Livestock and poultry manure treatment technology question and answer system based on large model and use method thereof

Abstract

The invention discloses a large model-based livestock and poultry manure treatment technology question-answering system and a use method thereof, and relates to the technical field of environmental protection and governance. In the system, firstly, a trained word segmentation model is used for performing word segmentation processing on a retrieval question input by a user; then retrieval is conducted in a text database and a vector database according to a word segmentation result, the retrieved text blocks are aggregated and deduplicated to obtain a plurality of retrieval text blocks highly related to a retrieval problem, and then large-model cue words are generated in combination with a cue word template; and finally, outputting a reply result according to the trained livestock and poultry manure treatment technology intelligent question and answer model. According to the method, the defects existing in a traditional general field large language model are overcome, the large language model can fully and accurately understand the retrieval problem of a user, a scientific and detailed reply result can be given, intelligent question answering of livestock and poultry excrement treatment knowledge is achieved, and the method has important significance in promoting development of an environment protection and treatment technology.

Description

Technical Field

[0001] The present application relates to the field of environmental protection and governance technology, and in particular to a large-scale model-based livestock and poultry manure treatment technology question-and-answer system and a method of using the system. Background Art

[0002] In the field of modern agricultural production, the rapid development of livestock and poultry farming has become a key factor in promoting agricultural economic growth. However, with the continuous expansion of the scale of farming, the amount of livestock and poultry manure has also increased sharply, which has brought great challenges to the environment. If livestock and poultry manure is not handled properly, it may cause water pollution, soil quality deterioration and air quality decline, which will have a negative impact on the ecological environment and human health. Traditional treatment methods, such as composting, anaerobic digestion and fertilizer, although they can alleviate environmental pressure to a certain extent, still have shortcomings in treatment efficiency, economic cost and adaptability. Therefore, it is particularly urgent to research and develop efficient livestock and poultry manure treatment and resource utilization technology. This research and development process often requires a long cycle and requires repeated experimental adjustments, which is not only time-consuming, but also requires extensive collection of literature, knowledge collation, comparison and summary when looking for the best experimental conditions, in order to form a feasible solution. In addition, the existing technology lacks intelligent management means and cannot monitor and adjust the treatment process in real time, which limits the efficiency of livestock and poultry manure treatment and resource utilization to a certain extent.

[0003] With the continuous advancement of artificial intelligence and big data model technology, a new perspective has been brought to the research on livestock and poultry manure treatment technology. Solutions based on artificial intelligence and big data models can intelligently retrieve and optimize livestock and poultry manure treatment and resource utilization technologies in a data-driven way. However, due to the existence of a certain illusion effect in general-purpose large language models and the timeliness of training data, these general-purpose large language models may not have comprehensive knowledge in livestock and poultry manure treatment and resource utilization. Therefore, improving the defects of traditional general-purpose large language models and developing an intelligent question-answering system are of great significance to promoting the development of environmental protection and governance technologies. Summary of the invention

[0004] The purpose of this application is to provide a large-scale model-based livestock and poultry manure treatment technology question-answering system and its use method, which can fully and accurately understand the search questions entered by users and give scientific and detailed answer results.

[0005] To achieve the above objectives, this application provides the following solutions: In the first aspect, the present application provides a large-scale model-based livestock and poultry manure treatment technology question-answering system, including: The automatic word segmentation module for search questions is used to obtain the search questions input by the user, and use the trained word segmentation model to segment the search questions to obtain keyword text and semantic query vectors.

[0006] The text block dual matching module is used to search in the text database and the vector database based on the keyword text and the semantic query vector, respectively, to determine a number of text blocks based on keyword matching and a number of text blocks based on semantic matching; the text database stores a number of text blocks and the TF-IDF codes corresponding to each text block, and the vector database stores a number of text blocks and the vectors corresponding to each text block.

[0007] The retrieval text block determination module is used to aggregate and remove duplicates of a number of text blocks based on keyword matching and a number of text blocks based on semantic matching to obtain a number of retrieval text blocks that are highly relevant to the retrieval question.

[0008] The large model intelligent reply generation module is used to generate large model prompts based on several search text blocks and prompt templates, and use the trained livestock and poultry manure treatment technology smart question and answer model to output the large model intelligent reply results according to the large model prompts; the livestock and poultry manure treatment technology smart question and answer model is a large language model obtained after reinforcement learning training based on knowledge in the field of livestock and poultry manure treatment technology.

[0009] Optionally, the large-model-based livestock and poultry manure treatment technology question and answer system also includes: a multi-source data acquisition and processing module, which is used to acquire livestock and poultry manure treatment related data from different knowledge retrieval platforms, obtain multi-source livestock and poultry manure treatment related data, and process the acquired multi-source livestock and poultry manure treatment related data to obtain a word segmentation training data set and a large language model training corpus; the processing includes data cleaning, data segmentation, data structuring and data review; livestock and poultry manure treatment related data include literature data, business data and experimental data related to livestock and poultry manure treatment technology.

[0010] Optionally, the livestock and poultry manure treatment technology question and answer system based on a large model also includes: a word segmentation model training module, which is used to train the initial word segmentation model using a word segmentation training data set obtained by processing multi-source livestock and poultry manure treatment related data to obtain a trained word segmentation model.

[0011] Optionally, the livestock and poultry manure treatment technology question and answer system based on a large model also includes: a large language model intensive training module, which is used to train the general domain large language model using the large language model training corpus obtained by processing multi-source livestock and poultry manure treatment related data, and obtain a trained livestock and poultry manure treatment technology intelligent question and answer model.

[0012] Optionally, the large language model reinforcement training module specifically includes: an incremental pre-training unit, a model fine-tuning unit and a reinforcement learning unit.

[0013] The incremental pre-training unit is used to pre-train the general-domain large language model using the large language model training corpus to obtain a pre-trained large language model; the pre-trained large language model initially possesses knowledge and language comprehension capabilities in the field of livestock and poultry manure treatment technology.

[0014] The model fine-tuning unit is used to fine-tune the parameters of the pre-trained large language model using high-quality large language model training corpus to obtain the large language model with fine-tuned parameters; the large language model with fine-tuned parameters can adapt to the specific language and knowledge requirements in the field of livestock and poultry manure treatment technology; the high-quality large language model training corpus is a large language model training corpus in the form of question and answer.

[0015] The reinforcement learning unit is used to perform reinforcement learning training on the large language model after fine-tuning the parameters through manual feedback and reward mechanism, continuously optimize the performance and output results of the large language model until the preset conditions are met, and obtain a trained livestock and poultry manure treatment technology intelligent question and answer model.

[0016] Optionally, the large-model-based livestock and poultry manure treatment technology question and answer system also includes: a database update module, which is used to regularly call the multi-source data acquisition and processing module to obtain livestock and poultry manure treatment related data from different knowledge retrieval platforms and update the text database and vector database after processing.

[0017] Optionally, the livestock and poultry manure treatment technology question and answer system based on the big model also includes: a response result tracing module, which is used to output the big model intelligent response results according to the big model prompts for the trained livestock and poultry manure treatment technology smart question and answer model, and indicate the original source documents of each technical point.

[0018] In a second aspect, the present application provides a method for using the large-model-based livestock and poultry manure treatment technology question-and-answer system as described above, comprising the following steps: Get the search question input by the user, use the trained word segmentation model to segment the search question, and obtain the keyword text and semantic query vector.

[0019] Based on the keyword text and the semantic query vector, searches are performed in the text database and the vector database respectively to determine a number of text blocks based on keyword matching and a number of text blocks based on semantic matching; the text database stores a number of text blocks and the TF-IDF codes corresponding to each text block, and the vector database stores a number of text blocks and the vectors corresponding to each text block.

[0020] Several text blocks based on keyword matching and several text blocks based on semantic matching are aggregated and deduplicated to obtain several retrieval text blocks that are highly relevant to the retrieval question.

[0021] A large model prompt is generated based on several search text blocks and prompt templates, and the trained livestock and poultry manure treatment technology smart question and answer model is used to output the large model intelligent reply result according to the large model prompt; the livestock and poultry manure treatment technology smart question and answer model is a large language model obtained after reinforcement learning training based on knowledge in the field of livestock and poultry manure treatment technology.

[0022] Optionally, before generating a large model prompt based on a plurality of search text blocks and prompt templates, and using the trained livestock and poultry manure treatment technology intelligent question-answering model to output a large model intelligent answer result according to the large model prompt, the method of use further includes the following steps: The data related to livestock and poultry manure treatment are obtained from different knowledge retrieval platforms, and multi-source data related to livestock and poultry manure treatment are obtained. The obtained multi-source data related to livestock and poultry manure treatment are processed to obtain word segmentation training data sets and large language model training corpus.

[0023] The large language model training corpus obtained by processing multi-source livestock and poultry manure treatment related data is used to train the general field large language model to obtain a trained livestock and poultry manure treatment technology intelligent question and answer model.

[0024] Optionally, a large language model training corpus obtained by processing multi-source livestock and poultry manure treatment related data is used to train a general domain large language model to obtain a trained livestock and poultry manure treatment technology intelligent question-answering model, which specifically includes the following steps: The large language model training corpus is used to pre-train the general domain large language model to obtain a pre-trained large language model; the pre-trained large language model initially possesses knowledge and language comprehension capabilities in the field of livestock and poultry manure treatment technology.

[0025] The parameters of the pre-trained large language model are fine-tuned using high-quality large language model training corpus to obtain a large language model with fine-tuned parameters; the large language model with fine-tuned parameters can adapt to the specific language and knowledge requirements in the field of livestock and poultry manure treatment technology; the high-quality large language model training corpus is a large language model training corpus in the form of question and answer.

[0026] Through manual feedback and reward mechanisms, the large language model with fine-tuned parameters is subjected to reinforcement learning training, and the performance and output results of the large language model are continuously optimized until the preset conditions are met, thus obtaining a trained intelligent question-and-answer model for livestock and poultry manure treatment technology.

[0027] According to the specific embodiments provided in this application, this application discloses the following technical effects: The present application provides a large-model-based livestock and poultry manure treatment technology question-and-answer system and its use method. In the system, the search question input by the user is first obtained, and the search question is segmented using a trained word segmentation model to obtain keyword text and semantic query vectors. Subsequently, a search is performed in a text database and a vector database based on the two, and several text blocks based on keyword matching and semantic matching are determined respectively. Subsequently, these text blocks are aggregated and deduplicated to obtain several search text blocks that are highly relevant to the search question, and a large-model prompt is generated in combination with a prompt template. Finally, the trained livestock and poultry manure treatment technology intelligent question-and-answer model is used to output a large-model intelligent reply result based on the large-model prompt. The above-mentioned scheme of the present application adopts a large language model obtained through reinforcement learning training based on knowledge in the field of livestock and poultry manure treatment technology, and constructs large model prompts based on text blocks obtained by matching keywords and semantics, so that the large language model can fully and accurately understand the search questions input by users, and can give scientific, detailed and industry-specific response results, which improves the defects of traditional general-purpose large language models, realizes intelligent question and answer about livestock and poultry manure treatment knowledge, saves researchers time and manpower consumption in searching and reading large amounts of literature, and is of great significance for promoting the development of environmental protection and governance technologies. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0029] Figure 1 A schematic diagram of the functional modules of a large-model-based livestock and poultry manure treatment technology question-and-answer system provided in one embodiment of the present application.

[0030] Figure 2 A schematic diagram of the initial human-computer interaction interface of a large-model-based livestock and poultry manure treatment technology question-and-answer system provided in one embodiment of the present application.

[0031] Figure 3 A schematic diagram of the human-computer interaction interface of a large-model-based livestock and poultry manure treatment technology question-and-answer system after outputting a large-model intelligent reply result provided in one embodiment of the present application.

[0032] Figure 4 A flowchart of a method for using a large-model-based livestock and poultry manure treatment technology question and answer system provided in one embodiment of the present application.

[0033] Figure 5A detailed flow chart of step B2 in a method for using a large-model-based livestock and poultry manure treatment technology question-and-answer system provided in one embodiment of the present application.

[0034] Figure 6 A schematic diagram of the structure of a computer device provided in one embodiment of the present application. DETAILED DESCRIPTION

[0035] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0036] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below with reference to the accompanying drawings and specific implementation methods.

[0037] In an exemplary embodiment, Figure 1 The functional module diagram shown in the figure provides a large-scale model-based livestock and poultry manure treatment technology question-answering system, including the following functional modules: The automatic word segmentation module M1 for search questions is used to obtain the search questions input by the user, and use the trained word segmentation model to perform word segmentation on the search questions to obtain keyword text and semantic query vectors.

[0038] The text block dual matching module M2 is used to search in the text database and the vector database based on the keyword text and the semantic query vector, respectively, to determine a number of text blocks based on keyword matching and a number of text blocks based on semantic matching; the text database stores a number of text blocks and the TF-IDF codes corresponding to each text block, and the vector database stores a number of text blocks and the vectors corresponding to each text block.

[0039] The search text block determination module M3 is used to aggregate and remove duplicates of a number of text blocks based on keyword matching and a number of text blocks based on semantic matching to obtain a number of search text blocks that are highly relevant to the search question.

[0040] The large model intelligent reply generation module M4 is used to generate large model prompts based on several search text blocks and prompt templates, and use the trained livestock and poultry manure treatment technology smart question and answer model to output the large model intelligent reply results according to the large model prompts; the livestock and poultry manure treatment technology smart question and answer model is a large language model obtained after reinforcement learning training based on knowledge in the field of livestock and poultry manure treatment technology.

[0041] like Figure 2As shown, it is a schematic diagram of the human-computer interaction interface of the livestock and poultry manure treatment technology question-answering system based on a large model provided in this embodiment at the initial stage, which displays the welcome words and question examples for user reference, and provides a text box to receive the search questions entered by the user. When the user presses the Enter key or clicks the Send button, the above system is called to analyze and automatically reply according to the search questions entered by the user. It can be understood that Figure 2 The content displayed in the initial human-computer interaction interface is for illustrative purposes only and should not be regarded as an additional limitation to the technical solution of the present application.

[0042] Before calling the automatic word segmentation module M1 to obtain the search question input by the user, and using the trained word segmentation model to perform word segmentation processing on the search question to obtain the keyword text and semantic query vector, relevant means are needed to obtain livestock and poultry manure treatment related data, and train a trained word segmentation model and a trained livestock and poultry manure treatment technology intelligent question and answer model; therefore, in another exemplary embodiment of the present application, the livestock and poultry manure treatment technology question and answer system based on the large model also includes: The multi-source data acquisition and processing module M5 is used to obtain livestock and poultry manure treatment related data from different knowledge retrieval platforms, obtain multi-source livestock and poultry manure treatment related data, and process the obtained multi-source livestock and poultry manure treatment related data to obtain a word segmentation training data set and a large language model training corpus; the processing includes data cleaning, data segmentation, data structuring and data review; the livestock and poultry manure treatment related data includes literature data, business data and experimental data related to livestock and poultry manure treatment technology.

[0043] The source of relevant literature data can come from resource websites such as Springer Nature, ScienceDirect, Web of Science, China National Knowledge Infrastructure, ACS, Sci-Hub and Scopus. Among them, the data cleaning is to remove noise, errors and duplicate information in the data to ensure the quality and accuracy of the data, such as removing format errors and duplicate citations in the literature; data segmentation is to divide the data into appropriate small blocks to facilitate subsequent vector processing and retrieval; data structuring is to convert unstructured data into a structured form so that it has certain structure and semantic information; data review is to review and proofread the processed data to ensure the reliability and availability of the data. After the collected data is stored in the text database (MySQL), it is also stored in the vector database (Milvus) and the index database (ElasticSearch) to facilitate retrieval of data from different dimensions.

[0044] Before using these data specifically, the collected data is also hybrid sorted to optimize the data sorting so that the collected data is screened and optimized before entering the model, ensuring that the results of knowledge fusion are more accurate and reliable, so as to improve the system retrieval efficiency and the accuracy of the answers.

[0045] The word segmentation model training module M6 is used to train the initial word segmentation model using the word segmentation training data set obtained by processing multi-source livestock and poultry manure treatment related data to obtain a trained word segmentation model. In this embodiment, Byte-Pair Encoding (BPE) in Sentence Piece is selected as the word segmentation algorithm, and a word segmentation model is trained. After multiple considerations of word segmentation granularity and encoding efficiency, a Sentence Piece model containing 50k words is finally trained.

[0046] The large language model intensive training module M7 is used to train the general field large language model using the large language model training corpus obtained by processing multi-source livestock and poultry manure treatment related data to obtain a trained livestock and poultry manure treatment technology intelligent question-answering model. In this embodiment, the Qwen model is selected as the base model. The Qwen model has outstanding advantages in language understanding, generation accuracy and fluency, and can better meet the needs of users in various scenarios.

[0047] Specifically in an exemplary embodiment of the present application, the large language model reinforcement training module M7 specifically includes: an incremental pre-training unit D1, a model fine-tuning unit D2 and a reinforcement learning unit D3.

[0048] The incremental pre-training unit D1 is used to pre-train the general domain large language model using the large language model training corpus to obtain a pre-trained large language model; the pre-trained large language model initially possesses knowledge and language comprehension capabilities in the field of livestock and poultry manure treatment technology.

[0049] The model fine-tuning unit D2 is used to fine-tune the parameters of the pre-trained large language model using high-quality large language model training corpus to obtain a large language model with fine-tuned parameters; the large language model with fine-tuned parameters can adapt to the specific language and knowledge requirements in the field of livestock and poultry manure treatment technology; the high-quality large language model training corpus is a large language model training corpus in the form of question and answer.

[0050] The reinforcement learning unit D3 is used to conduct reinforcement learning training on the large language model after fine-tuning the parameters through manual feedback and reward mechanisms, and continuously optimize the performance and output results of the large language model until the preset conditions are met to obtain a trained livestock and poultry manure treatment technology intelligent question-answering model. During the reinforcement learning training process, the model parameters and strategies are adjusted according to the user's feedback on the answer to improve the accuracy and satisfaction of the answer.

[0051] In order to ensure that the answers given by the intelligent question and answer model of livestock and poultry manure treatment technology can be more in line with the technological development of the current era, relevant means are also needed to update the database in the livestock and poultry manure treatment technology question and answer system. Therefore, in another exemplary embodiment of the present application, the livestock and poultry manure treatment technology question and answer system based on the large model also includes: a database update module M8, which is used to periodically call the multi-source data acquisition processing module to obtain livestock and poultry manure treatment related data from different knowledge retrieval platforms and update the text database and vector database after processing.

[0052] In the existing solutions of applying traditional general-purpose big models, generally only simple replies are given to users' questions, and the reference literature cannot be clearly traced, making it difficult to verify whether the model's reply results are in line with reality. Therefore, in another exemplary embodiment of the present application, the livestock and poultry manure treatment technology question and answer system based on the big model also includes: a reply result tracing module M9, which is used to output the big model intelligent reply results according to the big model prompts for the trained livestock and poultry manure treatment technology smart question and answer model, and indicate the original source documents of each technical point.

[0053] like Figure 3 As shown, it is a schematic diagram of the human-computer interaction interface of the livestock and poultry manure treatment technology question-and-answer system based on a large model provided in this embodiment after outputting the large model intelligent reply results, which displays the display form of the large model intelligent reply results generated by the livestock and poultry manure treatment technology intelligent question-and-answer model for the search questions input by the user, such as "the relationship between the initial pH value and the amount of ammonia volatilization", and you can see that there is a "View More" option in the lower right corner of the relevant reply generated by the large model. After clicking on this option, the relevant information of the literature, journals and other documents referenced by the reply result is displayed in detail below the reply to realize the function of knowledge tracing. It can be understood that Figure 3 The content displayed in the human-computer interaction interface is for illustrative purposes only and should not be regarded as an additional limitation to the technical solution of the present application.

[0054] in addition, Figure 1~Figure 3 The above-mentioned system architecture and human-computer interaction interface are only exemplary. When implementing different functions, one or at least two components in the system or human-computer interaction interface shown in the figure can be omitted according to actual needs.

[0055] The above-mentioned livestock and poultry manure treatment technology question-and-answer system based on a large model can not only efficiently and accurately retrieve information from the industry knowledge base, but also generate high-quality answers to solve the illusion problem existing in large models in general fields. Through document traceability, the transparency and credibility of the answers are ensured, allowing users to verify the authenticity of the information themselves. In addition, the system can be continuously optimized as the industry knowledge data grows and changes, ensuring that users are always provided with the latest and most accurate document information. At the same time, it can support cross-document evidence and can retrieve information from multiple related documents, ensuring that the generated answers are based on a wide range of reliable data sources, rather than a single or isolated information point.

[0056] Based on the same inventive concept, the embodiment of the present application also provides a method for using a large-scale model-based livestock and poultry manure treatment technology question-and-answer system, such as Figure 4 The method comprises the following steps: A1. Obtain the search question input by the user, use the trained word segmentation model to segment the search question, and obtain the keyword text and semantic query vector.

[0057] A2. Based on the keyword text and the semantic query vector, a search is performed in the text database and the vector database respectively to determine a number of text blocks based on keyword matching and a number of text blocks based on semantic matching; the text database stores a number of text blocks and the TF-IDF codes corresponding to each text block, and the vector database stores a number of text blocks and the vectors corresponding to each text block. In this embodiment, the Best Matching 25 algorithm is used to find the text blocks based on keyword matching, and the semantic query vector and the vector library are used to perform cosine similarity calculation to find the text blocks based on semantic matching.

[0058] A3. Aggregate and remove duplicates of several text blocks based on keyword matching and several text blocks based on semantic matching to obtain several search text blocks that are highly relevant to the search question. In this embodiment, multi-way recall fusion technology is used to aggregate and remove duplicates of results based on keyword matching and semantic matching.

[0059] A4. Generate a large model prompt based on several search text blocks and prompt templates, and use the trained livestock and poultry manure treatment technology intelligent question and answer model to output the large model intelligent answer result according to the large model prompt; the livestock and poultry manure treatment technology intelligent question and answer model is a large language model obtained after reinforcement learning training based on knowledge in the field of livestock and poultry manure treatment technology. In this embodiment, the topk most relevant text blocks obtained after aggregation and deduplication are fused with the prompt template (PromptTemplate) to generate a large model prompt (Prompt), which is input to the large model to obtain the output result. For example, for the search question of "optimal parameters for aerobic composting", the system can provide detailed parameters and suggestions on raw material selection, ventilation mode, pH value adjustment, composting cycle, temperature control, ammonia volatilization control, nitrogen conversion, etc. by integrating the contents of multiple documents.

[0060] Before step A4, relevant means are required to obtain a trained livestock and poultry manure treatment technology intelligent question-answering model. Specifically in this embodiment, the method further includes the following steps: B1. Obtain livestock and poultry manure treatment related data from different knowledge retrieval platforms, obtain multi-source livestock and poultry manure treatment related data, and process the obtained multi-source livestock and poultry manure treatment related data to obtain a word segmentation training data set and a large language model training corpus.

[0061] B2. Using the large language model training corpus obtained by processing multi-source livestock and poultry manure treatment related data, the general domain large language model is trained to obtain a trained livestock and poultry manure treatment technology intelligent question-answering model. Figure 5 In the flowchart shown, step B2 specifically includes the following steps: B21. Use the large language model training corpus to pre-train the general domain large language model to obtain a pre-trained large language model; the pre-trained large language model initially possesses knowledge and language comprehension capabilities in the field of livestock and poultry manure treatment technology.

[0062] B22. Use high-quality large language model training corpus to fine-tune the parameters of the pre-trained large language model to obtain a large language model with fine-tuned parameters; the large language model with fine-tuned parameters can adapt to the specific language and knowledge requirements in the field of livestock and poultry manure treatment technology; the high-quality large language model training corpus is a large language model training corpus in the form of question and answer.

[0063] B23. Through artificial feedback and reward mechanisms, the large language model with fine-tuned parameters is subjected to reinforcement learning training, and the performance and output results of the large language model are continuously optimized until the preset conditions are met, thereby obtaining a trained intelligent question-and-answer model for livestock and poultry manure treatment technology.

[0064] The implementation solution for solving the problem provided in this method is similar to the implementation solution recorded in the above-mentioned system. Therefore, the specific limitations in one or more method embodiments provided in this embodiment can be found in the above limitations on the large-model-based livestock and poultry manure treatment technology question and answer system, which will not be repeated here.

[0065] The above-mentioned livestock and poultry manure treatment technology question-and-answer system based on a large model and its use method provided by the present invention trains the traditional general field large model with specific industry knowledge, so that it can fully and accurately understand the search questions input by the user, and give scientific, detailed and industry-specific answer results, realizing intelligent question-and-answer about livestock and poultry manure treatment knowledge, saving the time and manpower consumption of researchers to query and read a large amount of literature, and significantly improving the data collection, analysis and retrieval efficiency in the process of livestock and poultry manure treatment technology research. The large model technology can be used to realize intelligent analysis of various environmental parameters in the process of livestock and poultry manure treatment. At the same time, through deep mining and integration of multi-source data, the system can accurately identify the key factors affecting the treatment effect and provide scientific and optimized decision support. In addition, the system uses large model technology for data analysis and association mining, which can be used to optimize the livestock and poultry manure treatment process, and can significantly improve the quality of products such as organic fertilizers, biogas, liquid fertilizers and soil conditioners generated after livestock and poultry manure treatment, greatly improve the efficiency of resource utilization, realize the green environmental protection and sustainable development of the livestock and poultry manure treatment process, and promote the modernization process of solid and liquid waste treatment and resource utilization.

[0066] In addition, the system can automatically identify the characteristics of different types of livestock and poultry manure, and dynamically adjust the treatment strategy based on the integration of massive environmental data, breeding data, and treatment process parameters, so as to achieve the best resource utilization effect. The application of the livestock and poultry manure treatment technology question-and-answer system based on large models not only improves the level of refined management of livestock and poultry manure, but also promotes the green and environmentally friendly development of the livestock and poultry manure treatment industry, truly realizing the intelligent, refined and sustainable development of livestock and poultry manure treatment, and has broad promotion value and application prospects.

[0067] In another exemplary embodiment, a computer device is also provided. The computer device may be a server or a terminal. The internal structure diagram thereof may be as follows: Figure 6As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. Among them, the processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store technical knowledge in the field of livestock and poultry manure treatment, as well as custom-built text databases and vector databases. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, the functions of the livestock and poultry manure treatment technology question and answer system based on a large model provided in the above embodiment can be realized.

[0068] Those skilled in the art will understand that Figure 6 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0069] In an exemplary embodiment, a computer device is further provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps in the above-mentioned method embodiments when executing the computer program.

[0070] In an exemplary embodiment, a computer-readable storage medium is provided, storing a computer program, and when the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0071] In an exemplary embodiment, a computer program product is provided, including a computer program, and when the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0072] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.

[0073] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM may be in various forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM).

[0074] The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. The non-relational database may include a distributed database based on blockchain, etc., but is not limited thereto. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but is not limited thereto.

[0075] The technical features of the above embodiments may be arbitrarily combined. To make the description concise, not all possible combinations of the technical features in the above 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.

[0076] This article uses specific examples to illustrate the principles and implementation methods of this application. The description of the above embodiments is only used to help understand the method and core ideas of this application. At the same time, for those skilled in the art, according to the ideas of this application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting this application.

Claims

1. A large-scale model-based livestock and poultry manure treatment technology question-answering system, characterized in that: include: The automatic word segmentation module for search questions is used to obtain the search questions input by the user, and use the trained word segmentation model to perform word segmentation processing on the search questions to obtain keyword text and semantic query vectors; A text block dual matching module is used to search in a text database and a vector database based on the keyword text and the semantic query vector, respectively, to determine a number of text blocks based on keyword matching and a number of text blocks based on semantic matching; the text database stores a number of text blocks and the TF-IDF codes corresponding to each text block, and the vector database stores a number of text blocks and the vectors corresponding to each text block; A search text block determination module is used to aggregate and remove duplicates of a number of text blocks based on keyword matching and a number of text blocks based on semantic matching to obtain a number of search text blocks that are highly relevant to the search question; The large model intelligent reply generation module is used to generate a large model prompt based on several of the search text blocks and prompt templates, and use the trained livestock and poultry manure treatment technology smart question and answer model to output the large model intelligent reply result according to the large model prompt; the livestock and poultry manure treatment technology smart question and answer model is a large language model obtained after reinforcement learning training based on knowledge in the field of livestock and poultry manure treatment technology.

2. The livestock and poultry manure treatment technology question-answering system based on a large model according to claim 1 is characterized in that: The large-model-based livestock and poultry manure treatment technology question and answer system also includes: a multi-source data acquisition and processing module, which is used to acquire livestock and poultry manure treatment related data from different knowledge retrieval platforms, obtain multi-source livestock and poultry manure treatment related data, and process the acquired multi-source livestock and poultry manure treatment related data to obtain a word segmentation training data set and a large language model training corpus; the processing includes data cleaning, data segmentation, data structuring and data review; the livestock and poultry manure treatment related data includes literature data, business data and experimental data related to livestock and poultry manure treatment technology.

3. The livestock and poultry manure treatment technology question-answering system based on a large model according to claim 2 is characterized in that: The large-model-based livestock and poultry manure treatment technology question-and-answer system also includes: a word segmentation model training module, which is used to train the initial word segmentation model using a word segmentation training data set obtained by processing multi-source livestock and poultry manure treatment related data to obtain a trained word segmentation model.

4. The livestock and poultry manure treatment technology question-answering system based on a large model according to claim 2 is characterized in that: The large-model-based livestock and poultry manure treatment technology question and answer system also includes: a large language model intensive training module, which is used to train the general-domain large language model using the large language model training corpus obtained by processing multi-source livestock and poultry manure treatment related data to obtain a trained livestock and poultry manure treatment technology intelligent question and answer model.

5. The livestock and poultry manure treatment technology question-answering system based on a large model according to claim 4 is characterized in that: The large language model reinforcement training module specifically includes: an incremental pre-training unit, a model fine-tuning unit and a reinforcement learning unit; The incremental pre-training unit is used to pre-train the general domain large language model using the large language model training corpus to obtain a pre-trained large language model; the pre-trained large language model preliminarily possesses knowledge and language comprehension capabilities in the field of livestock and poultry manure treatment technology; The model fine-tuning unit is used to fine-tune the parameters of the pre-trained large language model using high-quality large language model training corpus to obtain a large language model with fine-tuned parameters; the large language model with fine-tuned parameters can adapt to the specific language and knowledge requirements in the field of livestock and poultry manure treatment technology; the high-quality large language model training corpus is a large language model training corpus in the form of question and answer; The reinforcement learning unit is used to perform reinforcement learning training on the large language model after fine-tuning the parameters through manual feedback and reward mechanism, continuously optimize the performance and output results of the large language model until the preset conditions are met, and obtain a trained livestock and poultry manure treatment technology intelligent question and answer model.

6. The livestock and poultry manure treatment technology question-answering system based on a large model according to claim 2 is characterized in that: The large-model-based livestock and poultry manure treatment technology question and answer system also includes: a database update module, which is used to regularly call the multi-source data acquisition and processing module to obtain livestock and poultry manure treatment related data from different knowledge retrieval platforms and update the text database and the vector database after processing.

7. The livestock and poultry manure treatment technology question-answering system based on a large model according to claim 1 is characterized in that: The large-model-based livestock and poultry manure treatment technology question and answer system also includes: a response result tracing module, which is used to output the large-model intelligent response results according to the large-model prompts for the trained livestock and poultry manure treatment technology intelligent question and answer model, and indicate the original source documents of each technical point.

8. A method for using the large-scale model-based livestock and poultry manure treatment technology question-answering system as described in any one of claims 1 to 7, characterized in that: include: Obtaining a search question input by a user, and performing word segmentation processing on the search question using a trained word segmentation model to obtain a keyword text and a semantic query vector; Based on the keyword text and the semantic query vector, searching is performed in a text database and a vector database respectively to determine a plurality of text blocks based on keyword matching and a plurality of text blocks based on semantic matching; the text database stores a plurality of text blocks and TF-IDF codes corresponding to each text block, and the vector database stores a plurality of text blocks and vectors corresponding to each text block; Aggregate and remove duplicates of several text blocks based on keyword matching and several text blocks based on semantic matching to obtain several search text blocks that are highly relevant to the search question; A large model prompt is generated based on several of the search text blocks and prompt templates, and a trained livestock and poultry manure treatment technology smart question and answer model is used to output a large model intelligent reply result according to the large model prompt; the livestock and poultry manure treatment technology smart question and answer model is a large language model obtained after reinforcement learning training based on knowledge in the field of livestock and poultry manure treatment technology.

9. The method for using the large-scale model-based livestock and poultry manure treatment technology question-answering system according to claim 8 is characterized in that: Before generating a large model prompt based on a plurality of the search text blocks and the prompt templates, and outputting a large model intelligent answer result according to the large model prompt using the trained livestock and poultry manure treatment technology intelligent question-answering model, the use method further includes: Acquire livestock and poultry manure treatment related data from different knowledge retrieval platforms to obtain multi-source livestock and poultry manure treatment related data, and process the obtained multi-source livestock and poultry manure treatment related data to obtain a word segmentation training data set and a large language model training corpus; The large language model training corpus obtained by processing multi-source livestock and poultry manure treatment related data is used to train the general field large language model to obtain a trained livestock and poultry manure treatment technology intelligent question and answer model.

10. The method for using the large-scale model-based livestock and poultry manure treatment technology question-answering system according to claim 9 is characterized in that: The large language model training corpus obtained by processing multi-source livestock and poultry manure treatment related data is used to train the general field large language model to obtain a trained livestock and poultry manure treatment technology intelligent question-answering model, which specifically includes: Pre-training a general-domain large language model using a large language model training corpus to obtain a pre-trained large language model; the pre-trained large language model initially possesses knowledge and language comprehension capabilities in the field of livestock and poultry manure treatment technology; Using high-quality large language model training corpus to fine-tune the parameters of the pre-trained large language model to obtain a large language model with fine-tuned parameters; the large language model with fine-tuned parameters can adapt to the specific language and knowledge requirements in the field of livestock and poultry manure treatment technology; the high-quality large language model training corpus is a large language model training corpus in the form of question and answer; Through manual feedback and reward mechanisms, the large language model with fine-tuned parameters is subjected to reinforcement learning training, and the performance and output results of the large language model are continuously optimized until the preset conditions are met, thus obtaining a trained intelligent question-and-answer model for livestock and poultry manure treatment technology.

Images