantic relationships; The receiving unit is further used to receive the modification operation instruction of the artificial expert on the initial knowledge graph through the human-computer interaction interface of the large language model; The processing unit is also used to dynamically adjust and optimize the initial knowledge graph in response to the operation instruction, and generate a virtual expert with a professional field knowledge graph.

9. An electronic device, characterized in that: include: At least one memory, at least one processor, the at least one processor is used to execute instructions stored in the at least one memory, so that the electronic device performs the method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, wherein a program is stored in the computer-readable storage medium, characterized in that: When the program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images