Policy information intelligent pushing system and method based on big data analysis

Abstract

The invention discloses a policy information intelligent pushing system and method based on big data analysis, and relates to the technical field of policy information intelligent pushing. Comprising a data acquisition and preprocessing module, a dynamic keyword and semantic analysis module, a cross-domain information analysis and classification module, a user portrait and push strategy adjustment module and an intelligent push and feedback optimization module. According to the method, through multi-source data collection and NLP preprocessing, the data quality and availability are ensured; a keyword library and a semantic model are dynamically generated through deep learning, and accurate matching is achieved; policy information is intelligently analyzed and classified through a cross-domain information fusion technology, so that the correlation of pushed contents is ensured; the pushing strategy is dynamically adjusted in combination with the user portrait and real-time feedback, so that the system adapts to the interest change of the user, the pushing model is continuously optimized, the personalization and the accuracy of policy information pushing are remarkably improved, and the information obtaining efficiency and the decision-making ability of the user are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent policy information push, and in particular to a policy information intelligent push system and method based on big data analysis. Background Art

[0002] The policy information intelligent push system based on big data analysis is a system that integrates multiple advanced technologies to achieve personalized information distribution. The system first aggregates and processes multiple data sources such as government policy documents, industry regulations, and market supervision information through the collection and analysis of massive data. Using natural language processing (NLP) and machine learning technology, the system can automatically identify, classify and analyze these policy information to generate an information library for decision-making. Based on user needs and areas of concern, the system intelligently matches and accurately pushes the most relevant policy information. This not only helps users to grasp the latest policy trends in real time, but also promotes users' rapid response and strategic adjustments under policy changes. The system continuously optimizes algorithms in the background to improve the accuracy and timeliness of push information, ensuring that users always obtain the most valuable policy information, thereby gaining an advantage in the competition. The system is suitable for enterprises, government agencies and individual users, and can effectively reduce information overload and improve the efficiency of information acquisition.

[0003] The prior art has the following deficiencies:

[0004] In the prior art, policy information push systems usually rely on preset keywords or rules to classify and push information. However, when dealing with complex, cross-domain policy information, such static rule-based systems often have difficulty accurately understanding and parsing the contextual relationships and implicit meanings of policy documents, resulting in the information being pushed being inaccurate or inconsistent with user needs. Especially when policies are frequently updated and information sources are diverse, it is difficult for existing systems to adapt and adjust push strategies in a timely manner, and they are unable to meet users' high requirements for personalization, real-time, and accuracy, which in turn affects users' effective use of policy information and decision-making efficiency.

[0005] The above information disclosed in this Background section is only for enhancement of understanding of the background of the present disclosure and therefore it may contain information that does not constitute the prior art that is already known to one of ordinary skill in the art. Summary of the invention

[0006] The purpose of the present invention is to provide a policy information intelligent push system and method based on big data analysis, which ensures data quality and availability through multi-source data collection and NLP preprocessing; uses deep learning to dynamically generate keyword libraries and semantic models to achieve accurate matching; uses cross-domain information fusion technology to intelligently parse and classify policy information to ensure the relevance of pushed content; combines user portraits and real-time feedback to dynamically adjust push strategies so that the system adapts to changes in user interests, continuously optimizes push models, significantly improves the personalization and accuracy of policy information push, and improves the efficiency of user information acquisition and decision-making ability, so as to solve the problems in the above-mentioned background technology.

[0007] In order to achieve the above-mentioned purpose, the present invention provides the following technical solutions: a policy information intelligent push system based on big data analysis, including a data collection and preprocessing module, a dynamic keyword and semantic analysis module, a cross-domain information analysis and classification module, a user portrait and push strategy adjustment module, and an intelligent push and feedback optimization module;

[0008] The data collection and preprocessing module automatically collects various forms of policy-related raw data through multiple data sources. It uses natural language processing technology to perform preliminary preprocessing on the collected raw data, eliminate noise data, convert formats, and unify coding formats to ensure high data quality and availability.

[0009] The dynamic keyword and semantic analysis module uses deep learning technology to build dynamic keyword and semantic analysis models based on preprocessed data, and generates and updates adaptive keyword libraries and semantic analysis models in real time according to users' historical behavior data and interest preferences;

[0010] The cross-domain information analysis and classification module uses cross-domain information fusion technology to intelligently analyze and multi-dimensionally classify policy information from different sources. By building a multi-level topic model, it automatically identifies the implicit contextual relationships and cross-domain related information in policy documents, classifies information into different policy topics, and intelligently matches them based on the user's field of interest.

[0011] The user portrait and push strategy adjustment module uses multi-source data to analyze user portraits and dynamically adjust push strategies based on users' real-time needs and feedback. Through deep learning analysis of user behavior data, it accurately identifies changes in user interests and adjusts push content in real time based on changes in user portraits and needs.

[0012] The intelligent push and feedback optimization module sends the processed information to users through the intelligent push module, and has an embedded feedback mechanism to collect users' interactive data on the pushed content. Based on user feedback data, it further optimizes the push model, continuously learns and adjusts the push strategy, and enhances the accuracy of future pushes and user satisfaction.

[0013] Preferably, natural language processing technology is used for preliminary preprocessing, and the specific steps are as follows:

[0014] Perform noise detection and filtering on the collected raw data;

[0015] After eliminating noisy data, identify and standardize data formats from different sources;

[0016] After format standardization is completed, the encoding format of text data is unified;

[0017] The text data after format standardization and unified encoding is segmented and POS tagged. The text is segmented according to a specific language model, and the continuous text stream is broken down into separate vocabulary units. At the same time, POS tagging is performed, that is, the corresponding part of speech is assigned to each vocabulary.

[0018] Preferably, deep learning technology is used to build a dynamic keyword and semantic analysis model, and an adaptive keyword library and semantic analysis model are generated and updated in real time according to the user's historical behavior data and interest preferences. The specific steps are as follows:

[0019] Comprehensively collect and pre-process users’ historical behavior data;

[0020] After collecting user behavior data, we use deep learning technology to build and train keyword and semantic analysis models;

[0021] Generate a dynamic keyword library based on the trained deep learning model. The model will automatically generate a set of keywords that best reflect the user's current interests based on the user's current behavior data and historical interest preferences. The keywords include the content that the user explicitly searches for, as well as potential related words and topics.

[0022] As user behavior and policy information continue to change, the semantic analysis model is constantly updated in real time. By continuously introducing new user behavior data and policy texts, the system incrementally trains the semantic analysis model and updates the model parameters, enabling the semantic analysis model to adapt to new contexts and content.

[0023] Preferably, cross-domain information fusion technology is applied to intelligently analyze and multi-dimensionally classify policy information from different sources. By building a multi-level topic model, the implicit contextual relationship and cross-domain related information in the policy documents are automatically identified, and the information is classified into different policy topics. Intelligent matching is performed based on the user's field attention. The specific steps are as follows:

[0024] Collect cross-cutting policy-relevant information from multiple heterogeneous data sources;

[0025] After semantic integration is completed, a multi-level topic model is constructed to intelligently analyze and classify cross-domain policy information;

[0026] With the help of the built topic model, the contextual relationships and cross-domain related information in the policy documents are automatically identified. The context of words and sentences in the policy text is analyzed through natural language processing technology to identify implicit relationships.

[0027] After parsing out cross-domain related information, the information is intelligently classified based on the constructed topic model, and matched and pushed according to the user's area of ​​interest.

[0028] Preferably, through deep learning analysis of user behavior data, the user's interest changes can be accurately identified, and the push content can be adjusted in real time according to the changes in user portraits and needs. The specific steps are as follows:

[0029] Collect behavioral data and preference information about users from multiple data sources;

[0030] After integrating multi-source data, deep learning technology is used to analyze user data and build detailed user portraits;

[0031] Based on user portraits and combined with users' real-time needs and feedback, the push strategy is constantly adjusted. When users conduct new searches or interactions on the platform, the system will instantly capture these behaviors and analyze them.

[0032] Based on user portraits and real-time feedback, the system dynamically adjusts push strategies to ensure that the pushed content is highly personalized and accurate. The system updates the pushed content in real time according to the user's latest interests and behavior patterns, and prioritizes policy information that is most relevant to the user's current needs. In addition, the system uses feedback data to continuously optimize the push model, adjust the push frequency, content type and display format through A / B testing and deep learning algorithms, and improve user engagement and satisfaction.

[0033] Preferably, collect the user's interactive data on the pushed content, further optimize the push model based on the user's feedback data, and continuously learn and adjust the push strategy to enhance the accuracy of future push and user satisfaction. The specific steps are as follows:

[0034] Generate intelligent push content based on user portraits and real-time behavior data. By analyzing the user's interests, historical preferences, and current needs, the system selects policy information that is highly relevant to the user from the processed information database. In order to achieve personalized matching, the system will prioritize the topics that the user is interested in and take into account the user's usage habits.

[0035] After the push content is generated, the information is distributed to users through the intelligent push module;

[0036] After push content is distributed, the system has a built-in feedback mechanism to collect user interaction data on the push content in real time;

[0037] Use the collected feedback data to further optimize the push model and adjust future push strategies. Through deep learning algorithms, identify the patterns and associations between push content and user responses, thereby adjusting content generation, distribution channels, push frequency, and display formats.

[0038] The intelligent push method of policy information based on big data analysis includes the following steps:

[0039] Automatically collect various forms of policy-related raw data through multiple data sources, and use natural language processing technology to pre-process the collected raw data, eliminate noise data, convert the format and unify the coding format to ensure the high quality and availability of the data;

[0040] Based on the preprocessed data, deep learning technology is used to build dynamic keywords and semantic analysis models. According to the user's historical behavior data and interest preferences, adaptive keyword libraries and semantic analysis models are generated and updated in real time.

[0041] Preferably, it also includes:

[0042] By applying cross-domain information fusion technology, policy information from different sources is intelligently analyzed and classified in multiple dimensions. By building a multi-level topic model, the implicit contextual relationships and cross-domain related information in policy documents are automatically identified, and the information is classified into different policy topics. Intelligent matching is then performed based on the user's field attention.

[0043] Preferably, it also includes:

[0044] Use multi-source data to analyze user profiles, and dynamically adjust push strategies based on users' real-time needs and feedback. Through deep learning analysis of user behavior data, accurately identify changes in user interests, and adjust push content in real time based on changes in user profiles and needs.

[0045] Preferably, it also includes:

[0046] The processed information is sent to users through the intelligent push module, and a feedback mechanism is embedded to collect users' interaction data on the pushed content. Based on user feedback data, the push model is further optimized, and the push strategy is continuously learned and adjusted to enhance the accuracy of future pushes and user satisfaction.

[0047] In the above technical solution, the technical effects and advantages provided by the present invention are:

[0048] The present invention ensures the high quality and availability of data through automatic collection of multi-source data and preprocessing with NLP technology, dynamically generates and updates keyword libraries and semantic analysis models with the help of deep learning technology, and achieves accurate matching of user needs. Through cross-domain information fusion technology, it intelligently parses and classifies policy information to ensure the relevance and contextual consistency of pushed content. Combined with user portrait analysis and real-time feedback mechanism, it dynamically adjusts the push strategy, enables the system to adapt to changes in user interests, continuously optimizes the push model, significantly improves the personalization and accuracy of policy information push, and improves the efficiency of users in obtaining policy information and their decision-making ability. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0050] Figure 1 This is a module diagram of the policy information intelligent push system based on big data analysis of the present invention.

[0051] Figure 2 This is a method flow chart of the method for intelligently pushing policy information based on big data analysis of the present invention. DETAILED DESCRIPTION

[0052] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be implemented in a variety of forms and should not be construed as limited to the examples set forth herein; rather, these example embodiments are provided so that the description of the present disclosure will be more comprehensive and complete, and the concept of the example embodiments will be fully conveyed to those skilled in the art.

[0053] The present invention provides Figure 1 The policy information intelligent push system based on big data analysis shown in the figure includes a data collection and preprocessing module, a dynamic keyword and semantic analysis module, a cross-domain information analysis and classification module, a user portrait and push strategy adjustment module, and an intelligent push and feedback optimization module;

[0054] The data collection and preprocessing module automatically collects various forms of policy-related raw data through multiple data sources, including but not limited to text, images, audio and video data. The collected raw data is preprocessed using natural language processing (NLP) technology to eliminate noise data, convert formats and unify coding formats to ensure high data quality and availability.

[0055] Natural language processing (NLP) technology is used for preliminary preprocessing to eliminate noise data, convert the format and unify the encoding format. The specific steps are as follows:

[0056] Perform noise detection and filtering on the collected raw data;

[0057] The core of this step is to identify and remove irrelevant information, redundant data, and noise that may affect subsequent processing. For example, policy documents may contain repeated content, meaningless symbols, or format errors. By applying specific filtering algorithms, such as regular expression matching and redundancy detection algorithms, the system can automatically identify noise data. The accuracy of noise filtering directly affects the quality of subsequent data processing, so this step is crucial. Through effective noise data detection and filtering, the system can ensure the purity and effectiveness of input data, laying a solid foundation for subsequent processing.

[0058] After eliminating noisy data, identify and standardize data formats from different sources;

[0059] Policy information usually exists in multiple formats, including PDF, Word documents, HTML pages, and even images or audio files. The system first uses a format recognition tool to automatically detect the original format of each piece of data, and then converts these different formats of data into a unified standard text format, such as plain text or JSON format. This standardization process involves OCR (optical character recognition) technology for image text extraction, speech recognition technology for audio transcription, and file parsing tools for document format conversion. Through format standardization, the system eliminates the complexity of data processing caused by format diversity and ensures the consistency and operability of data in subsequent processing stages.

[0060] After format standardization is completed, the encoding format of text data is unified;

[0061] Since policy information may come from different countries or regions, different character sets may be used, such as UTF-8, ISO-8859-1, etc. This step first detects the character encoding of the original data, and then converts it to the system standard encoding format (usually UTF-8). At the same time, the system also handles possible encoding errors to ensure that all text data is stored and processed under a unified encoding standard. Through unified encoding, the system avoids character display errors or information loss problems that may occur in the subsequent analysis process, ensuring the continuity and stability of data processing.

[0062] Perform word segmentation and part-of-speech tagging on text data that has been standardized and uniformly encoded. Segment the text according to a specific language model (such as the Chinese word segmentation model), break the continuous text stream into separate vocabulary units, and perform part-of-speech tagging, that is, assign a corresponding part of speech (such as noun, verb, adjective, etc.) to each vocabulary word.

[0063] This processing makes subsequent text analysis more accurate and can better understand the structure and semantics of the text content. Through word segmentation and part-of-speech tagging, the system can extract the key elements in the policy document and prepare for further semantic analysis and information extraction.

[0064] This step solves the problem that the existing system cannot efficiently handle diverse data sources, and improves the comprehensiveness and accuracy of information collection.

[0065] The dynamic keyword and semantic analysis module uses deep learning technology to build dynamic keyword and semantic analysis models based on preprocessed data, and generates and updates adaptive keyword libraries and semantic analysis models in real time according to users' historical behavior data and interest preferences;

[0066] This dynamic generation mechanism enables the system to adaptively adjust keywords and semantic weights, avoiding the defect of traditional static rule systems that cannot accurately parse the context of policy documents, thereby improving the relevance and accuracy of pushed information;

[0067] Use deep learning technology to build dynamic keywords and semantic analysis models. Generate and update adaptive keyword libraries and semantic analysis models in real time based on users' historical behavior data and interest preferences. The specific steps are as follows:

[0068] Comprehensively collect and pre-process users’ historical behavior data;

[0069] These data include users’ browsing history, search queries, click behaviors, reading time, favorites and shared content on the policy information platform, and users’ interactions with push information (such as whether they have opened, read, and provided feedback). To ensure the accuracy and consistency of the data, the system cleans and formats the collected data to eliminate noise and redundant data. Through the preprocessing of behavioral data, the system can accurately capture users’ interest preferences and behavioral patterns, laying the foundation for the generation of subsequent keyword and semantic analysis models.

[0070] After collecting user behavior data, we use deep learning technology to build and train keyword and semantic analysis models;

[0071] First, based on historical data, we select a suitable neural network architecture, such as a convolutional neural network (CNN) or a recurrent neural network (RNN), to process and understand the complex semantic relationships in text data. By training on large-scale policy texts and user behavior data, the model can gradually learn how to associate user behavior with text semantics. The key to this step is the continuous optimization and adjustment of the model to ensure that it accurately reflects the interests and needs of users.

[0072] Generate a dynamic keyword library based on the trained deep learning model. The model will automatically generate a set of keywords that best reflect the user's current interests based on the user's current behavior data and historical interest preferences. The keywords include the content that the user explicitly searches for, as well as potential related words and topics.

[0073] The system will monitor changes in user behavior in real time. When the user's interest area changes, the model will dynamically adjust the keyword library, adding or removing keywords that are no longer relevant to ensure the accuracy and timeliness of the keyword library.

[0074] As user behavior and policy information continue to change, the semantic analysis model is constantly updated in real time. By continuously introducing new user behavior data and policy texts, the system incrementally trains the semantic analysis model and updates the model parameters so that it can adapt to new contexts and content.

[0075] During this process, the model automatically adjusts the semantic relationship between words, re-evaluates and weights the relevance of keywords to ensure that the policy information pushed is always highly matched with the user's interests and needs. Through this adaptive update, the system can continue to provide accurate semantic analysis and information push in a dynamically changing environment.

[0076] The cross-domain information analysis and classification module uses cross-domain information fusion technology to intelligently analyze and multi-dimensionally classify policy information from different sources. By building a multi-level topic model, it automatically identifies the implicit contextual relationships and cross-domain related information in policy documents, classifies information into different policy topics, and intelligently matches them based on the user's field of interest.

[0077] Apply cross-domain information fusion technology to intelligently analyze and multi-dimensionally classify policy information from different sources. By building a multi-level topic model, the implicit contextual relationships and cross-domain related information in policy documents are automatically identified, and the information is classified into different policy topics. Intelligent matching is performed based on the user's field attention. The specific steps are as follows:

[0078] Collect cross-cutting policy-relevant information from multiple heterogeneous data sources;

[0079] Policy-related cross-domain information includes but is not limited to laws and regulations, economic policies, industry standards, technical specifications, etc. This information usually comes from various channels such as government websites, industry reports, academic papers, and news media. The system uses semantic integration technology to unify the representation of information in different fields to ensure that policy data in various fields can be processed and understood under the same semantic framework. This step uses knowledge graphs and ontologies to standardize cross-domain terms, so that information from different sources can be compared and integrated in a unified semantic space, laying the foundation for subsequent intelligent analysis.

[0080] After semantic integration is completed, a multi-level topic model is constructed to intelligently analyze and classify cross-domain policy information;

[0081] These models use deep learning technology to automatically extract key topics from a large amount of policy texts and organize these topics in a hierarchical manner. For example, the first level may contain macro policy areas (such as economy, environmental protection, education, etc.), and the second level is further subdivided into specific policy issues (such as tax policies in the economy, carbon emission standards in environmental protection, etc.). Through training, the model can gradually learn how to identify and classify topics at different levels, and can capture the implicit contextual relationships in policy documents. This step ensures that the system can accurately understand the deep meaning and relevance of policy texts.

[0082] With the help of the built topic model, the contextual relationships and cross-domain related information in the policy documents are automatically identified. The context of words and sentences in the policy text is analyzed through natural language processing technology to identify implicit relationships.

[0083] For example, the potential impact of a policy clause on another field. The system also uses cross-domain knowledge graphs to detect and associate interdependencies or conflicts between different policies. For example, how environmental protection policies affect economic policies, or how education policies are related to technological innovation policies. This intelligent parsing step enables the system to effectively process complex, cross-domain policy information, match them with related topics, and provide users with a more comprehensive policy interpretation.

[0084] After parsing out cross-domain related information, intelligently classify the information based on the constructed topic model, and match and push it according to the user's interest areas;

[0085] The system will prioritize pushing policy information that is most relevant to the user's area of ​​concern to the user, and further refine the push content based on the user's historical behavior. During this process, the system dynamically adjusts the classification strategy to ensure that the information received by the user not only meets their current needs, but also covers new policies in related fields or cross-domain related information. Through this refined intelligent matching, the system not only improves the accuracy of information push, but also enhances the user's decision-making support capabilities, enabling users to respond to policy changes in a timely manner.

[0086] This process overcomes the limitations of existing systems in processing complex cross-domain policy information and improves the depth and breadth of information analysis.

[0087] The user portrait and push strategy adjustment module uses multi-source data to analyze user portraits and dynamically adjust push strategies based on users' real-time needs and feedback. Through deep learning analysis of user behavior data, it accurately identifies changes in user interests and adjusts push content in real time based on changes in user portraits and needs.

[0088] Through deep learning analysis of user behavior data, we can accurately identify changes in user interests and adjust push content in real time based on changes in user profiles and needs. The specific steps are as follows:

[0089] Collect behavioral data and preference information about users from multiple data sources;

[0090] These data sources include users' browsing history, search history, click behavior, content interaction (such as likes, comments, sharing, etc.), subscription content, social media activities, and user behavior data in other related applications. The system integrates these multi-source data and forms a complete user behavior data set through deduplication, cleaning and standardization. This step ensures that the system has a comprehensive and accurate understanding of users' behaviors and interests, providing a solid data foundation for subsequent user portrait analysis.

[0091] After integrating multi-source data, deep learning technology is used to analyze user data and build detailed user portraits;

[0092] Deep learning models, such as convolutional neural networks (CNN) or long short-term memory networks (LSTM), can automatically extract high-dimensional features from a large amount of user behavior data. These features include users' long-term interests, short-term preferences, typical behavior patterns, and potential needs. Through model training and optimization, the system can accurately depict the user's interest distribution, behavior habits, and preference tendencies in the user portrait. This step allows the user portrait to not only cover current interests, but also predict the user's future needs and possible changes in interests.

[0093] Based on user portraits and combined with users' real-time needs and feedback, the push strategy is constantly adjusted. When users conduct new searches or interactions on the platform, the system will instantly capture these behaviors and analyze them.

[0094] At the same time, the system will also record and analyze the user's feedback on the pushed content (such as whether it is clicked or ignored, whether there is a negative evaluation). Through real-time data capture and feedback analysis, the system can quickly identify changes in user interests, such as a certain field suddenly attracting a lot of attention from users. This process ensures that the system can respond quickly when user needs change and adjust the relevance and accuracy of the pushed content.

[0095] Based on user portraits and real-time feedback, the system dynamically adjusts push strategies to ensure that the push content is highly personalized and accurate. The system will update push content in real time based on the user's latest interests and behavior patterns, and prioritize policy information that is most relevant to the user's current needs. In addition, the system also uses feedback data to continuously optimize the push model, and adjusts push frequency, content type, and display format through A / B testing and deep learning algorithms to improve user engagement and satisfaction.

[0096] Through this dynamic adjustment mechanism, the system can continuously adapt to changes in user needs and ensure that the pushed content always matches the user's interests and needs.

[0097] This step solves the problem that the existing system is difficult to adapt to changes in user needs in real time, ensures high personalization and accuracy of pushed information, and improves user experience and decision-making efficiency.

[0098] The intelligent push and feedback optimization module sends the processed information to users through the intelligent push module, and has an embedded feedback mechanism to collect user interaction data on the pushed content. Based on user feedback data, the push model is further optimized, and the push strategy is continuously learned and adjusted to enhance the accuracy of future push and user satisfaction.

[0099] Collect user interaction data on pushed content, further optimize the push model based on user feedback data, and continuously learn and adjust the push strategy to enhance the accuracy of future push and user satisfaction. The specific steps are as follows:

[0100] Generate intelligent push content based on user portraits and real-time behavior data. By analyzing the user's interests, historical preferences, and current needs, the system selects policy information that is highly relevant to the user from the processed information database. In order to achieve personalized matching, the system will prioritize the topics that the user is interested in and take into account the user's usage habits.

[0101] The user's usage habits, such as preferred push time periods and information display formats. This step ensures that the content pushed each time can meet the user's needs to the greatest extent possible and enhance the user's willingness to interact and participate.

[0102] After the push content is generated, the information is distributed to users through the intelligent push module. The system will not only select the best push time, but also intelligently select the push channel, such as mobile app notification, email, SMS or social media platform, based on the user's device preferences and platform usage habits. In addition, the system can dynamically adjust the push frequency and channel based on user feedback data to avoid frequent interruptions to users while ensuring that important information is delivered in a timely manner;

[0103] Through multi-channel coverage, the system can reach users more widely and improve the efficiency and effectiveness of information delivery.

[0104] After push content is distributed, the system has a built-in feedback mechanism to collect user interaction data on the push content in real time;

[0105] This data includes whether the user opens the push notifications, how long they read, clicks on links, likes, comments, shares, and any proactive feedback that the user may give (such as satisfaction ratings or complaints about the content). Through this interactive data, the system can fully understand the user's response to the push content, identify which content is popular with users, and which content is ignored or generates negative feedback. Collecting this data is an important basis for optimizing push strategies.

[0106] Use the collected feedback data to further optimize the push model and adjust future push strategies. Through deep learning algorithms, identify patterns and correlations between push content and user reactions, and thus adjust content generation, distribution channels, push frequency, and display formats.

[0107] For example, if a certain type of policy information continues to receive a high click-through rate and reading time, the system will increase the push frequency of such information; on the contrary, if a certain type of information is frequently ignored, the system will reduce the push of such content or change its display method. With the continuous accumulation and analysis of feedback data, the system's push model can continue to learn and improve, further improving the accuracy of push content and user satisfaction.

[0108] This step continuously optimizes system performance through a closed-loop feedback mechanism, ensuring that the pushed information is always highly consistent with user needs, and solving the defects of the existing technology that the push strategy is rigid and cannot adapt to changes.

[0109] The present invention ensures the high quality and availability of data through automatic collection of multi-source data and preprocessing with NLP technology, dynamically generates and updates keyword libraries and semantic analysis models with the help of deep learning technology, and achieves accurate matching of user needs. Through cross-domain information fusion technology, it intelligently parses and classifies policy information to ensure the relevance and contextual consistency of pushed content. Combined with user portrait analysis and real-time feedback mechanism, it dynamically adjusts the push strategy, enables the system to adapt to changes in user interests, continuously optimizes the push model, significantly improves the personalization and accuracy of policy information push, and improves the efficiency of users in obtaining policy information and their decision-making ability.

[0110] The present invention provides Figure 2 The method for intelligently pushing policy information based on big data analysis includes the following steps:

[0111] Automatically collect various forms of policy-related raw data through multiple data sources, and use natural language processing technology to pre-process the collected raw data, eliminate noise data, convert the format and unify the coding format to ensure the high quality and availability of the data;

[0112] Based on the preprocessed data, deep learning technology is used to build dynamic keyword and semantic analysis models. According to the user's historical behavior data and interest preferences, adaptive keyword libraries and semantic analysis models are generated and updated in real time.

[0113] Apply cross-domain information fusion technology to intelligently analyze and multi-dimensionally classify policy information from different sources. By building a multi-level topic model, it automatically identifies the implicit contextual relationships and cross-domain related information in policy documents, classifies information into different policy topics, and intelligently matches them based on the user's field of interest.

[0114] Utilize multi-source data to analyze user profiles, and dynamically adjust push strategies based on users' real-time needs and feedback. Through deep learning analysis of user behavior data, accurately identify changes in user interests, and adjust push content in real time based on changes in user profiles and needs.

[0115] The processed information is sent to users through the intelligent push module, and a feedback mechanism is embedded to collect user interaction data on the pushed content. Based on user feedback data, the push model is further optimized, and push strategies are continuously learned and adjusted to enhance the accuracy of future push and user satisfaction.

[0116] The intelligent policy information push method based on big data analysis provided in an embodiment of the present invention is realized by the above-mentioned intelligent policy information push system based on big data analysis. The specific methods and processes of the intelligent policy information push system based on big data analysis are detailed in the embodiment of the intelligent policy information push method based on big data analysis, which will not be repeated here.

[0117] The above description is only by way of illustration of certain exemplary embodiments of the present invention. It is undoubted that those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the above drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

Claims

1. A policy information intelligent push system based on big data analysis, characterized in that: It includes data collection and preprocessing module, dynamic keyword and semantic analysis module, cross-domain information analysis and classification module, user portrait and push strategy adjustment module, and intelligent push and feedback optimization module; The data collection and preprocessing module automatically collects various forms of policy-related raw data through multiple data sources. It uses natural language processing technology to perform preliminary preprocessing on the collected raw data, eliminate noise data, convert formats, and unify coding formats to ensure high data quality and availability. The dynamic keyword and semantic analysis module uses deep learning technology to build dynamic keyword and semantic analysis models based on preprocessed data, and generates and updates adaptive keyword libraries and semantic analysis models in real time according to users' historical behavior data and interest preferences; The cross-domain information analysis and classification module uses cross-domain information fusion technology to intelligently analyze and multi-dimensionally classify policy information from different sources. By building a multi-level topic model, it automatically identifies the implicit contextual relationships and cross-domain related information in policy documents, classifies information into different policy topics, and intelligently matches them based on the user's field of interest. The user portrait and push strategy adjustment module uses multi-source data to analyze user portraits and dynamically adjust push strategies based on users' real-time needs and feedback. Through deep learning analysis of user behavior data, it accurately identifies changes in user interests and adjusts push content in real time based on changes in user portraits and needs. The intelligent push and feedback optimization module sends the processed information to users through the intelligent push module, and has an embedded feedback mechanism to collect users' interactive data on the pushed content. Based on user feedback data, it further optimizes the push model, continuously learns and adjusts the push strategy, and enhances the accuracy of future pushes and user satisfaction.

2. The policy information intelligent push system based on big data analysis according to claim 1 is characterized in that: Natural language processing technology is used for preliminary preprocessing. The specific steps are as follows: Perform noise detection and filtering on the collected raw data; After eliminating noisy data, identify and standardize data formats from different sources; After format standardization is completed, the encoding format of text data is unified; The text data after format standardization and unified encoding is segmented and POS tagged. The text is segmented according to the language model, and the continuous text stream is broken down into separate vocabulary units. At the same time, POS tagging is performed, that is, the corresponding part of speech is assigned to each word.

3. The policy information intelligent push system based on big data analysis according to claim 1 is characterized in that: Use deep learning technology to build dynamic keywords and semantic analysis models. Generate and update adaptive keyword libraries and semantic analysis models in real time based on users' historical behavior data and interest preferences. The specific steps are as follows: Comprehensively collect and pre-process users’ historical behavior data; After collecting user behavior data, we use deep learning technology to build and train keyword and semantic analysis models; Generate a dynamic keyword library based on the trained deep learning model. The model will automatically generate a set of keywords that best reflect the user's current interests based on the user's current behavior data and historical interest preferences. The keywords include the content that the user explicitly searches for, as well as potential related words and topics. As user behavior and policy information continue to change, the semantic analysis model is constantly updated in real time. By continuously introducing new user behavior data and policy texts, the system incrementally trains the semantic analysis model and updates the model parameters, enabling the semantic analysis model to adapt to new contexts and content.

4. The policy information intelligent push system based on big data analysis according to claim 1 is characterized in that: Apply cross-domain information fusion technology to intelligently analyze and multi-dimensionally classify policy information from different sources. By building a multi-level topic model, the implicit contextual relationships and cross-domain related information in policy documents are automatically identified, and the information is classified into different policy topics. Intelligent matching is performed based on the user's field attention. The specific steps are as follows: Collect cross-cutting policy-relevant information from multiple heterogeneous data sources; After semantic integration is completed, a multi-level topic model is constructed to intelligently analyze and classify cross-domain policy information; With the help of the built topic model, the contextual relationships and cross-domain related information in the policy documents are automatically identified. The context of words and sentences in the policy text is analyzed through natural language processing technology to identify implicit relationships. After parsing out cross-domain related information, the information is intelligently classified based on the constructed topic model, and matched and pushed according to the user's area of ​​interest.

5. The policy information intelligent push system based on big data analysis according to claim 1 is characterized in that: Through deep learning analysis of user behavior data, we can accurately identify changes in user interests and adjust push content in real time based on changes in user profiles and needs. The specific steps are as follows: Collect behavioral data and preference information about users from multiple data sources; After integrating multi-source data, deep learning technology is used to analyze user data and build detailed user portraits; Based on user portraits and combined with users' real-time needs and feedback, the push strategy is constantly adjusted. When users conduct new searches or interactions on the platform, the system will instantly capture these behaviors and analyze them. Based on user portraits and real-time feedback, the system dynamically adjusts push strategies to ensure that the pushed content is highly personalized and accurate. The system updates the pushed content in real time according to the user's latest interests and behavior patterns, and prioritizes policy information that is most relevant to the user's current needs. In addition, the system uses feedback data to continuously optimize the push model, adjust the push frequency, content type and display format through A / B testing and deep learning algorithms, and improve user engagement and satisfaction.

6. The policy information intelligent push system based on big data analysis according to claim 1 is characterized in that: Collect user interaction data on pushed content, further optimize the push model based on user feedback data, and continuously learn and adjust the push strategy to enhance the accuracy of future push and user satisfaction. The specific steps are as follows: Generate intelligent push content based on user portraits and real-time behavior data. By analyzing the user's interests, historical preferences, and current needs, the system selects policy information that is highly relevant to the user from the processed information database. In order to achieve personalized matching, the system will prioritize the topics that the user is interested in and take into account the user's usage habits. After the push content is generated, the information is distributed to users through the intelligent push module; After push content is distributed, the system has a built-in feedback mechanism to collect user interaction data on the push content in real time; Use the collected feedback data to further optimize the push model and adjust future push strategies. Through deep learning algorithms, identify the patterns and associations between push content and user responses, thereby adjusting content generation, distribution channels, push frequency, and display formats.

7. A method for intelligently pushing policy information based on big data analysis, implemented by the intelligently pushing policy information based on big data analysis as claimed in any one of claims 1 to 6, characterized in that: The following steps are involved: Automatically collect various forms of policy-related raw data through multiple data sources, and use natural language processing technology to pre-process the collected raw data, eliminate noise data, convert the format and unify the coding format to ensure the high quality and availability of the data; Based on the preprocessed data, deep learning technology is used to build dynamic keywords and semantic analysis models. According to the user's historical behavior data and interest preferences, adaptive keyword libraries and semantic analysis models are generated and updated in real time.

8. The intelligent policy information push method based on big data analysis according to claim 1 is characterized in that: Also includes: By applying cross-domain information fusion technology, policy information from different sources is intelligently analyzed and classified in multiple dimensions. By building a multi-level topic model, the implicit contextual relationships and cross-domain related information in policy documents are automatically identified, and the information is classified into different policy topics. Intelligent matching is then performed based on the user's field attention.

9. The intelligent policy information push method based on big data analysis according to claim 1 is characterized in that: Also includes: Use multi-source data to analyze user profiles, and dynamically adjust push strategies based on users' real-time needs and feedback. Through deep learning analysis of user behavior data, accurately identify changes in user interests, and adjust push content in real time based on changes in user profiles and needs.

10. The intelligent policy information push method based on big data analysis according to claim 1 is characterized in that: Also includes: The processed information is sent to users through the intelligent push module, and a feedback mechanism is embedded to collect users' interaction data on the pushed content. Based on user feedback data, the push model is further optimized, and the push strategy is continuously learned and adjusted to enhance the accuracy of future pushes and user satisfaction.

Images