Method and system for carrying out artificial intelligence inquiry on pet diseases

Abstract

The invention is suitable for the technical field of inquiry, and particularly relates to a method and system for carrying out artificial intelligence inquiry on pet diseases, and the method comprises the steps: S100, creating a distributed architecture composed of a cloud end and terminal equipment, and deploying a pre-training language model and a knowledge base in the cloud end; s200, pet case data in different formats are collected, the case data at least comprise symptom description, diagnosis results and treatment schemes, the pet case data are preprocessed and segmented into a plurality of single items, semantic extraction is carried out on the single items to obtain key information, and a preset multi-modal fusion model is utilized to carry out multi-modal fusion on the key information; and fusing the single items in different formats, and configuring a corresponding relationship among all the single items. The multi-dimensional knowledge graph is generated, the accuracy and comprehensiveness of diagnosis are greatly improved, the pre-training language model and the knowledge base are integrated, so that the distributed architecture can continuously self-evolve, and the accuracy of diagnosis is gradually improved.

Description

Technical Field

[0001] The present invention relates to the field of medical consultation technology, and in particular to a method and system for performing artificial intelligence medical consultation on pet diseases. Background Art

[0002] As people's living standards improve, the number of pets is increasing, and the demand for pet consultation and treatment is also increasing. Online consultation for pets has gradually become a new development trend. The emergence of intelligent consultation methods has also greatly improved the efficiency of pet consultation.

[0003] Intelligent diagnosis of pet diseases is to assist pet owners in quickly identifying and diagnosing their pets' health problems by using artificial intelligence technology, which can not only reduce waiting time but also provide personalized diagnosis solutions; therefore, "how to use pet case data to conduct intelligent diagnosis of pets" is the technical problem that the present invention needs to solve. Summary of the invention

[0004] The purpose of the present invention is to provide a method and system for artificial intelligence diagnosis of pet diseases, so as to solve the problem of "how to use pet case data to conduct intelligent diagnosis of pets" raised in the above background technology.

[0005] To achieve the above object, the present invention provides the following technical solutions:

[0006] A method for performing artificial intelligence diagnosis of pet diseases, the method comprising:

[0007] S100: creating a distributed architecture consisting of a cloud and terminal devices, and deploying a pre-trained language model and a knowledge base in the cloud;

[0008] S200: collecting pet case data in different formats, wherein the case data at least includes symptom description, diagnosis result and treatment plan, preprocessing the pet case data, dividing it into a plurality of items, and performing semantic extraction on the items to obtain key information, using a preset multimodal fusion model to fuse the items in different formats, configuring the corresponding relationships between all the items, integrating the items and the corresponding relationships, generating a multi-dimensional pet disease knowledge graph, and establishing a link with the knowledge base;

[0009] S300: receiving a consultation request uploaded by a user through the terminal device, establishing a dialogue, inputting the consultation request into a pre-trained language model, outputting key information in the consultation request, defining it as target information, transferring the target information into a multi-dimensional pet disease knowledge graph for similarity matching, and replying to the dialogue using the matching result;

[0010] S400: partitioning the knowledge base to obtain a correct knowledge base, a knowledge base to be verified, and an incorrect knowledge base, extracting knowledge points in the dialogue, and transferring them to the knowledge base to be verified, integrating a multi-level expert review mechanism in the knowledge base to be verified, and setting a trigger frequency;

[0011] S500: If the trigger frequency is met at the current moment, the multi-level expert review mechanism is activated to verify the knowledge base to be verified, and according to the verification result, the knowledge point is synchronized to the correct knowledge base or the wrong knowledge base, and a communication link between the correct knowledge base and the existing medical literature database is established;

[0012] S600: Based on the target information, search the correct knowledge base, and use the retrieved content to correct the dialogue.

[0013] Furthermore, the S100 includes:

[0014] Determine a certain number of virtual servers, build a distributed architecture, and embed a load balancing mechanism;

[0015] In the cloud, a processing environment of a virtual server is configured, and a pre-trained language model and a knowledge base are deployed.

[0016] Further, the S200 includes:

[0017] Preprocessing the pet case data, wherein the preprocessing at least includes: cleaning, word segmentation and entity recognition;

[0018] The key information is used to replace the data in the single item.

[0019] Furthermore, the S200 further includes:

[0020] Classifying the key information into several subcategories, and inserting pointers between the subcategories, wherein the pointers are used to characterize the symptom relationship between the subcategories, and the pointers at least include: triggering pointers, treatment pointers, and concurrent pointers;

[0021] The pointing is synchronized to the single items, and the corresponding relationship between the single items is determined.

[0022] Further, the S300 includes:

[0023] Using a preset word embedding model, similarity matching is performed between the target information and individual items in the multi-dimensional pet disease knowledge graph;

[0024] Sort the matched items in descending order, and define the first ranked item as the matching result;

[0025] Find out the keywords in the matching results, use the keywords to traverse the associated knowledge in the knowledge base, and link the associated knowledge to the matching results.

[0026] Furthermore, the S400 includes:

[0027] Classifying the difficulty of the knowledge points, splitting the multi-level expert review mechanism into several levels, and establishing a mapping relationship between the difficulty classification and the levels;

[0028] The error knowledge base is used to construct a verification strategy, wherein the verification strategy is: determining whether there is data in the dialogue that is the same as that in the error knowledge base, and if so, correcting the dialogue.

[0029] Furthermore, the S600 includes:

[0030] Adding tags to the conversation and establishing a knowledge traceability mechanism;

[0031] Using the attention visualization technology in the prior art, the focus points in the conversation are marked, and the confidence analysis of the conversation is performed to determine the trust value;

[0032] A tag is generated using the trust value, and the tag is inserted into the conversation.

[0033] Furthermore, the system comprises:

[0034] A deployment module, used to create a distributed architecture consisting of a cloud and terminal devices, and deploy a pre-trained language model and a knowledge base in the cloud;

[0035] Establish a module for collecting pet case data in different formats, wherein the case data at least includes symptom description, diagnosis result and treatment plan, pre-process the pet case data, divide it into several items, and perform semantic extraction on the items to obtain key information, use a preset multimodal fusion model to fuse the items in different formats, configure the corresponding relationship between all the items, integrate the items and the corresponding relationship, generate a multi-dimensional pet disease knowledge graph, and establish a link with the knowledge base;

[0036] A matching module is used to receive a consultation request uploaded by a user through the terminal device, establish a dialogue, input the consultation request into a pre-trained language model, output key information in the consultation request, define it as target information, transfer the target information into a multi-dimensional pet disease knowledge graph for similarity matching, and reply to the dialogue using the matching result;

[0037] A setting module is used to partition the knowledge base to obtain a correct knowledge base, a knowledge base to be verified and an incorrect knowledge base, extract knowledge points in the dialogue and transfer them to the knowledge base to be verified, integrate a multi-level expert review mechanism in the knowledge base to be verified, and set a trigger frequency;

[0038] A building module is used to activate the multi-level expert review mechanism if the trigger frequency is met at the current moment, verify the knowledge base to be verified, and synchronize the knowledge points to the correct knowledge base or the wrong knowledge base according to the verification result, and build a communication link between the correct knowledge base and the existing medical literature database;

[0039] The correction module is used to search in the correct knowledge base according to the target information and correct the dialogue using the retrieved content.

[0040] Furthermore, the deployment module includes:

[0041] The embedding unit is used to determine a certain number of virtual servers, build a distributed architecture, and embed a load balancing mechanism;

[0042] The configuration unit is used to configure a processing environment of a virtual server in the cloud and deploy a pre-trained language model and a knowledge base.

[0043] Furthermore, the establishment module includes:

[0044] A preprocessing unit, used for preprocessing the pet case data, wherein the preprocessing at least includes: cleaning, word segmentation and entity recognition;

[0045] A replacement unit, used to replace the data in the single item using the key information;

[0046] An inserting unit, used to classify the key information into several subcategories and insert pointers between the subcategories, wherein the pointers are used to characterize the symptom relationship between the subcategories, and the pointers at least include: triggering pointers, treatment pointers and concurrent pointers;

[0047] The synchronization unit is used to synchronize the pointing to the single item and determine the corresponding relationship between the single items.

[0048] Compared with the prior art, the present invention has the following beneficial effects:

[0049] By utilizing the preset multimodal fusion model, pet case data in different formats can be fused, greatly expanding the data source. By building a pre-trained language model, the language comprehension and reasoning capabilities of pet case data can be greatly improved. By generating a multi-dimensional knowledge graph, the accuracy and comprehensiveness of the diagnosis are greatly improved. By integrating the pre-trained language model with the knowledge base, the distributed architecture can continue to evolve and gradually improve the accuracy of diagnosis. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 A flowchart of a method for performing artificial intelligence diagnosis of pet diseases provided in an embodiment of the present invention;

[0051] Figure 2 A first sub-flow chart of the method for performing artificial intelligence diagnosis of pet diseases provided by an embodiment of the present invention;

[0052] Figure 3 A second sub-flow chart of the method for performing artificial intelligence diagnosis of pet diseases provided by an embodiment of the present invention;

[0053] Figure 4 A third sub-flow chart of the method for performing artificial intelligence diagnosis of pet diseases provided by an embodiment of the present invention;

[0054] Figure 5 A fourth sub-flow chart of the method for performing artificial intelligence diagnosis of pet diseases provided by an embodiment of the present invention;

[0055] Figure 6 A fifth sub-flow chart of the method for performing artificial intelligence diagnosis of pet diseases provided by an embodiment of the present invention;

[0056] Figure 7 A block diagram of the composition of a system for performing artificial intelligence diagnosis of pet diseases provided by an embodiment of the present invention;

[0057] Figure 8 A block diagram of the components of the deployment modules in the system for performing artificial intelligence diagnosis of pet diseases provided by an embodiment of the present invention;

[0058] Fig. 9 A block diagram of the components of the establishment module in the system for performing artificial intelligence diagnosis of pet diseases provided by an embodiment of the present invention;

[0059] Fig.10 A block diagram of the composition of a matching module in a system for performing artificial intelligence diagnosis of pet diseases provided by an embodiment of the present invention;

[0060] Fig.11 A block diagram of the components of the modules in the system for performing artificial intelligence diagnosis of pet diseases provided by an embodiment of the present invention;

[0061] Fig.12 A block diagram of the composition of the correction module in the system for performing artificial intelligence diagnosis of pet diseases provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0062] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0063] In Example 1, Figure 1 The implementation process of the method for performing artificial intelligence diagnosis of pet diseases provided by an embodiment of the present invention is shown and described in detail below:

[0064] S100: Create a distributed architecture consisting of a cloud and terminal devices, and deploy a pre-trained language model and a knowledge base in the cloud.

[0065] A distributed architecture is built using cloud servers and user terminal devices, and pre-trained language models and knowledge bases are deployed in the cloud servers. The pre-trained language models are mainly used for semantic extraction of pet case data, and the knowledge base is mainly used to store knowledge points in pet case data.

[0066] S200: Collect pet case data in different formats, wherein the case data at least includes symptom description, diagnosis result and treatment plan, pre-process the pet case data, divide it into several items, and perform semantic extraction on the items to obtain key information, use a preset multimodal fusion model to fuse the items in different formats, configure the corresponding relationships between all the items, integrate the items and the corresponding relationships, generate a multi-dimensional pet disease knowledge graph, and establish a link with the knowledge base.

[0067] Collect pet case data in different formats, where the specific formats can be pictures, texts, tables, etc. The pet case data at least includes: symptom description, diagnosis results, treatment plan and other data, and then pre-process the pet case data; divide the pet case data into several items, and the specific segmentation method can be: divide the pet case data into symptom description items, diagnosis results items and treatment plan items, etc.

[0068] Semantic extraction is performed on each single item to obtain the key information of each single item, and the multimodal fusion model in the prior art is used to fuse the key information with the same content but different formats, wherein the multimodal fusion model is a model that can process and fuse data from multiple modalities (such as text, image, audio and video, etc.), and merge the corresponding single items at the same time, configure the correspondence between the fused single items, and use the single items and the corresponding relationship to generate a multidimensional pet disease knowledge graph, and establish a link between the multidimensional pet disease knowledge graph and the knowledge base, wherein the benefit of building a link is to expand the knowledge graph; wherein the knowledge graph is a structure that organizes knowledge based on a graph, and in this embodiment, the knowledge graph is used to intuitively display the relationship between single items.

[0069] S300: Receive a consultation request uploaded by the user through the terminal device and establish a dialogue, input the consultation request into the pre-trained language model, output the key information in the consultation request, define it as target information, transfer the target information into the multi-dimensional pet disease knowledge graph for similarity matching, and use the matching result to reply to the dialogue.

[0070] Receive the consultation request uploaded by the user through the terminal device, where the terminal device can be a mobile terminal or a fixed terminal of the user, establish a dialogue with the user, input the consultation request uploaded by the user into the pre-trained model, perform semantic understanding of the consultation request, output the key information therein, and define this key information as target information, transfer the target information into the multi-dimensional pet disease knowledge graph for similarity matching, and use the matching results to reply to the dialogue.

[0071] S400: partition the knowledge base to obtain a correct knowledge base, a knowledge base to be verified and an incorrect knowledge base, extract the knowledge points in the conversation and transfer them to the knowledge base to be verified, integrate a multi-level expert review mechanism in the knowledge base to be verified, and set a trigger frequency.

[0072] The knowledge base is divided into a correct knowledge base, a knowledge base to be verified and an incorrect knowledge base, wherein the correct knowledge base is used to store correct knowledge points, and the incorrect knowledge base is used to store incorrect knowledge points. The function of the incorrect knowledge points is to verify the matching results and find out the incorrect parts. In the dialogue with the user, the knowledge points are extracted and transferred to the knowledge base to be verified, and the transferred knowledge points are verified using a multi-level expert review mechanism. For the existing knowledge points in the knowledge base to be verified, it is necessary to verify them after the trigger frequency is met. The multi-level expert review mechanism is composed of expert review mechanisms at multiple difficulty levels, and different levels are responsible for reviewing knowledge points of different difficulty levels.

[0073] S500: If the trigger frequency is met at the current moment, the multi-level expert review mechanism is activated to verify the knowledge base to be verified, and based on the verification result, the knowledge starting point is synchronized to the correct knowledge base or the incorrect knowledge base, and a communication link is established between the correct knowledge base and the existing medical literature database.

[0074] If the trigger frequency is met at the current moment, the multi-level expert review mechanism is activated to verify the knowledge base to be verified. If the verification result is correct, the corresponding knowledge point will be transferred to the correct knowledge base. If the verification result is incorrect, the corresponding knowledge point will be transferred to the incorrect knowledge base. After the correct knowledge base and the incorrect knowledge base are constructed, a communication link between the correct knowledge base and the existing medical literature database is established. That is to say, in the process of replying to users, not only the multi-dimensional pet disease knowledge graph and the correct knowledge base will be used, but also the existing medical literature database, where the existing medical literature database is directly obtained from public documents.

[0075] S600: Based on the target information, search the correct knowledge base, and use the retrieved content to correct the dialogue.

[0076] After replying to the conversation, the target information is transferred to the correct knowledge base for retrieval, and the conversation is corrected based on the retrieval results.

[0077] In Example 2, Figure 2 The implementation process of the method for performing artificial intelligence diagnosis of pet diseases provided by an embodiment of the present invention is shown, and S100 is described in detail below, as follows:

[0078] S101: Determine a certain number of virtual servers, build a distributed architecture, and embed a load balancing mechanism.

[0079] Determine the virtual server that has been created, where the virtual server is a node virtualized in the cloud for data processing, and use the virtual server to form a distributed architecture, while embedding a load balancing mechanism, where the load balancing mechanism is a specific method of coordinating and balancing the load.

[0080] S102: In the cloud, a processing environment of a virtual server is configured. The specific configuration process should include: operating system preparation, hardware resource allocation and network configuration, etc., and deployment of a pre-trained language model and knowledge base.

[0081] Configure the processing environment of the virtual server and deploy the pre-trained language model and knowledge base in the virtual server.

[0082] In Example 3, Figure 3The implementation process of the method for performing artificial intelligence diagnosis of pet diseases provided by an embodiment of the present invention is shown, and S200 is described in detail below, as follows:

[0083] S201: Preprocess the pet case data, wherein the preprocessing at least includes: cleaning, word segmentation and entity recognition.

[0084] The pet case data is preprocessed by first cleaning it and then performing word segmentation. Word segmentation is to divide the questions raised by users into independent words. Finally, specific entities are identified from them, including names, places, and organizations.

[0085] S202: Using the key information, replace the data in the single item.

[0086] Replace pet case data in single item with key information.

[0087] In Example 4, Figure 3 The implementation process of the method for performing artificial intelligence diagnosis of pet diseases provided by an embodiment of the present invention is shown, and S200 is further described in detail as follows:

[0088] S203: Classify the key information into several subcategories, and insert pointers between the subcategories, wherein the pointers are used to characterize the symptom relationship between the subcategories, and the pointers at least include: triggering pointers, treatment pointers, and concurrent pointers.

[0089] Divide the key information into several subcategories, which are also categories. It can be simply understood as dividing the key information into multiple subcategories according to different symptoms and treatment methods, and inserting pointers between different subcategories. In real life, a certain symptom of pet disease (such as vomiting) may correspond to multiple diagnostic results (such as gastroenteritis, food allergies), and also to multiple treatment methods (feeding antiemetics, changing food). For each symptom, diagnostic result and treatment method, a set is created, namely a subcategory. The advantage of doing this is that it can intuitively show the connection between pet diseases and better provide decision support for pet disease diagnosis.

[0090] S204: Synchronize the pointing to the single item and determine the corresponding relationship between the single items.

[0091] Synchronize the pointing to the single item and determine the corresponding relationship between the single items, where the corresponding relationship is the relationship between vomiting-gastroenteritis-feeding antiemetic drugs mentioned above.

[0092] In Example 5, Figure 4 The implementation process of the method for performing artificial intelligence diagnosis of pet diseases provided by an embodiment of the present invention is shown, and S300 is described in detail below, as follows:

[0093] S301: Using a preset word embedding model, similarity matching is performed between the target information and individual items in the multi-dimensional pet disease knowledge graph.

[0094] The preset word embedding model is used to map the single item in the target information into the vector space, and the cosine similarity formula in the prior art is used to calculate the similarity between the target information and the single item.

[0095] S302: Sort the matched items in descending order, and define the first ranked item as the matching result.

[0096] The matched items are sorted in descending order of similarity, and the item ranked first is defined as the matching result.

[0097] S303: Find out the keywords in the matching result, use the keywords to traverse the associated knowledge in the knowledge base, and link the associated knowledge to the matching result.

[0098] Find the keywords in the matching results, traverse the associated knowledge related to the keywords in the correct knowledge base, and collect this associated knowledge into the matching results.

[0099] In Example 6, Figure 5 The implementation process of the method for performing artificial intelligence diagnosis of pet diseases provided by an embodiment of the present invention is shown, and S400 is described in detail below, as follows:

[0100] S401: Classify the difficulty of the knowledge points, split the multi-level expert review mechanism into several levels, and establish a mapping relationship between difficulty classification and levels.

[0101] After acquiring new knowledge points in the conversation, the knowledge points are graded in difficulty, and the multi-level expert review mechanism is split into several levels. Each level processes knowledge points of different difficulty levels, and the processed knowledge points are classified into the correct knowledge base or the incorrect knowledge base.

[0102] S402: constructing a verification strategy using the error knowledge base, wherein the verification strategy is: determining whether there is data in the dialogue that is identical to data in the error knowledge base, and if so, correcting the dialogue.

[0103] Verify the dialogue using the verification strategy to determine whether there is data in the dialogue that is identical to the data in the error knowledge base. If so, it proves that the dialogue is wrong and needs to be corrected.

[0104] In Example 7, Figure 6The implementation process of the method for performing artificial intelligence diagnosis of pet diseases provided by an embodiment of the present invention is shown, and S600 is described in detail below, as follows:

[0105] S601: Add tags to the conversation and establish a knowledge traceability mechanism.

[0106] When using matching results to reply to conversations, tags are added to the matching results and a knowledge traceability mechanism is established, where the knowledge traceability mechanism is also the traceability method for the matching results.

[0107] S602: Using the attention visualization technology in the prior art, focus points in the conversation are marked, and confidence analysis is performed on the conversation to determine a trust value.

[0108] During the conversation, focus points are marked and confidence analysis is performed to quantify the confidence and obtain a trust value. The higher the trust value, the higher the matching similarity, which means the more accurate the consultation result.

[0109] S603: Generate a tag using the trust value, and insert the tag into the conversation.

[0110] Generate tags using trust values, insert tags into conversations, and update trust values ​​based on the conversations.

[0111] Figure 7 The structure block diagram of the system for performing artificial intelligence diagnosis of pet diseases provided by an embodiment of the present invention is shown. The system 1 for performing artificial intelligence diagnosis of pet diseases includes:

[0112] A deployment module 11, used to create a distributed architecture consisting of a cloud and terminal devices, and deploy a pre-trained language model and a knowledge base in the cloud;

[0113] Establishing module 12, for collecting pet case data in different formats, wherein the case data at least includes symptom description, diagnosis result and treatment plan, preprocessing the pet case data, dividing it into a plurality of items, and performing semantic extraction on the items to obtain key information, using a preset multimodal fusion model to fuse the items in different formats, configuring the corresponding relationship between all the items, integrating the items and the corresponding relationship, generating a multidimensional pet disease knowledge graph, and establishing a link with the knowledge base;

[0114] The matching module 13 is used to receive the consultation request uploaded by the user through the terminal device, establish a dialogue, input the consultation request into the pre-trained language model, output the key information in the consultation request, define it as target information, transfer the target information into the multi-dimensional pet disease knowledge graph for similarity matching, and reply to the dialogue using the matching result;

[0115] A setting module 14 is used to partition the knowledge base to obtain a correct knowledge base, a knowledge base to be verified and an incorrect knowledge base, extract knowledge points in the dialogue, and transfer them to the knowledge base to be verified, integrate a multi-level expert review mechanism in the knowledge base to be verified, and set a trigger frequency;

[0116] A building module 15 is used to activate the multi-level expert review mechanism if the trigger frequency is met at the current moment, verify the knowledge base to be verified, and synchronize the knowledge point to the correct knowledge base or the wrong knowledge base according to the verification result, and build a communication link between the correct knowledge base and the existing medical literature database;

[0117] The correction module 16 is used to search the correct knowledge base according to the target information and correct the dialogue using the retrieved content.

[0118] Figure 8 The structure diagram of the system for performing artificial intelligence diagnosis of pet diseases provided by an embodiment of the present invention is shown, and the deployment module 11 includes:

[0119] The embedding unit 111 is used to determine a certain number of virtual servers, and to form a distributed architecture and embed a load balancing mechanism;

[0120] The configuration unit 112 is used to configure a processing environment of a virtual server in the cloud and deploy a pre-trained language model and a knowledge base.

[0121] Fig. 9 The structure diagram of the system for performing artificial intelligence diagnosis of pet diseases provided by an embodiment of the present invention is shown, and the establishment module 12 includes:

[0122] A preprocessing unit 121 is used to preprocess the pet case data, wherein the preprocessing at least includes: cleaning, word segmentation and entity recognition;

[0123] A replacement unit 122, used to replace the data in the single item using the key information;

[0124] The inserting unit 123 is used to classify the key information into a plurality of subcategories and insert pointers between the subcategories, wherein the pointers are used to characterize the symptom relationship between the subcategories, and the pointers at least include: triggering pointers, treatment pointers and concurrent pointers;

[0125] The synchronization unit 124 is used to synchronize the pointing to the single item and determine the corresponding relationship between the single items.

[0126] Fig.10The structure diagram of the system for performing artificial intelligence diagnosis of pet diseases provided by an embodiment of the present invention is shown, and the matching module 13 includes:

[0127] A pairing unit 131 is used to perform similarity matching between the target information and a single item in the multi-dimensional pet disease knowledge graph using a preset word embedding model;

[0128] A sorting unit 132, used to sort the matched items in descending order, and define the first ranked item as a matching result;

[0129] The linking unit 133 is used to find out the keywords in the matching results, use the keywords to traverse the associated knowledge in the knowledge base, and link the associated knowledge to the matching results.

[0130] Fig.11 The structure diagram of the system for performing artificial intelligence diagnosis of pet diseases provided by an embodiment of the present invention is shown, and the setting module 14 includes:

[0131] The grading unit 141 is used to grade the difficulty of the knowledge points, split the multi-level expert review mechanism into several levels, and establish a mapping relationship between the difficulty grades and the levels;

[0132] The verification unit 142 is used to construct a verification strategy using the error knowledge base, wherein the verification strategy is: determining whether there is data in the dialogue that is the same as that in the error knowledge base, and if so, correcting the dialogue.

[0133] Fig.12 The structural block diagram of the system for performing artificial intelligence diagnosis of pet diseases provided by an embodiment of the present invention is shown, and the correction module 16 includes:

[0134] A tracing unit 161, used to add tags to the dialogue and establish a knowledge tracing mechanism;

[0135] A marking unit 162, used to mark the focus points in the conversation by using the attention visualization technology in the prior art, and perform confidence analysis on the conversation to determine the trust value;

[0136] The label unit 163 is used to generate a label using the trust value, and insert the label into the conversation.

[0137] The deployment module 11 is mainly used to complete step S100, the establishment module 12 is mainly used to complete step S200, the matching module 13 is mainly used to complete step S300, the setting module 14 is mainly used to complete step S400, the construction module 15 is mainly used to complete step S500, and the correction module 16 is mainly used to complete step S600;

[0138] The embedding unit 111 is mainly used to complete step S101, and the configuration unit 112 is mainly used to complete step S102;

[0139] The preprocessing unit 121 is mainly used to complete step S201, the replacing unit 122 is mainly used to complete step S202, the inserting unit 123 is mainly used to complete step S203, and the synchronization unit 124 is mainly used to complete step S204;

[0140] The pairing unit 131 is mainly used to complete step S301, the sorting unit 132 is mainly used to complete step S302, and the linking unit 133 is mainly used to complete step S303;

[0141] The classification unit 141 is mainly used to complete step S401, and the verification unit 142 is mainly used to complete step S402;

[0142] The tracing unit 161 is mainly used to complete step S601, the marking unit 162 is mainly used to complete step S602, and the label unit 163 is mainly used to complete step S603.

[0143] The technical features of the above-described embodiments may be arbitrarily combined. To make the description concise, not all possible combinations of the technical features in the above-described embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0144] The above-mentioned embodiments only express several implementation methods of the present invention, and the description thereof is relatively specific and detailed, but it cannot be understood as limiting the scope of the patent of the present invention. It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the concept of the present invention, which all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be subject to the attached claims.

[0145] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for artificial intelligence diagnosis of pet diseases, characterized in that: The method comprises: S100: creating a distributed architecture consisting of a cloud and terminal devices, and deploying a pre-trained language model and a knowledge base in the cloud; S200: collecting pet case data in different formats, wherein the case data at least includes symptom description, diagnosis result and treatment plan, preprocessing the pet case data, dividing it into a plurality of items, and performing semantic extraction on the items to obtain key information, using a preset multimodal fusion model to fuse the items in different formats, configuring the corresponding relationships between all the items, integrating the items and the corresponding relationships, generating a multi-dimensional pet disease knowledge graph, and establishing a link with the knowledge base; S300: receiving a consultation request uploaded by a user through the terminal device, establishing a dialogue, inputting the consultation request into a pre-trained language model, outputting key information in the consultation request, defining it as target information, transferring the target information into a multi-dimensional pet disease knowledge graph for similarity matching, and replying to the dialogue using the matching result; S400: partitioning the knowledge base to obtain a correct knowledge base, a knowledge base to be verified, and an incorrect knowledge base, extracting knowledge points in the dialogue, and transferring them to the knowledge base to be verified, integrating a multi-level expert review mechanism in the knowledge base to be verified, and setting a trigger frequency; S500: If the trigger frequency is met at the current moment, the multi-level expert review mechanism is activated to verify the knowledge base to be verified, and according to the verification result, the knowledge point is synchronized to the correct knowledge base or the wrong knowledge base, and a communication link between the correct knowledge base and the existing medical literature database is established; S600: Based on the target information, search the correct knowledge base, and use the retrieved content to correct the dialogue.

2. The method for performing artificial intelligence diagnosis of pet diseases according to claim 1, characterized in that: The S100 includes: Determine a certain number of virtual servers, build a distributed architecture, and embed a load balancing mechanism; In the cloud, a processing environment of a virtual server is configured, and a pre-trained language model and a knowledge base are deployed.

3. The method for performing artificial intelligence diagnosis of pet diseases according to claim 1, characterized in that: The S200 includes: Preprocessing the pet case data, wherein the preprocessing at least includes: cleaning, word segmentation and entity recognition; The key information is used to replace the data in the single item.

4. The method for conducting artificial intelligence diagnosis of pet diseases according to claim 3, characterized in that: The S200 further includes: Classifying the key information into several subcategories, and inserting pointers between the subcategories, wherein the pointers are used to characterize the symptom relationship between the subcategories, and the pointers at least include: triggering pointers, treatment pointers, and concurrent pointers; The pointing is synchronized to the single items, and the corresponding relationship between the single items is determined.

5. The method for conducting artificial intelligence diagnosis of pet diseases according to claim 4, characterized in that: The S300 includes: Using a preset word embedding model, similarity matching is performed between the target information and individual items in the multi-dimensional pet disease knowledge graph; Sort the matched items in descending order, and define the first ranked item as the matching result; Find out the keywords in the matching results, use the keywords to traverse the associated knowledge in the knowledge base, and link the associated knowledge to the matching results.

6. The method for artificial intelligence diagnosis of pet diseases according to claim 1, characterized in that: The S400 includes: Classifying the difficulty of the knowledge points, splitting the multi-level expert review mechanism into several levels, and establishing a mapping relationship between the difficulty classification and the levels; The error knowledge base is used to construct a verification strategy, wherein the verification strategy is: determining whether there is data in the dialogue that is the same as that in the error knowledge base, and if so, correcting the dialogue.

7. The method for artificial intelligence diagnosis of pet diseases according to claim 1, characterized in that: The S600 includes: Adding tags to the conversation and establishing a knowledge traceability mechanism; Using the attention visualization technology in the prior art, the focus points in the conversation are marked, and the confidence analysis of the conversation is performed to determine the trust value; A tag is generated using the trust value, and the tag is inserted into the conversation.

8. A system for artificial intelligence diagnosis of pet diseases, characterized in that: The system comprises: A deployment module, used to create a distributed architecture consisting of a cloud and terminal devices, and deploy a pre-trained language model and a knowledge base in the cloud; Establish a module for collecting pet case data in different formats, wherein the case data at least includes symptom description, diagnosis result and treatment plan, pre-process the pet case data, divide it into several items, and perform semantic extraction on the items to obtain key information, use a preset multimodal fusion model to fuse the items in different formats, configure the corresponding relationship between all the items, integrate the items and the corresponding relationship, generate a multi-dimensional pet disease knowledge graph, and establish a link with the knowledge base; A matching module is used to receive a consultation request uploaded by a user through the terminal device, establish a dialogue, input the consultation request into a pre-trained language model, output key information in the consultation request, define it as target information, transfer the target information into a multi-dimensional pet disease knowledge graph for similarity matching, and reply to the dialogue using the matching result; A setting module is used to partition the knowledge base to obtain a correct knowledge base, a knowledge base to be verified and an incorrect knowledge base, extract knowledge points in the dialogue and transfer them to the knowledge base to be verified, integrate a multi-level expert review mechanism in the knowledge base to be verified, and set a trigger frequency; A building module is used to activate the multi-level expert review mechanism if the trigger frequency is met at the current moment, verify the knowledge base to be verified, and synchronize the knowledge points to the correct knowledge base or the wrong knowledge base according to the verification result, and build a communication link between the correct knowledge base and the existing medical literature database; The correction module is used to search in the correct knowledge base according to the target information and correct the dialogue using the retrieved content.

9. The system for performing artificial intelligence diagnosis of pet diseases according to claim 8, wherein the deployment module comprises: The embedding unit is used to determine a certain number of virtual servers, build a distributed architecture, and embed a load balancing mechanism; The configuration unit is used to configure a processing environment of a virtual server in the cloud and deploy a pre-trained language model and a knowledge base.

10. The system for performing artificial intelligence diagnosis of pet diseases according to claim 8, wherein the establishment module comprises: A preprocessing unit, used for preprocessing the pet case data, wherein the preprocessing at least includes: cleaning, word segmentation and entity recognition; A replacement unit, used to replace the data in the single item using the key information; An inserting unit, used to classify the key information into several subcategories and insert pointers between the subcategories, wherein the pointers are used to characterize the symptom relationship between the subcategories, and the pointers at least include: triggering pointers, treatment pointers and concurrent pointers; The synchronization unit is used to synchronize the pointing to the single item and determine the corresponding relationship between the single items.

Images