A traditional village resilience comprehensive evaluation system

Abstract

The application discloses a traditional village resilience comprehensive evaluation system, which encodes the resilience aspect data of a traditional village to be evaluated by using a neural network model based on deep learning to extract semantic description coding features of multidimensional data of the resilience aspect of the traditional village to be evaluated, meanwhile, village resilience aspect data labeled as a first resilience evaluation label are called from a background database as reference features, semantic query matching analysis based on resilience aspect multi-source data is carried out on the traditional village to be evaluated and each village labeled as the first resilience evaluation label, whether the traditional village to be evaluated has a resilience level matched with the first resilience evaluation label is intelligently evaluated, and thus, through comparative analysis on the traditional village to be evaluated and the labeled village, the accuracy of the evaluation of the comprehensive resilience of the traditional village to be evaluated can be effectively improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of intelligent evaluation, and more specifically, to a traditional village resilience comprehensive evaluation system. BACKGROUND

[0002] Traditional villages are not only symbols of regional culture, but also important heritage of human civilization development. However, in the rapidly developing modern society, traditional villages are facing a series of challenges, such as frequent natural disasters, economic activity recession, population outflow, and imperfect management mechanism, etc. These problems seriously threaten the survival and development of traditional villages. Therefore, resilience evaluation of traditional villages has become the key to protect and develop traditional villages.

[0003] The existing traditional village resilience evaluation methods mainly include qualitative evaluation and quantitative evaluation. The qualitative evaluation method mainly relies on expert knowledge and experience to obtain evaluation conclusions in a qualitative way. Although this method can reveal the internal characteristics and potential advantages of the village to some extent, it is highly subjective and difficult to quantitatively compare the differences between different villages. The quantitative evaluation method measures the resilience of the village by constructing mathematical models or statistical index systems. This method can provide more objective evaluation results, but there are some limitations in practical application, such as complex index system construction and insufficient model generalization ability. Whether it is qualitative or quantitative evaluation, it is difficult to completely avoid the influence of human factors, especially in the process of setting evaluation standards and weight distribution, subjective judgment may lead to deviation of the evaluation results.

[0004] Therefore, an optimized traditional village resilience comprehensive evaluation system is expected. SUMMARY

[0005] In order to solve the above technical problems, the present application is proposed. The embodiments of the present application provide a traditional village resilience comprehensive evaluation system, which encodes the resilience aspect data of the traditional village to be evaluated by using a neural network model based on deep learning to extract semantic description coding features of multi-dimensional data of the resilience aspect of the traditional village to be evaluated, and simultaneously, retrieves the village resilience aspect data labeled as the first resilience evaluation label from the background database as reference features, and analyzes the semantic query matching based on the resilience aspect multi-source data of the traditional village to be evaluated and each village labeled as the first resilience evaluation label, so as to intelligently evaluate whether the traditional village to be evaluated has a resilience level matching the first resilience evaluation label. In this way, by comparing and analyzing the traditional village to be evaluated and the labeled villages, the accuracy of the comprehensive resilience evaluation of the traditional village to be evaluated can be effectively improved.

[0006] According to an aspect of the present application, a traditional village resilience comprehensive evaluation system is provided, which comprises:

[0007] a data acquisition module configured to acquire a resilience aspect dataset of a traditional village to be evaluated;

[0008] a semantic encoding module configured to perform semantic encoding and structural processing on the resilience aspect dataset of the traditional village to be evaluated to obtain a village resilience aspect multi-source semantic aggregation matrix to be evaluated;

[0009] a first resilience evaluation label extraction module configured to extract a set of village resilience aspect multi-source semantic aggregation reference matrices labeled as first resilience evaluation labels from a background database;

[0010] a semantic correlation encoding module configured to perform semantic correlation encoding on the village resilience aspect multi-source semantic aggregation matrix to be evaluated and the set of village resilience aspect multi-source semantic aggregation reference matrices respectively to obtain a village resilience aspect multi-source semantic aggregation implicit feature vector to be evaluated and a set of village resilience aspect multi-source semantic aggregation implicit reference feature vectors;

[0011] a query encoding module configured to perform cross-domain query encoding on the village resilience aspect multi-source semantic aggregation implicit feature vector to be evaluated relative to the set of village resilience aspect multi-source semantic aggregation implicit reference feature vectors based on the essential features of the set of village resilience aspect multi-source semantic aggregation implicit reference feature vectors to obtain a village resilience aspect semantic query response representation vector;

[0012] a resilience evaluation module configured to determine whether to label the resilience of the traditional village to be evaluated as the first resilience evaluation label based on the village resilience aspect semantic query response representation vector to be evaluated.

[0013] Compared with the prior art, the traditional village resilience comprehensive evaluation system provided by the present application can extract semantic description encoding features of multi-dimensional data of the resilience aspect of the traditional village to be evaluated by performing semantic encoding on the resilience aspect data of the traditional village to be evaluated based on a neural network model of deep learning, and meanwhile, can obtain village resilience aspect data labeled as first resilience evaluation labels as reference features from a background database, and can perform semantic query matching analysis based on the resilience aspect multi-source data on the traditional village to be evaluated and each village labeled as the first resilience evaluation label, so as to intelligently evaluate whether the traditional village to be evaluated has a resilience level matching the first resilience evaluation label. In this way, by performing comparative analysis on the traditional village to be evaluated and the labeled villages, the accuracy of the evaluation of the comprehensive resilience of the traditional village to be evaluated can be effectively improved. BRIEF DESCRIPTION OF DRAWINGS

[0014] The above and other objects, features and advantages of the present application will become more apparent from the following detailed description when taken in conjunction with the accompanying drawings in which:

[0015] Figure 1 A block diagram of a traditional village resilience comprehensive evaluation system according to an embodiment of the present application;

[0016] Figure 2 A data flow schematic diagram of a traditional village resilience comprehensive evaluation system according to an embodiment of the present application;

[0017] Figure 3 A block diagram of a query coding module in a traditional village resilience comprehensive evaluation system according to an embodiment of the present application. DETAILED DESCRIPTION

[0018] Hereinafter, example embodiments according to the present application will be described in detail with reference to the accompanying drawings. It should be apparent to those skilled in the art that the described embodiments are merely a portion of the embodiments of the present application and that the present application is not limited to the described embodiments. Further, in the description of the present application, same or similar components are designated by the same reference numerals throughout the several embodiments.

[0019] As used in the present application and in the claims, the articles "a", "an", and "the" are not limited to refer to only one of the referenced item in the definition, but rather can refer to one or more of the referenced item unless otherwise indicated. In general, the terms "including", "includes" or "comprising", "comprises" when used in the specification, specify the presence of the stated features, elements, steps or components but do not preclude the presence or addition of one or more other features, elements, steps, components or groups thereof.

[0020] Although the present application makes various references to certain modules in the system according to the embodiments of the present application, however, any number of different modules can be used and run on the user terminal and / or server. The modules are merely illustrative and different aspects of the system and method can use different modules.

[0021] Flowcharts have been used herein to illustrate the operations performed by the system according to the embodiments of the present application. It should be understood that the preceding or following operations are not necessarily performed in the exact order as shown. Rather, various steps can be performed in reverse order or simultaneously, as desired. Other operations can also be added to or removed from these processes, or one or more steps can be removed from these processes.

[0022] Hereinafter, example embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only a part but not all of the embodiments of the present application, and the present application can be implemented in many different forms. It should be understood that the present application is not limited to the described embodiments.

[0023] The existing traditional village resilience evaluation methods mainly include qualitative evaluation and quantitative evaluation. The qualitative evaluation method mainly relies on expert knowledge and experience, and obtains evaluation conclusions in a qualitative way. Although this method can reveal the inherent characteristics and potential advantages of the village to a certain extent, it is subjective and difficult to quantitatively compare the differences between different villages. The quantitative evaluation method measures the resilience of the village by constructing a mathematical model or a statistical index system. This method can provide more objective evaluation results, but also has some limitations in practical application, such as complex index system construction and insufficient model generalization ability. Whether it is qualitative or quantitative evaluation, it is difficult to completely avoid the influence of human factors, especially in the process of setting evaluation standards and weight distribution, subjective judgment may lead to deviation of the evaluation results. Therefore, an optimized traditional village resilience comprehensive evaluation system is expected.

[0024] In the technical solution of the present application, a traditional village resilience comprehensive evaluation system is proposed. Figure 1 The block diagram of the traditional village resilience comprehensive evaluation system according to the embodiment of the present application. Figure 2 The data flow schematic diagram of the traditional village resilience comprehensive evaluation system according to the embodiment of the present application. As Figure 1 and Figure 2As shown, the traditional village resilience comprehensive evaluation system 300 according to the embodiments of the present application comprises: a data acquisition module 310, configured to acquire a resilience aspect data set of a traditional village to be evaluated; a village resilience aspect multi-source semantic aggregation module 320, configured to perform semantic coding and structured processing on the resilience aspect data set of the traditional village to be evaluated to obtain a village resilience aspect multi-source semantic aggregation matrix to be evaluated; a first resilience evaluation label extraction module 330, configured to extract a set of village resilience aspect multi-source semantic aggregation reference matrices labeled as first resilience evaluation labels from a background database; a semantic correlation coding module 340, configured to perform semantic correlation coding on the village resilience aspect multi-source semantic aggregation matrix to be evaluated and the set of village resilience aspect multi-source semantic aggregation reference matrices respectively to obtain a village resilience aspect multi-source semantic aggregation implicit feature vector to be evaluated and a set of village resilience aspect multi-source semantic aggregation implicit reference feature vectors; a query coding module 350, configured to perform cross-domain query coding on the village resilience aspect multi-source semantic aggregation implicit feature vector to be evaluated relative to the set of village resilience aspect multi-source semantic aggregation implicit reference feature vectors based on the essential features of the set of village resilience aspect multi-source semantic aggregation implicit reference feature vectors to obtain a village resilience aspect semantic query response representation vector; and a resilience evaluation module 360, configured to determine whether to label the resilience of the traditional village to be evaluated as the first resilience evaluation label based on the village resilience aspect semantic query response representation vector.

[0025] Specifically, the data acquisition module 310 is used to obtain a resilience dataset for the traditional village to be assessed. This resilience data includes descriptions of the physical environment, economic conditions, social and cultural aspects, and management mechanisms. It should be understood that the resilience of a traditional village is influenced by multiple factors, which interact to determine the village's resilience and recovery capabilities. In one example, the physical environment is the foundation for the survival and development of a traditional village, including its geographical location, natural landscape, ecological environment, and infrastructure. By assessing the physical environment of the village to be assessed, we can understand the village's ability to withstand natural disasters, the health of the ecological environment, and the level of infrastructure, providing a basis for improving the physical environment. The economic situation reflects the village's economic vitality and development potential, including income levels, industrial structure, and employment opportunities. By assessing the economic status of the village to be assessed, we can identify its economic strengths and weaknesses. The quality of its economic situation directly affects the quality of life of residents and the sustainable development of the village. Social and cultural factors are the unique charm of traditional villages, including historical heritage, folk customs, and community relationships. By evaluating the social culture of the village to be evaluated, we can understand the status of cultural heritage protection, community cohesion and residents' sense of identity in the village. The management mechanism is the key to ensuring the orderly operation and sustainable development of the village, including governance structure, rules and regulations, community participation, etc. By evaluating the management mechanism of the village to be evaluated, we can discover problems and shortcomings in governance, so as to optimize the management structure and improve governance efficiency. In the technical solution of this application, by analyzing data from multiple dimensions, it is possible to more comprehensively reflect the actual situation of traditional villages and avoid the one-sidedness and limitations caused by single-dimensional evaluation.

[0026] Data on the physical environment primarily comes from geographic information system (GIS) data, remote sensing imagery, environmental monitoring data, and historical disaster records. This data can help us understand a village's location, topography, climate, natural landscape, vegetation cover, water distribution, air quality, water quality, soil quality, and past natural disasters. In one example, relevant data is first collected through various means, including field visits, questionnaires, remote sensing technology, and environmental monitoring equipment. For example, drones can be used for aerial photography to obtain high-resolution images of terrain and vegetation, while environmental monitoring stations can be installed to regularly collect air and water quality data. Next, the collected data is cleaned and standardized to ensure consistency and comparability. Specific steps include deduplication, imputing missing values, and addressing outliers. Finally, the processed data is stored in a database for subsequent analysis and access. A database table containing fields such as location, topography, climate, environmental monitoring data, and historical disaster records can be designed to ensure efficient data management and querying.

[0027] The data on economic conditions is mainly sourced from economic statistics provided by the Bureau of Statistics, local governments, the number and size of enterprises in the village, industry distribution information, and economic activities and consumption habits of villagers through questionnaires and interviews. In an example, first, obtain economic statistics through official channels, conduct a census or sample survey of enterprises in the village. For example, you can download the latest economic census data from the Bureau of Statistics, or collect household income and expenditure information from villagers through questionnaires. Second, analyze the collected economic data statistically to identify the strengths and weaknesses of the village economy. You can use statistical software such as SPSS or R to analyze the data, generate charts and reports, and visually display the economic situation. Finally, integrate economic data from different sources to form a complete economic situation description. You can design a database table containing fields such as per capita income, employment rate, number of enterprises, industrial structure, etc. to ensure data integrity and consistency.

[0028] The data on social culture is mainly sourced from the history of the village, cultural heritage, traditional festivals, and other information, through questionnaires, interviews, and other means to understand the social relationships, community cohesion, and cultural identity of villagers, as well as to consult local chronicles, academic papers, news reports, and other literature. In an example, first, collect social and cultural data through field research, literature review, and community surveys. For example, you can interview local elders to record their oral history, consult local chronicles to understand the historical changes of the village, and conduct a questionnaire survey to understand the villagers' awareness and participation in traditional culture. Second, classify and organize the collected data to ensure data integrity and accuracy. You can design a database table containing fields such as historical evolution, cultural heritage, traditional festivals, community cohesion, etc. to ensure efficient management and query of data. Finally, store the organized data in the database for subsequent analysis and retrieval. You can use a relational database such as MySQL or PostgreSQL to store data to ensure data security and integrity.

[0029] The data on management mechanisms mainly comes from the policy documents, regulations and systems of local governments on village management, community management records of villages (such as meeting minutes, activity records), and villagers' satisfaction and suggestions on community management through questionnaire survey, interview and other ways. In an example, first, obtain government documents through official channels, organize community management records, and carry out resident survey. For example, the latest policy documents can be downloaded from the local government website, community meeting minutes and activity records can be collected; through questionnaire survey, villagers' satisfaction and suggestions on community management can be understood. Second, analyze the collected management mechanism data, identify the problems and shortcomings in management. Text analysis tools such as Python's NLTK library can be used to conduct sentiment analysis on text data to understand villagers' attitudes and opinions on community management. Finally, integrate the management mechanism data from different sources to form a complete management mechanism description. A database table can be designed to include policy documents, community management records, resident satisfaction, etc. to ensure the integrity and consistency of the data.

[0030] In particular, the village resilience aspect multi-source semantic aggregation module 320 to be evaluated is used to perform semantic coding and structured processing on the resilience aspect data set of the to-be-evaluated traditional village to obtain a to-be-evaluated village resilience aspect multi-source semantic aggregation matrix. In a specific example of the present application, first, the description of the physical environment, the description of the economic situation, the description of the social culture and the description of the management mechanism are respectively coded to obtain a set of to-be-evaluated traditional village resilience aspect description semantic coding vectors, wherein the set of to-be-evaluated traditional village resilience aspect description semantic coding vectors includes a physical environment aspect description semantic coding vector, an economic situation aspect description semantic coding vector, a social culture aspect description semantic coding vector and a management mechanism aspect description semantic coding vector. Here, considering that the description of the physical environment, the description of the economic situation, the description of the social culture and the description of the management mechanism all exist in the form of natural language, computers cannot directly understand and process them. Therefore, in the technical solution of the present application, the natural language processing technology is used to code the description of the physical environment, the description of the economic situation, the description of the social culture and the description of the management mechanism respectively, so as to effectively extract and process the semantic information in the resilience aspect text description to obtain the physical environment aspect description semantic coding vector, the economic situation aspect description semantic coding vector, the social culture aspect description semantic coding vector and the management mechanism aspect description semantic coding vector. Then, the set of to-be-evaluated traditional village resilience aspect description semantic coding vectors is arranged in a matrix to organically combine the data from different fields of the to-be-evaluated village resilience aspect (physical environment, economic situation, social culture, management mechanism) to form a comprehensive representation of the multi-source data of the to-be-evaluated village resilience aspect, so as to obtain the to-be-evaluated village resilience aspect multi-source semantic aggregation matrix, so as to facilitate comprehensive analysis and evaluation.

[0031] In particular, the first resilience evaluation label extraction module 330 is configured to extract a set of village resilience aspect multi-source semantic aggregation reference matrices labeled as first resilience evaluation labels from a background database. That is, by obtaining a village resilience aspect dataset labeled as first resilience evaluation labels from the background database; similarly, the village resilience aspect dataset labeled as first resilience evaluation labels is subjected to semantic coding and structured processing to obtain a set of village resilience aspect multi-source semantic aggregation reference matrices labeled as first resilience evaluation labels. The background database stores a large amount of village resilience evaluation data, which has been verified by a large amount of historical experience and data and has reference value for village resilience evaluation. Therefore, in the technical solution of the present application, the resilience aspect data (physical environment, economic status, social culture, management mechanism) of each village labeled as first resilience evaluation labels is used as reference data, which is subjected to the above-mentioned semantic coding and matrix arrangement processing to form a set of village resilience aspect multi-source semantic aggregation reference matrices, thereby providing a comprehensive resilience evaluation benchmark and improving the accuracy of resilience evaluation of the traditional village to be evaluated.

[0032] In particular, the semantic correlation coding module 340 is configured to respectively perform semantic correlation coding on the village resilience aspect multi-source semantic aggregation matrix to be evaluated and the set of village resilience aspect multi-source semantic aggregation reference matrices to obtain a village resilience aspect multi-source semantic aggregation implicit feature vector to be evaluated and a set of village resilience aspect multi-source semantic aggregation implicit reference feature vectors. It should be understood that village resilience evaluation needs to consider multiple dimensions of data, including physical environment, economic status, social culture, governance ability, etc. Single-dimensional evaluation cannot comprehensively reflect the overall situation of the village, thereby affecting the accuracy of the model in analyzing the resilience of the village. Therefore, in the technical solution of the present application, the village resilience aspect multi-source semantic aggregation matrix to be evaluated is input into a resilience aspect multi-source data semantic correlation feature encoder based on a text convolutional neural network model to realize semantic correlation fusion of the multi-source data of the village to be evaluated in the resilience aspect, so as to extract the village resilience aspect multi-source semantic aggregation implicit feature of the traditional village to be evaluated and obtain a village resilience aspect multi-source semantic aggregation implicit feature vector. Similarly, the set of village resilience aspect multi-source semantic aggregation reference matrices is input into a resilience aspect multi-source data semantic correlation feature encoder based on a text convolutional neural network model to respectively extract the semantic correlation aggregation features between the multi-source data in each village resilience aspect multi-source semantic aggregation reference matrix and obtain a set of village resilience aspect multi-source semantic aggregation implicit reference feature vectors, thereby improving the accuracy of the reference data. In particular, the text convolutional neural network can effectively process and integrate multi-source data through sliding convolution operation, capture the complex correlation between the multi-source data of the village resilience aspect, and realize deeper semantic correlation fusion in the feature space, thereby providing a more accurate basis for resilience evaluation.

[0033] In particular, the query encoding module 350 is used to perform cross-domain query encoding of the multi-source semantic aggregation implicit feature vector of the village resilience to be evaluated relative to the set of multi-source semantic aggregation implicit reference feature vectors of the village resilience based on the essential features of the set of multi-source semantic aggregation implicit reference feature vectors of the village resilience to obtain a semantic query response representation vector of the village resilience to be evaluated. Since village resilience assessment involves data of multiple dimensions and multiple modalities, and different village resilience assessment multi-source semantic aggregation features have different feature distributions, traditional methods often find it difficult to effectively integrate these diverse resilience features. Therefore, in the technical solution of the present application, in order to reduce the noise interference between different village resilience features, the present application proposes a cross-domain query encoding method based on essential features, which learns the essential features of the set of multi-source semantic aggregation implicit reference feature vectors of the village resilience, and uses this to perform differential reinforcement and hierarchical modulation on the multi-source semantic aggregation implicit features of each village resilience to enhance the discrimination between the resilience features of different villages, thereby avoiding noise interference and improving the ability to evaluate village resilience. In a specific example of the present application, such as Figure 3 As shown, the query encoding module 350 includes: an essential feature extraction unit 351, which is used to input the set of multi-source semantically aggregated implicit reference feature vectors of the village resilience into the essential feature capture network to obtain the multi-source semantically aggregated reference essential feature representation vector of the village resilience; a semantic expression hierarchical modulation unit 352, which is used to perform semantic expression hierarchical modulation on each multi-source semantically aggregated implicit reference feature vector of the village resilience in the set of multi-source semantically aggregated implicit reference feature vectors of the village resilience based on the multi-source semantically aggregated reference essential feature representation vector of the village resilience to obtain a set of modulated multi-source semantically aggregated implicit reference feature vectors of the village resilience; a cross-domain query encoding unit 353, which is used to perform cross-domain query encoding on the multi-source semantically aggregated implicit feature vector of the village resilience to be evaluated and the set of the modulated multi-source semantically aggregated implicit reference feature vector of the village resilience to obtain the semantic query response representation vector of the village resilience to be evaluated.

[0034] Specifically, the essential feature extraction unit 351 is configured to input the set of village resilience aspect multi-source semantic aggregated implicit reference feature vectors into an essential feature capturing network to obtain a village resilience aspect multi-source semantic aggregated reference essential feature representation vector. That is, the set of village resilience aspect multi-source semantic aggregated implicit reference feature vectors is input into an essential feature capturing network designed to extract the inherent attributes or core structure from data. The essential feature capturing network is adapted to learn and extract complex reference state parameter multi-modal temporal correlation patterns from the set of village resilience aspect multi-source semantic aggregated implicit reference feature vectors and compress them into a fixed-length reference state parameter essential feature representation vector, thereby retaining important information, reducing noise, and providing a reference anchor for subsequent processing. In the technical solution of the present application, the essential feature capturing network can extract the essential features that best reflect the nature of village resilience from multi-source data. These features can more accurately describe the resilience level of the village, thereby improving the quality and credibility of the evaluation, to obtain a village resilience aspect multi-source semantic aggregated reference essential feature representation vector.

[0035] More specifically, in the embodiments of the present application, the specific steps of inputting the set of village resilience aspect multi-source semantic aggregated implicit reference feature vectors into the essential feature capturing network include: first, calculating the semantic difference coefficients of each village resilience aspect multi-source semantic aggregated implicit reference feature vector in the set of village resilience aspect multi-source semantic aggregated implicit reference feature vectors with respect to other village resilience aspect multi-source semantic aggregated implicit reference feature vectors to obtain a set of village resilience aspect multi-source semantic aggregated reference semantic difference coefficients; then, taking the opposite of each village resilience aspect multi-source semantic aggregated reference semantic difference coefficient in the set of village resilience aspect multi-source semantic aggregated reference semantic difference coefficients and performing normalization based on the Softmax function to obtain a set of village resilience aspect multi-source semantic aggregated reference essential semantic relevance factors; and then, taking the set of village resilience aspect multi-source semantic aggregated reference essential semantic relevance factors as a set of weights, calculating the position-weighted sum of the set of village resilience aspect multi-source semantic aggregated implicit reference feature vectors to obtain the village resilience aspect multi-source semantic aggregated reference essential feature representation vector.

[0036] Among them, the specific steps of calculating the semantic difference coefficient of each village resilience multi-source semantic aggregation implicit reference feature vector in the set of village resilience multi-source semantic aggregation implicit reference feature vectors relative to other village resilience multi-source semantic aggregation implicit reference feature vectors include: calculating the mean of the norm of the positional difference vectors between the village resilience multi-source semantic aggregation implicit reference feature vector and other village resilience multi-source semantic aggregation implicit reference feature vectors in the set of village resilience multi-source semantic aggregation implicit reference feature vectors as the village resilience multi-source semantic aggregation reference semantic difference coefficient of the village resilience multi-source semantic aggregation implicit reference feature vector.

[0037] In the above embodiment, the set of implicit reference feature vectors of multi-source semantic aggregation of village resilience is subjected to essential feature extraction using the following essential feature extraction formula to obtain a multi-source semantic aggregation reference essential feature representation vector of village resilience; wherein the essential feature extraction formula is:

[0038] X={v1,v2,...,v i ,...,v n}

[0039]

[0040] Where X represents the set of implicit reference feature vectors of multi-source semantic aggregation in terms of village resilience, v1, v2, v i and v n denote the first, second, i-th, and n-th implicit reference feature vectors of the multi-source semantic aggregation of village resilience in the set of implicit reference feature vectors of the multi-source semantic aggregation of village resilience, n is the number of implicit reference feature vectors of the multi-source semantic aggregation of village resilience, exp(·) denotes exponential operation, ‖·‖1 denotes the vector norm, and D i represents the semantic difference coefficient of the multi-source semantic aggregation reference of village resilience corresponding to the i-th multi-source semantic aggregation implicit reference feature vector, P k A multi-source semantic aggregation reference essential feature representation vector representing the resilience of the village.

[0041] Specifically, the semantic expression hierarchical modulation unit 352 is configured to perform semantic expression hierarchical modulation on each village resilience aspect multi-source semantic aggregation implicit reference feature vector in the set of village resilience aspect multi-source semantic aggregation implicit reference feature vectors based on the village resilience aspect multi-source semantic aggregation reference essential feature representation vector to obtain a set of modulated village resilience aspect multi-source semantic aggregation implicit reference feature vectors. That is, in the embodiment of the present application, first, the village resilience aspect multi-source semantic aggregation reference semantic contribution coefficient between each village resilience aspect multi-source semantic aggregation implicit reference feature vector in the set of village resilience aspect multi-source semantic aggregation implicit reference feature vectors and the village resilience aspect multi-source semantic aggregation reference essential feature representation vector is calculated to determine the specific contribution of each village resilience aspect multi-source semantic aggregation implicit reference feature vector to the overall village resilience aspect multi-source semantic aggregation expression, to obtain a set of village resilience aspect multi-source semantic aggregation reference semantic contribution coefficients. Further, based on the set of village resilience aspect multi-source semantic aggregation reference semantic contribution coefficients, the set of village resilience aspect multi-source semantic aggregation implicit reference feature vectors is hierarchically modulated to pull apart the semantic essential expression differences between each village resilience aspect multi-source semantic aggregation implicit reference feature vector in the set of village resilience aspect multi-source semantic aggregation implicit reference feature vectors, to obtain the set of modulated village resilience aspect multi-source semantic aggregation implicit reference feature vectors. By hierarchically modulating the set of village resilience aspect multi-source semantic aggregation implicit reference feature vectors, important resilience features in the village resilience aspect multi-source semantic aggregation implicit reference feature vector can be given higher weights to enhance the expressiveness of important resilience features, while suppressing the influence of secondary resilience features. By enhancing the expressiveness of important resilience features, the discrimination between different resilience features can be improved, making it easier for the model to identify and distinguish different village resilience features. In this way, the model can better focus on those features that are crucial to the final decision, thereby improving robustness and generalization ability.

[0042] wherein the specific steps of calculating the village resilience aspect multi-source semantic aggregation reference semantic contribution coefficient between each village resilience aspect multi-source semantic aggregation implicit reference feature vector in the set of village resilience aspect multi-source semantic aggregation implicit reference feature vectors and the village resilience aspect multi-source semantic aggregation reference essential feature representation vector include: calculating the Mahalanobis distance between each village resilience aspect multi-source semantic aggregation implicit reference feature vector in the set of village resilience aspect multi-source semantic aggregation implicit reference feature vectors and the village resilience aspect multi-source semantic aggregation reference essential feature representation vector to obtain the set of village resilience aspect multi-source semantic aggregation reference semantic contribution coefficients.

[0043] More specifically, based on the set of village resilience aspect multi-source semantic aggregation reference semantic contribution coefficients, the specific step of hierarchically modulating the set of village resilience aspect multi-source semantic aggregation implicit reference feature vectors comprises: setting a first mask threshold and a second mask threshold, wherein the second mask threshold is twice the first mask threshold; if the feature value in the village resilience aspect multi-source semantic aggregation implicit reference feature vector is greater than the second mask threshold, it is amplified by two times; if the feature value in the village resilience aspect multi-source semantic aggregation implicit reference feature vector is greater than the first mask threshold and less than or equal to the second mask threshold, it is kept unchanged; if the feature value in the village resilience aspect multi-source semantic aggregation implicit reference feature vector is less than or equal to the first mask threshold, it is reduced by two times, thereby obtaining the set of modulated village resilience aspect multi-source semantic aggregation implicit reference feature vectors.

[0044] In the above embodiment, based on the village resilience aspect multi-source semantic aggregation reference essential feature representation vector, each village resilience aspect multi-source semantic aggregation implicit reference feature vector in the set of village resilience aspect multi-source semantic aggregation implicit reference feature vectors is hierarchically modulated by a semantic expression hierarchical modulation to obtain the set of modulated village resilience aspect multi-source semantic aggregation implicit reference feature vectors; wherein the semantic expression hierarchical modulation formula is:

[0045]

[0046] wherein S represents the covariance matrix, C i represents the village resilience aspect multi-source semantic aggregation reference semantic contribution coefficient corresponding to the i-th village resilience aspect multi-source semantic aggregation implicit reference feature vector, mask(·) is the mask processing, θ is the predetermined threshold, v i ’ represents the set of modulated village resilience aspect multi-source semantic aggregation implicit reference feature vectors.

[0047] Specifically, the cross-domain query encoding unit 353 is configured to perform cross-domain query encoding on the to-be-evaluated village resilience aspect multi-source semantic aggregated implicit feature vector and the set of modulated village resilience aspect multi-source semantic aggregated implicit reference feature vectors to obtain the to-be-evaluated village resilience aspect semantic query response representation vector. That is, in the embodiment of the present application, first, the to-be-evaluated village resilience aspect multi-source semantic aggregated implicit feature vector and each modulated village resilience aspect multi-source semantic aggregated implicit reference feature vector in the set of modulated village resilience aspect multi-source semantic aggregated implicit reference feature vectors are input into a cross-domain query encoding attention network to generate attention scores according to the interaction between the to-be-evaluated village resilience aspect multi-source semantic aggregated implicit feature vector and the modulated village resilience aspect multi-source semantic aggregated implicit reference feature vector, reflecting the correlation of different resilience feature combinations, to obtain a set of to-be-evaluated village resilience aspect multi-source semantic aggregated cross-domain query attention coefficients. Then, to ensure the numerical stability of the to-be-evaluated village resilience aspect multi-source semantic aggregated cross-domain query attention weights and make them a smooth and interpretable probability distribution, further, the set of to-be-evaluated village resilience aspect multi-source semantic aggregated cross-domain query attention coefficients is input into a weight conversion network based on the Sigmoid activation function to convert them into weight values within a normalized range to obtain a set of to-be-evaluated village resilience aspect multi-source semantic aggregated cross-domain query attention weights. In this way, it can be ensured that the features in each modulated village resilience aspect multi-source semantic aggregated implicit reference feature vector are properly considered, while avoiding the problem of gradient vanishing or explosion in extreme cases. Subsequently, the set of to-be-evaluated village resilience aspect multi-source semantic aggregated cross-domain query attention weights is used as the set of weights to perform element-wise multiplication and then accumulation operation on the set of village resilience aspect multi-source semantic aggregated implicit reference feature vectors to obtain the to-be-evaluated village resilience aspect semantic query response representation vector. Here, through attention weights, the correlation between the multi-source semantic aggregated features of the to-be-evaluated village and the labeled village can be enhanced, the real-time state optimization query feature representation can be optimized, and the accuracy of subsequent classifier classification can be improved.

[0048] In the above embodiment, the to-be-evaluated village resilience aspect multi-source semantic aggregated implicit feature vector and the set of modulated village resilience aspect multi-source semantic aggregated implicit reference feature vectors are cross-domain query encoded to obtain the to-be-evaluated village resilience aspect semantic query response representation vector by using the following cross-domain query encoding formula; wherein the cross-domain query encoding formula is:

[0049]

[0050] wherein v a represents the to-be-evaluated village resilience aspect multi-source semantic aggregated implicit feature vector, wi represents the i-th said modulated village resilience aspect multi-source semantic aggregated implicit reference feature vector corresponding to the said to-be-evaluated village resilience aspect multi-source semantic aggregated cross-domain query attention coefficient, a i represents the i-th said to-be-evaluated village resilience aspect multi-source semantic aggregated cross-domain query attention weight in the set of said to-be-evaluated village resilience aspect multi-source semantic aggregated cross-domain query attention weights, sigmoid(·) is a sigmoid function, V f represents the said to-be-evaluated village resilience aspect semantic query response representation vector.

[0051] In particular, the said resilience evaluation module 360 is configured to determine whether to label the resilience of the to-be-evaluated traditional village as a first resilience evaluation label based on the said to-be-evaluated village resilience aspect semantic query response representation vector. In one specific example of the present application, the said to-be-evaluated village resilience aspect semantic query response representation vector is input into a classifier-based resilience comprehensive evaluation module to obtain an evaluation result, which is used to represent whether to label the resilience of the to-be-evaluated traditional village as a first resilience evaluation label. In particular, the said classifier mainly learns the features of different evaluation category labels through training data based on machine learning and statistical methods, and classifies the said to-be-evaluated village resilience aspect semantic query response representation vector. Specifically, in the technical solution of the present application, the labels of the classifier include labeling the resilience of the to-be-evaluated traditional village as a first resilience evaluation label (first classification label) and not labeling the resilience of the to-be-evaluated traditional village as a first resilience evaluation label (second classification label), wherein the classifier determines which classification label the said to-be-evaluated village resilience aspect semantic query response representation vector belongs to through a soft-max function. It is worth noting that the said first classification label p1 and the said second classification label p2 here do not contain artificial set concepts. In fact, in the training process, the computer model does not have the concept of “whether to label the resilience of the to-be-evaluated traditional village as a first resilience evaluation label”. It only has two classification labels and outputs the probabilities of the features under these two classification labels, i.e., the sum of p1 and p2 is one. Therefore, the evaluation result of whether to label the resilience of the to-be-evaluated traditional village as a first resilience evaluation label is actually converted into a binary classification probability distribution in accordance with the natural law, and the physical meaning of the natural probability distribution of the label is essentially used, rather than the linguistic text meaning of “whether to label the resilience of the to-be-evaluated traditional village as a first resilience evaluation label”. In this way, multi-dimensional comprehensive evaluation of the resilience of the to-be-evaluated traditional village can be realized, the scientificity and reliability of the evaluation result can be ensured, classification and labeling management can be realized, and scientific basis and support can be provided for policy making and resource allocation.

[0052] In the technical solution of the present application, the village resilience aspect multi-source semantic aggregation implicit feature vector to be evaluated and the set of village resilience aspect multi-source semantic aggregation implicit reference feature vectors represent the semantic correlation features of the resilience aspect data of the traditional village to be evaluated and the set of village resilience aspect multi-source semantic correlation features labeled as the first resilience evaluation label, respectively. However, considering that there may be interference signals or noise components in the set of village resilience aspect multi-source semantic aggregation reference matrix labeled as the first resilience evaluation label, when the village resilience aspect multi-source semantic aggregation implicit feature vector to be evaluated and the set of village resilience aspect multi-source semantic aggregation implicit reference feature vectors are subjected to cross-domain query coding based on intrinsic feature hierarchical modulation, the obtained village resilience aspect semantic query response representation vector may be mixed with interference components, which may cause the dynamic deviation of the overall feature vector class target probability mapping of the village resilience aspect semantic query response representation vector in the high-dimensional feature space, affect the convergence consistency of the classifier, and affect the accuracy of the evaluation result obtained by inputting the village resilience aspect semantic query response representation vector into the resilience comprehensive evaluation module based on the classifier.

[0053] In the technical solution of the present application, the village resilience aspect semantic query response representation vector to be evaluated is input into the resilience comprehensive evaluation module based on the classifier to obtain an evaluation result, which includes:

[0054] Determining the village resilience aspect semantic query response probability value p obtained by inputting the village resilience aspect semantic query response representation vector to be evaluated into the resilience comprehensive evaluation module based on the classifier, wherein the village resilience aspect semantic query response probability value represents the probability of labeling the resilience of the traditional village to be evaluated as the first resilience evaluation label;

[0055] Multiplying the feature mean value of the village resilience aspect semantic query response representation vector to be evaluated by the village resilience aspect semantic query response probability value to obtain the village resilience aspect semantic query response statistical field value n = μp, wherein μ represents the feature mean value of the village resilience aspect semantic query response representation vector to be evaluated;

[0056] Subtracting one from the village resilience aspect semantic query response statistical field value and dividing the village resilience aspect semantic query response statistical field value to obtain the village resilience aspect semantic query response partial probability value ρ = (n-1) / n;

[0057] Calculating the power function V of the village resilience aspect semantic query response representation vector to be evaluated with the village resilience aspect semantic query response partial probability value as the exponent ⊙ρand the semantic query response of the village resilience aspect to be evaluated is multiplied by the partial probability value point to obtain the micro-representation vector V1 of the semantic query response of the village resilience aspect to be evaluated V1 = p o V ⊙ρ wherein, o represents point multiplication;

[0058] After the semantic query response of the village resilience aspect to be evaluated is multiplied by the partial probability value point, the exponential function with a natural constant as the base is calculated to obtain the macro-mapping vector V2 of the semantic query response of the village resilience aspect to be evaluated V2 = exp(V o p);

[0059] After the 2-based logarithmic value of the micro-representation vector of the semantic query response of the village resilience aspect to be evaluated is calculated, the weighted sum of the macro-mapping vector of the semantic query response of the village resilience aspect to be evaluated is obtained to obtain the optimized semantic query response of the village resilience aspect to be evaluated wherein, a and b represent weighted hyperparameters, represents point addition;

[0060] The optimized semantic query response of the village resilience aspect to be evaluated is input into the resilience comprehensive evaluation module based on the classifier to obtain the evaluation result.

[0061] Therefore, the statistical distribution field partial low-order derivative corresponding to the semantic query response of the village resilience aspect to be evaluated is used as the non-overlapping macro-feature representation behavior patch of the semantic query response of the village resilience aspect to be evaluated, so as to strengthen the dynamic sensitivity of the long program sequence micro-complex information distribution of the semantic query response of the village resilience aspect to be evaluated to the macro-representation behavior of the class probability under the non-isotropic main structure of the semantic query response of the village resilience aspect to be evaluated, thereby promoting the class target iterative dynamic consistency between the classification target and the extracted features in the feature space-class probability mapping, so as to improve the accuracy of the evaluation result obtained by inputting the semantic query response of the village resilience aspect to be evaluated into the resilience comprehensive evaluation module based on the classifier.

[0062] As described above, the traditional village resilience comprehensive evaluation system 300 according to the embodiments of the present application can be implemented in various wireless terminals, such as a server with a traditional village resilience comprehensive evaluation algorithm, etc. In one possible implementation, the traditional village resilience comprehensive evaluation system 300 according to the embodiments of the present application can be integrated into a wireless terminal as a software module and / or a hardware module. For example, the traditional village resilience comprehensive evaluation system 300 can be a software module in the operating system of the wireless terminal, or can be an application program developed for the wireless terminal; of course, the traditional village resilience comprehensive evaluation system 300 can also be one of the many hardware modules of the wireless terminal.

[0063] Alternatively, in another example, the traditional village resilience comprehensive evaluation system 300 and the wireless terminal can also be separate devices, and the traditional village resilience comprehensive evaluation system 300 can be connected to the wireless terminal through a wired and / or wireless network, and transmit interactive information in an agreed data format.

[0064] The above has described the embodiments of the present disclosure, and the above description is exemplary, not exhaustive, and is not limited to the disclosed embodiments. Many modifications and changes are obvious to those skilled in the art without departing from the scope and spirit of the described embodiments. The choice of terms used herein is intended to best explain the principles, practical application, or improvement of technology in the market of the embodiments, or to enable other ordinary skilled in the art to understand the embodiments disclosed herein.

Claims

1. A comprehensive evaluation system for the resilience of traditional villages, characterized by: include: The data acquisition module is used to obtain the resilience dataset of the traditional villages to be assessed; A multi-source semantic aggregation module for the resilience of the village to be evaluated, configured to perform semantic coding and structural processing on the resilience dataset of the traditional village to be evaluated to obtain a multi-source semantic aggregation matrix for the resilience of the village to be evaluated; A first resilience evaluation label extraction module is used to extract a set of multi-source semantic aggregation reference matrices of village resilience labeled with the first resilience evaluation label from a backend database; a semantic association coding module, configured to perform semantic association coding on the set of the multi-source semantic aggregation matrix of the village resilience to be evaluated and the set of the multi-source semantic aggregation reference matrix of the village resilience to obtain a set of multi-source semantic aggregation implicit feature vectors of the village resilience to be evaluated and a set of multi-source semantic aggregation implicit reference feature vectors of the village resilience; a query encoding module for encoding a cross-domain query of the multi-source semantically aggregated implicit feature vector of the village resilience to be evaluated relative to the multi-source semantically aggregated implicit reference feature vector of the village resilience to obtain a semantic query response representation vector of the village resilience to be evaluated based on the essential characteristics of the set of multi-source semantically aggregated implicit reference feature vectors of the village resilience; a resilience assessment module, configured to determine whether to mark the resilience of the traditional village to be assessed as a first resilience assessment label based on the semantic query response representation vector of the resilience aspect of the village to be assessed; The query encoding module includes: An essential feature extraction unit is configured to input the set of implicit reference feature vectors of multi-source semantic aggregation of village resilience into an essential feature capture network to obtain a multi-source semantic aggregation reference essential feature representation vector of village resilience; a semantic expression hierarchical modulation unit for performing semantic expression hierarchical modulation on each of the village resilience multi-source semantic aggregation implicit reference feature vectors in the set of village resilience multi-source semantic aggregation implicit reference feature vectors based on the village resilience multi-source semantic aggregation reference essential feature representation vector to obtain a set of modulated village resilience multi-source semantic aggregation implicit reference feature vectors; a cross-domain query encoding unit, configured to perform cross-domain query encoding on a set of the multi-source semantically aggregated implicit feature vector of the village resilience to be evaluated and the modulated multi-source semantically aggregated implicit reference feature vector of the village resilience to be evaluated, so as to obtain a semantic query response representation vector of the village resilience to be evaluated; The essential feature extraction unit is used to: Calculating the semantic difference coefficient of each village resilience multi-source semantic aggregation implicit reference feature vector in the set of village resilience multi-source semantic aggregation implicit reference feature vectors relative to other village resilience multi-source semantic aggregation implicit reference feature vectors to obtain a set of village resilience multi-source semantic aggregation reference semantic difference coefficients; Taking the inverse of each village resilience multi-source semantic aggregation reference semantic difference coefficient in the set of village resilience multi-source semantic aggregation reference semantic difference coefficients and performing normalization processing based on the Softmax function to obtain a set of village resilience multi-source semantic aggregation reference essential semantic correlation factors; Taking the set of multi-source semantic aggregation reference essential semantic relevance factors of the village resilience as the set of weights, the position-weighted sum of the set of multi-source semantic aggregation implicit reference feature vectors of the village resilience is calculated to obtain the multi-source semantic aggregation reference essential feature representation vector of the village resilience.

2. The comprehensive evaluation system for traditional village resilience according to claim 1 is characterized in that: The resilience data include descriptions of the physical environment, economic conditions, social culture and management mechanisms.

3. The comprehensive evaluation system for traditional village resilience according to claim 2 is characterized in that: The multi-source semantic aggregation of the resilience of the villages to be assessed includes: a semantic coding unit, configured to semantically encode the description of the physical environment, the description of the economic situation, the description of the social and cultural aspects, and the description of the management mechanism, respectively, to obtain a set of semantic coding vectors describing the resilience of the traditional village to be evaluated, wherein the set of semantic coding vectors describing the resilience of the traditional village to be evaluated includes a semantic coding vector describing the physical environment, a semantic coding vector describing the economic situation, a semantic coding vector describing the social and cultural aspects, and a semantic coding vector describing the management mechanism; A matrix arrangement unit is used to perform matrix arrangement on the set of semantic coding vectors describing the resilience of the traditional village to be evaluated to obtain a multi-source semantic aggregation matrix of the resilience of the village to be evaluated.

4. The comprehensive evaluation system for traditional village resilience according to claim 3 is characterized in that: The semantic association encoding module is used to: A semantic association feature encoder for multi-source data on resilience based on a text convolutional neural network model is used to perform semantic association encoding on the multi-source semantic aggregation matrix of the village resilience to be evaluated and the set of multi-source semantic aggregation reference matrix of the village resilience to obtain the set of multi-source semantic aggregation implicit feature vectors of the village resilience to be evaluated and the set of multi-source semantic aggregation implicit reference feature vectors of the village resilience.

5. The comprehensive evaluation system for traditional village resilience according to claim 4 is characterized in that: The semantic expression hierarchical modulation unit includes: a semantic contribution coefficient calculation subunit, configured to calculate a village resilience multi-source semantic aggregation reference semantic contribution coefficient between each village resilience multi-source semantic aggregation implicit reference feature vector in the set of village resilience multi-source semantic aggregation implicit reference feature vectors and the village resilience multi-source semantic aggregation reference essential feature representation vector to obtain a set of village resilience multi-source semantic aggregation reference semantic contribution coefficients; The feature modulation subunit is used to hierarchically modulate the set of multi-source semantic aggregation implicit reference feature vectors of the village resilience based on the set of multi-source semantic aggregation reference semantic contribution coefficients of the village resilience to obtain the set of modulated multi-source semantic aggregation implicit reference feature vectors of the village resilience.

6. The comprehensive evaluation system for traditional village resilience according to claim 5 is characterized in that: The cross-domain query encoding unit includes: a cross-domain query attention coefficient calculation subunit, configured to input each modulated village resilience multi-source semantic aggregation implicit reference feature vector in the set of the multi-source semantic aggregation implicit feature vector of the village resilience to be evaluated and the modulated village resilience multi-source semantic aggregation implicit reference feature vector into a cross-domain query encoding attention network to obtain a set of cross-domain query attention coefficients of the multi-source semantic aggregation cross-domain query of the village resilience to be evaluated; A weight conversion subunit, configured to input the set of attention coefficients of the multi-source semantic aggregation cross-domain query on the resilience of the village to be evaluated into a weight conversion network based on a Sigmoid activation function to obtain a set of attention weights of the multi-source semantic aggregation cross-domain query on the resilience of the village to be evaluated; The query optimization sub-unit is used to use the set of multi-source semantically aggregated cross-domain query attention weights of the village resilience to be evaluated as a set of weights, calculate the position-weighted sum of the set of multi-source semantically aggregated implicit reference feature vectors of the village resilience to obtain the semantic query response representation vector of the village resilience to be evaluated.

7. The comprehensive evaluation system for traditional village resilience according to claim 6 is characterized in that: The resilience assessment module is used to: The semantic query response representation vector of the resilience of the village to be evaluated is input into the classifier-based comprehensive resilience evaluation module to obtain an evaluation result, and the evaluation result is used to indicate whether the resilience of the traditional village to be evaluated is marked as the first resilience evaluation label.

Images