County country planting industry whole industry chain development resource allocation optimization method and system

Abstract

The invention discloses a county country planting industry whole industry chain development resource allocation optimization method and system, and the method comprises the steps: constructing a county country planting industry whole industry chain knowledge graph based on the county country planting industry whole industry chain data, and receiving a natural language problem input by a user in the aspect of resource allocation optimization; the method comprises the following steps of: performing semantic analysis on a question input by a user by adopting a natural language processing technology based on deep learning, searching a corresponding question template from a question template library based on similarity matching of a semantic level, and performing answer search in a knowledge graph based on a query statement of the question template so as to obtain an answer corresponding to the question input by the user. Through the mode, the quality and speed of resource allocation decision making of county country planting industry can be effectively improved, the resource utilization efficiency is improved, waste is reduced, and meanwhile intelligent transformation of agricultural production is promoted.

Description

Technical Field

[0001] The present application relates to the field of intelligent configuration, and more specifically, to a method and system for optimizing resource configuration for the development of the entire industrial chain of county-level rural planting industry. Background Art

[0002] With the global population growth and the acceleration of urbanization, the demand for food and other agricultural products continues to increase, which brings huge challenges to agricultural production and resource utilization. Since traditional agricultural production methods often rely on empiricism and extensive management, the planting industry in counties and villages usually faces the problem of uneven resource allocation, resource waste and frequent environmental problems, especially in the allocation of water resources, fertilizers, seeds and labor.

[0003] At the same time, due to the limited access to information in rural areas of counties, it is difficult for growers to keep up with the latest market trends, advanced planting techniques and effective resource allocation strategies. This information asymmetry often leads to planting decision errors, such as blindly following the trend, over-fertilization and unreasonable irrigation measures, which not only cause a huge waste of resources, but also may pollute the environment and lead to a decline in the quality of agricultural products.

[0004] Therefore, in order to improve agricultural productivity and ensure sustainable development, a resource allocation optimization method and system for the development of the entire industrial chain of county-level rural planting industry is expected. Summary of the invention

[0005] In order to solve the above technical problems, the present application is proposed. The embodiment of the present application provides a method and system for optimizing resource allocation for the development of the entire industrial chain of county-level rural planting industry. It first constructs a knowledge graph of the entire industrial chain of county-level rural planting industry based on the data of the entire industrial chain of county-level rural planting industry. After receiving the natural language questions on resource allocation optimization input by the user, it uses the natural language processing technology based on deep learning to perform semantic analysis on it, finds the corresponding question template from the question template library based on the similarity matching at the semantic level, and searches for answers in the knowledge graph based on the query statement of the question template, so as to obtain the answer corresponding to the question input by the user. In this way, the quality and speed of resource allocation decision-making in county-level rural planting industry can be effectively improved, resource utilization efficiency can be improved, waste can be reduced, and the intelligent transformation of agricultural production can also be promoted.

[0006] According to one aspect of the present application, a method for optimizing resource allocation for the development of the entire industrial chain of a county-level rural planting industry is provided, which includes: Obtain data on the entire industry chain of the county and rural planting industry; Based on the data of the entire industry chain of the county-level rural planting industry, a knowledge graph of the entire industry chain of the county-level rural planting industry is constructed; receiving a natural language question regarding resource allocation optimization input by a user; Based on the knowledge graph of the entire industrial chain of the county-level rural planting industry, answers to natural language questions about resource allocation optimization are searched for, and answers corresponding to the natural language questions about resource allocation optimization are output.

[0007] According to another aspect of the present application, a resource allocation optimization system for the development of the entire industrial chain of county-level rural planting industry is provided, which includes: Data acquisition module, used to obtain data on the entire industry chain of county and rural planting industry; A knowledge graph construction module is used to construct a knowledge graph of the entire industrial chain of the county-level rural planting industry based on the data of the entire industrial chain of the county-level rural planting industry; A question accepting module, used for receiving natural language questions on resource allocation optimization input by a user; The answer search module is used to search for answers to the natural language questions about resource allocation optimization based on the knowledge graph of the entire industrial chain of the county-level rural planting industry, and output answers corresponding to the natural language questions about resource allocation optimization.

[0008] Compared with the prior art, the present application provides a method and system for optimizing resource allocation for the development of the entire industrial chain of county-level rural planting industry. First, the knowledge graph of the entire industrial chain of county-level rural planting industry is constructed based on the data of the entire industrial chain of county-level rural planting industry. After receiving the natural language questions on resource allocation optimization input by the user, the natural language processing technology based on deep learning is used to perform semantic analysis, and the corresponding question template is found from the question template library based on the similarity matching at the semantic level, and the answer search is performed in the knowledge graph based on the query statement of the question template, so as to obtain the answer corresponding to the question input by the user. In this way, the quality and speed of resource allocation decision-making in county-level rural planting industry can be effectively improved, resource utilization efficiency can be improved, waste can be reduced, and the intelligent transformation of agricultural production can also be promoted. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0010] Figure 1 It is a flow chart of a method for optimizing resource allocation for the development of the entire industrial chain of county-level rural planting industry according to an embodiment of the present application; Figure 2A data flow diagram of a method for optimizing resource allocation for the entire industrial chain of county-level rural planting industry according to an embodiment of the present application; Figure 3 It is a flowchart of sub-step S4 of the method for optimizing resource allocation for the whole industrial chain development of the county rural planting industry according to an embodiment of the present application; Figure 4 It is a flowchart of sub-step S42 of the method for optimizing resource allocation for the entire industrial chain development of the county-level rural planting industry according to an embodiment of the present application; Figure 5 This is a block diagram of a resource allocation optimization system for the development of the entire industrial chain of county-level rural planting industry according to an embodiment of the present application. DETAILED DESCRIPTION

[0011] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.

[0012] As shown in this application and claims, unless the context clearly indicates an exception, the words "a", "an", "an" and / or "the" do not refer to the singular and may also include the plural. Generally speaking, the terms "include" and "comprise" only indicate the inclusion of the steps and elements that have been clearly identified, and these steps and elements do not constitute an exclusive list. The method or device may also include other steps or elements.

[0013] Although the present application makes various references to certain modules in the system according to the embodiments of the present application, any number of different modules can be used and run on the user terminal and / or server. The modules are only illustrative, and different aspects of the system and method can use different modules.

[0014] Flowcharts are used in the present application to illustrate the operations performed by the system according to the embodiments of the present application. It should be understood that the preceding or following operations are not necessarily performed accurately in order. On the contrary, various steps may be processed in reverse order or simultaneously as required. Meanwhile, other operations may also be added to these processes, or a certain step or several steps of operations may be removed from these processes.

[0015] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.

[0016] In the technical solution of this application, a method for optimizing resource allocation for the development of the entire industrial chain of county-level rural planting industry is proposed. Figure 1This is a flow chart of a method for optimizing resource allocation for the entire industry chain development of county-level rural planting industry according to an embodiment of the present application. Figure 2 This is a data flow diagram of the method for optimizing resource allocation for the entire industrial chain of county-level rural planting industry according to an embodiment of the present application. Figure 1 and Figure 2 As shown, according to the embodiment of the present application, the method for optimizing resource allocation for the development of the entire industrial chain of county-level rural planting industry includes the following steps: S1, acquiring the data of the entire industrial chain of county-level rural planting industry; S2, constructing a knowledge graph of the entire industrial chain of county-level rural planting industry based on the data of the entire industrial chain of county-level rural planting industry; S3, receiving natural language questions on resource allocation optimization input by users; S4, searching for answers to the natural language questions on resource allocation optimization based on the knowledge graph of the entire industrial chain of county-level rural planting industry, and outputting answers corresponding to the natural language questions on resource allocation optimization.

[0017] In particular, the S1 obtains the data of the entire industry chain of the county and rural planting industry. In the embodiment of the present application, the data of the entire industry chain of the county and rural planting industry covers data from pre-planting preparation, planting process management to post-harvest processing and sales, including but not limited to soil type, climate conditions, planting varieties, planting techniques, yield, market demand, market price fluctuations, planting costs, agricultural policies, water resources and other information.

[0018] In a specific example, in terms of data collection, web crawler tools are used to automatically capture web page content, especially unstructured or semi-structured data that exists on the Web. Noise removal technology is used to remove irrelevant information to retain useful structured data. For some data that is not easy to obtain automatically, such as some sensitive internal corporate information or first-hand information obtained from field research, manual collection is required. The cleaned structured data will be imported into a relational database management system (RDBMS), such as MySQL. This not only facilitates management and query, but also provides a solid foundation for the subsequent construction of domain ontology.

[0019] In particular, the S2, based on the county rural planting industry full industry chain data, constructs the county rural planting industry full industry chain knowledge map. Considering that the county rural planting industry involves many disciplines and technical fields, such as biology, ecology, economics, etc., and the various types of data are interrelated and influenced. Therefore, in order to effectively integrate and associate the planting industry full industry chain data of different sources and types, this application further constructs the county rural planting industry full industry chain knowledge map based on the county rural planting industry full industry chain data. It should be understood that the knowledge map is an efficient structured information representation form, which graphically expresses entities and their relationships in the form of nodes and edges, and can effectively integrate scattered knowledge points and form an organic whole. For the county rural planting industry, by constructing the county rural planting industry full industry chain knowledge map, it means that the data from production factors (such as soil analysis, water source analysis, seed selection) to production activities (such as farming, fertilization, irrigation, pest control), and then to market sales and other links will be systematically sorted, and the association model between the elements will be established. In this way, we can achieve a comprehensive understanding and in-depth analysis of the entire industry chain of the planting industry, and provide growers with more accurate resource allocation recommendations.

[0020] In an embodiment of the present application, based on the county-level rural planting industry full industrial chain data, a knowledge graph of the county-level rural planting industry full industrial chain is constructed, including: first, the data in the county-level rural planting industry full industrial chain data are structured to obtain a set of industrial chain data structured data, and the set of industrial chain data structured data is stored in a relational database; here, the data in the county-level rural planting industry full industrial chain (including but not limited to semi-structured and unstructured data) are cleaned, converted and structured, and the processed structured data set is stored in a relational database to ensure that the data for subsequent analysis and use has a unified format and high quality. Then, all the industrial chain data structured data are analyzed to construct a domain ontology of the county-level rural planting industry full industrial chain; specifically, by analyzing all structured industrial chain data, terms, concepts and their relationships can be extracted from them to construct a domain ontology suitable for the county-level rural planting industry full industrial chain. This ontology defines the core concepts in the industry and the relationships between these concepts. Then, the domain ontology of the county-level rural planting industry full industrial chain is used as the model layer of the knowledge graph of the county-level rural planting industry full industrial chain. In this way, a logical framework can be provided for the entire knowledge graph to help understand the association between various entities. Then, the mapping relationship between the domain ontology and the structured data of the industrial chain data in the relational database is configured; the specific fields or tables in the relational database are mapped to the corresponding classes, attributes or instances in the ontology by configuring the consistency mapping rules between the domain ontology and the relational database. In one example, this can be achieved by using a mapping tool such as OnTop, which allows direct query and conversion of data in RDF format from the relational database without physically migrating the data. In this way, the position of the original data in the relational database can be kept unchanged, while the powerful functions of the knowledge graph can be used for complex queries and reasoning. Finally, based on the mapping relationship, the structured data of the industrial chain data stored in the relational database is exported as triple data, and the triple data is used as the data layer of the knowledge graph of the entire industrial chain of the county rural planting industry. That is, according to the pre-set mapping relationship, the structured data stored in the relational database is exported as triples in the form of resource description framework (RDF), and these triples constitute the data layer of the knowledge graph, which are the manifestation of specific facts and record various relationships between entities.

[0021] In particular, S3 receives a natural language question about resource configuration optimization input by a user. It should be understood that in the technical solution of the present application, natural language is used as an interactive method, so that users can ask questions in a more natural and intuitive way without having to master complex professional knowledge or terminology, which effectively lowers the technical threshold.

[0022] In particular, the S4 searches for answers to the natural language questions about resource allocation optimization based on the knowledge graph of the entire industry chain of the county-level rural planting industry, and outputs answers corresponding to the natural language questions about resource allocation optimization. In a specific example of the present application, Figure 3 As shown, the S4 includes: S41, semantically encoding the natural language problem of resource allocation optimization to obtain a semantic encoding vector of the problem description of resource allocation optimization; S42, determining the question template with the highest matching degree based on the semantic encoding vector of the problem description of resource allocation optimization; S43, extracting the query statement of the question template with the highest matching degree, and querying the knowledge graph of the entire industrial chain of the county rural planting industry according to the query statement, and outputting the answer through a response function.

[0023] Specifically, the S41 semantically encodes the natural language problem in resource configuration optimization to obtain a semantic encoding vector describing the problem in resource configuration optimization. That is, in the technical solution of the present application, a semantic encoder based on the BERT model is used to semantically encode the natural language problem in resource configuration optimization to obtain a semantic encoding vector describing the problem in resource configuration optimization. Here, considering that there are many limitations of traditional keyword matching technology, such as difficulty in dealing with semantic ambiguity and polysemy, the present application uses a natural language processing technology based on deep learning to semantically encode the natural language problem in resource configuration optimization input by the user, so as to extract the contextual semantic information of the user input question, thereby improving the accuracy of the answer query through query matching at the semantic level. In a specific example of the present application, a semantic encoder based on the BERT model is used to semantically encode the natural language problem in resource configuration optimization to obtain a semantic encoding vector describing the problem in resource configuration optimization. Those of ordinary skill in the art should know that the BERT model uses a two-way training mechanism to enable BERT to simultaneously understand the left and right context information of each word in a sentence, thereby more accurately capturing the complex relationship between words. Specifically, BERT is built on the encoder part of the Transformer architecture. Unlike traditional unidirectional language models, it implements bidirectional training through a masked language model. During the pre-training phase, BERT randomly masks certain words in the input text and tries to predict these masked words based on the context. This approach breaks the sequential dependency of traditional language models and allows the model to use information from all positions at once for richer representations. In addition, in order to enhance the understanding of the logical relationship between sentences, BERT also adds the next sentence prediction task during the pre-training process to help the model determine whether two sentences appear consecutively, which is particularly useful for tasks such as question answering and textual entailment.

[0024] Specifically, the S42 determines the problem template with the highest matching degree based on the semantic coding vector of the problem description in the resource configuration optimization aspect. Figure 4 As shown, the S42 includes: S421, extracting a set of question templates from a question template library; S422, semantically encoding each question template in the set of question templates to obtain a set of question template semantic embedding coding vectors; S423, semantically distinguishing and enhancing the set of question template semantic embedding coding vectors to obtain a set of question template semantic distinguishing and enhancing coding vectors; S424, respectively calculating the semantic matching degree between the semantic coding vector of the problem description of the resource configuration optimization aspect and each question template semantic distinguishing and enhancing coding vector in the set of question template semantic distinguishing and enhancing coding vectors to obtain a set of semantic matching degrees; S425, taking the question template corresponding to the maximum value in the set of semantic matching degrees as the question template with the highest matching degree.

[0025] More specifically, the S421 extracts a set of question templates from the question template library. Considering that the query questions input by the user may have different expressions, the accuracy of the query results will be affected by noise and ambiguity when semantic matching is performed directly from the knowledge graph. To this end, the present application further covers various possible resource allocation optimization problems by pre-building a question template library. Among them, the question template library contains a series of standardized question templates, each template is designed with a standardized query statement for a specific resource allocation optimization problem, which is used to search for answers in the knowledge graph, thereby improving the accuracy of the query.

[0026] More specifically, the S422 semantically encodes each question template in the set of question templates to obtain a set of semantic embedding coding vectors of the question templates. Here, in order to effectively semantically match each question template in the set of question templates with the user's input questions, the present application also uses natural language processing technology to semantically encode each question template in the set of question templates, and maps each question template to a high-dimensional semantic feature space, thereby obtaining a set of semantic embedding coding vectors of question templates. In a specific example of the present application, a semantic encoder based on the BERT model is also used to implement semantic encoding of each question template. In this way, by adopting the same semantic encoding model, it can be ensured that the question template is comparable with the user input question in the semantic space, reducing the semantic deviation introduced by inconsistent data processing methods, thereby improving the accuracy of question template matching.

[0027] More specifically, in S423, the set of semantic embedding coding vectors of the question template is semantically distinguishable and enhanced to obtain a set of semantically distinguishable and enhanced coding vectors of the question template. In order to more accurately match the user question with the closest question template, the present application further performs feature distinguishable and enhanced processing on the set of semantic embedding coding vectors of the question template to enhance the differences between the various question templates, avoid selection errors caused by multiple question templates being too similar, and thus improve the relevance and accuracy of the final answer.

[0028] Specifically, the specific process of performing semantic distinguishable enhancement on the set of semantic embedding coding vectors of the question template includes: first, extracting the essential features of the set of semantic embedding coding vectors of the question template to obtain the semantic field essential feature vector of the question template. That is, in the technical solution of the present application, first, a field mapper is used to perform field mapping on each semantic embedding coding vector of the question template in the set of semantic embedding coding vectors of the question template to obtain a set of semantic field feature vectors of the question template; considering that each semantic embedding coding vector of the question template may have different feature dimensions, therefore, first, a field mapper is used to perform field mapping operation on each semantic embedding coding vector of the question template to map it to a unified field space, thereby unifying the feature dimensions and better reflecting the intrinsic semantic association between each question template. Then, the field depth factor of each semantic field feature vector of the question template in the set of semantic field feature vectors of the question template is calculated to obtain a set of semantic field depth factors of the question template; then, based on the set of semantic field depth factors of the question template, the field essential features of the set of semantic field feature vectors of the question template are calculated to obtain the semantic field essential feature vector of the question template. That is, by calculating the field depth factor of the semantic field feature vector of each question template obtained after mapping, we can measure its information density and feature importance in the field space, and get a set of semantic field depth factors of the question template. Based on this, we use the field effect normalization mechanism to perform position-weighted aggregation on the semantic field feature vectors of each question template, thereby extracting the semantic feature distribution that comprehensively reflects the core attributes of the question template set, that is, the essential feature vector of the semantic field of the question template.

[0029] Among them, based on the set of semantic field depth factors of the question template, the specific process of calculating the field essential features of the set of semantic field feature vectors of the question template includes: inputting the set of semantic field depth factors of the question template into the softmax function for normalization processing to obtain a set of normalized semantic field depth factors of the question template; using the set of normalized semantic field depth factors of the question template as a set of weights, calculating the weighted sum of the set of semantic field feature vectors of the question template to obtain the semantic field essential feature vector of the question template. More specifically, the essential features of the set of semantic embedding coding vectors of the question template are extracted using the following essential feature extraction formula to obtain the semantic field essential feature vector of the question template; wherein, the essential feature extraction formula is: in, represents the first in the set of semantic embedding encoding vectors of the question template The semantic embedding encoding vector of the question template, and Represent the first linear mapping matrix and the second linear mapping matrix respectively, Represents the semantic field feature vector of the question template, represents the L2 norm of the vector, represents the logarithmic function with base 2, Indicates The semantic field depth factor of the question template, represents the natural exponential function with base e, Indicates Normalized question template semantic field depth factor, Represents the essential feature vector of the semantic field of the question template.

[0030] Furthermore, taking the intrinsic feature vector of the semantic field of the problem template as the conditional feature vector, each semantic embedding coding vector of the problem template in the set of the semantic embedding coding vector of the problem template is distinguishably enhanced to obtain the set of semantically distinguishable enhanced coding vectors of the problem template. That is, in the technical solution of the present application, first, the position-weighted sum of the intrinsic feature vector of the semantic field of the problem template and the semantic embedding coding vector of the problem template is calculated to obtain the essential optimization problem template semantic embedding coding vector; finally, the point-added vector between the essential optimization problem template semantic embedding coding vector and the bias vector is calculated, and the point-added vector is multiplied by the position with the semantic embedding coding vector of the problem template after passing the sigmoid activation function to obtain the semantically distinguishable enhanced coding vector of the problem template. That is, taking the intrinsic feature vector of the semantic field of the problem template as the conditional feature vector, each semantic embedding coding vector of the problem template is distinguishably enhanced element by element, and the unique information in each problem template is specifically amplified to obtain the set of semantically distinguishable enhanced coding vectors of the problem template, thereby ensuring that the subtle differences between different problem templates can be effectively distinguished during the semantic matching process.

[0031] More specifically, the semantic field essential feature vector of the question template is used as the conditional feature vector, and each question template semantic embedding coding vector in the set of the question template semantic embedding coding vector is distinguishably enhanced by the following distinguishable enhancement formula to obtain the set of the question template semantic distinguishable enhanced coding vector; wherein the distinguishable enhancement formula is: in, and Represent different weight parameters respectively. represents the bias vector, represents the sigmoid activation function, It means point multiplication by position. Indicates The semantically distinguishable reinforced encoding vector of each question template.

[0032] More specifically, the S424 calculates the semantic matching degree between the semantic coding vector of the problem description in resource configuration optimization and each semantically distinguishable and enhanced coding vector of the problem template in the set of semantically distinguishable and enhanced coding vectors of the problem template respectively to obtain a set of semantic matching degrees. That is, in the technical solution of the present application, the semantic matching degree between the semantic coding vector of the problem description in resource configuration optimization and each semantically distinguishable and enhanced coding vector of the problem template in the set of semantically distinguishable and enhanced coding vectors of the problem template is calculated respectively to obtain a set of semantic matching degrees. In a specific implementation of the present application, the semantic matching degree between the semantic coding vector of the problem description in resource configuration optimization and the semantically distinguishable and enhanced coding vector of the problem template is measured by calculating the cosine similarity between the two. It should be understood that cosine similarity is a commonly used similarity measurement method. The similarity between the two vectors is determined by calculating the cosine value of the angle between the two vectors. The closer the cosine similarity is to 1, the more similar the two are and the higher the matching degree.

[0033] More specifically, in S425, the question template corresponding to the maximum value in the set of semantic matching degrees is used as the question template with the highest matching degree. That is, based on the magnitude of each semantic matching degree, the question template with the highest matching degree with the user input question can be screened out from the question template library, thereby providing a more accurate answer.

[0034] Specifically, S43 extracts the query statement of the question template with the highest matching degree, and searches the knowledge graph of the entire industrial chain of the county rural planting industry according to the query statement, and outputs the answer through a response function.

[0035] In one example, each question template is usually associated with one or more query statements, which are SPARQL (a query language for querying RDF data) or other forms of query expressions used to retrieve relevant information from the knowledge graph. When executing the query, the system replaces the variables in the query statement with specific values ​​to ensure that the query is targeted at the specific situation that the user is concerned about. Continuing with the above example, the system will replace "specific crop name" with the specific crop type mentioned by the user, so that the query points to the correct data node. After the query is completed, the system needs to process the result set, extract useful information from it, and format it into an easy-to-understand form to feed back to the user. This step involves parsing, filtering, and further calculation of the raw data when necessary. For example, if the query returns a set of irrigation time recommendations, the system may select the best one or several as the final answer.

[0036] In order to provide a more friendly and personalized user experience, the system can also design a set of response functions to generate the final answer text. These functions can adjust the output content according to different contexts, such as adding explanatory text, providing additional reference links, etc. In addition, the response function can also be responsible for converting technical terms into easy-to-understand language to ensure that ordinary users can easily understand it. For example, the system can convert professional terms such as "optimal irrigation cycle" in the query results into more intuitive terms, such as "the best time for watering."

[0037] In summary, according to the embodiment of the present application, the resource allocation optimization method for the development of the entire industrial chain of the county-level rural planting industry is explained. It first constructs a knowledge graph of the entire industrial chain of the county-level rural planting industry based on the data of the entire industrial chain of the county-level rural planting industry. After receiving the natural language questions on resource allocation optimization input by the user, it uses the natural language processing technology based on deep learning to perform semantic analysis, finds the corresponding question template from the question template library based on the similarity matching at the semantic level, and searches for answers in the knowledge graph based on the query statement of the question template, so as to obtain the answer corresponding to the question input by the user. In this way, the quality and speed of resource allocation decisions in the county-level rural planting industry can be effectively improved, the efficiency of resource utilization can be improved, waste can be reduced, and the intelligent transformation of agricultural production can also be promoted. Furthermore, a resource allocation optimization system for the development of the entire industrial chain of the county-level rural planting industry is also provided.

[0038] Figure 5 This is a block diagram of a resource allocation optimization system for the development of the entire industrial chain of county-level rural planting industry according to an embodiment of the present application. Figure 5 As shown, according to the embodiment of the present application, the resource allocation optimization system 300 for the development of the entire industrial chain of county-level rural planting industry includes: a data acquisition module 310, which is used to acquire the entire industrial chain data of county-level rural planting industry; a knowledge graph construction module 320, which is used to construct a knowledge graph of the entire industrial chain of county-level rural planting industry based on the entire industrial chain data of county-level rural planting industry; a question acceptance module 330, which is used to receive natural language questions about resource allocation optimization input by users; an answer search module 340, which is used to search for answers to the natural language questions about resource allocation optimization based on the knowledge graph of the entire industrial chain of county-level rural planting industry, and output answers corresponding to the natural language questions about resource allocation optimization.

[0039] As described above, the resource allocation optimization system 300 for the development of the entire industrial chain of the county-level rural planting industry according to the embodiment of the present application can be implemented in various wireless terminals, such as a server with a resource allocation optimization algorithm for the development of the entire industrial chain of the county-level rural planting industry. In one possible implementation, the resource allocation optimization system 300 for the development of the entire industrial chain of the county-level rural planting industry according to the embodiment of the present application can be integrated into the wireless terminal as a software module and / or a hardware module. For example, the resource allocation optimization system 300 for the development of the entire industrial chain of the county-level rural planting industry can be a software module in the operating system of the wireless terminal, or it can be an application developed for the wireless terminal; of course, the resource allocation optimization system 300 for the development of the entire industrial chain of the county-level rural planting industry can also be one of the many hardware modules of the wireless terminal.

[0040] Alternatively, in another example, the county-level rural planting industry full industrial chain development resource allocation optimization system 300 and the wireless terminal may also be separate devices, and the county-level rural planting industry full industrial chain development resource allocation optimization system 300 may be connected to the wireless terminal via a wired and / or wireless network, and transmit interactive information in accordance with an agreed data format.

[0041] The embodiments of the present disclosure have been described above, and the above description is exemplary, not exhaustive, and is not limited to the disclosed embodiments. Many modifications and changes will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the described embodiments. The selection of terms used herein is intended to best explain the principles of the embodiments, practical applications, or improvements to the technology in the market, or to enable other persons of ordinary skill in the art to understand the embodiments disclosed herein.

Claims

1. A method for optimizing resource allocation for the development of the entire industrial chain of county-level rural planting industry, characterized in that: include: Obtain data on the entire industry chain of the county and rural planting industry; Based on the data of the entire industry chain of the county-level rural planting industry, a knowledge graph of the entire industry chain of the county-level rural planting industry is constructed; receiving a natural language question regarding resource allocation optimization input by a user; Based on the knowledge graph of the entire industrial chain of the county-level rural planting industry, answers to natural language questions about resource allocation optimization are searched for, and answers corresponding to the natural language questions about resource allocation optimization are output.

2. The method for optimizing resource allocation for the development of the entire industrial chain of county-level rural planting industry according to claim 1 is characterized in that: Based on the data of the entire industry chain of the county-level rural planting industry, a knowledge graph of the entire industry chain of the county-level rural planting industry is constructed, including: Structuring the various data in the county-level rural planting industry full industrial chain data to obtain a set of industrial chain data structured data, and storing the set of industrial chain data structured data in a relational database; Analyze all the structured data of the industrial chain and construct the domain ontology of the entire industrial chain of the county and rural planting industry; The domain ontology of the entire industrial chain of the county-level rural planting industry is used as the model layer of the knowledge graph of the entire industrial chain of the county-level rural planting industry; Configuring a mapping relationship between the domain ontology and the structured data of the industrial chain data in the relational database; Based on the mapping relationship, the structured data of the industrial chain data stored in the relational database is exported as triple data, and the triple data is used as the data layer of the knowledge graph of the entire industrial chain of the county rural planting industry.

3. The method for optimizing resource allocation for the development of the entire industrial chain of county-level rural planting industry according to claim 2 is characterized in that: Based on the knowledge graph of the entire industry chain of the county-level rural planting industry, the answer to the natural language question on resource allocation optimization is searched, and the answer corresponding to the natural language question on resource allocation optimization is output, including: Semantically encoding the natural language problem in the resource allocation optimization aspect to obtain a semantic encoding vector describing the problem in the resource allocation optimization aspect; Determine the problem template with the highest matching degree based on the semantic coding vector of the problem description in the resource configuration optimization aspect; The query statement of the question template with the highest matching degree is extracted, and a query is performed on the knowledge graph of the entire industrial chain of the rural planting industry in the county according to the query statement, and the answer is output through a response function.

4. The method for optimizing resource allocation for the development of the entire industrial chain of county-level rural planting industry according to claim 3 is characterized in that: Semantically encoding the natural language problem in the resource configuration optimization aspect to obtain a semantic encoding vector of the problem description in the resource configuration optimization aspect includes: A semantic encoder based on the BERT model is used to semantically encode the natural language problem in the resource configuration optimization aspect to obtain a semantic encoding vector describing the problem in the resource configuration optimization aspect.

5. The method for optimizing resource allocation for the development of the entire industrial chain of county-level rural planting industry according to claim 4 is characterized in that: Based on the semantic coding vector of the problem description in the resource configuration optimization aspect, determining the problem template with the highest matching degree includes: Extracting a collection of question templates from a question template library; Performing semantic encoding on each question template in the set of question templates to obtain a set of question template semantic embedding encoding vectors; Performing semantically distinguishable enhancement on the set of semantically embedded coding vectors of the question template to obtain a set of semantically distinguishable enhanced coding vectors of the question template; Calculating the semantic matching degree between the semantic coding vector of the problem description in the resource configuration optimization aspect and each problem template semantic distinguishable and enhanced coding vector in the set of the problem template semantic distinguishable and enhanced coding vectors to obtain a set of semantic matching degrees; The question template corresponding to the maximum value in the set of semantic matching degrees is used as the question template with the highest matching degree.

6. The method for optimizing resource allocation for the development of the entire industrial chain of county-level rural planting industry according to claim 5 is characterized in that: The set of semantically distinguishable enhancement of the question template semantic embedding coding vectors is performed to obtain a set of semantically distinguishable enhanced coding vectors of the question template, including: Extracting the essential features of the set of semantic embedding coding vectors of the question template to obtain the essential feature vector of the semantic field of the question template; Taking the essential feature vector of the semantic field of the question template as the conditional feature vector, each question template semantic embedding coding vector in the set of the question template semantic embedding coding vector is distinguishably enhanced to obtain the set of the question template semantic distinguishable enhanced coding vectors.

7. The method for optimizing resource allocation for the development of the entire industrial chain of county-level rural planting industry according to claim 6 is characterized in that: Extracting the essential features of the set of semantic embedding coding vectors of the question template to obtain the essential feature vector of the semantic field of the question template includes: Using a field mapper to perform field mapping on each question template semantic embedding coding vector in the set of question template semantic embedding coding vectors to obtain a set of question template semantic field feature vectors; Calculating the field depth factor of each question template semantic field feature vector in the set of question template semantic field feature vectors to obtain a set of question template semantic field depth factors; Based on the set of semantic field depth factors of the question template, the field essential features of the set of semantic field feature vectors of the question template are calculated to obtain the semantic field essential feature vector of the question template.

8. The method for optimizing resource allocation for the development of the entire industrial chain of county-level rural planting industry according to claim 7 is characterized in that: Based on the set of semantic field depth factors of the question template, calculating the field essential features of the set of semantic field feature vectors of the question template to obtain the semantic field essential feature vector of the question template, including: Inputting the set of semantic field depth factors of the question template into a softmax function for normalization processing to obtain a set of normalized semantic field depth factors of the question template; The set of normalized question template semantic field depth factors is used as a set of weights, and the weighted sum of the set of question template semantic field feature vectors is calculated to obtain the question template semantic field essential feature vector.

9. The method for optimizing resource allocation for the development of the entire industrial chain of county-level rural planting industry according to claim 8 is characterized in that: Taking the semantic field essential feature vector of the question template as a conditional feature vector, each question template semantic embedding coding vector in the set of the question template semantic embedding coding vector is distinguishably enhanced to obtain the set of the question template semantic distinguishable enhanced coding vectors, including: Calculating the position-weighted sum of the semantic field essential feature vector of the question template and the semantic embedding coding vector of the question template to obtain the semantic embedding coding vector of the essential optimization question template; The point-added vector between the semantic embedding coding vector of the essential optimization problem template and the bias vector is calculated, and the point-added vector is passed through a sigmoid activation function and multiplied by the position with the semantic embedding coding vector of the problem template to obtain the semantically distinguishable enhanced coding vector of the problem template.

10. A resource allocation optimization system for the development of the entire industrial chain of county-level rural planting industry, characterized in that: include: Data acquisition module, used to obtain data on the entire industry chain of county and rural planting industry; A knowledge graph construction module is used to construct a knowledge graph of the entire industrial chain of the county-level rural planting industry based on the data of the entire industrial chain of the county-level rural planting industry; A question accepting module, used for receiving natural language questions on resource allocation optimization input by a user; The answer search module is used to search for answers to the natural language questions about resource allocation optimization based on the knowledge graph of the entire industrial chain of the county-level rural planting industry, and output answers corresponding to the natural language questions about resource allocation optimization.

Images