A method for constructing an under-forest economy sub-region development expert system

Abstract

This invention provides a method for constructing an expert system for regional development of understory economy, relating to the field of forestry monitoring. The method includes: dividing the target forestry area into geographically coded zones based on its geographical characteristics; storing forestry data related to forestry economy within each zone according to the structure of forestry data, establishing a standardized forestry database, which includes at least a forestry text vector database and a forestry relation database; responding to user search operations and interactive operations on the front-end map, converting the search input into search vectors, and obtaining a list of original forestry texts by calculating the semantic similarity between the search vectors and forestry vectors in the forestry text vector database; and visually displaying the forestry data using the list of original forestry texts, the forestry relation database, and the geographical codes from the interactive operations.

Description

Technical Field

[0001] This invention relates to the field of forestry monitoring, and more specifically to a method for constructing an expert system for regional development of understory economy. Background Technology

[0002] In recent years, multidisciplinary, multi-source data integration, and multi-dimensional visual display technologies have emerged one after another. AI technology has sprung up like mushrooms after rain, Web GIS technology has become increasingly mature, and natural language processing technology has been put into production practice. This will inevitably drive the rapid advancement and innovation of various expert systems.

[0003] On the other hand, for expert systems in vertical domains, due to the specialization and particularity of the domain, the experts in the sub-domain are often not information technology experts, and information technology experts are not clear about the knowledge characteristics of the sub-domain. This results in the scarcity of relevant vertical expert systems.

[0004] Therefore, in the context of establishing a broad view of agriculture and food and creating distinctive agricultural industrial clusters, how to construct a knowledge decision-making system in an intuitive, vivid, and scientific way to serve the development of the under-forest economy is an urgent problem to be solved. Summary of the Invention

[0005] In view of the above problems, the present invention provides a method for constructing an expert system for the regional development of forest economy, which provides a one-stop solution for the unified management, intelligent retrieval, and scientific decision-making of the overall development picture and planning knowledge of forest economy in different regions.

[0006] This invention provides a method for constructing an expert system for regional development of understory economy, comprising:

[0007] Based on the geographical characteristics of the target forestry area, the target forestry area is divided into geographically coded areas, and the geographical characteristics include at least topographic features and administrative boundary features.

[0008] Based on the structure of forestry data, store forestry data related to forestry economy in each region, and establish a standardized forestry database. The forestry database shall include at least a forestry text vector database and a forestry relationship database.

[0009] In response to the user's search operation and interaction with the front-end map, the search input is converted into a search vector. By calculating the semantic similarity between the search vector and the forestry vector in the forestry text vector database, a list of original forestry texts is obtained.

[0010] Forestry data is visualized using forestry original text lists, forestry relationship databases, and geocoding in interactive operations.

[0011] According to an embodiment of the present invention, the target forestry area is divided into geographically coded zones based on its geographical characteristics, including:

[0012] Obtain topographic and administrative division data for the target forestry area;

[0013] Based on topographic data, areas with continuous topographic features are divided into candidate regions;

[0014] Based on the administrative boundaries defined by the administrative division data, the candidate areas are divided into zones, and a unique geocode is generated for each zone.

[0015] According to an embodiment of the present invention, based on the structure of forestry data, forestry data related to forestry economy within each region is stored to establish a standardized forestry database, including:

[0016] The tabular and graphic data in forestry data are standardized using a preset template and then stored in the forestry relationship database. The preset template is used to standardize the format of tabular and graphic data according to the needs of forestry data.

[0017] Text data from forestry data is vectorized and stored in a forestry text vector database. The text data includes text identified from files.

[0018] According to an embodiment of the present invention, storing tabular data in forestry data into a forestry relationship database after unifying it using a preset template includes:

[0019] In response to user input of new table data, the table data content is populated into the table according to the table format template defined based on the JSON specification. The table data includes at least a table of the current status of understory economic development and a line graph of understory economic development. The table format template includes at least a regional geographic code, column definition, data type and display style.

[0020] Configure access permissions and publishing scope for table data.

[0021] According to an embodiment of the present invention, the graphic and textual data in forestry data are unified and stored in a forestry relationship database after being processed by a preset template, including:

[0022] Using a rich text editor, text descriptions of specific forestry economic models, images of display scenarios, and related tables are stored and presented according to rich text format templates, forming structured forestry knowledge entries. Forestry economic models include at least forest poultry, forest medicinal herbs, and forest fungi models, and rich text format templates include at least regional geographic codes, economic characteristics, and practices and measures.

[0023] Forestry knowledge entries are stored in the forestry relationship database.

[0024] According to an embodiment of the present invention, the text data in forestry data is vectorized and stored in a forestry text vector database, including:

[0025] Extract text data from a file;

[0026] The content of the text data is identified, and semantic segmentation is performed using the long text extension algorithm to obtain short text segments with complete semantics.

[0027] The text clips are vectorized and normalized using a word embedding model to obtain text vectors;

[0028] The text vectors are stored in a forestry text vector database, and a vector index is created.

[0029] Extracting text data from the file includes:

[0030] For HTML files, regular expression rules are used to remove HTML tags and useless information, resulting in clean text data.

[0031] For flat text files, text data can be read directly through code;

[0032] For other files, a dedicated third-party package is used to read the text data. Other files include at least one of the following formats: Word document, Excel spreadsheet, PowerPoint presentation, and PDF document.

[0033] For non-text content in the document, OCR or convolutional neural networks are used for content recognition to obtain text data.

[0034] According to an embodiment of the present invention, the retrieval input is converted into a retrieval vector, and a list of forestry original texts is obtained by calculating the semantic similarity between the retrieval vector and forestry vectors in a forestry text vector database, including:

[0035] The user's search input is converted into a search vector using a word embedding model, and the search vector is then L2 normalized.

[0036] In the forestry text vector database, the K-nearest neighbor algorithm is used to calculate the semantic similarity between the normalized retrieval vector and the text vector in the database based on the cosine distance, and the most similar preset number of text vectors are retrieved.

[0037] Based on a preset number of text vectors, find the corresponding text clips to form a list of original forestry texts.

[0038] According to embodiments of the present invention, forestry data is visualized using a list of forestry original texts, a forestry relational database, and geocoding in interactive operations, including:

[0039] In response to user interaction with the target area on the front-end map, the system retrieves and displays the image and text data and table indexes of the target area from the forestry relation database based on the target area's geocode.

[0040] In response to a user clicking on an index, the corresponding tabular or graphic data is retrieved from the forestry relation database and displayed.

[0041] In response to a user's search operation for a target area on the front-end map, the original forestry text list will be displayed in list format.

[0042] According to an embodiment of the present invention, the method further includes:

[0043] Constructing a knowledge graph for the understory economy sector, and answering user questions based on the knowledge graph, specifically including:

[0044] Based on forestry data, an ontology for the understory economy is defined. The entity classes of the ontology include at least topographic regions, administrative regions, understory economic models, planting and breeding categories, and operating entities.

[0045] Entities, attributes, and relationships between entities are extracted from structured data to form the first set of triples. For unstructured data, entities, attributes, and relationships between entities are extracted using an entity recognition and relationship extraction model to form the second set of triples.

[0046] The first set of triples and the second set of triples are merged and disambiguated, and the processed triples are stored in a graph database to form a knowledge graph of the understory economy.

[0047] Predefined question-and-answer intents and answer templates related to the understory economy;

[0048] The system receives user input questions, performs intent recognition and entity extraction on the input questions, and generates query statements for the knowledge graph of the understory economy field based on the question and answer intent and entities.

[0049] Execute the query statement, retrieve the query results, and populate the corresponding answer reply template with the query results to generate the answer.

[0050] A second aspect of the present invention provides an expert system for the regional development of understory economy, which can be used to implement the above-mentioned method for constructing the expert system for the regional development of understory economy, including:

[0051] The segmentation module is used to divide the target forestry area into geographically coded areas based on the regional characteristics of the target forestry area. The regional characteristics include at least topographic features and administrative boundary features.

[0052] The storage module is used to store forestry data related to forestry economy in each area according to the structure of forestry data, and to establish a standardized forestry database. The forestry database includes at least a forestry text vector database and a forestry relation database.

[0053] The retrieval module, in response to user retrieval operations and interactive operations on the front-end map, converts the retrieval input into a retrieval vector. By calculating the semantic similarity between the retrieval vector and the forestry vectors in the forestry text vector database, a list of original forestry texts is obtained.

[0054] The display module is used to visualize forestry data by utilizing forestry original text lists, forestry relation databases, and geocoding in interactive operations.

[0055] A third aspect of the present invention provides an electronic device comprising: one or more processors; and a memory for storing one or more programs, wherein, when the one or more programs are executed by the one or more processors, the one or more processors cause the one or more processors to execute the above-described method for constructing an expert system for regional development of understory economy.

[0056] A fourth aspect of the present invention also provides a computer-readable storage medium having executable instructions stored thereon, which, when executed by a processor, cause the processor to perform the above-described method for constructing an expert system for regional development of understory economy.

[0057] According to the construction method of the expert system for regional development of understory economy provided by this invention, a structured database is constructed by organizing and integrating structured and unstructured data such as text, tables, images, and videos. By vectorizing text data and providing a fast retrieval function based on semantic distance, and by integrating Web GIS concepts with HTML5, this system vividly presents relevant knowledge of forestry regional development planning and displays current statistical data in a combination of charts and graphs. Therefore, it at least partially solves the technical problem of low retrieval efficiency caused by unreasonable traditional forestry data management. This allows users to understand the current status and overall planning of understory economic zones in a vivid and intuitive way, providing a one-stop solution for the management and retrieval of knowledge related to regional development and planning of understory economy. Attached Figure Description

[0058] Figure 1 A flowchart illustrating a method for constructing an expert system for regional development of understory economy according to an embodiment of the present invention is shown.

[0059] Figure 2 This schematically illustrates a fast retrieval flowchart based on semantic distance according to an embodiment of the present invention;

[0060] Figure 3This diagram schematically illustrates the system architecture of an expert consultation system constructed according to an embodiment of the present invention.

[0061] Figure 4 This illustration schematically shows a data processing flowchart of an expert consultation system according to an embodiment of the present invention;

[0062] Figure 5 This illustration schematically shows the backend management interface of an expert consultation system according to an embodiment of the present invention;

[0063] Figure 6 This diagram illustrates an example of publishing an arbitrary planar text-based knowledge base in an expert consultation system according to an embodiment of the present invention.

[0064] Figure 7 This diagram illustrates an example of a visual definition of an arbitrary table in an expert consultation system according to an embodiment of the present invention.

[0065] Figure 8 This illustration schematically demonstrates the visual editing of arbitrary tabular data in an expert consultation system according to an embodiment of the present invention;

[0066] Figure 9 This illustration schematically demonstrates the multidimensional display of arbitrary tabular data in an expert consultation system according to an embodiment of the present invention.

[0067] Figure 10 This illustration schematically demonstrates an example of planar knowledge base construction and text segmentation in an expert consultation system according to an embodiment of the present invention;

[0068] Figure 11 This illustration schematically shows the front-end user display interface of an expert consultation system according to an embodiment of the present invention;

[0069] Figure 12 A block diagram of an electronic device suitable for constructing an expert system for regional development of understory economy according to an embodiment of the present invention is shown schematically. Detailed Implementation

[0070] Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of the invention. In the following detailed description, numerous specific details are set forth to provide a thorough understanding of the embodiments of the invention for ease of explanation. However, it will be apparent that one or more embodiments may be practiced without these specific details. Furthermore, descriptions of well-known structures and techniques are omitted in the following description to avoid unnecessarily obscuring the concept of the invention.

[0071] The terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the invention. The terms “comprising,” “including,” etc., as used herein indicate the presence of the stated features, steps, operations, and / or components, but do not exclude the presence or addition of one or more other features, steps, operations, or components.

[0072] All terms used herein (including technical and scientific terms) have the meanings commonly understood by those skilled in the art, unless otherwise defined. It should be noted that the terms used herein are to be interpreted in a manner consistent with the context of this specification, and not in an idealized or overly rigid way.

[0073] When using expressions such as "at least one of A, B and C", they should generally be interpreted in accordance with the meaning that is commonly understood by those skilled in the art (e.g., "a system having at least one of A, B and C" should include, but is not limited to, a system having A alone, a system having B alone, a system having C alone, a system having A and B, a system having A and C, a system having B and C, and / or a system having A, B and C, etc.).

[0074] The development of expert systems can be roughly divided into three stages: the initial stage (1920s-1970s), characterized by high specialization and strong ability to solve problems in specific domains, but with incomplete system structure, poor portability, and lack of interpretability; the mature stage (1970s-1980s), characterized by single-discipline specialized systems with complete structure, improved portability, the introduction of heuristic reasoning and imprecise reasoning, and support for human-computer interaction; and the development stage (1980s to present), characterized by multi-disciplinary integrated systems that employ various knowledge representation methods, separate the inference engine from the knowledge base, and support dynamic integrated databases and knowledge acquisition.

[0075] To further improve the intelligent management of knowledge, documents, and other data related to the regional development of understory economy and lower the threshold for system and data use, this invention provides a method for constructing an expert system for the regional development of understory economy, which can be further used for the construction of a smart expert system for the regional development of understory economy.

[0076] Figure 1 A flowchart illustrating a method for constructing an expert system for regional development of understory economy according to an embodiment of the present invention is shown, such as... Figure 1As shown, embodiments of the present invention provide a method for constructing an expert system for regional development of understory economy, comprising: dividing the target forestry area into geographically coded areas based on the geographical characteristics of the target forestry area, wherein the geographical characteristics include at least topographic features and administrative boundary features; storing forestry data related to forestry economy within each area according to the structure of forestry data, and establishing a standardized forestry database, wherein the forestry database includes at least a forestry text vector database and a forestry relation database; responding to user search operations and interactive operations on the front-end map, converting the search input into search vectors, and obtaining a list of forestry original texts by calculating the semantic similarity between the search vectors and forestry vectors in the forestry text vector database; and visually displaying the forestry data using the list of forestry original texts, the forestry relation database, and the geographical codes in the interactive operations.

[0077] It should be noted that the forestry data refers to forestry data related to understory economy.

[0078] Through the embodiments of this invention, a human-computer interaction interface design is adopted, which is easy to operate, highly scalable and compatible, and can adapt to multi-source data that may appear in the future. Web GIS is used as the main line for data publishing, combined with HTML5 and LayUI technologies, to achieve a good human-computer interaction effect for knowledge publishing and integrated knowledge management for the regional development of under-forest economy. It includes the comprehensive application of knowledge such as information management, GIS, HTML5, algorithms, and AI, to realize a comprehensive presentation of the regional development of under-forest economy, and ultimately empower intelligent scientific decision-making, contributing to the development of under-forest economy.

[0079] Based on the above embodiments, the target forestry area is divided into geographically coded areas according to its geographical characteristics. This includes: acquiring topographic data and administrative division data of the target forestry area; dividing areas with continuous topographic features into candidate areas based on the topographic data; dividing the candidate areas into sub-areas based on the administrative boundaries defined by the administrative division data, thereby obtaining the areas, and generating a unique geographical code for each area.

[0080] The geocoding has at least two identifiers: the first identifier represents a candidate area divided into continuous regions based on topographic features, and the second identifier represents an area further divided according to administrative divisions.

[0081] An administrative region may contain various discontinuous terrains; for example, a county may contain mountains, hills, and plains simultaneously. In practice, the primary classification of a region should be determined based on the dominant terrain features or the similarity of its understory economic model. For instance, if a county is mostly mountainous with a small portion of plains, it will be classified as a candidate area for the corresponding mountainous region.

[0082] Through the embodiments of the present invention, topographic and administrative division data are obtained, and candidate areas are divided based on the continuity of topography. Then, adjustments and codes are made in combination with administrative boundaries, which takes into account both the continuity of natural geographical features and the reality of administrative management units. This solves the problems of management inconvenience or planning that are not in line with reality caused by dividing areas solely according to administrative divisions or purely natural landforms. Each area is assigned a unique geographic code, which lays the foundation for the spatial correlation and accurate display of subsequent data.

[0083] Based on the above embodiments, according to the structure of forestry data, forestry data related to forestry economy in each area is stored to establish a standardized forestry database, including: storing tabular data and graphic data in forestry data into a forestry relationship database after unifying them through a preset template, the preset template being used to unify the format of tabular data and graphic data according to forestry data requirements; and storing text data in forestry data into a forestry text vector database after vectorization, the text data including text identified from files.

[0084] That is, it includes structured information processing and unstructured text data vector processing.

[0085] The former uses the MVC pattern, RESTful as the technical standard, and JSON as the format to achieve structured data input and output via HTTP standardization. Considering the system's scalability and future adaptability, the system innovatively adopts the JSON specification to customize the table format and published content, and stipulates that the column data types are text, date, and list types, facilitating subsequent data collection. These three types basically meet the needs of most application scenarios.

[0086] The latter uses text stored in relational databases and uploaded file data in formats such as txt, pdf, word, excel, PowerPoint, video, and images as raw materials for content recognition, semantic segmentation, and vectorization. The vectorization results are stored in the FAISS database, and a vector index is built for subsequent rapid retrieval of information based on semantic distance. Unstructured data is stored in the relational database as HTML or attachments.

[0087] Through the embodiments of the present invention, the data is stored in relational databases and vectorized data in vector databases according to different characteristics. The most suitable storage and indexing methods are selected for different types of data, which solves the problem that forestry data is difficult to manage and utilize in a unified and efficient manner due to its diverse formats and different structures. The standardized storage and management of multi-source forestry data is realized.

[0088] Based on the above embodiments, the tabular data in the forestry data is uniformly stored in the forestry relationship database after being processed by a preset template. This includes: responding to the user input of newly added tabular data content, filling the tabular data content into the table according to the tabular data format template defined based on the JSON specification. The tabular data includes at least a table of the current status of under-forest economic development and a line graph of under-forest economic development. The tabular format template includes at least a regional geographic code, column definition, data type and display style; and setting the access permissions and publishing scope of the tabular data.

[0089] In this embodiment, Figure 8 This illustration schematically demonstrates the visual editing of arbitrary tabular data in an expert consultation system according to an embodiment of the present invention, such as... Figure 8 As shown, the format template of any two-dimensional table can be customized to achieve arbitrary adaptation of columns and data types. That is, through the JSON technology standard, the format template, data content, display style and chart analysis of any table can be maintained and managed in a visual or text code manner, and access permissions and publishing scope can be verified.

[0090] Specifically, the system manages tabular data through a JSON-based, customizable template mechanism. Taking the statistical table of the under-forest economy development status in the Qinba Mountains of Northwest Hubei (geographic code HB-01) as an example, the administrator first defines a JSON-formatted table template in the backend. This template explicitly defines the table structure (including fields such as county / city name, main under-forest economic models, annual output value, and their data types), display style, and associated line charts. When users enter specific data through the data maintenance interface, the system converts the data into structured records based on this template. These records, associated with the template information and the area's geographic code, are then stored in a standardized PostgreSQL relational database. Simultaneously, the access permissions and publishing scope set in the template control the visibility and operability of the data. On the front end, when users click on the corresponding area on the map or access it through the menu, the system dynamically renders the table and automatically generates visual charts such as annual output value trend graphs based on the data.

[0091] Through embodiments of the present invention, the format template, column definition, data type and display style of the table can be flexibly defined using a visual or code-based method based on the JSON specification. Due to the flexibility and scalability of the JSON template, the system can adapt to various types of two-dimensional table data that may appear in the future, realizing convenient customization, maintenance and publishing of table data such as the current status and development trend of forestry economy, and supporting flexible access control.

[0092] Based on the above embodiments, the text and image data in the forestry data are unified through a preset template and stored in the forestry relationship database. This includes: using a rich text editor, storing text descriptions of specific forestry economic models, images of display scenarios, and related tables according to rich text format templates to form structured forestry knowledge entries. The forestry economic models include at least forest poultry, forest medicinal herbs, and forest fungi models. The rich text format templates include at least regional geographic codes, economic characteristics, and practices and measures. The forestry knowledge entries are then stored in the forestry relationship database.

[0093] In this embodiment, the image and text data is managed by integrating a self-developed asynchronous web file management module with the open-source rich text editor UEditor to facilitate the management of various media styles such as images, text, and tables. This unit enables flexible expansion of the managed content and management type, and finally stores the formatted and sorted data content into the open-source PostgreSQL relational database.

[0094] Through the embodiments of the present invention, textual and graphic knowledge entries are managed in a structured manner according to a template containing fields such as geographic codes and economic characteristics. Unstructured textual and graphic descriptions are organized and managed, solving the problem that textual and graphic knowledge is not easy to associate and retrieve, and facilitating association, calling and display with map regions.

[0095] Based on the above embodiments, the text data in the forestry data is vectorized and stored in the forestry text vector database, including: extracting text data from files; identifying the content of the text data, performing semantic segmentation using a long text extension algorithm to obtain short text segments with complete semantics; using a word embedding model to vectorize and normalize the short text segments to obtain text vectors; storing the text vectors in the forestry text vector database and establishing a vector index; wherein, extracting text data from files includes: for HTML files, removing HTML tags and useless information using regular expression rules to clean the text data; for flat text files, directly reading the text data through code; for other files, calling a dedicated third-party package to read the text data, other files including at least one of Word documents, Excel spreadsheets, PowerPoint presentations, and PDF documents; for non-text content in the files, using OCR or convolutional neural networks to perform content recognition and obtain text data.

[0096] In some feasible approaches, various types of target data need to be preprocessed before vectorization. Preprocessing includes functions such as removing duplicate data, handling missing values, handling outliers, and data format conversion to ensure that the converted data can be easily provided by subsequent vectorization processing units. The preprocessed target data is then fused according to standards to obtain standardized and usable data. Based on structured data, indexing, backup, recovery, and access control are performed to lay the foundation for further data utilization.

[0097] In this embodiment, the text content is extracted from the original text data, which includes:

[0098] (1) HTML to text content conversion unit: For HTML data stored in PostgreSQL, HTML tags and useless information are removed using regular expressions and other rules to clean up the text content.

[0099] (2) The document to text content conversion unit reads the text content directly through code for flat text such as txt; for files such as word, excel, PowerPoint, and pdf, it calls a dedicated third-party package to read the content; for non-text content such as images that may exist in the document, it uses OCR or convolutional neural network AI model to recognize the content and finally obtain the text content.

[0100] In this embodiment, data includes input / output standardization and knowledge vectorization, enabling the input and output of data or knowledge according to technical specifications. Input / output standardization employs the MVC development model based on the .NET Core technology roadmap and the RESTful technical standard for formatted data input and output; if the output is a data type, JSON format data is returned. Knowledge vectorization uses natural language processing principles to vectorize text. This embodiment uses bge-large-zh-v1.5 for vectorization, performs L2 normalization on the vectorization results, and finally stores the vectorization results in the FAISS database for semantic retrieval.

[0101] In this embodiment, vectorization specifically includes:

[0102] (1) Processing textual content to obtain semantic short clips that can reflect the local semantic features of the original text, which include:

[0103] The text content is converted into semantic short pieces. However, the text content obtained from the file-to-text conversion unit is often too long, leading to problems such as slow retrieval speed, large semantic distances, large matching scales, and neglecting latent semantics in subsequent semantic retrieval. Therefore, it is necessary to segment the complete text content into multiple semantic short pieces of varying lengths. These short pieces become the raw materials for subsequent semantic retrieval. The segmentation algorithm in this system uses a fixed-length text extension method, as shown in the following formula:

[0104]

[0105] In the formula Let `PaddingLeft` and `PaddingRight` be the i-th fixed-length segment of the original full text, representing the left and right extensions, respectively. Based on text features, `PaddingLeft` or `PaddingRight` extends left and right word by word or character by character until punctuation marks appear. Punctuation marks can be commas, periods, semicolons, quotation marks, question marks, colons, etc., in both Chinese and English. This algorithm ensures that the final segment contains clauses with complete semantic meaning, thus guaranteeing semantic integrity. Furthermore, since there may be local overlap between segments within the same context, this algorithm ensures that no semantic elements are omitted.

[0106] Based on the general fixed-length text extension method, this paper introduces boundary recognition rules for forestry entities and key phrases to ensure that the segmentation can maintain the semantic integrity of forestry terminology, regional names, and economic patterns to the greatest extent possible. The optimized algorithm is shown in the following equation:

[0107]

[0108] in, This refers to the i-th initial text block of the original forestry document, divided according to a preset length. This represents a specific set of boundary rules for the forestry sector. The generation of this set is based on domain knowledge extracted from the forestry database, including but not limited to: geographical entities, such as the names of regions and administrative divisions such as the Qinba Mountains in northwestern Hubei, Fang County, and Shennongjia Forestry District; understory economic models, such as the forest medicinal herb model, the forest fungus model, and the forest poultry model; planting and breeding categories, such as Polygonatum, Coptis, shiitake mushrooms, and Chinese honeybees; and operating entities and brands, such as professional cooperatives, Wudang Seven Flavors, and National Reserve Forest+. ( ) is the extension function, and its operation rule is as follows: Starting from this point, extend character by character in both the left and right directions. The extension process stops when it encounters the following priority boundaries: regular punctuation marks, matched... The boundaries of forestry entities or key phrases defined in the document are used to ensure that the integrity of such professional concepts is not disrupted.

[0109] (2) A short video vectorization processing unit, which includes:

[0110] For STexti vectorization, following the general natural language processing approach, the tokenizer (segmenter) is used for digitization first, followed by word embedding, and finally L2 normalization. This system compared various word embedding components and selected the bge-large-zh-v1.5 model, which is more suitable for Chinese semantic scenarios, as the word embedding model.

[0111] (3) Vector storage unit: The vector data of the above short film is stored in the FAISS database in combination with the API provided by FAISS (Facebook AISimilarity Search).

[0112] Through embodiments of this invention, text is extracted from files of various formats, semantically segmented using a fixed-length text extension algorithm, vectorized and normalized using a specific word embedding model (bge-large-zh-v1.5), and finally stored in a vector database (FAISS) and indexed. Natural language processing is employed to process textual data, segmenting long texts semantically and establishing vector storage based on cosine distance in the FAISS database. This ensures the accuracy of knowledge retrieval, solves the problems of low efficiency in direct retrieval of long texts, inaccurate semantic matching, and difficulty in extracting content from documents of different formats, and achieves efficient semantic retrieval of massive amounts of unstructured data such as forestry documents.

[0113] Figure 2 A flowchart illustrating a fast retrieval process based on semantic distance according to an embodiment of the present invention is shown, such as... Figure 2 As shown, the search input is converted into a search vector. By calculating the semantic similarity between the search vector and forestry vectors in the forestry text vector database, a list of original forestry texts is obtained. This includes: using a word embedding model to convert the user's search input into a search vector, and L2 normalizing the search vector; in the forestry text vector database, using the K-nearest neighbor algorithm, calculating the semantic similarity between the normalized search vector and the text vectors in the database based on cosine distance, and retrieving the most similar preset number of text vectors; and based on the preset number of text vectors, finding the corresponding text clips to form a list of original forestry texts.

[0114] In this embodiment, a fast retrieval function based on semantic distance is provided. The user enters search keywords or phrases in a form. The system calls the backend to receive a POST request, embeds the input keywords or phrases into vectors using the bge-large-zh-v1.5 model, performs L2 normalization, and then applies the KNN algorithm to retrieve the Top K most similar vectors from the FAISS database. Based on the retrieved vector list, the system reverse-engineers the short video sequence and the corresponding original document list, ultimately retrieving the corresponding knowledge document list and providing it back to the user.

[0115] The KNN retrieval unit completes the Top K sequence solution. This unit adopts the classic KNN algorithm idea and retrieves the similar Top K sequences from the FAISS database. The algorithm process is described as follows: preprocess the input text; calculate the distance from the input sample to each sample in the knowledge base; sort each distance, and then select the K vectors with the smallest distance as the output.

[0116]

[0117] The formula for calculating the cosine distance of a short film is shown above, where the short film vector SText a =(a1,a2,a3,…,a n ), SText b =(b1,b2,b3,…,b n ), where a i and b i The smaller the cosθ value, the better the short video SText is. a and SText b The closer the distance, the better. FAISS does not directly support cosine distance calculation; by default, it supports IndexFlatIP or IndexIVFFlat. In this application, in order to use cosine distance to solve semantic similarity, SText... a and SText b L2 normalization was performed, and the calculation formula is as follows:

[0118]

[0119] Where x is a text vector.

[0120] Through the embodiments of this invention, user search keywords are also vectorized, and the cosine similarity is calculated in the vector database using the K-nearest neighbor algorithm to find the top K most similar text vectors. Then, the original text is located in reverse, and similarity calculation is performed in a unified vector space (after L2 normalization), rather than simple keyword matching. This solves the problem of missed detection and false detection caused by differences in word expression in traditional keyword retrieval, and realizes accurate and fast knowledge retrieval based on semantic content. Users can ask questions in natural language, and the system can understand their intentions and find relevant content.

[0121] Based on the above embodiments, forestry data is visualized using a forestry original text list, a forestry relation database, and geocoding in interactive operations. This includes: responding to a user's interactive operation on a target area on the front-end map, retrieving and displaying the image and text data and table index of the target area from the forestry relation database according to the geocoding of the target area; responding to a user clicking on an index, retrieving and displaying the corresponding table data or image and text data from the forestry relation database; and responding to a user's search operation on a target area on the front-end map, displaying the forestry original text list in list form.

[0122] This multi-dimensional display module integrates technologies such as HTML5, Web GIS, LayUI, eChart, and flexible templates to achieve a user-friendly and efficient display of multi-source knowledge, including text and structured data, on the regional development of Hubei Province's under-forest economy. The integration of HTML5 technology enhances the human-computer interaction interface; the system's design utilizes HTML5 as the foundation for a responsive layout. Web GIS technology is also integrated, providing a clear and intuitive display of Hubei Province's under-forest economy by region. The system frontend uses an MVC pattern to call the backend API, reading geographic data from counties and districts. Combined with LayUI frontend components, it clearly displays the number of under-forest economy zones in Hubei Province, the cities and counties included in each zone, regional characteristics, county and city under-forest economy models, under-forest planting (breeding) categories, and key practices and measures implemented by counties and cities. The system adapts to the client's monitor resolution and device media type, delivering optimal rendering effects for the under-forest economy development plan while maintaining high efficiency and stability in performance. Combining LayUI and custom template concepts, it can provide post-customization capabilities for publishing arbitrarily adapted 2D table data and transforming 2D table data into chart displays, based on subsequent display requirements.

[0123] Through embodiments of this invention, using Web GIS concepts as the main framework and integrating UI technologies such as HTML5 and LayUI, a vivid knowledge display effect is achieved. This embodiment can intuitively and clearly display the regional distribution of under-forest economy in Hubei Province in a map format. When the mouse hovers over a county or district, it can promptly and comprehensively display, in a multi-dimensional manner combining text, images, audio, and video, the under-forest economy zone to which that county or district belongs, its regional characteristics, suitable under-forest economic varieties for development, and the measures and policies adopted. Structured tables can be displayed through backend definition, asynchronous frontend display, and a combination of charts. Document-type data can be displayed through remote loading using HTML or iFrame.

[0124] Based on the above embodiments, the method further includes: constructing a knowledge graph for the understory economy field, and answering user questions based on the knowledge graph. Specifically, this includes: defining an ontology for the understory economy field based on forestry data, wherein the entity classes of the ontology include at least topographic regions, administrative regions, understory economic models, planting and breeding categories, and operating entities; extracting entities, attributes, and relationships between entities from structured data to form a first set of triples; extracting entities, attributes, and relationships between entities from unstructured data using an entity recognition and relationship extraction model to form a second set of triples; fusing and disambiguating the first and second sets of triples, and storing the processed triples in a graph database to form a knowledge graph for the understory economy field; predefining question-and-answer intents and answer response templates related to the understory economy field; receiving user input questions, performing intent recognition and entity extraction on the input questions, and generating a query statement for the knowledge graph of the understory economy field based on the question-and-answer intent and entities; executing the query statement, obtaining the query results, and filling the query results into the corresponding answer response template to generate the answer.

[0125] In this embodiment, the system first defines core entity classes and their relationships based on forestry data, including topographic regions, administrative regions, understory economic models, planting and breeding categories, and operating entities, forming a domain ontology. Subsequently, the system extracts entities, attributes, and relationships from structured tabular data (such as regional division tables and county / district detail tables), for example, "belongs to," "located in," "suitable for development," "specific category is," "main practice is," and "operated by," forming the first set of triplets. Simultaneously, using an entity recognition and relationship extraction model, the system extracts the second set of triplets from unstructured policy documents, technical reports, and other texts. For example, from the text "Zhuxi County relies on the 'Wudang Seven Flavors' brand to vigorously develop authentic medicinal herbs such as Coptis chinensis and Polygonatum sibiricum," relationships such as "Zhuxi County, brand-driven, Wudang Seven Flavors" and "Coptis chinensis, is, authentic medicinal herb" can be extracted. Next, the system merges and disambiguates these two sets of triplets. For example, if both structured and unstructured data mention Zhuxi County and Polygonatum sibiricum, the system will identify them as the same entity and associate them with all relevant information. The processed and interconnected knowledge is stored in a graph database, thereby constructing a structured knowledge graph for the understory economy.

[0126] When a user asks a question, such as "What under-forest economic models are suitable for development in the Wuling Mountains of southwestern Hubei?", the system will first identify the user's question and answer intent and extract key entities, and then generate a query statement targeting the knowledge graph. After executing the query, the system can accurately locate the entity "Wuling Mountains in Southwest Hubei" from the map and deduce a series of related answers such as "belongs to," "suitable for development," and "specific categories" along relational paths such as "Zhuxi County," "forest-medicinal herb model," and "Polygonatum sibiricum, Coptis chinensis, and Gastrodia elata." Finally, the results are filled into a predefined answer template ({region_name} is mainly suitable for developing the following understory economic models: {model_list}. For example, in {county_example}, the main development is {model_example}, and the specific categories include {category_example}), generating a clearly structured and direct answer such as "The Wuling Mountains in Southwest Hubei are mainly suitable for developing the following understory economic models: forest-medicinal herb model and forest-fungus model. For example, in Zhuxi County, the main development is the forest-medicinal herb model, with specific categories including Polygonatum sibiricum, Coptis chinensis, and Gastrodia elata; it also develops the forest-fungus model, planting shiitake mushrooms and morels. In addition, Fang County also focuses on developing the forest-fungus model, with the main categories being shiitake mushrooms and Hericium erinaceus." This achieves precise question-and-answering for complex forestry issues.

[0127] Through embodiments of the present invention, by constructing a knowledge graph in the field of understory economy, entities and relationships are extracted from structured and unstructured data, and user questions are answered based on the graph and predefined question-and-answer templates, thus accurately answering complex forestry questions.

[0128] Based on the same inventive concept, this invention also provides an expert system for the regional development of understory economy, which can be used to implement the above-mentioned method for constructing the expert system for the regional development of understory economy, including: a division module, used to divide the target forestry area into geographically coded areas according to the regional characteristics of the target forestry area, the regional characteristics including at least topographic features and administrative boundary features; a storage module, used to store forestry data related to forestry economy in each area according to the structure of forestry data, and establish a standardized forestry database, the forestry database including at least a forestry text vector database and a forestry relation database; a retrieval module, used to convert the retrieval input into a retrieval vector in response to the user's retrieval operation and interactive operation on the front-end map, and obtain a list of forestry original texts by calculating the semantic similarity between the retrieval vector and the forestry vectors in the forestry text vector database; and a display module, used to visualize the forestry data using the list of forestry original texts, the forestry relation database, and the geographical codes in the interactive operation.

[0129] Taking Hubei Province as an example, the system development and installation, such as Figure 3 , Figure 4 The architecture and data flow are designed accordingly. The backend function organization and information management adopt the LayUI interface style, and related functions are organized using a tree-like menu, such as... Figure 5 As shown. The functional modules are described as follows: The system includes three main aspects: a multi-source data management module, which organizes and integrates structured and unstructured data such as text, tables, images, and videos, ultimately providing a unified management interface for disorganized multi-source data, offering an effective way for current data archiving, future data storage, and comprehensive knowledge utilization; a processing module, which standardizes the constructed structured knowledge base, using the currently most widely accepted RESTful development paradigm to realize the input and output of knowledge and data, vectorizes text-type data, and adopts a fast retrieval function based on semantic distance calculation to improve data accuracy; and a multi-dimensional display module, based on the currently popular HTML5, integrating Web GIS concepts to vividly display regional development and planning knowledge, and displaying current statistical data in any combination of maps and tables. In summary, this application provides a one-stop solution for the unified management and intuitive display of data such as the current status and planning of forest under-forest economy in Hubei Province, providing technical support for knowledge management, intelligent retrieval, and scientific decision-making in the regional development and planning of forest under-forest economy.

[0130] Based on the analysis and research of the regional characteristics of counties and districts in Hubei Province, the under-forest economy in Hubei Province is divided into six preliminary regions: the Qinba Mountains in northwestern Hubei, the Wuling Mountains in southwestern Hubei, the hilly areas in northern Hubei, the Dabie Mountains in northeastern Hubei, the Mufu Mountains in southeastern Hubei, and the hilly plains along the Yangtze River. Their characteristics are shown in Table 1.

[0131] Table 1

[0132]

[0133]

[0134]

[0135] All counties and districts are divided into zones based on their characteristics, as shown in Table 2. The characteristics of the understory economy in each county and district are maintained. For example, the Qinba Mountains in northwestern Hubei Province include Maojian District, Zhangwan District, Yunyang District, Yunxi County, Zhushan County, Zhuxi County, Fang County, and Danjiangkou City in Shiyan City; Nanzhang County, Baokang County, and Gucheng County in Xiangyang City; Xiling District, Wujiagang District, Xiaoting District, Dianjun District, Yiling District, Yuan'an County, and Xingshan County in Yichang City; and Shennongjia Forestry District, etc. Some characteristics of this region are described in Table 2 (not all regions). By clicking the "County and District Management" link in the function menu, settings such as the county and district directory, zones, characteristics of the understory economy in each county and district, and practices and measures can be configured. The management interface is as follows: Figure 6 As shown.

[0136] Table 2

[0137]

[0138]

[0139]

[0140]

[0141]

[0142]

[0143] Any Table: This feature allows you to maintain and publish tables related to the understory economy in Hubei Province. You can access the data publishing directory by clicking the "Any Table" link in the function menu. Figure 5 As shown; in the data publishing directory, click the "Add" button, and enter the name of the data publishing in the pop-up dialog box to create the data publishing directory; in the directory list, click the "Edit" button to maintain the data format definition and data content management in JSON format; click the "Edit Column" or "Edit Data" button in the data publishing directory to start visual column or data content editing, such as... Figure 7 As shown; click the "Template Edit" button in the data publishing directory to start the data template, display template, and icon template settings.

[0144] Knowledge System Management: The knowledge system is a knowledge base function built upon the maintained and organized knowledge data of the expert consultation system. This function is activated by clicking the "Knowledge System" link in the function menu. Clicking the "New" button in the knowledge base directory triggers the knowledge base upload function. When the user selects the corresponding knowledge base or the corresponding county / district characteristic text, the system asynchronously triggers the background API to complete a series of actions, including file-to-text conversion, automatic text segmentation into semantic fragments, short-text vectorization, L2 normalization of vectorization, and vector storage. Ultimately, this achieves the entire process from file to vector index. The relevant algorithms and processes are described earlier in this application, and the final effect is as follows: Figure 10 As shown ( Figure 10 This is only a schematic diagram of a partial knowledge base construction, not the entire knowledge base.

[0145] Multi-dimensional display module: The system will be based on HTML5 technology, combined with LayUI and Web GIS to display relevant knowledge about the development of forest understory areas, such as... Figure 11 As shown. In Figure 11 In the interface, data knowledge, policies and regulations, technical standards, and results are displayed in a fixed window on the left. When a link in the data knowledge is clicked, the system displays any table data published in the backend in a floating window. The data is displayed in tables and charts according to the display template and chart objectives. Figure 9 As shown; for policies, regulations, technical standards, and results materials, in addition to displaying the full textual knowledge, semantic retrieval is also supported, using algorithms such as the KNN algorithm described above; for the dissemination of knowledge such as regional features and zoning, a Web GIS technology approach is adopted, such as... Figure 11 In the middle section, when the user's mouse hovers over the corresponding district or county, the information, including its regional characteristics, understory economic model, and the main practices and measures for understory planting (breeding), is displayed in a ToopTip format with both text and images.

[0146] Figure 12 A block diagram of an electronic device suitable for constructing an expert system for regional development of understory economy according to an embodiment of the present invention is shown schematically.

[0147] like Figure 12As shown, an electronic device 1200 according to an embodiment of the present invention includes a processor 1201, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 1202 or a program loaded from a storage portion 1208 into a random access memory (RAM) 1203. The processor 1201 may include, for example, a general-purpose microprocessor (e.g., a CPU), an instruction set processor and / or an associated chipset and / or a special-purpose microprocessor (e.g., an application-specific integrated circuit (ASIC)), etc. The processor 1201 may also include onboard memory for caching purposes. The processor 1201 may include a single processing unit or multiple processing units for performing different actions of the method flow according to an embodiment of the present invention.

[0148] RAM 1203 stores various programs and data required for the operation of electronic device 1200. Processor 1201, ROM 1202, and RAM 1203 are interconnected via bus 1204. Processor 1201 executes various operations of the method flow according to embodiments of the present invention by executing programs in ROM 1202 and / or RAM 1203. It should be noted that the programs may also be stored in one or more memories other than ROM 1202 and RAM 1203. Processor 1201 may also execute various operations of the method flow according to embodiments of the present invention by executing programs stored in said one or more memories.

[0149] According to an embodiment of the present invention, the electronic device 1200 may further include an input / output (I / O) interface 1205, which is also connected to the bus 1204. The electronic device 1200 may also include one or more of the following components connected to the I / O interface 1205: an input section 1206 including a keyboard, mouse, etc.; an output section 1207 including a cathode ray tube (CRT), liquid crystal display (LCD), etc., and a speaker, etc.; a storage section 1208 including a hard disk, etc.; and a communication section 1209 including a network interface card such as a LAN card, modem, etc. The communication section 1209 performs communication processing via a network such as the Internet. A drive 1210 is also connected to the I / O interface 1205 as needed. A removable medium 1211, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., is installed on the drive 1210 as needed so that computer programs read from it can be installed into the storage section 1208 as needed.

[0150] The present invention also provides a computer-readable storage medium, which may be included in the device / apparatus / system described in the above embodiments; or it may exist independently and not assembled into the device / apparatus / system. The computer-readable storage medium carries one or more programs, which, when executed, implement the method according to the embodiments of the present invention.

[0151] According to embodiments of the present invention, a computer-readable storage medium may be a non-volatile computer-readable storage medium, such as including, but not limited to: portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In the present invention, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. For example, according to embodiments of the present invention, a computer-readable storage medium may include ROM 1202 and / or RAM 1203 and / or one or more memories other than ROM 1202 and RAM 1203 described above.

[0152] Embodiments of the present invention also include a computer program product comprising a computer program containing program code for performing the methods shown in the flowchart. When the computer program product is run on a computer system, the program code is used to cause the computer system to implement the methods provided in the embodiments of the present invention.

[0153] When the computer program is executed by the processor 1201, it performs the functions defined in the system / apparatus of this invention. According to embodiments of the invention, the systems, apparatuses, modules, units, etc., described above can be implemented by computer program modules.

[0154] In one embodiment, the computer program may rely on a tangible storage medium such as an optical storage device or a magnetic storage device. In another embodiment, the computer program may also be transmitted and distributed in the form of signals over a network medium, and may be downloaded and installed via the communication section 1209, and / or installed from the removable medium 1211. The program code contained in the computer program can be transmitted using any suitable network medium, including but not limited to: wireless, wired, etc., or any suitable combination thereof.

[0155] In such an embodiment, the computer program can be downloaded and installed from a network via the communication section 1209, and / or installed from the removable medium 1211. When the computer program is executed by the processor 1201, it performs the functions defined in the system of this embodiment of the invention. According to embodiments of the invention, the systems, devices, apparatuses, modules, units, etc., described above can be implemented by computer program modules.

[0156] According to embodiments of the present invention, program code for executing the computer programs provided in the embodiments of the present invention can be written in any combination of one or more programming languages. Specifically, these computational programs can be implemented using high-level procedural and / or object-oriented programming languages, and / or assembly / machine languages. Programming languages ​​include, but are not limited to, languages ​​such as Java, C++, Python, "C", or similar programming languages. The program code can be executed entirely on the user's computing device, partially on the user's device, partially on a remote computing device, or entirely on a remote computing device or server. In cases involving remote computing devices, the remote computing device can be connected to the user's computing device via any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computing device (e.g., via the Internet using an Internet service provider).

[0157] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of the present invention. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram or flowchart, and combinations of blocks in a block diagram or flowchart, may be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0158] Those skilled in the art will understand that the features described in the various embodiments and / or claims of the present invention can be combined or combined in various ways, even if such combinations or combinations are not explicitly described in the present invention. In particular, the features described in the various embodiments and / or claims of the present invention can be combined or combined in various ways without departing from the spirit and teachings of the present invention. All such combinations and / or combinations fall within the scope of the present invention.

[0159] The embodiments of the present invention have been described above. However, these embodiments are merely illustrative and not intended to limit the scope of the invention. Although various embodiments have been described above, this does not mean that the measures in the various embodiments cannot be used advantageously in combination. The scope of the invention is defined by the appended claims and their equivalents. Various substitutions and modifications can be made by those skilled in the art without departing from the scope of the invention, and all such substitutions and modifications should fall within the scope of the invention.

Claims

1. A method for constructing an expert system for regional development of understory economy, characterized in that, include: Based on the geographical characteristics of the target forestry area, the target forestry area is divided into geographically coded areas, and the geographical characteristics include at least topographic features and administrative boundary features; Based on the structure of forestry data, forestry data related to forestry economy in each region is stored, and a standardized forestry database is established. The forestry database includes at least a forestry text vector database and a forestry relation database. The forestry text vector database stores vectorized text data. The forestry relation database is used to store at least structured tabular data and unstructured graphic data. In response to the user's search operation and interaction with the front-end map, the search input is converted into a search vector. By calculating the semantic similarity between the search vector and the forestry vectors in the forestry text vector database, a list of original forestry texts is obtained. The forestry data is visualized using the forestry original text list, the forestry relationship database, and the geocoding in the interactive operation.

2. The method according to claim 1, wherein, The process of dividing the target forestry area into geographically coded zones based on its geographical characteristics includes: Obtain the topographic data and administrative division data of the target forestry area; Based on the terrain data, areas with continuous terrain features are divided into candidate regions; Based on the administrative boundaries defined by the administrative division data, the candidate areas are divided into zones to obtain zones, and a unique geographic code is generated for each zone.

3. The method according to claim 1, wherein, The process of storing forestry data related to the forestry economy within each region, based on the structure of forestry data, and establishing a standardized forestry database includes: The tabular and graphic data in forestry data are unified by a preset template and then stored in the forestry relationship database. The preset template is used to unify the format of the tabular and graphic data according to the needs of forestry data. Text data from forestry data is vectorized and stored in a forestry text vector database. The text data includes text identified from files.

4. The method according to claim 3, wherein, The tabular data in the forestry data is standardized using a preset template and then stored in the forestry relationship database, including: In response to user input of new table data, the table data content is filled into the table according to the table format template defined based on the JSON specification. The table data includes at least a table of the current status of under-forest economic development and a line graph of under-forest economic development. The table format template includes at least a regional geographic code, column definition, data type and display style. Configure access permissions and publishing scope for the table data.

5. The method according to claim 3, wherein, The text and image data in forestry data are standardized using a preset template and then stored in the forestry relationship database, including: Using a rich text editor, text descriptions of specific forestry economic models, images of display scenarios, and related tables are stored and displayed according to rich text format templates to form structured forestry knowledge entries. The forestry economic models include at least forest poultry, forest medicinal herbs, and forest fungi models, and the rich text format templates include at least regional geographic codes, economic characteristics, and practices and measures. The forestry knowledge entries are stored in the forestry relationship database.

6. The method according to claim 3, wherein, The process of vectorizing text data in forestry data and storing it in a forestry text vector database includes: Extract text data from a file; The content of the text data is identified, and semantic segmentation is performed using a long text extension algorithm and forestry boundary recognition rules to obtain short text segments with complete semantics. The forestry boundary recognition rules include at least entities in the forestry field. The text clips are vectorized and normalized using a word embedding model to obtain text vectors; The text vectors are stored in the forestry text vector database, and a vector index is established. Extracting text data from the file includes: For HTML files, regular expression rules are used to remove HTML tags and useless information, resulting in clean text data. For flat text files, text data can be read directly through code; For other files, a dedicated third-party package is used to read the text data. These other files include at least one of the following formats: Word documents, Excel spreadsheets, PowerPoint presentations, and PDF documents. For non-text content in the document, OCR or convolutional neural networks are used for content recognition to obtain text data.

7. The method according to claim 1, wherein, The process involves converting the retrieval input into a retrieval vector, and then calculating the semantic similarity between the retrieval vector and forestry vectors in the forestry text vector database to obtain a list of original forestry texts, including: The user's search input is converted into a search vector using a word embedding model, and the search vector is then L2 normalized. In the forestry text vector database, the K-nearest neighbor algorithm is used to calculate the semantic similarity between the normalized retrieval vector and the text vector in the database based on the cosine distance, and the most similar preset number of text vectors are retrieved. Based on the preset number of text vectors, the corresponding text clips are found to form a list of original forestry texts.

8. The method according to claim 1, wherein, The forestry data is visualized using the forestry original text list, the forestry relation database, and the geocoding in the interactive operation, including: In response to the user's interactive operation on the front-end map of the target area, the index of the text and image data and table data of the target area is retrieved from the forestry relationship database according to the geocode of the target area; In response to a user clicking an index, the corresponding tabular or graphic data is retrieved from the forestry relationship database and displayed. In response to a user's search operation for a target area on the front-end map, the original forestry text list is displayed in list form.

9. The method according to claim 1, characterized in that, The method further includes: Constructing a knowledge graph for the understory economy sector, and answering user questions based on the knowledge graph, specifically including: Based on the forestry data, an ontology for the understory economy is defined. The entity classes of the ontology include at least topographic regions, administrative regions, understory economic models, planting and breeding categories, and operating entities. Entities, attributes, and relationships between entities are extracted from structured data to form the first set of triples. For unstructured data, entities, attributes, and relationships between entities are extracted using an entity recognition and relationship extraction model to form the second set of triples. The first set of triples and the second set of triples are merged and disambiguated, and the processed triples are stored in a graph database to form a knowledge graph of the understory economy. Predefined question-and-answer intents and answer templates related to the understory economy; The system receives user input questions, performs intent recognition and entity extraction on the input questions, and generates query statements for the knowledge graph of the understory economy field based on the question-and-answer intent and entities. Execute the query statement, obtain the query results, and fill the corresponding answer reply template with the query results to generate the answer.

10. A regional development expert system for understory economy, characterized in that, The method for constructing an expert system for regional development of understory economy as described in claim 1 includes: The segmentation module is used to divide the target forestry area into geographically coded areas based on the regional characteristics of the target forestry area. The regional characteristics include at least topographic features and administrative boundary features. The storage module is used to store forestry data related to forestry economy in each area according to the structure of forestry data, and to establish a standardized forestry database. The forestry database includes at least a forestry text vector database and a forestry relation database. The retrieval module, in response to user retrieval operations and interactive operations on the front-end map, converts the retrieval input into a retrieval vector, and obtains a list of original forestry texts by calculating the semantic similarity between the retrieval vector and the forestry vectors in the forestry text vector database. The display module is used to visualize the forestry data using the forestry original text list, the forestry relationship database, and the geocoding in the interactive operation.

Images