An archive data management system based on semantic analysis

Abstract

The present invention relates to the field of data management technology, and specifically to an archive data management system based on semantic analysis. A data collection unit collects historical data and real-time multi-source data in the field of urban greening, including structured data and unstructured data; a data preprocessing unit preprocesses the data to obtain standardized data. A semantic analysis unit constructs a semantic analysis model for greening data to identify standardized data and generate semantic labels for urban greening; a semantic association unit first analyzes semantic labels to obtain preliminary associations for urban greening, and then analyzes historical data to obtain implicit associations, spatiotemporal associations, and causal associations between urban greening semantic labels, and generates urban greening associated data. A greening graph construction unit associates real-time multi-source data with real-time urban greening associated data to generate a dynamic knowledge graph for urban greening. A decision support unit generates greening management suggestions and issues early warning information through the dynamic knowledge graph for urban greening and input data.

Description

Technical Field

[0001] The present invention relates to the technical field of data management, and in particular to an archive data management system based on semantic analysis. Background Art

[0002] With the acceleration of global urbanization, urban greening plays an important role in improving the ecological environment, regulating the climate, and improving the quality of life of citizens. Urban greening projects usually include the planting, maintenance, and monitoring of plants, which generate a large amount of data. These data include structured data, such as plant species, planting time, and maintenance plans, as well as unstructured data, such as citizen feedback, plant growth images, and environmental monitoring data. Traditional archive management systems mainly rely on manual entry and classification, which makes it difficult to efficiently manage and utilize massive, complex, and heterogeneous data. In addition, with the rapid development of the Internet of Things, big data, and artificial intelligence technologies, more and more sensor devices can monitor dynamic data related to urban greening in real time, further increasing the complexity and difficulty of data management.

[0003] Traditional management methods often have problems with incomplete information and long time consumption when retrieving information, making it difficult to efficiently analyze the correlation between data, affecting the accuracy and timeliness of decision-making. At the same time, a large amount of duplicate information further increases the difficulty of management, resulting in low efficiency in data processing and updating.

[0004] Therefore, this paper proposes an archive data management system based on semantic analysis. Summary of the invention

[0005] The purpose of the present invention is to provide an archive data management system based on semantic analysis. The data collection unit collects historical data and real-time multi-source data in the field of urban greening, including structured data and unstructured data; the data preprocessing unit preprocesses the data to obtain standardized data. The semantic analysis unit constructs a semantic analysis model for greening data to identify standardized data and generate semantic labels for urban greening; the semantic association unit first analyzes the semantic labels to obtain preliminary associations of urban greening, and then analyzes the historical data to obtain implicit associations, spatiotemporal associations and causal associations between urban greening semantic labels, and generates urban greening associated data. The greening graph construction unit associates real-time multi-source data with real-time urban greening associated data to generate a dynamic knowledge graph for urban greening. The decision support unit generates greening management suggestions and issues early warning information through the dynamic knowledge graph of urban greening and input data.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] An archive data management system based on semantic analysis, comprising:

[0008] Data collection unit: used to collect historical data and real-time multi-source data of urban greening data, wherein the historical data and real-time multi-source data include structured data and unstructured data;

[0009] Data preprocessing unit: used for performing semantic preprocessing on the real-time multi-source data to obtain standardized data;

[0010] Semantic analysis unit: used to construct a semantic analysis model for greening data to analyze and identify the standardized data and generate urban greening semantic labels;

[0011] Semantic association unit: used to obtain preliminary urban greening association by analyzing the urban greening semantic tags; to obtain implicit association, spatiotemporal association and causal association between urban greening semantic tags by constructing an urban greening semantic association model to analyze the historical data; to generate urban greening association data through preliminary urban greening association, implicit association, spatiotemporal association and causal association;

[0012] A greening graph construction unit is used to analyze the real-time multi-source data and the real-time urban greening related data to generate a dynamic knowledge graph of urban greening;

[0013] Decision support unit: used to generate greening management suggestions and issue early warning information based on the greening dynamic knowledge graph and input data; the input data is urban greening information data.

[0014] Preferably, the structured data is basic file information of greening management projects, including basic plant information, planting information, maintenance plans and management projects;

[0015] The basic plant information includes the plant species, name, health status and growth stage;

[0016] The planting information includes the planting time, geographical location, planting density and area of ​​the greening area;

[0017] The maintenance plan includes regular maintenance measures for green areas, such as irrigation, pruning, fertilization and pest control;

[0018] The management projects include the project names, task arrangements, timetables and responsible persons related to urban greening management;

[0019] The unstructured data includes citizen feedback, picture and video data, image data and environmental monitoring data;

[0020] Said citizen feedback includes citizens’ opinions, complaints and improvement suggestions on green areas;

[0021] The image and video data include plant growth conditions and pest monitoring;

[0022] The environmental monitoring data includes soil moisture, temperature, wind speed and air quality.

[0023] Preferably, the semantic preprocessing includes data cleaning, formatting and word vector representation;

[0024] The data cleaning includes removing noise, standardizing data types and removing outliers;

[0025] The formatting process includes segmenting the greening text data and removing stop words, extracting features from the greening picture data through a convolutional neural network, and normalizing the environmental data;

[0026] The word vector representation includes converting greening text data into greening numerical vectors through the word vector model GloVe;

[0027] The standardized data includes greening text word vectors, greening images and greening environment monitoring data.

[0028] Preferably, the greening data semantic analysis model includes a greening text word vector feature extraction layer, a greening image feature extraction layer, a greening environment monitoring data feature extraction layer, a multimodal data analysis layer and an urban greening semantic label generation layer;

[0029] The greening text word vector feature extraction layer performs feature extraction on the greening text word vector to obtain the greening text word feature vector;

[0030] The greening image feature extraction layer extracts features from the greening image to obtain a greening image feature vector;

[0031] The greening environment monitoring data feature extraction layer extracts features from the greening environment monitoring data to obtain a greening environment feature vector;

[0032] The multimodal data analysis layer analyzes the greening text word feature vector, the greening image feature vector and the greening environment feature vector to generate a first urban greening semantic label;

[0033] The urban greening semantic label generation layer updates the first urban greening semantic label through the real-time multi-source data to generate an urban greening semantic label; the urban greening semantic label includes a plant information label, an environmental status label, a risk warning label and a management demand label.

[0034] Preferably, the greening image feature vector acquisition process includes:

[0035] Input the image feature vector into a 3×3 convolution layer with a convolution kernel size of 16 to obtain a first image feature map; input the first image feature map into a 3×3 convolution layer with a convolution kernel size of 32 to obtain a second image feature map; perform batch normalization on the second image feature map, and pass the batch normalized feature map to a Sigmoid activation function to obtain a third image feature map; splice the second image feature map and the third image feature map according to the channel dimension and input them into a 1×1 convolution layer with a convolution kernel size of 16, and add a channel attention module after the convolution layer to obtain a fourth image feature map;

[0036] The second image feature map, the third image feature map and the fourth image feature map are input into the image feature fusion layer to generate a greening image feature vector; the greening image feature vector includes the color, texture, shape, plant height and pest status of the plant.

[0037] Preferably, the real-time multi-source data and the real-time urban greening related data are analyzed to generate a dynamic knowledge graph of urban greening;

[0038] The urban greening dynamic knowledge graph generation process includes:

[0039] The entities and entity relationships in the real-time multi-source data are identified through a deep learning model for the urban greening related data; an urban greening dynamic knowledge graph is generated by acquiring the entities and entity relationships in the semantic associations between the data; the entities are represented as nodes in the urban greening dynamic knowledge graph, and the entity relationships are represented as edges in the urban greening dynamic knowledge graph.

[0040] Preferably, the urban greening semantic relevance model includes a semantic similarity layer, a spatiotemporal correlation analysis layer, and a causal relationship analysis layer;

[0041] The semantic similarity layer obtains a semantic similarity matrix based on the similarity between the historical data, and generates implicit semantic associations of urban greening;

[0042] The spatiotemporal correlation analysis layer analyzes the temporal and spatial correlation of urban greening based on the time series data in the historical data, and generates the semantic spatiotemporal correlation of urban greening;

[0043] The causal relationship analysis layer analyzes the causal relationship of urban greening based on the historical data and generates semantic causal associations of urban greening.

[0044] Preferably, the real-time multi-source data and the real-time urban greening related data are analyzed to generate a dynamic knowledge graph of urban greening;

[0045] The urban greening dynamic knowledge graph generation process includes:

[0046] The entities and entity relationships in the real-time multi-source data are identified by using a deep learning model for the urban greening related data; the graph establishment module generates a knowledge graph by acquiring the entities and entity relationships in the semantic associations between the data; the entities are represented as nodes in the knowledge graph, and the entity relationships are represented as edges in the knowledge graph.

[0047] Preferably, the input data is semantically analyzed by the greening data semantic analysis model to obtain urban greening related data associated with the input data in the urban greening archive data management system;

[0048] Generate greening management suggestions based on the urban greening related data.

[0049] Preferably, sending an early warning message to the administrator according to the greening management suggestion;

[0050] The warning information includes: analysis of potential pests and diseases, impact of environmental abnormalities on plants, greening areas that need priority treatment, climate change warnings, and warnings of plant health deterioration.

[0051] Compared with the prior art, the present invention has the following beneficial effects:

[0052] 1. The present invention constructs a semantic analysis model for greening data; obtains greening text word feature vectors, greening image feature vectors and greening environment feature vectors by identifying preprocessed standardized data; analyzes the feature vectors through multimodal data fusion technology to generate a first urban greening semantic label, and combines real-time multi-source data to generate an urban greening semantic label, thereby effectively improving the management efficiency of urban greening data.

[0053] 2. The present invention constructs a semantic correlation model for urban greening; obtains preliminary correlations of urban greening by analyzing the semantic labels of urban greening; analyzes historical data through the semantic correlation model of urban greening to obtain implicit correlations, spatiotemporal correlations and causal correlations between semantic labels of urban greening; comprehensively considers correlations at different levels to provide support for scientific planning and management of urban greening.

[0054] 3. The present invention uses a greening data semantic analysis model to perform in-depth semantic analysis on input data, and obtains urban greening related data associated with the urban greening archive data management system. Based on the related data and the greening dynamic knowledge graph, corresponding greening management suggestions are generated. Potential risks are communicated in a timely manner through early warning information, aiming to improve the efficiency and responsiveness of greening management and promote the optimization of urban greening data management. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 It is a structural schematic diagram of an archive data management system based on semantic analysis of the present invention;

[0056] Figure 2 It is a structural schematic diagram of the semantic analysis model of greening data of the present invention;

[0057] Figure 3 It is a structural schematic diagram of the urban greening semantic relevance model of the present invention. DETAILED DESCRIPTION

[0058] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0059] The present invention proposes an archive data management system based on semantic analysis, which is suitable for urban greening data management. Embodiment 1

[0060] The present invention proposes an archive data management system based on semantic analysis, the structural diagram of the system is as follows: Figure 1 As shown, it includes: a data collection unit, a data preprocessing unit, a semantic analysis unit, a semantic association unit, a greening map construction unit and a decision support unit.

[0061] A data collection unit, used to collect historical data and real-time multi-source data of urban greening data, wherein the historical data and real-time multi-source data include structured data and unstructured data;

[0062] Furthermore, the structured data is basic file information of greening management projects, including basic plant information, planting information, maintenance plans and management projects;

[0063] The basic plant information includes the plant species, name, health status and growth stage;

[0064] The planting information includes the planting time, geographical location, planting density and area of ​​the greening area;

[0065] The maintenance plan includes regular maintenance measures for green areas, such as irrigation, pruning, fertilization and pest control;

[0066] The management projects include the project names, task arrangements, timetables and responsible persons related to urban greening management;

[0067] The unstructured data includes citizen feedback, picture and video data, image data and environmental monitoring data;

[0068] Said citizen feedback includes citizens’ opinions, complaints and improvement suggestions on green areas;

[0069] The image and video data include plant growth conditions and pest monitoring;

[0070] The environmental monitoring data includes soil moisture, temperature, wind speed, light and air quality.

[0071] In this embodiment, by collecting structured data and unstructured data, the multi-dimensional information of urban greening management can be fully covered, so that the system can evaluate the greening status from different angles and levels, improve management effect and efficiency, and promote the sustainable development of the urban ecological environment.

[0072] A data preprocessing unit, used for performing semantic preprocessing on the real-time multi-source data to obtain standardized data;

[0073] Furthermore, the semantic preprocessing includes data cleaning, formatting and word vector representation;

[0074] The data cleaning includes removing noise, standardizing data types and removing outliers;

[0075] The formatting process includes segmenting the greening text data and removing stop words, extracting features from the greening picture data through a convolutional neural network, and normalizing the environmental data;

[0076] The word vector representation includes converting greening text data into greening numerical vectors through the word vector model GloVe;

[0077] The standardized data includes greening text word vectors, greening images and greening environment monitoring data.

[0078] The specific steps of the word vector model GloVe to convert greening text data into greening numerical vectors are:

[0079] S1. Construct a co-occurrence matrix, where each element Expressive words and words The number of co-occurrences in the context window is used to reflect the correlation between words. For example, the co-occurrence frequency between tree and forest is higher than that between tree and car. The size of the co-occurrence matrix is , is the total number of words in the corpus. A co-occurrence matrix between terms such as plant species, pests and diseases, and air quality is generated from citizen feedback texts to obtain the relationships between different greening data.

[0080] S2. Co-occurrence frequency Logarithmic processing can help highlight low-frequency abnormal data, such as rare pest and disease reports, and avoid common words, such as plants and green spaces, dominating the semantic construction of the overall model.

[0081] S3. Optimize the objective function to enable the word vector to maintain the co-occurrence relationship between words. The objective function is as follows:

[0082] ;

[0083] in, is the logarithm of the co-occurrence of similar words, Words The word vector of Words The transpose of the word vector, Words The word vector of Words The corresponding word bias of Words The corresponding word bias of Words and words The logarithm of the number of co-occurrences of is a weighting function used to control the influence of different word pairs on the model.

[0084] S4. Generate word vectors. Generate a word vector for each word by optimizing the objective function. The dimension of the word vector is set to 100 in this embodiment.

[0085] In this embodiment, through data cleaning and formatting steps, noise is removed, data types are standardized, and word segmentation and stop word removal are performed to reduce the interference of invalid information and ensure high data quality. At the same time, the word vector generation process can provide a reliable data foundation for subsequent greening semantic label generation, dynamic knowledge graph construction, and management suggestion generation through the extraction and representation of global semantic information.

[0086] A semantic analysis unit, used to construct a semantic analysis model for greening data to analyze and identify the standardized data and generate urban greening semantic labels;

[0087] Furthermore, the greening data semantic analysis model includes a greening text word vector feature extraction module, a greening image feature extraction module, a greening environment monitoring data feature extraction module, a multimodal data analysis module and an urban greening semantic label generation module, see Figure 2 ;

[0088] The greening text word vector feature extraction module performs feature extraction on the greening text word vector to obtain the greening text word feature vector;

[0089] The greening image feature extraction module extracts features from the greening image to obtain a greening image feature vector;

[0090] The greening environment monitoring data feature extraction module extracts features from the greening environment monitoring data to obtain a greening environment feature vector;

[0091] The multimodal data analysis module analyzes the greening text word feature vector, the greening image feature vector and the greening environment feature vector to generate a first urban greening semantic label;

[0092] The urban greening semantic label generation module updates the first urban greening semantic label through the real-time multi-source data to generate an urban greening semantic label; the urban greening semantic label includes a plant information label, an environmental status label, a risk warning label and a management demand label.

[0093] The plant information label includes plant species, health status, growth stage, and pest and disease conditions.

[0094] The environmental status labels include environmental influences such as soil moisture, air quality, temperature, light, and wind speed.

[0095] The risk warning labels include potential pest and disease risks, the impact of environmental abnormalities on plants, and green areas that require priority treatment.

[0096] The Management Needs tab includes recommended maintenance measures such as irrigation, pruning, and fertilization.

[0097] In this embodiment, feature extraction is performed on different types of data, such as text, image, and environmental monitoring data, through the greening text word vector feature extraction module, the greening image feature extraction module, and the greening environment monitoring data feature extraction module. The fusion analysis method can break through the limitation of a single data type, so that different types of data complement each other, thereby revealing the status of urban greening in more depth.

[0098] Furthermore, the greening image feature vector acquisition process includes:

[0099] Input the image feature vector into a 3×3 convolution layer with a convolution kernel size of 16 to obtain a first image feature map; input the first image feature map into a 3×3 convolution layer with a convolution kernel size of 32 to obtain a second image feature map; perform batch normalization on the second image feature map, and pass the batch normalized feature map to a Sigmoid activation function to obtain a third image feature map; splice the second image feature map and the third image feature map according to the channel dimension and input them into a 1×1 convolution layer with a convolution kernel size of 16, and add a channel attention module after the convolution layer to obtain a fourth image feature map;

[0100] The second image feature map, the third image feature map and the fourth image feature map are input into the image feature fusion layer to generate a greening image feature vector; the greening image feature vector includes the color, texture, shape and plant height of the plant.

[0101] In this embodiment, the image feature vector acquisition process is combined with the channel attention mechanism through a convolutional neural network to improve the extraction accuracy of plant image features. Through multi-layer convolution operations, the feature representation is richer. The introduction of the Sigmoid activation function ensures the stability of the model and enhances the expression of nonlinear features. The channel attention mechanism further focuses on key features, filters irrelevant information, and improves the effectiveness of features. The feature fusion of different convolutional layers ensures the retention and extraction of multidimensional information. The image feature vector finally generated can effectively reflect the color, texture, shape and plant height of the plant, thereby carefully monitoring the state of the plant and providing data support for urban greening management.

[0102] A semantic association unit, used to obtain urban greening association data by analyzing the urban greening semantic tags;

[0103] Furthermore, the urban greening related data generation process includes:

[0104] By analyzing the semantic tags of urban greening, a preliminary association of urban greening is obtained; the preliminary association includes the association between plant information tags and environmental status tags, the association between plant information tags and risk warning tags, the association between environmental status tags and management demand tags, and the association between risk warning tags and management demand tags;

[0105] By constructing an urban greening semantic relevance model to analyze the historical data, implicit associations, spatiotemporal associations, and causal associations between urban greening semantic labels are obtained.

[0106] In this embodiment, the urban greening related data generation process reveals the preliminary associations between plant information and environmental status, risk warning and management needs, etc. by analyzing urban greening semantic tags. The urban greening semantic association model further constructed analyzes historical data to extract implicit associations, spatiotemporal associations and causal associations, providing a scientific basis for decision makers. This ability to deeply mine data enables urban greening management to be more targeted.

[0107] Furthermore, the urban greening semantic correlation model includes a semantic similarity module, a spatiotemporal correlation analysis module and a causal relationship analysis module. Figure 3 ;

[0108] The semantic similarity module obtains a semantic similarity matrix based on the similarity between the historical data, and generates implicit semantic associations of urban greening;

[0109] The spatiotemporal correlation analysis module analyzes the temporal and spatial correlation of urban greening based on the time series data in the historical data, and generates the semantic spatiotemporal correlation of urban greening;

[0110] The causal relationship analysis module analyzes the causal relationship of urban greening based on the historical data and generates semantic causal associations of urban greening.

[0111] In this embodiment, the urban greening semantic relevance model improves the analysis ability of urban greening data through the semantic similarity module, the spatiotemporal association analysis module and the causal relationship analysis module. By using the similarity matrix constructed by historical data, the implicit association between urban greening elements is revealed, which promotes a deep understanding of plant species and their growth environment. The spatiotemporal association analysis module analyzes the changing laws of urban greening at different times and spaces by studying time series data, which helps to identify the impact of seasonal and geographical factors on greening effects. The causal relationship analysis module clarifies the causal relationship between various factors and provides a basis for formulating scientific management strategies. The comprehensive application of these analyses enhances the scientificity and effectiveness of urban greening management.

[0112] A greening graph construction unit is used to analyze the real-time multi-source data and the real-time urban greening related data to generate a dynamic knowledge graph of urban greening;

[0113] Furthermore, the real-time multi-source data and the real-time urban greening related data are analyzed to generate a dynamic knowledge graph of urban greening;

[0114] The urban greening dynamic knowledge graph generation process includes:

[0115] The entities and entity relationships in the real-time multi-source data are identified by using a deep learning model for the urban greening related data; the graph establishment module generates a knowledge graph by acquiring the entities and entity relationships in the semantic associations between the data; the entities are represented as nodes in the knowledge graph, and the entity relationships are represented as edges in the knowledge graph.

[0116] Said entities include plant species, regional names, and climatic factors;

[0117] Said entity relationships include maintenance records and health status;

[0118] The semantic associations between the data include soil moisture, temperature changes and maintenance records, and plant health status text associations.

[0119] In this embodiment, the analysis of real-time multi-source data and urban greening related data generates a dynamic knowledge graph of urban greening, which improves the intelligent level of urban greening management. During the generation process, the deep learning model effectively identifies the entities and their relationships in the real-time multi-source data, so that the knowledge graph can truly reflect the complexity of urban greening. By analyzing semantic associations such as soil moisture and temperature changes, managers can better understand the impact of the environment on plant health, and thus formulate scientific and reasonable management strategies.

[0120] A decision support unit is used to generate greening management suggestions and issue early warning information based on the greening dynamic knowledge graph and input data; the input data is urban greening information data.

[0121] Further, the input data is semantically analyzed by the greening data semantic analysis model to obtain urban greening related data associated with the input data in the urban greening archive data management system;

[0122] Generate greening management suggestions based on the urban greening related data.

[0123] In this embodiment, the system performs semantic analysis on the input data through the semantic analysis model of greening data. The input data includes a variety of information related to urban greening, such as plant species, planting locations, climatic conditions, and maintenance history. Using semantic understanding technology, the system identifies key concepts in the input data and their relationships. After completing the semantic analysis, the system will automatically generate urban greening related data related to the input data, including plant suitability analysis, maintenance recommendations, pest and disease warnings, and ecological benefit evaluation. For example, the system recommends plant species suitable for planting based on climatic conditions and soil types, and provides the best maintenance plan, watering frequency, and fertilization timing based on historical maintenance records. At the same time, by analyzing historical data and current environmental conditions, the system can identify potential pest and disease risks and propose preventive measures. Based on the urban greening related data, the system generates a series of greening management recommendations, such as planting specific plants in a certain area to improve soil quality, or regularly watering a certain plant. Finally, all generated management recommendations will be presented to greening managers in an easy-to-understand format to help them effectively plan and implement urban greening projects.

[0124] In this embodiment, the management efficiency and decision-making accuracy of urban greening are improved by generating greening management suggestions based on the greening dynamic knowledge graph and input data. The greening data semantic analysis model performs in-depth semantic analysis on the input data to extract associated data related to urban greening. This intelligent decision support not only improves the management level of urban greening.

[0125] Further, according to the greening management suggestion, an early warning message is sent to the administrator;

[0126] The warning information includes: analysis of potential pests and diseases, impact of environmental abnormalities on plants, greening areas that need priority treatment, climate change warnings, and warnings of plant health deterioration.

[0127] In this embodiment, the early warning information sent to the administrator based on the greening management suggestions improves the response speed and effectiveness of urban greening management. These early warning information include potential pest and disease analysis, the impact of environmental abnormalities on plants, priority greening areas, climate change warnings, and reminders of plant health deterioration. By delivering these key information in a timely manner, managers can quickly take countermeasures to prevent the problem from further deteriorating.

[0128] The archive data management system based on semantic analysis of the present invention optimizes management measures and improves the maintenance efficiency of urban greening by integrating multiple data processing modules. The data acquisition unit collects historical and real-time structured and unstructured data, so that the system has a rich information foundation. The data preprocessing unit eliminates noise and outliers through semantic preprocessing, ensuring the availability of data. The semantic analysis model of greening data constructed by the semantic analysis unit can deeply identify and generate urban greening semantic tags, providing a multi-dimensional understanding of the greening status. The semantic association unit analyzes the preliminary association and implicit association between urban greening tags, constructs a semantic association model of urban greening, and reveals the spatiotemporal relationship and causal relationship between tags. This analysis lays the foundation for generating urban greening related data, and is further used to construct a dynamic knowledge graph. The greening graph construction unit captures entities and their relationships in urban greening through real-time data analysis, thereby realizing the visualization of knowledge. The decision support unit generates specific greening management suggestions based on the greening dynamic knowledge graph, and sends early warning information to the administrator in a timely manner. The content of the early warning information covers pest and disease analysis, environmental anomalies, priority treatment areas and climate change, helping managers to respond quickly, optimize management measures and improve the efficiency and effectiveness of urban greening maintenance.

[0129] The present invention uses an archive data management system based on semantic analysis to intelligently manage urban greening archive data, which is superior to traditional management systems in multiple dimensions and provides faster, more comprehensive and more efficient information management capabilities; as shown in Table 1;

[0130] Table 1 Multi-dimensional comparison of effects

[0131]

[0132] In Table 1, information retrieval time refers to the time required for users to find the required information in the archive data management system; information completeness refers to the comprehensiveness and accuracy of the information in the system, which is usually calculated by comparing the available data with the actual amount of data required; user satisfaction refers to the user's satisfaction with the system due to its quick response and rich information; data update frequency refers to the frequency with which the system updates data; data access frequency refers to the number of times users access the system within a specific time period; information repetition rate refers to the proportion of repeated information in the system; error information rate refers to the proportion of error information in the system, including outdated, erroneous or irrelevant data; data cleaning time refers to the time required to clean and process the original data in the system, including removing noise, standardizing data types and correcting errors; multi-source data fusion capability reflects the system's ability to integrate data from different sources. A high level of multi-source data fusion capability means that the system can effectively combine multiple types of information to provide more comprehensive data analysis and services. Embodiment 2

[0133] With the gradual increase in urban greening coverage, the amount of urban greening archive data is growing exponentially, and the archive content structure is complex, covering plant species, planting time, maintenance records, soil conditions and other information. The traditional archive management system mainly relies on keyword search and manual classification, which is difficult to effectively respond to the increasingly complex archival data query, retrieval and management needs. In addition, since greening management requires dynamic adjustment, how to quickly obtain relevant archival information to support decision-making has become a technical problem that needs to be solved urgently. To this end, this embodiment provides an archive data management system based on semantic analysis, which can perform intelligent analysis on greening archive data, optimize the classification, retrieval and management methods of information, thereby improving the efficiency and accuracy of archive management.

[0134] Data collection unit: used to collect historical data and real-time multi-source data of urban greening data, wherein the historical data and real-time multi-source data include structured data and unstructured data;

[0135] Furthermore, the structured data is basic file information of greening management projects, including basic plant information, planting information, maintenance plans and management projects;

[0136] The basic plant information includes the plant species, name, health status and growth stage;

[0137] The planting information includes the planting time, geographical location, planting density and area of ​​the greening area;

[0138] The maintenance plan includes regular maintenance measures for green areas, such as irrigation, pruning, fertilization and pest control;

[0139] The management projects include the project names, task arrangements, timetables and responsible persons related to urban greening management;

[0140] The unstructured data includes citizen feedback, picture and video data, image data and environmental monitoring data;

[0141] Said citizen feedback includes citizens’ opinions, complaints and improvement suggestions on green areas;

[0142] The image and video data include plant growth conditions and pest monitoring;

[0143] The environmental monitoring data includes soil moisture, temperature, wind speed and air quality.

[0144] Data preprocessing unit: used for performing semantic preprocessing on the real-time multi-source data to obtain standardized data;

[0145] Furthermore, the semantic preprocessing includes data cleaning, formatting and word vector representation;

[0146] The data cleaning includes removing noise, standardizing data types and removing outliers;

[0147] The formatting process includes segmenting the greening text data and removing stop words, extracting features from the greening picture data through a convolutional neural network, and normalizing the environmental data;

[0148] The word vector representation includes converting greening text data into greening numerical vectors through the word vector model GloVe;

[0149] The standardized data includes greening text word vectors, greening images and greening environment monitoring data.

[0150] Semantic analysis unit: used to construct a semantic analysis model for greening data to analyze and identify the standardized data and generate urban greening semantic labels;

[0151] Furthermore, the greening data semantic analysis model includes a greening text word vector feature extraction layer, a greening image feature extraction layer, a greening environment monitoring data feature extraction layer, a multimodal data analysis layer and an urban greening semantic label generation layer;

[0152] The greening text word vector feature extraction layer performs feature extraction on the greening text word vector to obtain the greening text word feature vector;

[0153] The greening image feature extraction layer extracts features from the greening image to obtain a greening image feature vector;

[0154] The greening environment monitoring data feature extraction layer extracts features from the greening environment monitoring data to obtain a greening environment feature vector;

[0155] The multimodal data analysis layer analyzes the greening text word feature vector, the greening image feature vector and the greening environment feature vector to generate a first urban greening semantic label;

[0156] The urban greening semantic label generation layer updates the first urban greening semantic label through the real-time multi-source data to generate an urban greening semantic label; the urban greening semantic label includes a plant information label, an environmental status label, a risk warning label and a management demand label.

[0157] Furthermore, the greening image feature vector acquisition process includes:

[0158] Input the image feature vector into a 3×3 convolution layer with a convolution kernel size of 16 to obtain a first image feature map; input the first image feature map into a 3×3 convolution layer with a convolution kernel size of 32 to obtain a second image feature map; perform batch normalization on the second image feature map, and pass the batch normalized feature map to a Sigmoid activation function to obtain a third image feature map; splice the second image feature map and the third image feature map according to the channel dimension and input them into a 1×1 convolution layer with a convolution kernel size of 16, and add a channel attention module after the convolution layer to obtain a fourth image feature map;

[0159] The second image feature map, the third image feature map and the fourth image feature map are input into the image feature fusion layer to generate a greening image feature vector; the greening image feature vector includes the color, texture, shape, plant height and pest status of the plant.

[0160] At the same time, the greening environment monitoring data feature extraction layer is responsible for extracting features from environmental monitoring data and generating greening environment feature vectors. This vector includes environmental parameters such as temperature, humidity, soil pH, etc., reflecting the ecological status of the greening area.

[0161] In the multimodal data analysis layer, the system conducts a comprehensive analysis of the greening text word feature vector, greening image feature vector and greening environment feature vector to generate the first urban greening semantic label, which includes plant information, environmental status, risk warning and management requirements.

[0162] The environmental status of urban greening refers to the various ecological and climatic factors that affect urban greening. These factors include temperature, humidity, soil conditions, and air quality. The specific formula for expressing the environmental status of urban greening is:

[0163] ;

[0164] in, is the environmental status index of urban greening, The weight of urban greening affected by temperature, is the temperature measured by the temperature sensor, The optimal temperature for urban greening growth. The influence of soil moisture on urban greening weight, is the humidity measured by the humidity sensor, is the field water holding capacity, The weight of urban greening affected by air quality, For the The concentration of pollutants, For the The maximum safe concentration of a pollutant.

[0165] In this embodiment, the formula can systematically integrate multiple environmental factors such as temperature, humidity, soil conditions and air quality, so as to comprehensively reflect the ecological status of urban greening. It provides managers with a quantitative basis so that they can more intuitively understand the impact of different factors on plant growth. The weight design of each parameter in the formula allows managers to identify which factors have a more significant impact on urban greening.

[0166] In addition, by analyzing the environmental status, the system is able to identify potential risks in advance, such as plant pests and diseases and extreme weather. By analyzing the environmental status through quantitative formulas, managers can quickly assess the impact of ecological and climate factors on urban greening, so as to take preventive measures and reduce losses.

[0167] Risk early warning for urban greening refers to the early identification and assessment of potential pests and diseases, climate change and other environmental pressures, which can provide early warning information for urban greening management. The specific formula for risk early warning for urban greening is:

[0168] ;

[0169] in, is the risk warning index for urban greening. The weight of urban greening affected by temperature, The influence of soil moisture on urban greening weight, The current weight of insect pests affecting urban greening. is a natural constant, For the current level of pests, The current climate conditions affect the weight of urban greening. is the current climate condition index, is the critical threshold of current climate conditions, The greening area affects the weight of urban greening. is the green area, is the area of ​​the total region.

[0170] The risk warning formula for urban greening comprehensively evaluates environmental pressure, identifies key influencing factors, identifies potential threats in advance, and promotes scientific decision-making. At the same time, the risk warning formula for urban greening makes management decisions more scientific and data-driven through quantitative analysis. This data-oriented management method can improve the accuracy of decision-making, optimize resource allocation, and thus improve the overall efficiency of urban greening management.

[0171] Semantic association unit: used to obtain preliminary urban greening association by analyzing the urban greening semantic tags; to obtain implicit association, spatiotemporal association and causal association between urban greening semantic tags by constructing an urban greening semantic association model to analyze the historical data; to generate urban greening association data through preliminary urban greening association, implicit association, spatiotemporal association and causal association;

[0172] Among them, preliminary associations are identified through direct analysis of urban greening semantic labels, such as plant species, environmental factors, management measures, etc. For example, specific plant species may be associated with specific soil types and climate conditions.

[0173] Implicit associations are constructed by building a semantic association model of urban greening and analyzing the potential patterns and relationships in historical data. These implicit associations are not easy to observe directly, but can be obtained through data mining technology.

[0174] Spatiotemporal correlation analysis reveals the dynamic relationship between plant growth and environmental changes by considering the temporal changes and spatial distribution characteristics of urban greening data.

[0175] Causal association is the process of analyzing historical data to identify the cause-effect relationship between different factors. For example, the growth of a certain plant may be affected by climate change, management measures, or pests and diseases.

[0176] By integrating the above various types of related data, the related data of urban greening can be generated; the urban greening related data can be expressed by the formula:

[0177] ;

[0178] in, To generate urban greening related data, is the initial associated data weight of urban greening, is a function of climatic conditions, soil type, and plant species. For climate conditions, For soil type, For plant species, is the implicit association data weight of urban greening, For the Class urban greening information data under conditions Next is the The probability that the urban greening information data is associated with is the weight of urban greening spatiotemporal association data, is the functional relationship between time variables, spatial variables and management measures, is the time variable, is a spatial variable, For management measures, is the causal association data weight of urban greening, For the The weight of the causal factors, For the Causal factors, For the The error term affects the causal factor.

[0179] The urban greening related data formula integrates multiple factors to form comprehensive urban greening related data, excavates the hidden associations of urban greening, reveals the potential connections between different urban greening information, and analyzes the impact of different time and space variables on greening effects. Comprehensive analysis enables managers to deeply understand the relationship between climate conditions, soil types and plant species, provides a basis for scientific management, and has a significant impact on improving the scientificity and effectiveness of urban greening management.

[0180] Furthermore, the urban greening semantic relevance model includes a semantic similarity layer, a spatiotemporal correlation analysis layer, and a causal relationship analysis layer;

[0181] The semantic similarity layer obtains a semantic similarity matrix based on the similarity between the historical data, and generates implicit semantic associations of urban greening;

[0182] The spatiotemporal correlation analysis layer analyzes the temporal and spatial correlation of urban greening based on the time series data in the historical data, and generates the semantic spatiotemporal correlation of urban greening;

[0183] The causal relationship analysis layer analyzes the causal relationship of urban greening based on the historical data and generates semantic causal associations of urban greening.

[0184] Greening graph construction unit: used to analyze the real-time multi-source data and real-time urban greening related data to generate a dynamic knowledge graph of urban greening.

[0185] Furthermore, the real-time multi-source data and the real-time urban greening related data are analyzed to generate a dynamic knowledge graph of urban greening;

[0186] The urban greening dynamic knowledge graph generation process includes:

[0187] The entities and entity relationships in the real-time multi-source data are identified through a deep learning model for the urban greening related data; an urban greening dynamic knowledge graph is generated by acquiring the entities and entity relationships in the semantic associations between the data; the entities are represented as nodes in the urban greening dynamic knowledge graph, and the entity relationships are represented as edges in the urban greening dynamic knowledge graph.

[0188] Decision support unit: used to generate greening management suggestions and issue early warning information based on the greening dynamic knowledge graph and input data; the input data is urban greening information data.

[0189] Further, the input data is semantically analyzed by the greening data semantic analysis model to obtain urban greening related data associated with the input data in the urban greening archive data management system;

[0190] Generate greening management suggestions based on the urban greening related data.

[0191] Further, according to the greening management suggestion, an early warning message is sent to the administrator;

[0192] The warning information includes: analysis of potential pests and diseases, impact of environmental abnormalities on plants, greening areas that need priority treatment, climate change warnings, and warnings of plant health deterioration.

[0193] The greening management suggestions include text, data, pictures and tables. Table 2 is a form of expression of greening management suggestions.

[0194] Table 2 Greening Management Suggestions

[0195]

[0196] This system optimizes the management of urban greening archive data through semantic analysis technology, and improves the efficiency of data retrieval and analysis. The data collection unit not only integrates structured historical data and real-time multi-source data, but also includes citizen feedback and environmental monitoring information, achieving comprehensive data coverage. The data preprocessing unit provides a data basis for subsequent analysis through cleaning, formatting and word vector representation. The system generates comprehensive urban greening semantic labels by constructing a greening data semantic analysis model. And through formulas to analyze environmental status and risk warnings, managers can quantify the impact of ecological factors, so as to more effectively identify potential risks and take preventive measures. In addition, the construction of a dynamic knowledge graph enables managers to monitor and adjust greening management strategies in real time, promote scientific decision-making, and improve the management efficiency and accuracy of urban greening.

[0197] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. An archive data management system based on semantic analysis, characterized in that: include: Data collection unit: used to collect historical data and real-time multi-source data of urban greening data, wherein the historical data and real-time multi-source data include structured data and unstructured data; Data preprocessing unit: used for performing semantic preprocessing on the real-time multi-source data to obtain standardized data; The semantic preprocessing includes data cleaning, formatting and word vector representation; the formatting includes segmenting the greening text data and removing stop words, extracting features from the greening picture data through a convolutional neural network, and normalizing the environmental data; the word vector representation includes converting the greening text data into a greening numerical vector through a word vector model GloVe; the standardized data includes greening text word vectors, greening images and greening environment monitoring data; Semantic analysis unit: used to construct a semantic analysis model for greening data to analyze and identify the standardized data and generate urban greening semantic labels; The greening data semantic analysis model includes a greening text word vector feature extraction layer, a greening image feature extraction layer, a greening environment monitoring data feature extraction layer, a multimodal data analysis layer and an urban greening semantic label generation layer; the greening text word vector feature extraction layer performs feature extraction on the greening text word vector to obtain a greening text word feature vector; the greening image feature extraction layer performs feature extraction on the greening image to obtain a greening image feature vector; the greening environment monitoring data feature extraction layer performs feature extraction on the greening environment monitoring data to obtain a greening environment feature vector; the multimodal data analysis layer analyzes the greening text word feature vector, the greening image feature vector and the greening environment feature vector to generate a first urban greening semantic label; the urban greening semantic label generation layer updates the first urban greening semantic label through the real-time multi-source data to generate an urban greening semantic label; the urban greening semantic label includes a plant information label, an environmental status label, a risk warning label and a management demand label; The greening image feature vector acquisition process includes: inputting the image feature vector into a 3×3 convolution layer with a convolution kernel size of 16 to obtain a first image feature map; inputting the first image feature map into a 3×3 convolution layer with a convolution kernel size of 32 to obtain a second image feature map; batch normalizing the second image feature map, and passing the batch normalized feature map to a Sigmoid activation function to obtain a third image feature map; splicing the second image feature map and the third image feature map according to the channel dimension and inputting them into a 1×1 convolution layer with a convolution kernel size of 16, and adding a channel attention module after the convolution layer to obtain a fourth image feature map; inputting the second image feature map, the third image feature map and the fourth image feature map into the image feature fusion layer to generate a greening image feature vector; the greening image feature vector includes the color, texture, shape, plant height and pest status of the plant; Semantic association unit: used to obtain preliminary urban greening association by analyzing the urban greening semantic tags; to obtain implicit association, spatiotemporal association and causal association between urban greening semantic tags by constructing an urban greening semantic association model to analyze the historical data; to generate urban greening association data through preliminary urban greening association, implicit association, spatiotemporal association and causal association; The urban greening associated data generation process includes: obtaining preliminary urban greening associations by analyzing the urban greening semantic tags; the preliminary associations include associations between plant information tags and environmental status tags, associations between plant information tags and risk warning tags, associations between environmental status tags and management demand tags, and associations between risk warning tags and management demand tags; analyzing the historical data by constructing an urban greening semantic association model to obtain implicit associations, spatiotemporal associations, and causal associations between urban greening semantic tags; The urban greening semantic association model includes a semantic similarity layer, a spatiotemporal association analysis layer and a causal relationship analysis layer; the semantic similarity layer obtains a semantic similarity matrix based on the similarity between the historical data, and generates the implicit semantic association of urban greening; the spatiotemporal association analysis layer analyzes the temporal and spatial correlation of urban greening based on the time series data in the historical data, and generates the spatiotemporal association of urban greening semantics; the causal relationship analysis layer analyzes the causal relationship of urban greening based on the historical data, and generates the semantic causal association of urban greening; A greening graph construction unit is used to analyze the real-time multi-source data and the real-time urban greening related data to generate a dynamic knowledge graph of urban greening; Decision support unit: used to generate greening management suggestions and issue early warning information based on the greening dynamic knowledge graph and input data; the input data is urban greening information data.

2. The archive data management system based on semantic analysis according to claim 1, characterized in that: The structured data is the basic file information of the greening management project, including basic plant information, planting information, maintenance plans and management projects; The basic plant information includes the plant species, name, health status and growth stage; The planting information includes the planting time, geographical location, planting density and area of ​​the greening area; The maintenance plan includes regular maintenance measures for green areas, irrigation, pruning, fertilization and pest control; The management projects include the project names, task arrangements, timetables and responsible persons related to urban greening management; The unstructured data includes citizen feedback, picture and video data, image data and environmental monitoring data; Said citizen feedback includes citizens’ opinions, complaints and improvement suggestions on green areas; The image and video data include plant growth conditions and pest monitoring; The environmental monitoring data includes soil moisture, temperature, wind speed and air quality.

3. The archive data management system based on semantic analysis according to claim 1, characterized in that: The semantic preprocessing includes data cleaning, formatting and word vector representation; The data cleaning includes removing noise, standardizing data types and removing outliers.

4. The archive data management system based on semantic analysis according to claim 1, characterized in that: Analyze the real-time multi-source data and the real-time urban greening related data to generate a dynamic knowledge graph of urban greening; The urban greening dynamic knowledge graph generation process includes: The entities and entity relationships in the real-time multi-source data are identified through a deep learning model for the urban greening related data; an urban greening dynamic knowledge graph is generated by acquiring the entities and entity relationships in the semantic associations between the data; the entities are represented as nodes in the urban greening dynamic knowledge graph, and the entity relationships are represented as edges in the urban greening dynamic knowledge graph.

5. The archive data management system based on semantic analysis according to claim 1, characterized in that: Perform semantic analysis on the input data through the greening data semantic analysis model to obtain urban greening related data associated with the input data in the urban greening archive data management system; Generate greening management suggestions based on the urban greening related data.

6. The archive data management system based on semantic analysis according to claim 1, characterized in that: Sending early warning information to the administrator based on the greening management suggestions; The warning information includes: analysis of potential pests and diseases, the impact of environmental abnormalities on plants, greening areas that need priority treatment, climate change warnings, and deterioration of plant health.

Images