Regionalization analysis-based scientific culture transmission feasibility evaluation system

Abstract

The invention discloses a scientific culture transmission feasibility evaluation system based on regionalization analysis. The method comprises the following steps: acquiring initial regionalization data in a specific region; the method comprises the steps of obtaining initial regionalization data, constructing four-dimensional feature tags including geographical, time, cultural features and scientific association degrees, carrying out multi-dimensional feature extraction on the initial regionalization data based on the four-dimensional feature tags, carrying out sentiment analysis on the feature regionalization data based on a RoBERTa fine tuning model, generating a regional language feature map in combination with a dialect sensitive word library, and obtaining the regional language feature map. Establishing a regionalized hybrid evaluation model based on GWR (geographically weighted regression), XGBoost and GNN graph neural networks; and inputting the depth regionalization data into the regionalization hybrid evaluation model, and generating a propagation potential thermodynamic diagram and a conflict early warning index. The cultural transmission possibility of regional features such as geographical environment, cultural context and time law is fully considered, and the feasibility and accuracy of cultural transmission are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data analysis, in particular to a scientific culture dissemination feasibility evaluation system based on regional analysis. BACKGROUND

[0002] Scientific knowledge and scientific methods, as an important bridge connecting scientific research innovation and social cognition, their dissemination efficiency directly affects the public scientific literacy and the social transformation of scientific and technological achievements. However, traditional scientific dissemination strategies rely on one-way content output and general evaluation indicators, lacking deep analysis of regional heterogeneity, resulting in unbalanced allocation of dissemination resources, insufficient cultural adaptation and other problems. SUMMARY

[0003] The purpose of the present application is to solve the above problems, and a scientific culture dissemination feasibility evaluation system based on regional analysis is designed.

[0004] To achieve the above purpose, the technical scheme of the present application is that, further, in the scientific culture dissemination feasibility evaluation system based on regional analysis, the scientific culture dissemination feasibility evaluation system comprises the following modules: A regional data acquisition module is used to acquire scientific research databases, social media, open data and dialect data in a specific region to obtain regional data, and the regional data is preprocessed to obtain initial regional data; A multi-dimensional feature extraction module is used to construct four-dimensional feature labels including geography, time, cultural characteristics and scientific correlation degree, and multi-dimensional feature extraction is performed on the initial regional data based on the four-dimensional feature labels to obtain feature regional data; A language deep analysis module is used to perform sentiment analysis on the feature regional data based on a RoBERTa fine-tuning model, and a regional language feature map is generated by combining a dialect sensitive word library to obtain deep regional data; An evaluation model establishment module is used to establish a regional hybrid evaluation model based on GWR geographic weighted regression, XGBoost and GNN graph neural network, and a spatiotemporal Transformer model is introduced to process sequence data with strong spatiotemporal dependence to obtain a target regional hybrid evaluation model; A feasibility evaluation module is used to input the deep regional data into the target regional hybrid evaluation model to generate a dissemination potential heat map and a conflict early warning index.

[0005] Further, in the scientific culture dissemination feasibility evaluation system based on regional analysis, the regional data acquisition module comprises the following sub-modules: A scientific research data acquisition sub-module is used to acquire open experimental data in databases and scientific research institutions to obtain regional scientific research subset data; The media data collection submodule is configured to collect text, picture, and video content data in a social platform in a region to obtain social media subset data. The open data acquisition submodule is configured to connect to an open data platform to acquire population distribution, education level, and cultural facility layout data in the region to obtain open subset data. The dialect data collection submodule is configured to acquire dialect voice data in the region and language data in a dialect database to obtain dialect subset data. The data integration submodule is configured to integrate the regional research subset data, the social media subset data, the open subset data, and the dialect subset data to obtain regionalization data.

[0006] Further, in the scientific and cultural dissemination feasibility evaluation system based on regionalization analysis, the regionalization data acquisition module further includes the following submodules: The cleaning and deduplication submodule is configured to acquire the regionalization data, remove duplicate data and invalid data in the regionalization data, filter noise content irrelevant to scientific and cultural dissemination in the regionalization data through an NLP natural language model, and obtain cleaned regionalization data. The standardization processing submodule is configured to unify time fields in the cleaned regionalization data into ISO standards, convert geographic coordinates into a WGS84 coordinate system, and vectorize and encode unstructured data in the cleaned regionalization data to obtain standardized regionalization data. The missing value filling submodule is configured to fill missing fields in the standardized regionalization data through adjacent regional data interpolation and an ARIMA time series prediction model to obtain initial regionalization data.

[0007] Further, in the scientific and cultural dissemination feasibility evaluation system based on regionalization analysis, the multidimensional feature extraction module includes the following submodules: The spatial feature extraction submodule is configured to construct four-dimensional feature labels including geography, time, cultural features, and scientific correlation degree, perform grid division on the initial regionalization data based on a GIS geographic information system, calculate geographic attribute statistical values in each grid, and generate a spatial feature matrix. The time feature extraction submodule is configured to perform sliding window analysis on time series data in the initial regionalization data, extract trend items, periodic items, and abnormal points, and obtain a time feature matrix. The cultural feature extraction submodule is configured to extract cultural keywords and scientific field keywords from the initial regionalization data based on an LDA topic model, calculate word frequency co-occurrence frequency and regional specificity index, and obtain a scientific and cultural feature matrix. A feature integration module is configured to integrate the spatial feature matrix, the temporal feature matrix, and the scientific and cultural feature matrix to obtain feature regionalization data.

[0008] Further, in the scientific and cultural dissemination feasibility evaluation system based on regionalization analysis, the language depth analysis module comprises the following sub-modules: An emotion recognition sub-module is configured to use a pre-trained RoBERTa fine-tuning model to score social media text in terms of positive and negative emotions, identify the emotional attitude of the public towards scientific topics, calculate the emotional intensity value, distinguish between slight attention and strong emotional expression, and obtain emotion data. A language recognition sub-module is configured to build a regional dialect voice recognition model, convert dialect audio into text through an LSTM deep neural network, and obtain deep regionalization data.

[0009] Further, in the scientific and cultural dissemination feasibility evaluation system based on regionalization analysis, the evaluation model establishment module comprises the following sub-modules: A weight determination sub-module is configured to use geographic coordinates as weights to build a local regression model and generate a spatial weight matrix. A data input sub-module is configured to input pre-processed multi-dimensional features, use an XGBoost tree model to process high-order interactions between features, and output feature importance ranking. A model establishment sub-module is configured to treat each geographic unit as a graph node, with node attributes including spatial and temporal features and cultural features, and edge attributes defined as the correlation between regions. The sub-module learns the propagation dependency relationship between regions through GCN graph convolution operation to obtain a regional hybrid evaluation model.

[0010] Further, in the scientific and cultural dissemination feasibility evaluation system based on regionalization analysis, the evaluation model establishment module further comprises the following sub-modules: A data division sub-module is configured to stratified sample according to regional geographic location, and divide the training set, the validation set, and the test set. A model optimization sub-module is configured to use Bayesian optimization to adjust the bandwidth parameter of WR, the learning rate and tree depth of XGBoost, and the number of layers and embedding dimension of GNN to obtain a regional hybrid evaluation model.

[0011] Further, in the method for implementing the scientific and cultural dissemination feasibility evaluation system based on regionalization analysis, the method comprises the following steps: Obtain scientific research databases, social media, open data, and dialect data in a specific region to obtain regionalization data, and perform data preprocessing on the regionalization data to obtain initial regionalization data. Construct a four-dimensional feature label including geography, time, cultural characteristics and scientific correlation, perform multi-dimensional feature extraction on the initial regionalization data based on the four-dimensional feature label, and obtain feature regionalization data; Perform sentiment analysis on the feature regionalization data based on a RoBERTa fine-tuning model, generate a regional language feature map in combination with a dialect sensitive word library, and obtain deep regionalization data; Establish a regionalization hybrid evaluation model based on GWR geographic weighted regression, XGBoost and GNN graph neural network, introduce a spatiotemporal Transformer model to process sequence data with strong spatiotemporal dependence, and obtain a target regionalization hybrid evaluation model; Input the deep regionalization data into the target regionalization hybrid evaluation model to generate a propagation potential heat map and a conflict early warning index.

[0012] Further, in the method for implementing the scientific and cultural dissemination feasibility evaluation system based on regionalization analysis, the method comprises the following steps: Obtain regional scientific research subset data from open experimental data in databases and scientific research institutions; Obtain social media subset data from text, picture and video content data in social platforms in the region; Obtain open subset data from population distribution, education level and cultural facility layout data in the region by connecting to an open data platform; Obtain dialect subset data from dialect voice data in the region and language data in a dialect database; Integrate the regional scientific research subset data, social media subset data, open subset data and dialect subset data to obtain regionalization data.

[0013] Further, in the method for implementing the scientific and cultural dissemination feasibility evaluation system based on regionalization analysis, the method comprises the following steps: Obtain regionalization data, remove duplicate data and invalid data in the regionalization data, filter noise content irrelevant to scientific and cultural dissemination in the regionalization data through an NLP natural language model, and obtain cleaned regionalization data; Uniformly convert time fields in the cleaned regionalization data into ISO standards, convert geographic coordinates into WGS84 coordinate system, and vectorize and encode unstructured data in the cleaned regionalization data to obtain standard regionalization data; Fill in missing fields in the standard regionalization data by interpolation of adjacent regional data and an ARIMA time series prediction model to obtain initial regionalization data.

[0014] The beneficial effect is that, by obtaining scientific research databases, social media, open data and dialect data in a specific region, obtaining regionalized data, data preprocessing is performed on the regionalized data to obtain initial regionalized data, four-dimensional feature labels including geography, time, cultural characteristics and scientific correlation degree are constructed, multi-dimensional feature extraction is performed on the initial regionalized data based on the four-dimensional feature labels to obtain feature regionalized data, sentiment analysis is performed on the feature regionalized data based on a RoBERTa fine-tuning model, a regional language feature map is generated in combination with a dialect sensitive word library to obtain deep regionalized data, a regionalized hybrid evaluation model is established based on GWR geographic weighted regression, XGBoost and GNN graph neural network, and the deep regionalized data is input into the target regionalized hybrid evaluation model to generate a propagation potential heat map and a conflict early warning index. 1. The cultural transmission possibility of regional characteristics such as geographical environment, cultural context and time law is fully considered, and the feasibility and accuracy of cultural transmission are improved. 2. Combined with dialect and language analysis, scientific cognitive bias behind emotional metaphor can be avoided, and emotional fluctuation characteristics can be focused on, so that the transmission content is matched with the regional cultural psychological needs. 3. According to the acceptance difference of different economic level regions, specific scientific and cultural transmission can be carried out, and the transmission efficiency is improved. BRIEF DESCRIPTION OF DRAWINGS

[0015] Various other advantages and benefits will become apparent to those of ordinary skill in the art upon reading the following detailed description of the preferred embodiments. The drawings are for purposes of illustration only and are not intended to limit the present application.

[0016] Figure 1 The first embodiment schematic diagram of the scientific and cultural transmission feasibility evaluation system based on regionalized analysis in the embodiments of the present application; Figure 2 The second embodiment schematic diagram of the scientific and cultural transmission feasibility evaluation system based on regionalized analysis in the embodiments of the present application; Figure 3 The third embodiment schematic diagram of the scientific and cultural transmission feasibility evaluation system based on regionalized analysis in the embodiments of the present application. DETAILED DESCRIPTION

[0017] In order to make the objects, technical solutions and advantages of the present application clearer, the following further describes the present application in combination with the drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and do not limit the present application.

[0018] As will be understood by one of ordinary skill in the art, the singular forms "a," "an," and "the" used herein include plural referents unless the context clearly dictates otherwise. It should be further understood that the terms "comprises", "comprising", "includes", "including", "has", "having", "contains" and "containing" when used herein, specify the presence of stated features, integers, steps, operations, elements, and / or components but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof.

[0019] The present application will be described in detail below with reference to the accompanying drawings, such as Figure 1 The scientific and cultural dissemination feasibility evaluation system based on regionalization analysis includes the following modules: 101、Regionalization data acquisition module, for acquiring scientific research database, social media, open data and dialect data in a specific region, obtaining regionalization data, and performing data preprocessing on the regionalization data to obtain initial regionalization data; Specifically, the present embodiment further includes a scientific research data acquisition sub-module for acquiring open experimental data disclosed in databases and scientific research institutions to obtain regional scientific research subset data; A media data acquisition sub-module is configured to acquire text, picture, and video content data in social platforms within the region to obtain social media subset data; An open data acquisition sub-module is configured to interface with an open data platform to acquire population distribution, education level, and cultural facility layout data in the region to obtain open subset data; A dialect data acquisition sub-module is configured to acquire dialect voice data in the region and language data in a dialect database to obtain dialect subset data; A data integration sub-module is configured to integrate the regional scientific research subset data, social media subset data, open subset data, and dialect subset data to obtain regionalization data.

[0020] A cleaning and deduplication sub-module is configured to acquire regionalization data, eliminate duplicate data and invalid data in the regionalization data, filter noise content irrelevant to scientific and cultural dissemination in the regionalization data through an NLP natural language model, and obtain cleaned regionalization data; A standardization processing sub-module is configured to unify time fields in the cleaned regionalization data into ISO standards, convert geographic coordinates into a WGS84 coordinate system, and vectorize and encode unstructured data in the cleaned regionalization data to obtain standardized regionalization data; A missing value filling sub-module is configured to fill missing fields in the standardized regionalization data through adjacent regional data interpolation and an ARIMA time series prediction model to obtain initial regionalization data.

[0021] Constructing a stereoscopic regional feature image and breaking through the "dissemination context barrier" Multi-source data integration realizes global coverage and innovative integration of scientific research data, social media UGC, government open data and dialect corpus. In particular, field investigation and complementary collection of existing databases are adopted for dialect data, filling the gap of regional cultural semantic analysis.

[0022] For example, when disseminating agricultural science and technology in the first region, the dialect recognition technology is used to analyze local expressions such as "Bao Valley planting", so that the popular science content is closer to the cognitive habits of farmers 102、Multi-dimensional feature extraction module for constructing four-dimensional feature labels including geography, time, cultural characteristics and scientific correlation degree, and performing multi-dimensional feature extraction on the initial regionalized data based on the four-dimensional feature labels to obtain feature regionalized data; Specifically, the embodiment also includes a spatial feature extraction submodule for constructing four-dimensional feature labels including geography, time, cultural characteristics and scientific correlation degree, and performing grid division on the initial regionalized data based on a GIS geographic information system, calculating the geographical attribute statistical value in each grid, and generating a spatial feature matrix; The time feature extraction submodule is used to perform sliding window analysis on the time series data in the initial regionalized data, extract trend items, periodic items and abnormal points, and obtain a time feature matrix; The cultural feature extraction submodule is used to extract cultural keywords and scientific field keywords from the initial regionalized data based on an LDA topic model, calculate the word frequency co-occurrence frequency and regional specificity index, and obtain a scientific and cultural feature matrix; The feature integration submodule is used to integrate the spatial feature matrix, the time feature matrix and the scientific and cultural feature matrix to obtain feature regionalized data.

[0023] The four-dimensional feature label system accurately locates the dimensions of population density and urban-rural differences in pain point geography, festival cycle in time characteristics, dialect vocabulary distribution in cultural characteristics, and industrial technology matching degree in scientific correlation degree. Cross analysis can quantitatively identify the "advantages" and "constraints" of regional dissemination. For example, the second region locates the problem of "high demand for water resource popular science but low media contact rate in summer busy period", guiding the dissemination mechanism to choose radio, field and field explanation, etc. Adaptation channel, solving the problem of blindness of traditional channel delivery.

[0024] 103、Language depth analysis module for performing sentiment analysis on feature regionalized data based on a RoBERTa fine-tuning model, generating a regional language feature map in combination with a dialect sensitive word library, and obtaining deep regionalized data; Specifically, the embodiment also includes an emotion recognition submodule, which is used for scoring positive and negative emotions of social media text by using a pre-trained RoBERTa fine-tuning model, identifying the emotional attitude of the public towards scientific topics, calculating the emotional intensity value, distinguishing between slight attention and strong emotional expression, and obtaining emotional data. A language recognition submodule is used to build a regional dialect voice recognition model, convert dialect audio into text through an LSTM deep neural network, and obtain deep regionalization data.

[0025] Emotion computing captures the implicit needs of the public through sentiment intensity and tendency analysis of social media text, which can identify potential concerns or points of interest of the public in a region on a specific scientific topic.

[0026] Dialect recognition technology promotes culturally adapted communication. Dialect voice transcription and semantic analysis are supported to convert professional scientific terms into dialect expressions (e.g., "blockchain" is explained as "account book sharing technology"), and to identify scientific cognitive errors in dialects (e.g., incorrect attribution of natural phenomena by dialect proverbs).

[0027] 104. An evaluation model establishment module is used to establish a regionalized hybrid evaluation model based on GWR geographic weighted regression, XGBoost, and GNN graph neural network, introduce a spatiotemporal Transformer model to process sequence data with strong spatiotemporal dependence, and obtain a target regionalized hybrid evaluation model. Specifically, the embodiment also includes a weight determination submodule, which is used to use geographic coordinates as weights to build a local regression model and generate a spatial weight matrix. A data input submodule is used to input preprocessed multi-dimensional features, use the tree model of XGBoost to process high-order interactions between features, and output feature importance ranking. A model establishment submodule is used to regard each geographic unit as a graph node, the node attributes include spatiotemporal features and cultural features, the edge attributes are defined as the correlation between regions, the propagation dependence relationship between regions is learned through GCN graph convolution operation, and a regionalized hybrid evaluation model is obtained.

[0028] GWR captures the spatial heterogeneity of the propagation effect. A local model is built based on geographic weighted regression, which can quantify the differences in the effects of the same influencing factors in different regions.

[0029] GNN models regional correlation and propagation path. Through graph neural network analysis of the correlation between regions such as population flow and traffic connection, the cross-regional diffusion risk of scientific controversy events is predicted.

[0030] Core objective: Direct and efficient modeling and learning of dynamic sequence data with strong spatiotemporal dependence.

[0031] Key technology: Transformer's self-attention mechanism: Capturing Long-Term Dependencies: The Transformer's core advantage is its ability to model the relationship between any two elements in a sequence (regardless of how distant they are) through its self-attention mechanism. In spatiotemporal sequences, this means it can: Learn the correlation between the current sentiment of a region and its own sentiment in the distant past; Learn the correlation between the current sentiment of a region and the sentiment of another distant region at a specific point in the past.

[0032] Parallel computing efficiency: Compared to recurrent networks such as RNN / LSTM that require sequential processing, Transformer can process the entire sequence in parallel, making training faster.

[0033] 105. Feasibility assessment module, used to input deep regionalization data into the target regionalization hybrid assessment model to generate a propagation potential heat map and conflict warning index.

[0034] Specifically, this embodiment also includes a data partitioning submodule for stratified sampling based on regional geographic location, and dividing the data into training sets, validation sets, and test sets; The model optimization submodule is used to use Bayesian optimization to adjust the bandwidth parameters of WR, the learning rate and tree depth of XGBoost, the number of layers and embedding dimension of GNN to obtain a regionalized hybrid evaluation model.

[0035] Its beneficial effects include obtaining regionalized data by accessing scientific research databases, social media, open data, and dialect data from a specific region. This data is then preprocessed to obtain initial regionalized data. Four-dimensional feature labels, including geographic, temporal, cultural, and scientific relevance, are constructed. Multi-dimensional features are extracted from the initial regionalized data based on these four-dimensional feature labels to obtain feature regionalized data. Sentiment analysis is performed on the feature regionalized data using a fine-tuned RoBERTa model, and regional language feature maps are generated by combining them with a dialect-sensitive vocabulary to obtain deep regionalized data. A regionalized hybrid evaluation model is established using GWR geographically weighted regression, XGBoost, and GNN graph neural networks. The deep regionalized data is then input into the regionalized hybrid evaluation model to generate a communication potential heat map and a conflict warning index. 1. This model fully considers the cultural communication potential of regional characteristics such as geographic environment, cultural context, and temporal patterns, improving the feasibility and accuracy of cultural communication. 2. The integration of dialect and language analysis avoids the scientific cognitive bias underlying emotional metaphors, focusing on emotional fluctuations and aligning communication content with regional cultural and psychological needs. 3. It can carry out specific scientific and cultural dissemination based on the differences in acceptance among regions with different economic levels, thereby improving dissemination efficiency.

[0036] See also Figure 2In the feasibility assessment system for scientific and cultural communication based on regional analysis, the regional data acquisition module includes the following submodules: The scientific research data acquisition submodule is used to obtain the experimental open data published in databases and scientific research institutions to obtain regional scientific research subset data; The media data collection submodule is used to collect text, image, and video content data from social platforms in the region to obtain social media subset data; The open data acquisition submodule is used to connect to the public data platform to obtain data on population distribution, education level and cultural facilities layout in the region, and obtain open subset data; The dialect data collection submodule is used to obtain the dialect voice data in the region and the language data in the dialect database to obtain the dialect subset data; The data integration submodule is used to integrate regional scientific research subset data, social media subset data, open subset data and dialect subset data to obtain regionalized data.

[0037] See also Figure 3 ,In the feasibility evaluation system of scientific and cultural communication based on ,regionalization analysis, the multi-dimensional feature extraction module includes the ,following sub-modules: The spatial feature extraction submodule is used to construct four-dimensional feature labels including geographic, temporal, cultural characteristics and scientific relevance. It grids the initial regionalized data based on the GIS geographic information system, calculates the geographic attribute statistics within each grid, and generates a spatial feature matrix. The time feature extraction submodule is used to perform sliding window analysis on the time series data in the initial regionalized data, extract trend items, period items and abnormal points, and obtain the time feature matrix; The cultural feature extraction submodule is used to extract cultural keywords and scientific keywords from the initial regionalized data based on the LDA topic model, calculate the word co-occurrence frequency and regional specificity index, and obtain the scientific and cultural feature matrix; The feature integration submodule is used to integrate the spatial feature matrix, temporal feature matrix and scientific and cultural feature matrix to obtain feature regionalization data.

[0038] Specifically, the present invention can also be implemented using the following solutions: 1. Regionalized data acquisition and preprocessing: 1. Multi-source data collection: Scientific research database: Targetedly capture the academic achievement database (papers, patents, technical reports) of universities and research institutions in the region, integrate the open data of local key laboratories, and simultaneously access regional subset data of general databases such as CNKI and Web of Science.

[0039] Social media data: Collect text, images, and video content from mainstream social platforms (local forums, government new media, and vertical field communities) within the region. Focus on user-generated content (UGC) and official published content (OGC) related to scientific topics.

[0040] Open data: Integrate government open data platforms (regional statistics bureau, science and technology bureau, and cultural and tourism bureau data) to obtain structured data such as population distribution, education level, cultural facility layout, and science and technology policy documents.

[0041] Dialect data: Collect dialect audio corpus (local slang, and dialect expressions for popular science) through fieldwork, integrate existing dialect databases (China Language Resources Protection Engineering Data), and supplement regional minority languages or regional characteristic language materials.

[0042] 2. Data preprocessing: Cleaning and deduplication: Remove duplicate records and invalid data (garbled characters, format errors), and filter out noise content unrelated to science and culture dissemination (pure entertainment topics) through natural language processing (NLP).

[0043] Standardization: Unify data format (time field unified as ISO standard, geographic coordinates converted to WGS84 coordinate system), and vectorize non-structured data (text, audio) (TF-IDF, word embedding).

[0044] Missing value filling: Fill in missing fields through neighboring area data interpolation, time series prediction model (ARIMA), or expert knowledge rules to ensure data integrity.

[0045] II. Four-dimensional feature label design and spatiotemporal feature extraction: 1. Four-dimensional feature label system: Geographical features: Geographical coordinates (latitude and longitude), terrain (plain / mountainous area / water area), population density, and urban-rural differences (urbanization rate) for locating the physical carrier of dissemination and the audience base.

[0046] Temporal features: Time stamp (accurate to hour / day / month / year), periodicity (busy season, winter and summer vacation, local festivals) for capturing the timeliness and seasonality of dissemination.

[0047] Cultural features: Extract dialect features (vocabulary differences, voice tones), folk traditions, cultural identity (attention to local science celebrities), and education level to assess the adaptability of dissemination content to the local cultural context.

[0048] 2. Spatiotemporal feature extraction method: Spatial dimension: Based on geographic information systems (GIS), the data is grid-based (1 km x 1 km grid), and the statistical value of geographic attributes (average population density, cultural facility coverage) in each grid is calculated to generate a spatial feature matrix.

[0049] Temporal dimension: Sliding window analysis is performed on time series data (weekly / monthly aggregation), and trend items (annual growth of scientific topic discussion volume) and periodic items (peak transmission before and after holidays) are extracted.

[0050] Culture and science dimension: Through text mining technology (topic model LDA), cultural keywords (dialect characteristic words) and scientific field keywords ("artificial intelligence" "agricultural technology") are extracted from unstructured data, and word frequency, co-occurrence frequency and regional specificity index are calculated.

[0051] Three, language depth analysis, sentiment calculation and dialect recognition: 1. Sentiment calculation module: Sentiment classification: Use pre-trained sentiment analysis model (BERT sentiment classifier) to score social media text for positive and negative sentiment, and identify public sentiment towards scientific topics.

[0052] Emotion intensity quantification: Combine emoticons, adverbs, punctuation marks (exclamation mark frequency) to calculate emotion intensity value, distinguish between mild attention and strong emotional expression (anger or excitement), and identify potentially controversial sensitive content.

[0053] Sentiment propagation path analysis: Track sentiment polarization (positive and negative views of scientific controversial topics spread), and locate key transmission nodes (influence of opinion leaders' sentiment on the group).

[0054] 2. Dialect recognition technology: Speech-text conversion: Build regional dialect speech recognition model, convert dialect audio to text through deep neural network (LSTM-CTC), support multi-modal data (dialect explanation in short video) semantic analysis.

[0055] Dialect vocabulary detection: Based on rule engine (dialect dictionary matching) and statistical model (n-gram language model) to identify dialect vocabulary in text, mark its regional attribute, and distinguish between Mandarin and dialect mixed expression.

[0056] Cultural semantic analysis: Combine local knowledge base to analyze the deep cultural meaning of dialect text, such as identifying the localization metaphor of scientific concepts in dialect.

[0057] Four, regional hybrid evaluation model construction: 1. Model architecture design: Geographically Weighted Regression (GWR): Use geographical coordinates as weights to build a local regression model, capture the spatial differences in the dissemination of scientific and cultural effects in different regions (the difference in transmission efficiency between mountainous and plain areas), and generate a spatial weight matrix.

[0058] Non-linear feature fusion (XGBoost): Input pre-processed multi-dimensional features (geography, time, culture, scientific correlation), use XGBoost tree model to process high-order interactions between features ("dialect science popularization video + slack season"), output feature importance ranking, and select key influencing factors.

[0059] Regional association modeling (GNN graph neural network): Treat each geographical unit (township) as a graph node, node attributes include spatio-temporal features and cultural features, edge attributes define the correlation between regions (population flow, traffic connection), and learn the propagation dependency relationship between regions (scientific controversy event diffusion risk in adjacent areas) through graph convolution operation (GCN).

[0060] 2、Model training and optimization: Data division: stratified sampling by regional geographical location, divide into training set (70%), validation set (20%), and test set (10%), to avoid model bias caused by spatial autocorrelation.

[0061] Parameter tuning: Use grid search or Bayesian optimization to adjust the bandwidth parameter of GWR, the learning rate and tree depth of XGBoost, and the number of layers and embedding dimension of GNN, with propagation effect prediction accuracy (root mean square error RMSE) and conflict warning accuracy as the objective function.

[0062] Enhance interpretability: Analyze the decision logic of the XGBoost model through SHAP values, visualize the spatial weight distribution of GWR, and generate a report on the contribution of each regional feature to the propagation potential.

[0063] Five, result output and application: 1、Propagation potential heat map: Visualization form: Based on GIS, superimpose geographical base map, show the scientific and cultural propagation potential value of each region through color gradient (red represents high potential, blue represents low potential), superimpose population density, cultural facility distribution and other layers, intuitively present "high potential-high demand" "low potential-high demand" and other key areas.

[0064] Application scenario: Assist in formulating regional dissemination strategies and identifying "dissemination blind areas".

[0065] 2、Conflict warning index: Index system: The early warning index of 0-100 is constructed by the comprehensive emotional polarization degree (the difference of positive and negative emotion proportion), the cultural conflict probability and the scientific controversy heat (the higher the value, the greater the risk).

[0066] Early warning mechanism: When the three-level early warning threshold is triggered, the conflict type (cultural misunderstanding type, technical controversy type) and the occurrence area are automatically associated, the risk report is pushed to the transmission management department, and the intervention measures are prompted.

[0067] The above shows and describes the basic principles, main features and advantages of the present application. Those skilled in the art should understand that the present application is not limited to the above examples, and the above examples and descriptions in the specification are only preferred examples of the present application and are not intended to limit the present application. Without departing from the spirit and scope of the present application, various changes and improvements can be made to the present application, and these changes and improvements all fall within the scope of the claimed present application. The scope of protection of the present application is defined by the appended claims and their equivalents.

Claims

1. A system for assessing the feasibility of scientific and cultural dissemination based on regionalization analysis, characterized by, The scientific and cultural dissemination feasibility evaluation system comprises the following modules: A regionalized data acquisition module, configured to acquire scientific research databases, social media, open data and dialect data in a specific region, obtain regionalized data, and perform data preprocessing on the regionalized data to obtain initial regionalized data; A multi-dimensional feature extraction module, configured to construct a four-dimensional feature label comprising geographical, temporal, cultural feature and scientific correlation degree, perform multi-dimensional feature extraction on the initial regionalized data based on the four-dimensional feature label, and obtain feature regionalized data; A language depth analysis module, configured to perform sentiment analysis on the feature regionalized data based on a RoBERTa fine-tuning model, generate a regional language feature map in combination with a dialect sensitive word library, and obtain deep regionalized data; An evaluation model establishment module, configured to establish a regionalized hybrid evaluation model based on a GWR geographical weighted regression, an XGBoost and a GNN graph neural network, introduce a spatiotemporal Transformer model to process sequence data with strong spatiotemporal dependence, and obtain a target regionalized hybrid evaluation model; A feasibility evaluation module, configured to input the deep regionalized data into the target regionalized hybrid evaluation model to generate a dissemination potential heat map and a conflict early warning index.

2. The regionalized analysis based scientific cultural dissemination feasibility evaluation system according to claim 1, wherein, The regionalized data acquisition module comprises the following sub-modules: A scientific research data acquisition sub-module, configured to acquire open experimental data disclosed in databases and scientific research institutions to obtain regional scientific research subset data; A media data acquisition sub-module, configured to acquire text, picture and video content data in social platforms in a region to obtain social media subset data; An open data acquisition sub-module, configured to connect to a public data platform to acquire population distribution, education level and cultural facility layout data in a region to obtain open subset data; A dialect data acquisition sub-module, configured to acquire dialect voice data in a region and language data in a dialect database to obtain dialect subset data; A data integration sub-module, configured to integrate the regional scientific research subset data, the social media subset data, the open subset data and the dialect subset data to obtain regionalized data.

3. The regionalized analysis based scientific cultural dissemination feasibility evaluation system according to claim 2, wherein, The regionalized data acquisition module further comprises the following sub-modules: A cleaning and deduplication sub-module, configured to acquire regionalized data, eliminate duplicate data and invalid data in the regionalized data, filter noise content irrelevant to scientific and cultural dissemination in the regionalized data through an NLP natural language model, and obtain cleaned regionalized data; A standardization processing sub-module, configured to unify time fields in the cleaned regionalized data into ISO standards, convert geographical coordinates into a WGS84 coordinate system, and vectorize and encode unstructured data in the cleaned regionalized data to obtain standardized regionalized data; A missing value filling sub-module, configured to fill missing fields in the standardized regionalized data through neighboring regional data interpolation and an ARIMA time series prediction model to obtain initial regionalized data.

4. The regionalized analysis based scientific cultural dissemination feasibility evaluation system according to claim 1, wherein, The multi-dimensional feature extraction module comprises the following sub-modules: The spatial feature extraction submodule is configured to construct a four-dimensional feature label including geographical, temporal, cultural features and scientific correlation, perform grid division on the initial regionalized data based on a GIS (Geographic Information System), calculate geographical attribute statistical values in each grid, and generate a spatial feature matrix; The temporal feature extraction submodule is configured to perform sliding window analysis on time series data in the initial regionalized data, extract trend items, periodic items and abnormal points, and obtain a temporal feature matrix; The cultural feature extraction submodule is configured to extract cultural keywords and scientific field keywords from the initial regionalized data based on an LDA (Latent Dirichlet Allocation) topic model, calculate word frequency co-occurrence frequency and regional specificity index, and obtain a scientific and cultural feature matrix; The feature integration submodule is configured to integrate the spatial feature matrix, the temporal feature matrix and the scientific and cultural feature matrix, and obtain feature regionalized data.

5. The regionalized analysis based scientific cultural dissemination feasibility evaluation system according to claim 1, wherein, The language depth analysis module includes the following submodules: The emotion recognition submodule is configured to use a pre-trained RoBERTa fine-tuning model to perform positive and negative sentiment scoring on social media text, identify the emotional attitude of the public towards scientific topics, calculate sentiment intensity values, distinguish between slight attention and strong emotional expression, and obtain emotion data; The language recognition submodule is configured to construct a regional dialect speech recognition model, convert dialect audio to text through an LSTM (Long Short-Term Memory) deep neural network, and obtain deep regionalized data.

6. The regionalized analysis based scientific cultural dissemination feasibility evaluation system according to claim 1, wherein, The evaluation model establishment module includes the following submodules: The weight determination submodule is configured to use geographical coordinates as weights to construct a local regression model and generate a spatial weight matrix; The data input submodule is configured to input preprocessed multi-dimensional features, use an XGBoost tree model to process high-order interactions between features, and output feature importance ranking; The model establishment submodule is configured to regard each geographical unit as a graph node, with node attributes including spatial and temporal features and cultural features, and edge attributes defined as correlation between regions, learn propagation dependency between regions through GCN (Graph Convolutional Network) graph convolution operation, and obtain a regionalized hybrid evaluation model.

7. The regionalized analysis based scientific cultural dissemination feasibility evaluation system according to claim 1, wherein, The evaluation model establishment module further includes the following submodules: The data division submodule is configured to perform stratified sampling according to regional geographical locations, and divide training set, validation set and test set; The model optimization submodule is configured to use Bayesian optimization to adjust the bandwidth parameter of WR, the learning rate and tree depth of XGBoost, and the number of layers and embedding dimension of GNN, and obtain a regionalized hybrid evaluation model.

8. The method of realizing the scientific cultural transmission feasibility evaluation system based on regionalization analysis according to claim 1, characterized in that, The method includes the following steps: Obtain scientific research databases, social media, open data and dialect data in a specific region to obtain regionalized data, and perform data preprocessing on the regionalized data to obtain initial regionalized data; Construct a four-dimensional feature label including geographical, temporal, cultural features and scientific correlation, perform multi-dimensional feature extraction on the initial regionalized data based on the four-dimensional feature label, and obtain feature regionalized data; Perform sentiment analysis on the feature regionalized data based on a RoBERTa fine-tuning model, generate a regional language feature map in combination with a dialect sensitive word library, and obtain deep regionalized data; A regionalized hybrid evaluation model is established based on GWR, XGBoost and GNN, a spatio-temporal Transformer model is introduced to process sequence data with strong spatio-temporal dependence, and a target regionalized hybrid evaluation model is obtained; The deep regionalized data is input into the target regionalized hybrid evaluation model to generate a propagation potential heat map and a conflict early warning index.

9. The method of realizing the scientific cultural transmission feasibility evaluation system based on regionalization analysis according to claim 1, characterized in that, The method further comprises the following steps: Obtain experimental open data in databases and scientific research institutions to obtain regional scientific research subset data; Collect text, picture and video content data in social platforms in the region to obtain social media subset data; Interface with an open data platform to obtain population distribution, education level and cultural facility layout data in the region to obtain open subset data; Obtain dialect voice data in the region and language data in a dialect database to obtain dialect subset data; Integrate the regional scientific research subset data, social media subset data, open subset data and dialect subset data to obtain regionalized data.

10. A method of implementing the system for assessing the feasibility of science and culture communication based on regionalization analysis according to claim 1, characterized in that, The method further comprises the following steps: Obtain regionalized data, remove duplicate data and invalid data in the regionalized data, filter noise content unrelated to scientific and cultural communication in the regionalized data through an NLP natural language model, and obtain cleaned regionalized data; Uniformly set a time field in the cleaned regionalized data to an ISO standard, convert geographic coordinates to a WGS84 coordinate system, vectorize unstructured data in the cleaned regionalized data, and obtain standard regionalized data; Fill in missing fields in the standard regionalized data through adjacent regional data interpolation and an ARIMA time series prediction model to obtain initial regionalized data.

Images