Business text abstract generation method and system based on intelligent platform

Abstract

The embodiment of the invention relates to the technical field of artificial intelligence, in particular to a business text summary generation method and system based on an intelligent platform, and the method comprises the steps: obtaining an intelligent government affair business session text flow; mining a service interaction state knowledge vector chain from the intelligent government affair service session text flow according to a service interaction state mining node set; based on the business element thermal evaluation data of each business interaction state knowledge vector in the business interaction state knowledge vector chain, determining the state knowledge confidence of each business interaction state knowledge vector; based on the state knowledge confidence, the intelligent government affair platform system determines the target key session text semantics from the business interaction state knowledge vector chain, and can realize an automatic abstract generation function for the online business session to be processed, thereby improving the efficiency and quality of government affair processing.

Description

Technical Field

[0001] This application relates to the field of artificial intelligence technology, and more specifically, to a method and system for generating business text summaries based on a smart platform. Background Technology

[0002] With the rapid development and widespread adoption of information technology, smart government service platform systems have become an important means for government platforms to provide public services and achieve transparency and efficiency in government affairs. In these systems, processing and analyzing the text flow of government business conversations is a core task, involving accurate understanding and efficient response to interactions between citizens, businesses, and other users and the government service platform.

[0003] Traditional smart government platforms often rely on simple keyword matching or rule-based judgment when processing government business conversations. This approach proves inadequate when dealing with complex and ever-changing government interactions. It fails to fully understand the user's true intentions and needs, and cannot accurately identify key information in the conversation text, thus impacting the efficiency and quality of government services. Summary of the Invention

[0004] To address the technical problems existing in related technologies, this application provides a business text summarization method and system based on an intelligent platform.

[0005] In a first aspect, embodiments of this application provide a method for generating business text summaries based on a smart platform, applied to a smart government affairs platform system. The method includes: acquiring a smart government affairs business conversation text stream; mining a business interaction state knowledge vector chain from the smart government affairs business conversation text stream based on a business interaction state mining node set; wherein the business interaction state mining node set is obtained by integrating at least one business interaction state mining node based on the target government affairs handling task; determining the state knowledge confidence level of each business interaction state knowledge vector based on the business element heat map evaluation data of each business interaction state knowledge vector in the business interaction state knowledge vector chain; wherein the business element heat map evaluation data is used to reflect the heat map statistical results of the business interaction state knowledge vector, and the state knowledge confidence level is used to reflect the importance of the corresponding business interaction state knowledge vector; and determining the target key conversation text semantics from the business interaction state knowledge vector chain based on the state knowledge confidence level.

[0006] Secondly, this application also provides a smart government affairs platform system, including: a memory for storing program instructions and data; and a processor for coupling with the memory to execute the instructions in the memory to implement the method described above.

[0007] Thirdly, this application also provides a computer storage medium containing instructions that, when executed on a processor, implement the above-described method.

[0008] To address these issues, several smart government service platform systems based on deep learning and natural language processing technologies have emerged in recent years. These systems attempt to understand users' language by training numerous models, thus improving processing accuracy to some extent. However, these systems often face challenges such as high model complexity, high computational resource consumption, and difficulty in adapting to changes in government affairs.

[0009] To address the shortcomings of existing technologies, this application proposes a smart government affairs platform system and its processing method. This method introduces the concept of business interaction state mining node sets to extract business interaction state knowledge vector chains from the text stream of smart government affairs business conversations. These node sets are flexibly integrated according to the target government affairs processing tasks, enabling the system to more accurately understand the user's interaction intent and needs. Simultaneously, by introducing business element heat map assessment data and state knowledge confidence levels, the system can further optimize resource allocation and improve processing efficiency.

[0010] In summary, the smart government affairs platform system and its processing method proposed in this application aim to solve the problems and challenges existing in the prior art through innovative technical means, and improve the efficiency and quality of government services. Attached Figure Description

[0011] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the specification, serve to explain the principles of this application.

[0012] Figure 1 This is a flowchart illustrating a business text summary generation method based on a smart platform provided in an embodiment of this application.

[0013] Figure 2 This is a structural block diagram of a smart government affairs platform system provided in an embodiment of this application. Detailed Implementation

[0014] The technical solutions in this application will now be described with reference to the accompanying drawings.

[0015] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application.

[0016] It should be noted that the terms "first," "second," etc., in the specification and the above-mentioned figures of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence.

[0017] The methods and embodiments provided in this application can be executed in a smart government platform system, computer equipment, or similar computing device. Taking a smart government platform system as an example, the smart government platform system may include one or more processors (processors may include, but are not limited to, microprocessors such as MCUs or programmable logic devices such as FPGAs) and a memory for storing data. Optionally, the smart government platform system may also include a transmission device for communication functions. Those skilled in the art will understand that the above structures are merely illustrative and do not limit the structure of the smart government platform system. For example, the smart government platform system may also include more or fewer components than shown above, or have a different configuration than shown above.

[0018] The memory can be used to store computer programs, such as application software programs and modules, like the computer program corresponding to a business text digest generation method based on a smart platform in this application embodiment. The processor executes various functional applications and data processing by running the computer program stored in the memory, thus implementing the above-described method. The memory may include high-speed random access memory, and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory may further include memory remotely located relative to the processor, and these remote memories can be connected to the smart government platform system via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0019] The transmission device is used to receive or send data via a network. Specific examples of the network mentioned above may include a wireless network provided by the communication provider of the smart government platform system. In one example, the transmission device includes a Network Interface Controller (NIC), which can connect to other network devices via a base station to communicate with the Internet. In another example, the transmission device may be a Radio Frequency (RF) module, used for wireless communication with the Internet.

[0020] Based on this, please refer to Figure 1 , Figure 1 This is a flowchart illustrating a business text summary generation method based on a smart platform provided in this application embodiment. The method is applied to a smart government affairs platform system and may further include steps 110-130.

[0021] Step 110: The smart government affairs platform system acquires the smart government affairs business conversation text stream; and mines the business interaction state knowledge vector chain from the smart government affairs business conversation text stream based on the business interaction state mining node set; wherein, the business interaction state mining node set is obtained by integrating at least one business interaction state mining node based on the target government affairs handling task.

[0022] Step 120: The smart government affairs platform system determines the state knowledge confidence level of each business interaction state knowledge vector based on the business element heat map evaluation data of each business interaction state knowledge vector in the business interaction state knowledge vector chain; wherein, the business element heat map evaluation data is used to reflect the heat map statistical results of the business interaction state knowledge vector, and the state knowledge confidence level is used to reflect the importance of the corresponding business interaction state knowledge vector.

[0023] Step 130: Based on the state knowledge confidence level, the smart government affairs platform system determines the semantics of the target key conversation text from the business interaction state knowledge vector chain.

[0024] In its daily operation, the smart government affairs platform system processes a large volume of text streams containing government business conversations. These text streams contain various interactive information between users, businesses, and the government, and are an important component of government services. The following section details how the smart government affairs platform system processes these text streams and extracts valuable information from them.

[0025] First, the smart government affairs platform system acquires these smart government affairs business conversation text streams. These text streams may originate from various channels such as online government service platforms, social media, and email. After acquiring these text streams, the system processes them using a pre-defined set of business interaction state mining nodes. These node sets are integrated based on the target government affairs handling tasks, and they can identify and process various business interaction states in the text streams.

[0026] By processing the node set of business interaction states, the smart government platform system can extract a knowledge vector chain of business interaction states from the text stream. This vector chain consists of a series of business interaction state knowledge vectors, each representing a business interaction state in the text stream.

[0027] Next, the system will determine the state knowledge confidence level of each vector in the business interaction state knowledge vector chain based on the business element heatmap assessment data. The business element heatmap assessment data is obtained by statistically analyzing business elements in the text stream, reflecting the statistical results of the heatmap of the business interaction state knowledge vectors. The state knowledge confidence level is calculated based on these statistical results and reflects the importance of the corresponding business interaction state knowledge vector.

[0028] Finally, the smart government platform system determines the semantics of target key conversation texts from the business interaction state knowledge vector chain based on the confidence level of the state knowledge. These target key conversation texts are the most important parts of the text stream, potentially containing crucial information such as user and enterprise needs, questions, and suggestions. By determining these target key conversation texts, the smart government platform system can more accurately understand the needs of users and enterprises, thereby providing more efficient and precise government services.

[0029] In summary, the smart government affairs platform system acquires and processes the text stream of smart government affairs business conversations, mines the knowledge vector chain of business interaction states, and then determines the semantics of key target conversation texts based on the confidence level of the vector state knowledge. This achieves a deep understanding and efficient processing of government affairs business conversations. This not only helps improve the efficiency and quality of government services but also provides strong data support for government decision-making.

[0030] In other application scenarios, we can explore more deeply how the smart government affairs platform system processes the text stream of smart government affairs business conversations and extract valuable information from it.

[0031] In step 110, the smart government affairs platform system first acquires smart government affairs business conversation text streams from multiple channels (such as online government service platforms, social media, email, etc.) in real time or periodically. These text streams are raw records of interactions between users, enterprises, and government departments, containing a wealth of information. Once these text streams are acquired, the system processes them using a business interaction state mining node set. This node set is built based on advanced technologies such as machine learning and natural language processing (NLP), and it can identify and process various business interaction states in the text, such as inquiries, applications, and complaints. The mining node set performs preprocessing operations on the text stream, such as word segmentation, part-of-speech tagging, and named entity recognition, to extract key information from the text. Then, based on this information, the system constructs a business interaction state knowledge vector chain. Each knowledge vector represents a business interaction state in the text stream, containing information such as the characteristics and attributes of that state.

[0032] In step 120, after obtaining the business interaction state knowledge vector chain, the smart government platform system further analyzes these vectors to determine their state knowledge confidence levels. This is achieved by evaluating the heatmap of business elements within the vectors. The business element heatmap evaluation data is obtained by statistically analyzing business elements (such as keywords and phrases) in the text stream. The frequency, position, and contextual relationships of these elements in the text are all taken into account. Displaying the statistical results of these elements using a heatmap allows for a visual understanding of which elements are more important in the text. The state knowledge confidence level is calculated based on the business element heatmap evaluation data. It reflects the importance of the corresponding business interaction state knowledge vector in the text stream. Vectors with high confidence levels represent core information in the text, while vectors with low confidence levels may represent secondary or irrelevant information.

[0033] In step 130, the smart government platform system finally determines the target key conversation text semantics from the business interaction state knowledge vector chain based on the state knowledge confidence level. These semantics represent the most valuable information in the text stream, directly reflecting the needs and intentions of users and businesses. Determining the target key conversation text semantics may involve comprehensive analysis and comparison of multiple vectors. The system selects the most important semantics based on factors such as vector confidence and relevance. These semantics may include user inquiries, business applications, and complaint content. Once the target key conversation text semantics are determined, the smart government platform system can take corresponding actions based on this information. For example, it can automatically respond to user inquiries, process business applications, and forward complaint content to relevant departments. In this way, the system can process government business conversation text streams more efficiently, improving the efficiency and quality of government services.

[0034] The detailed steps outlined above provide a clearer understanding of how the smart government platform system processes the text stream of smart government business conversations and extracts valuable information from it. This process involves several advanced technologies and methods, such as machine learning, natural language processing, and statistical analysis. The application of these technologies enables the system to more accurately understand the needs of users and businesses, providing strong support for government services.

[0035] The above technical solution will be introduced through a more practical application scenario.

[0036] Application scenario: Intelligent handling of citizen complaints Mr. Li, a resident, found the park near his community to be dirty and messy, with garbage left uncollected for a long time. He decided to submit a complaint through the smart government affairs platform, hoping that the relevant departments could handle it as soon as possible.

[0037] 1) Submit a complaint Mr. Li accessed the online complaint system of the smart government affairs platform via his mobile phone or computer, filled in the complaint details, including the specific location of the park, a description of the garbage accumulation, and uploaded photos of the scene as evidence.

[0038] 2) Obtain the text stream After receiving Mr. Li's complaint, the smart government affairs platform system incorporated it into the smart government affairs business conversation text stream. This text stream contained Mr. Li's complaint content, a timestamp, and other relevant metadata.

[0039] 3) Mining knowledge vector chains The system uses business interaction state mining node sets to process complaint texts. These node sets identify key information in the text, such as complaint type (environmental sanitation), complaint object (park), problem description (garbage accumulation), etc., and construct a business interaction state knowledge vector chain based on this information.

[0040] 4) Determine the confidence level of state knowledge The system further analyzes each vector in the knowledge vector chain and determines their state knowledge confidence level based on the business element heat map data. In this example, because the complaint is clear and the environmental sanitation issues involved are also a hot topic of public concern, the relevant vectors have a high confidence level.

[0041] 5) Determine the semantics of key conversation texts Based on the confidence level of state knowledge, the system determines the semantics of the target key conversation text from the knowledge vector chain, namely, the citizen's complaint about the park's environmental sanitation. This is the most valuable information in the text stream, directly reflecting the citizen's needs and intentions.

[0042] 6) Automatic classification and forwarding Based on the identified key semantic information in the conversation, the smart government affairs platform system automatically categorizes the complaint as an environmental sanitation issue and forwards it to the relevant departments for processing. Simultaneously, the system generates an automated reply informing Mr. Li that his complaint has been received and is being processed.

[0043] 7) Processing and Feedback Upon receiving the complaint, the relevant departments will promptly take action to clean up the park's litter and improve its environmental sanitation. Once the matter is resolved, the department will send feedback to Mr. Li through the smart government platform, informing him of the outcome and expressing gratitude.

[0044] This application scenario demonstrates how the smart government platform system intelligently handles citizen complaints using the aforementioned technological solutions. The system accurately identifies and understands citizens' needs and intentions, automatically categorizes and forwards complaints to relevant departments for processing, thereby improving the efficiency and quality of government services. Simultaneously, the system can provide timely feedback to citizens, enhancing interaction and trust between the government and citizens.

[0045] Next, the technical terms appearing in steps 110-130 above will be explained.

[0046] Smart government business conversation text streams refer to the streams of text information generated during business interactions between users (including citizens and businesses) and government staff or service systems on smart government service platforms provided by the government. These text streams include various business conversation content such as user inquiries, applications, feedback, and complaints. For example, in an online tax consultation system, there is a continuous text stream consisting of questions raised by taxpayers, answers from tax officials, and automatically generated prompts from the system.

[0047] Business interaction state mining node set: A set of processing nodes set up to analyze and mine the business interaction state in the text stream of smart government business conversations. Each node is responsible for identifying, extracting, or processing specific information or patterns in the text stream. For example, in a social security business processing system, the business interaction state mining node set may include nodes for identifying the type of user inquiry, nodes for extracting key information, and nodes for judging user intent, etc.

[0048] Business interaction state knowledge vector chain: This is a data structure composed of a series of knowledge vectors representing different business interaction states, obtained after processing the business interaction state mining node set. Each knowledge vector contains information such as features and attributes of the corresponding business interaction state. For example, when processing a conversational text stream of a citizen applying for a passport, the business interaction state knowledge vector chain may contain knowledge vectors representing different interaction states such as "inquiring about the passport application process," "submitting application materials," and "confirming application information."

[0049] Target government service tasks: These are specific government services provided by the government, such as household registration, tax declaration, and social security payment. These tasks typically require users to submit logical text through the smart government service platform and undergo processing by the government department before completion. For example, a business user's tax declaration task through the smart government service platform includes steps such as filling out the declaration form, uploading relevant supporting documents, and confirming the declaration information.

[0050] Business Interaction State Mining Node: A single node in the business interaction state mining node set, responsible for processing specific parts or types of business interaction states in the smart government business session text stream. Each node has its unique function and processing logic. For example, in a housing provident fund query system, a business interaction state mining node might be specifically responsible for extracting housing provident fund account information from the user's query statement for subsequent query operations.

[0051] In smart government platform systems, integration specifically refers to the process of unified management and collaborative work among multiple business interaction state mining nodes. This means that different nodes need to achieve data sharing, exchange, and interoperability to ensure that the entire system can efficiently and accurately handle complex government business conversation text flows. For example, a smart government platform needs to handle business inquiries involving multiple fields such as taxation, social security, and housing provident fund. To achieve this goal, the system needs to integrate mining nodes responsible for identifying tax-related interaction states, social security-related interaction states, and housing provident fund-related interaction states. In this way, when a user raises a cross-domain inquiry, the system can coordinate the work of various nodes to extract all key information related to the question and provide a comprehensive and accurate answer. By integrating business interaction state mining nodes, smart government platform systems can better cope with complex and ever-changing government business needs, improving service quality and efficiency. At the same time, this also requires the system to have powerful data processing capabilities, flexible architecture design, and clear business process planning.

[0052] Business interaction state knowledge vectors: In a smart government platform system, business interaction state knowledge vectors are data structures used to represent and store knowledge related to business interaction states. Each knowledge vector contains information such as the characteristics, attributes, and relationships of a specific business interaction state, which helps the system understand and process complex government business conversations. For example, in the business process of a citizen applying for a passport, an interaction state of "submitting application materials" can be represented as a knowledge vector, which contains attribute information such as the type, quantity, and format of the application materials.

[0053] Business Element Heatmap Assessment Data: Business element heatmap assessment data refers to data used to assess the importance and attention of business elements (such as keywords and phrases) within the text flow of smart government business conversations by analyzing statistical information such as frequency, location, and contextual relationships. This data is typically presented as a heatmap, visually reflecting the popularity of each business element. For example, in a tax consultation text flow, "tax filing deadline" and "tax incentive policies" might be frequently occurring and highly relevant to users; their heatmap assessment data would reflect their significant position within the text.

[0054] State knowledge confidence: State knowledge confidence refers to the degree of trust or reliability that a smart government platform system has in a knowledge vector representing the state of a specific business interaction. It is calculated based on information such as business element heatmap data and is used to assign different weights to different knowledge vectors during system decision-making. For example, if a knowledge vector represents a business interaction state that appears frequently in the text stream and is closely related to the user's core needs, then its state knowledge confidence will be relatively high.

[0055] Heatmap Statistical Results: Heatmap statistical results are statistical analysis data presented in a heatmap format to visually reflect the popularity distribution of different business elements in the text flow of smart government service conversations. The intensity of the color or the magnitude of the value in the heatmap represents the importance and attention level of each business element. For example, in the heatmap statistical results of a social security service consultation text flow, it can be clearly seen that elements such as "pension eligibility conditions" and "social security contribution ratio" are the most frequently asked questions by users.

[0056] Importance: In the context of a smart government platform system, importance typically refers to the significance and influence of a particular business interaction state, business element, or information within the overall business process or user needs. It can be quantified and assessed in various ways, such as frequency statistics and user feedback. For example, in the visa application process, information such as the "visa application materials checklist" and "visa processing time" are of high importance to users because they directly relate to the success of the application and the user's travel arrangements.

[0057] Target key conversation text semantics: Target key conversation text semantics refers to the most representative and valuable core semantic information identified through system analysis and mining within the text flow of smart government business conversations. This information typically reflects the user's real needs, concerns, or key aspects of the business process. For example, in a citizen complaint text flow, target key conversation text semantics might include key information points such as "complaint content," "complaint recipient," and "expected resolution method," which are crucial for handling complaints and resolving problems.

[0058] The smart government platform system of this application has the following beneficial effects in processing smart government business conversation text streams: Improving the efficiency of government affairs processing: By introducing a set of business interaction state mining nodes, the system can automatically mine the knowledge vector chain of business interaction states from the text stream of smart government affairs business conversations. This processing method greatly reduces the workload of manual analysis and processing of text streams, thereby improving the efficiency of government affairs processing; Ensuring information accuracy: The construction of the business interaction state knowledge vector chain is based on a set of nodes mined from the business interaction state. These node sets are integrated according to the target government affairs handling tasks. This design ensures that the mined information is closely related to the target government affairs, thus improving the accuracy of the information. Optimized resource allocation: The system determines the state knowledge confidence level of each knowledge vector based on the heat map assessment data of business elements in the business interaction state knowledge vector. This allows the system to prioritize processing those more important and urgent business interaction states, optimizing resource allocation and further improving processing efficiency; Enhancing User Experience: By accurately and quickly determining the semantics of key conversational texts, the system can more precisely understand users' needs and intentions, thereby providing more personalized and timely service responses, significantly improving the user experience. Enhanced system flexibility: The smart government affairs platform system of this application can adapt to the changing needs of different government affairs handling tasks by integrating and consolidating different business interaction status mining nodes, demonstrating good scalability and flexibility.

[0059] In summary, the smart government affairs platform system of this application significantly improves the efficiency and quality of government affairs processing through automated data mining, optimized resource allocation, and precise understanding of user needs, while also enhancing the system's flexibility and user-friendliness.

[0060] In some preferred embodiments, the business interaction state mining node set includes at least one or more business interaction state mining nodes selected from periodic linkage processing nodes, item linkage processing nodes, user linkage processing nodes, scenario linkage processing nodes, or adjustable processing nodes. The step of mining a business interaction state knowledge vector chain from the smart government business session text stream based on the business interaction state mining node set includes at least one of the following: mining first business interaction state information continuously generated by online business sessions within a target activation period from the smart government business session text stream through periodic linkage processing nodes, and determining the business interaction state knowledge vector of the first business interaction state information; or, mining second business interaction state information continuously generated between different online business sessions from the smart government business session text stream through item linkage processing nodes, and determining... The second business interaction state information is a business interaction state knowledge vector; or, the third business interaction state information, which is continuously and repeatedly generated in the online business session, is mined from the smart government business session text stream through a user linkage processing node, and the business interaction state knowledge vector of the third business interaction state information is determined; or, the fourth business interaction state information, in which the online business session achieves question-and-answer dialogue interaction in the target task process, is mined from the smart government business session text stream through a scene linkage processing node, and the business interaction state knowledge vector of the fourth business interaction state information is determined; or, the fifth business interaction state information generated by the online business session under the target government affairs handling task is mined from the smart government business session text stream through an adjustable processing node, and the business interaction state knowledge vector of the fifth business interaction state information is determined.

[0061] In some preferred embodiments, the business interaction state mining node set of the smart government platform system is designed to be more detailed and comprehensive to address business interaction needs under different circumstances. This node set includes multiple business interaction state mining nodes, such as periodic linkage processing nodes, item linkage processing nodes, user linkage processing nodes, scenario linkage processing nodes, and adjustable processing nodes. These nodes can mine corresponding business interaction state knowledge vector chains from the smart government business conversation text stream based on different business interaction characteristics.

[0062] Specifically, the system can extract the first business interaction status information continuously generated within a target activation period from the text stream of smart government business conversations by periodically linking processing nodes. This "target activation period" can be a preset time period, such as one day, one week, or one month. Within this time period, the system continuously monitors and analyzes online business conversations, extracts the interaction status information related to the period, and determines the corresponding business interaction status knowledge vector accordingly.

[0063] In addition, the system can also extract second business interaction status information continuously generated between different online business sessions from the text stream of smart government business sessions through the linked processing nodes. Here, "different online business sessions" refers to multiple business sessions initiated by multiple users or multiple business sessions initiated by the same user at different time periods. The system will analyze the correlation and interaction status between these sessions, extract the interaction status information related to the matter, and determine the corresponding business interaction status knowledge vector accordingly.

[0064] In addition, the user-linked processing node is responsible for extracting continuously and repeatedly generated third-party business interaction state information from the text stream of smart government business sessions. Here, "continuously and repeatedly generated" refers to the interaction states that the same user repeatedly mentions or performs within the same business session. The system identifies and analyzes these repeated interaction states and information, extracting interaction state knowledge vectors related to user behavior.

[0065] The scenario-linked processing node focuses on extracting the fourth business interaction state information from the text stream of smart government business conversations, specifically the question-and-answer dialogue interaction within the target task process. Here, "target task process" refers to the specific procedures and steps a user takes to handle government affairs. The system analyzes the user's question-and-answer dialogue interaction states and information within these procedures and steps, extracting scenario-related interaction state knowledge vectors.

[0066] Finally, adjustable processing nodes are a more flexible type of node, which can be customized and adjusted according to the specific needs of handling government affairs tasks. Through adjustable processing nodes, the system can extract the fifth business interaction state information generated by the online business session under the target government affairs task from the text stream of the smart government affairs business conversation. This information may involve multiple aspects such as user needs, business processes, policies and regulations, and the system will determine the corresponding business interaction state knowledge vector based on this information.

[0067] In summary, by mining these different types of business interaction state nodes, the smart government platform system can more comprehensively and accurately extract business interaction state knowledge vector chains from the text stream of smart government business conversations. This provides strong data support for subsequent business processing and summary generation, helping to improve the efficiency and quality of government services.

[0068] In some other possible embodiments, the step of mining the first business interaction state information continuously generated by the online business session within the target activation period from the smart government business session text stream through the periodic linkage processing node, and determining the business interaction state knowledge vector of the first business interaction state information, includes at least one of the following: setting the periodic constraint value of the periodic linkage processing node; obtaining the detection results of each business interaction state information continuously generated by the online business session within the target activation period of the periodic constraint value; and determining the business interaction state knowledge vector of the first business interaction state information based on the detection results; wherein the detection results include at least one of the following: the number of updates of each business interaction state information, the maximum state maintenance stage of each business interaction state information, the minimum state maintenance stage of each business interaction state information, or the generation result of each business interaction state information.

[0069] In some other possible embodiments, the smart government affairs platform system uses periodically linked processing nodes to mine the first business interaction state information continuously generated by online business sessions within the target activation period, and determines the corresponding business interaction state knowledge vector accordingly. This process specifically includes the following steps.

[0070] First, the system sets a periodic constraint value for the periodically linked processing nodes. This periodic constraint value is a preset time period that defines the length of the target activation period. For example, if the periodic constraint value is set to one day, the system will monitor the business interaction status information generated continuously by online business sessions throughout the day.

[0071] Next, the system will acquire the detection results of the various business interaction state information continuously generated by the online business session within the target activation period of the periodic constraint value. These detection results are obtained through real-time monitoring and analysis of the online business session, and they reflect the changes in the various business interaction states within the target activation period.

[0072] The detection results may include various data such as the number of updates to the state information of each business interaction, the maximum state maintenance period, the minimum state maintenance period, and the generated result. Among them, the number of updates indicates how many times a certain business interaction state information is updated within the target activation period; the maximum state maintenance period and the minimum state maintenance period indicate the longest and shortest state period that a certain business interaction state information is maintained within the target activation period, respectively; and the generated result reflects the final state of a certain business interaction state information within the target activation period.

[0073] Then, the system determines the business interaction state knowledge vector of the first business interaction state information based on these detection results. This process may involve operations such as statistical analysis and summarization of the detection results to extract key interaction state information related to the target government affairs handling task. For example, the system may determine the importance and urgency of a certain business interaction state information based on the number of updates and the data of the state maintenance phase, and then generate the corresponding business interaction state knowledge vector.

[0074] Finally, these business interaction state knowledge vectors will be used in subsequent business processing and summary generation tasks. They can help the system more accurately understand user needs and intentions, improving the efficiency and quality of government services. Furthermore, because these knowledge vectors are derived from actual online business session data, they are also highly practical and targeted.

[0075] To facilitate understanding of the above embodiments, a more detailed exemplary description is provided below.

[0076] In a smart government platform system, a periodic linkage processing node is a key component. It is responsible for extracting business interaction status information within a specific period from the text stream of online business sessions. This information is crucial for understanding user behavior and optimizing service processes.

[0077] 1. Set the periodic constraint value for the periodic linkage processing node. Periodic Constraint Value: This is a time parameter used to define the "target activation period" of online business sessions that the system is interested in. In other words, it specifies the time range within which the system analyzes the session text stream. For example, if the periodic constraint value is set to 24 hours, the system will only analyze session data from the past 24 hours.

[0078] System administrators or automated scripts will set this periodic constraint value based on business needs. Different government affairs may require different periodic constraint values ​​in order to more accurately capture dynamic changes in user interactions.

[0079] 2. Obtain the detection results of online business sessions. Once the periodic constraint value is set, the system will begin monitoring and analyzing online business sessions within the target activation period. These sessions may originate from multiple channels (such as websites, mobile applications, social media, etc.) and involve different users and business scenarios.

[0080] Business interaction status information: This refers to the information left by users' interactions with the system or other users during the process of handling government affairs. This information may include various forms such as text, voice, and images, reflecting the user's operation, query, and feedback status.

[0081] Detection Results: These are the results obtained by the system after analyzing the business interaction status information. It may contain data from multiple dimensions, such as the number of updates and the state maintenance phase.

[0082] Update frequency: This refers to the number of times a certain business interaction status information is modified or updated within the target activation period. Frequent updates may indicate that users are highly concerned about the matter or have encountered some problems.

[0083] Maximum / Minimum State Maintenance Phase: This refers to the longest / shortest time period during which the state information of a certain business interaction remains unchanged within the target activation cycle. This helps the system identify the stable and changing periods of user interaction.

[0084] 3. Determine the knowledge vector of business interaction status based on the detection results. Business interaction state knowledge vector: This is a multi-dimensional data structure used to represent key interaction state information mined from online business sessions. It may include multiple aspects such as user behavior patterns, demand preferences, and service satisfaction.

[0085] The system will construct or update the knowledge vector of business interaction status based on the detection results. For example, if the update frequency of a certain business interaction status information is abnormally high, the system may mark it as "high concern" or "potential problem" and add the corresponding weight to the knowledge vector.

[0086] These knowledge vectors are crucial for subsequent business processing. They help the system more accurately understand user needs, predict user behavior, and optimize service processes. For example, by identifying frequently asked questions and needs, the system can automatically recommend relevant solutions or guide users to more suitable service channels.

[0087] Through the detailed explanations and definitions above, we can see the crucial role of the periodic linkage processing node in the smart government platform system. It not only extracts valuable information from massive online business sessions but also transforms this information into actionable knowledge vectors, providing strong support for improving the quality and efficiency of government services.

[0088] In an alternative design approach, determining the state knowledge confidence level of each business interaction state knowledge vector based on the business element heatmap evaluation data of each business interaction state knowledge vector in the business interaction state knowledge vector chain includes: creating a first reference key text content and a second reference key text content based on the target government affairs handling task, wherein the first reference key text content consists of business interaction state knowledge vectors that do not have popular topics in the target government affairs handling task, and the second reference key text content consists of business interaction state knowledge vectors that have popular topics in the target government affairs handling task; determining the business element heatmap evaluation data of each business interaction state knowledge vector in the business interaction state knowledge vector chain in the first reference key text content and the second reference key text content; and determining the state knowledge confidence level of each business interaction state knowledge vector based on the business element heatmap evaluation data of the business interaction state knowledge vectors.

[0089] Under an alternative design approach, the smart government platform system will evaluate the state knowledge confidence of business interaction state knowledge vectors using a specific method. This process is primarily based on the heatmap evaluation data of business elements within each business interaction state knowledge vector chain. The detailed implementation steps are as follows.

[0090] First, the system creates two sets of reference key text content based on the target government service task: the first reference key text content and the second reference key text content. These two sets of text content differ significantly in composition. The first reference key text content primarily consists of business interaction state knowledge vectors that are not relevant to the target government service task. In other words, it contains relatively niche or less-discussed information points. The second reference key text content, on the other hand, is composed of business interaction state knowledge vectors that are relevant to the target government service task, containing information points that are of general public concern or frequently discussed.

[0091] Next, the system will determine the business element heatmap assessment data for each business interaction state knowledge vector in the business interaction state knowledge vector chain within the two reference key text contents. Here, "business element heatmap assessment data" can be understood as the degree of matching or relevance between a certain business interaction state knowledge vector and the reference key text content. It may be obtained through algorithms such as text analysis and semantic comparison, and the specific value reflects the importance or influence of that knowledge vector within the corresponding text content.

[0092] Finally, based on the heatmap data of these business elements, the system determines the state knowledge confidence level of each business interaction state knowledge vector. This confidence level can be seen as a comprehensive evaluation index of the accuracy and reliability of a business interaction state knowledge vector. It may comprehensively consider the performance of the knowledge vector in multiple dimensions, such as its relevance to popular topics, its frequency of appearance in text content, and its influence on other information. In this way, the system can more comprehensively evaluate the value of each business interaction state knowledge vector, providing stronger support for subsequent business processing and decision-making.

[0093] It is important to note that the evaluation method based on this design relies on high-quality reference key text content and accurate and effective business element heatmap evaluation algorithms. Therefore, in practical applications, the system may need to continuously optimize and adjust these reference texts and evaluation algorithms to adapt to the needs and changes of different government affairs processing tasks. Furthermore, to improve the accuracy and reliability of the evaluation results, the system can also consider incorporating other auxiliary information or expert opinions for comprehensive analysis.

[0094] In the following steps, determining the business element heatmap evaluation data of each business interaction state knowledge vector in the business interaction state knowledge vector chain within the first and second reference key text contents includes: calculating the cumulative value of each business interaction state knowledge vector in the business interaction state knowledge vector chain existing in the first reference key text content as a first detection variable, and calculating the cumulative value of each business interaction state knowledge vector in the business interaction state knowledge vector chain existing in the second reference key text content as a second detection variable; determining the total number of business interaction state knowledge vectors contained in the first reference key text content as a third detection variable, and determining the total number of business interaction state knowledge vectors contained in the second reference key text content as a fourth detection variable; based on the... Based on the first and third detection variables, a first calculation result is determined between the cumulative value of each business interaction state knowledge vector in the business interaction state knowledge vector chain existing in the first reference key text content and the total number of business interaction state knowledge vectors contained in the first reference key text content; based on the second and fourth detection variables, a second calculation result is determined between the cumulative value of each business interaction state knowledge vector in the business interaction state knowledge vector chain existing in the second reference key text content and the total number of business interaction state knowledge vectors contained in the second reference key text content; based on the first and second calculation results, business element heat map evaluation data of each business interaction state knowledge vector in the first and second reference key text contents is determined.

[0095] In the following steps, the system will determine the business element heatmap assessment data in the first and second reference key text contents of the business interaction state knowledge vector through a series of calculations and evaluation operations. The detailed implementation steps are as follows.

[0096] First, the system will calculate the cumulative value of each business interaction state knowledge vector in the business interaction state knowledge vector chain within the first reference key text content. This value is called the first detection variable. This cumulative value may be obtained by calculating the frequency, duration, and importance weight of the knowledge vector in the text content, reflecting the coverage and influence of the knowledge vector in the first reference key text content.

[0097] Similarly, the system also calculates the cumulative value of each business interaction state knowledge vector in the business interaction state knowledge vector chain within the second reference key text content. This value is called the second detection variable. The calculation method for this cumulative value is similar to that of the first detection variable, but it applies to the second reference key text content.

[0098] Next, the system will determine the total number of business interaction state knowledge vectors contained in the first reference key text content; this number is called the third detection variable. This total number reflects the richness and scope of the first reference key text content.

[0099] Similarly, the system will also determine the total number of business interaction state knowledge vectors contained in the second reference key text content; this number is referred to as the fourth detection variable. This total number also reflects the richness and scope of the second reference key text content.

[0100] Then, based on the first and third detection variables, the system calculates the first operation result between the cumulative value of each business interaction state knowledge vector in the business interaction state knowledge vector chain existing in the first reference key text content and the total number of business interaction state knowledge vectors contained in the first reference key text content. This operation result may be a ratio, ranking, or other form of numerical value used to measure the relative importance and influence of the knowledge vector in the first reference key text content.

[0101] Similarly, based on the second and fourth detection variables, the system calculates a second operation result between the cumulative value of each business interaction state knowledge vector in the business interaction state knowledge vector chain present in the second reference key text content and the total number of business interaction state knowledge vectors contained in the second reference key text content. This operation result is also used to measure the relative importance and influence of the knowledge vector in the second reference key text content.

[0102] Finally, based on the results of the first and second calculations, the system will determine the business element heatmap assessment data for each business interaction state knowledge vector within the first and second reference key text contents. This assessment data may be a comprehensive score, rating, or other form of indicator used to comprehensively evaluate the overall performance and value of the knowledge vector within the two reference key text contents.

[0103] Through this calculation and evaluation process, the system can more accurately understand the importance and influence of each business interaction state knowledge vector in different text contents, providing stronger data support for subsequent business processing and decision-making.

[0104] To facilitate understanding, the calculation process can be explained in more detail using a formula. The following is a formulaic representation based on the aforementioned description: 1. Define variables (V_{ik}): The cumulative value of the (i)th knowledge vector in the business interaction state knowledge vector chain in the (k)th reference key text content ((k=1) is the first reference key text content, (k=2) is the second reference key text content).

[0105] (T_k): The total number of business interaction state knowledge vectors contained in the (k)th reference key text content.

[0106] 2. Calculate the detection variables For each knowledge vector (i), we need to calculate its cumulative value (V_{i1}) and (V_{i2}) in the two reference key text contents, as well as the total number of knowledge vectors in the two reference key text contents (T_1) and (T_2).

[0107] 3. Calculate the results The first result ((R_{i1})) represents the importance of the (i)th knowledge vector in the first reference key text content. It can be calculated using the following formula: [R_{i1}=frac{V_{i1}}{T_1}] Here, (R_{i1}) is the ratio of the cumulative value of the (i)th knowledge vector in the first reference key text content to the total number, reflecting the relative importance of the knowledge vector in the first text.

[0108] The second result ((R_{i2})) represents the importance of the (i)th knowledge vector in the second reference key text content. It can be calculated using the following formula: [R_{i2}=frac{V_{i2}}{T_2}] Similarly, (R_{i2}) is the ratio of the cumulative value of the (i)th knowledge vector in the second reference key text content to the total number, reflecting the relative importance of the knowledge vector in the second text.

[0109] 4. Determine the thermal assessment data for business elements. The business element heat map data ((H_i)) can be determined based on a combination of the first and second calculation results. For example, it can be a simple global average of the two, or a weighted average can be calculated by assigning different weights according to business needs.

[0110] [H_i=A·R_{i1}+B·R_{i2}] Where (A) and (B) are weighting coefficients, which are set according to the specific situation, and (A+B=1).

[0111] If both (A) and (B) are set to 0.5, it means that they are of equal importance. If a particular reference text is considered more important, its weight can be adjusted accordingly.

[0112] These formulas enable the system to quantify the importance of each business interaction state knowledge vector within different reference key text contents, and to conduct subsequent business processing and decision support accordingly.

[0113] In a preferred embodiment, determining the state knowledge confidence level of each business interaction state knowledge vector based on the business element heat map data of the business interaction state knowledge vector includes: performing interval numerical mapping processing on the business element heat map data of the business interaction state knowledge vector to obtain the interval numerical mapping result of the business element heat map data; performing feature transformation processing based on the interval numerical mapping result of the business element heat map data, and using the obtained feature transformation result as the state knowledge confidence level of each business interaction state knowledge vector.

[0114] In a preferred technical solution, the system determines the state knowledge confidence level of each business interaction state knowledge vector through a series of processing steps. The following are the detailed implementation steps.

[0115] First, the system performs interval numerical mapping on the business element heatmap assessment data of the business interaction state knowledge vector. Interval numerical mapping is a technique that maps raw data to a specific interval, typically used for data normalization or standardization. In this step, the system maps the business element heatmap assessment data of each business interaction state knowledge vector to a preset numerical interval, obtaining the interval numerical mapping result of the business element heatmap assessment data.

[0116] This preset numerical range can be set according to actual needs, such as between 0 and 1, or other numerical ranges. The mapping process can be achieved through linear transformation, nonlinear transformation, etc., depending on the distribution characteristics of the data and business requirements.

[0117] Next, the system will perform feature transformation based on the interval numerical mapping results of the business element heat map data. Feature transformation is a method to convert the original features into a feature representation that is more conducive to model learning. In this step, the system will further transform the interval numerical mapping results into a feature form that is more suitable for representing state knowledge confidence.

[0118] The specific method of feature transformation can be selected according to actual needs, such as normalization, standardization, or discretization. The transformed features will better reflect the importance and confidence of the business interaction state knowledge vector, providing stronger support for subsequent business processing and decision-making.

[0119] Finally, the system uses the obtained feature transformation results as the state knowledge confidence score for each business interaction state knowledge vector. This state knowledge confidence score is a comprehensive evaluation metric used to measure the accuracy and reliability of each business interaction state knowledge vector. It can be used in subsequent business decisions, recommendation systems, risk assessments, and other aspects to help the system better understand and utilize the information in the business interaction state knowledge vectors.

[0120] Through these processing steps, the system can more accurately assess the state knowledge confidence of each business interaction state knowledge vector, providing stronger data support for subsequent business processing and decision-making. At the same time, this technical solution also possesses a degree of flexibility and scalability, allowing for adjustments and optimizations based on different business needs and data characteristics.

[0121] In some exemplary embodiments, determining the target key conversation text semantics from the business interaction state knowledge vector chain based on the state knowledge confidence level includes: determining the semantic feature dimension of the target key conversation text semantics based on the target government affairs handling task; determining the confidence interval of the state knowledge confidence level of the candidate target key conversation text semantics based on the semantic feature dimension; and determining the target key conversation text semantics from each business interaction state knowledge vector in the business interaction state knowledge vector chain based on the confidence interval.

[0122] In some exemplary embodiments, the system determines the semantics of the target key session text based on state knowledge confidence. The following are the detailed implementation steps.

[0123] First, the system determines the semantic feature dimensions of the target key conversation text based on the target government affairs handling task. Semantic feature dimensions are multiple aspects or angles used to describe and measure the semantics of text, such as the text's theme, sentiment, and entities. In this step, the system selects relevant semantic feature dimensions based on the specific needs and characteristics of the government affairs handling task for subsequent text semantic analysis and determination.

[0124] Next, the system determines the confidence interval for the state knowledge confidence of the candidate target key conversation text semantics based on the determined semantic feature dimensions. A confidence interval is a statistical concept used to estimate the range within which the true value of a parameter might exist; here, it is used to measure the reliability and accuracy of the state knowledge confidence. The system calculates a reasonable confidence interval based on the selected semantic feature dimensions and the corresponding state knowledge confidence, which is then used to filter target key conversation text semantics that meet the criteria.

[0125] Finally, based on the determined confidence interval, the system identifies the target key session text semantics from each business interaction state knowledge vector in the business interaction state knowledge vector chain. Specifically, the system iterates through each business interaction state knowledge vector in the chain and calculates whether its state knowledge confidence score falls within the determined confidence interval. If the state knowledge confidence score of a business interaction state knowledge vector falls within the confidence interval, and its corresponding text semantics matches the selected semantic feature dimension, then the system identifies it as the target key session text semantics.

[0126] It is important to note that in practical applications, the system may flexibly adjust and optimize the above steps based on specific business needs and data characteristics. For example, more complex semantic analysis models may be used to improve the accuracy of text semantic recognition, or the confidence interval settings may be dynamically adjusted based on historical data.

[0127] Through this processing flow, the system can more accurately determine the key conversation text semantics related to the target government affairs handling task from the business interaction state knowledge vector chain, providing strong support for subsequent business processing and decision-making.

[0128] In some optional design examples, the method further includes: updating the priority of each smart government business session text in the smart government business session text stream according to the priority of the session output; performing joint semantic embedding on each smart government business session text after priority update; and cleaning up noisy business session text in the smart government business session text stream.

[0129] In some optional design examples, the system will further optimize the text stream of smart government business conversations, specifically including the following steps.

[0130] First, the system updates the priority of each smart government business conversation text in the smart government business conversation text stream based on the priority of the conversation output. The priority of the conversation output is determined by a combination of factors such as business needs, user level, and the urgency of the conversation content. In this step, the system evaluates each smart government business conversation text and updates its priority based on the evaluation results. The priority update can be dynamically adjusted to adapt to business needs under different circumstances.

[0131] For example, the system may prioritize smart government service conversations involving important matters or emergencies to ensure timely processing. Conversely, the system may lower the priority of routine or low-urgency conversations, scheduling them for later processing.

[0132] Secondly, the system performs joint semantic embedding on the smart government business conversation texts after priority updates. Joint semantic embedding is a method that maps multiple texts to the same semantic space to enable comparison and association at the semantic level. In this step, the system uses a pre-trained semantic embedding model or algorithm to convert the various smart government business conversation texts into vector representations and embeds these vectors into the same semantic space.

[0133] By using joint semantic embedding, the system can more accurately understand the semantic relationships and similarities between various smart government business conversation texts, providing richer semantic information for subsequent conversation processing. This helps the system better grasp user needs and intentions, improving the accuracy and efficiency of business processing.

[0134] Finally, the system will clean up noisy business session text in the smart government business session text stream. Noisy business session text refers to session text that is irrelevant to business processing and contains errors or redundant information. This text may interfere with the normal processing of the system and reduce the accuracy and efficiency of business processing. Therefore, the system needs to identify and clean it.

[0135] During the text cleaning process, the system can employ rule-based methods, machine learning-based methods, or a combination of both to identify and filter noisy business conversation text. For example, the system can formulate rules to identify common noisy text patterns, or use machine learning algorithms to train a classifier to automatically identify and filter noisy text. Through this text cleaning process, the system can retain valuable smart government business conversation text, improving the quality and efficiency of subsequent business processing.

[0136] In some optional embodiments, the method further includes steps 210-220.

[0137] Step 210: Input the target key conversation text semantics into a deep structured semantic model and create a business text summary generation network.

[0138] Step 220: Input the text of the proposed smart government business session into the business text summary generation network to obtain the business text summary of the proposed online business session.

[0139] In some optional embodiments, the smart government platform system further extends its functionality by including steps 210 and 220 to achieve automatic summary generation of the online business session to be processed. The following is a detailed illustrative explanation of these steps.

[0140] In step 210, the semantics of the target key conversation text are entered into a deep structured semantic model to create a business text summary generation network.

[0141] In this step, the smart government affairs platform system takes the previously identified target key conversation text semantics as input and feeds it into a deep structured semantic model. This model is a trained deep learning network capable of performing deep understanding and structured processing of the input text semantics. Through this process, the system can capture the core information and semantic structure of the text, providing a foundation for subsequent summary generation.

[0142] Deep structured semantic models contain multiple processing layers, each responsible for extracting features from the text at different levels. These features include, but are not limited to, lexical features, syntactic features, and semantic roles. By extracting and integrating these features layer by layer, the model can construct a comprehensive and refined text representation that retains the key information of the original text while facilitating subsequent processing and analysis.

[0143] After inputting the semantics of the target key conversation text into a deep structured semantic model, the system uses this model to create a business text summary generation network. This network is specifically designed for processing smart government business conversation text, and it can automatically generate concise and accurate business text summaries based on the input text semantics.

[0144] In step 220, the text of the proposed smart government business session to be processed is entered into the business text summary generation network to obtain the business text summary of the proposed online business session.

[0145] After creating the business text summary generation network, the smart government platform system can begin processing proposed online business sessions. When the system receives a new text for a proposed smart government business session, it will input this text into the previously created summary generation network.

[0146] The abstract generation network first preprocesses the input text, including word segmentation, stop word removal, and standardization, to convert the text into a format suitable for model processing. Then, the network uses a deep structured semantic model to perform in-depth semantic analysis and structured processing on the processed text, extracting key information and semantic structure from the text.

[0147] Next, the summary generation network generates a concise and accurate business text summary based on the extracted key information and semantic structure. This summary contains the core content and key information of the original text, but removes redundant and secondary parts, enabling users to quickly understand the main content and key points of the conversation.

[0148] Finally, the smart government affairs platform system will return the generated business text summary to the user or store it in the system for later use. This allows users to quickly understand the main content and processing results of the online business session by viewing the summary.

[0149] By implementing steps 210 and 220 above, the smart government affairs platform system can realize the function of automatically generating summaries of online business sessions to be processed, thereby improving the efficiency and quality of government affairs processing.

[0150] In some other embodiments, the method further includes: creating a first reference key text content and a second reference key text content based on the target government affairs handling task; determining a derived target key session text semantic based on the first reference key text content and the second reference key text content; wherein the derived target key session text semantic is used to create a derived business text summary generation network that matches the target government affairs handling task, so that the derived business text summary generation network determines the business text summary of the online business session to be processed under the target government affairs handling task.

[0151] In other embodiments, the smart government affairs platform system further enhances its functionality by creating reference key text content to determine the semantics of derived target key conversation text, and based on this, creating a derived business text summary generation network that matches specific government affairs handling tasks. Detailed examples are provided below.

[0152] First, the system creates a first reference key text and a second reference key text based on the target government service task. These reference key texts are derived from a deep understanding and analysis of the target government service task, and they include key information, business processes, policies, and regulations that may be involved in handling the task. The first and second reference key texts may emphasize different aspects; for example, the first reference key text may focus more on user needs and business processes, while the second reference key text may focus more on policies, regulations, and business standards.

[0153] Next, the system will determine the semantics of the derived target key conversation text based on the content of the first and second reference key texts. This process may involve techniques such as deep analysis of the reference key text content, information extraction, and semantic fusion. Through this process, the system can generate a comprehensive, accurate, and closely related derived target key conversation text semantics that is relevant to the target government affairs handling task.

[0154] This derived target key conversation text semantics plays a crucial role in subsequent steps. The system uses it to create a derived business text summary generation network that matches the target government affairs handling task. This network is specifically designed to handle business conversation text under specific government affairs handling tasks, and it can automatically generate concise and accurate business text summaries based on the input conversation text.

[0155] When the system receives a new online business session to be processed, it inputs the session text into a previously created derived business text summary generation network. The network first preprocesses and analyzes the input text, extracting key information and features. Then, it uses the semantics of the derived target key session text to guide the summary generation process, ensuring that the generated summary is closely related to the target government affairs processing task and accurately reflects the core content and key points of the original session text.

[0156] Finally, the system will return the generated business text summary to the user or store it in the system for later use. This allows users to quickly understand the main content and processing results of the online business session under the target government service task by viewing the summary.

[0157] In this way, the smart government affairs platform system can more accurately process and analyze business conversation texts under specific government affairs handling tasks, improving the efficiency and quality of government services. At the same time, because the summary generation network is derived for specific tasks, it has stronger targeting and adaptability, and can better meet user needs and expectations.

[0158] It is important to note that identifying the derived target key conversational text semantics is the core innovation of this application, which is crucial for improving the processing capabilities and service quality of the smart government platform system. This core inventive point will be described in more detail below to ensure readability and coherence.

[0159] In smart government platform systems, determining the semantics of derived target key conversation texts is a complex and meticulous process. This process aims to extract the most critical and representative semantic information from massive amounts of government business conversation texts to guide subsequent business processing and summary generation.

[0160] First, the system conducts an in-depth analysis and understanding of the target government service task. This includes a comprehensive review and interpretation of the task's nature, purpose, involved business processes, policies, and regulations. Through this process, the system establishes a comprehensive understanding of the target task, providing a solid foundation for determining the semantics of subsequent key conversational texts.

[0161] Next, based on its understanding of the target task, the system will create a first reference key text and a second reference key text. These two parts describe and define the target task from different angles and with different focuses. For example, the first reference key text might focus on describing user needs and business processes, while the second reference key text might focus more on policies, regulations, and business standards. These two parts complement each other, together forming a complete description of the target task.

[0162] Then, the system uses advanced natural language processing techniques and deep learning algorithms to perform in-depth analysis and information extraction on the first and second reference key text contents. This process includes multiple steps such as lexical analysis, syntactic analysis, and semantic role labeling, aiming to extract key information elements and semantic structures from the text.

[0163] After extracting key information, the system further performs semantic fusion and reasoning. It integrates and reasons with contextual information and domain knowledge to generate a comprehensive, accurate, and closely related derived target key conversation text semantics that is relevant to the target government task. This semantics not only includes the key information from the original text but also incorporates the system's understanding and reasoning results regarding the target task.

[0164] Finally, this derived target key conversation text semantics will be used to guide subsequent business processing and summary generation. The system will create a derived business text summary generation network based on this semantics, matching the target government affairs processing task. This network can automatically generate concise and accurate business text summaries based on the input conversation text, helping users quickly understand the main content and processing results of the conversation.

[0165] In summary, determining the derived target key conversation text semantics is the core innovation of this application. Through in-depth understanding and analysis of the target government affairs handling tasks, combined with natural language processing technology and deep learning algorithms, a comprehensive, accurate, and closely related derived target key conversation text semantics is generated. This semantics provides strong guidance for subsequent business processing and summary generation, improving the processing capabilities and service quality of the smart government affairs platform system.

[0166] In one independent embodiment, after inputting the text of the proposed smart government business session (to be processed) into the business text summary generation network to obtain the business text summary of the proposed online business session, the method includes: determining at least one structured semantic tree of the proposed online business session and storage guidance features for each structured semantic tree based on the business text summary, each structured semantic tree including a session semantic block of the proposed online business session; for each structured semantic tree, inputting the structured semantic tree and the corresponding storage guidance features into a text storage optimization network to obtain an initial structured session text updated to the structured semantic tree; performing text restatement on the corresponding initial structured session text based on the structured semantic tree to obtain a target structured session text updated to the structured semantic tree; and performing structured storage of the proposed online business session based on the target structured session text corresponding to each structured semantic tree.

[0167] In one standalone embodiment, the system specifies in detail the processing flow for the online business sessions to be handled. The following are the specific steps and explanations of this embodiment.

[0168] First, the system inputs the text of the proposed smart government business session (the online business session to be processed) into the business text summarization generation network to obtain a business text summary of the proposed online business session. The business text summarization generation network is a deep learning model that can perform semantic understanding and information extraction on the input session text, generating a concise summary containing key information.

[0169] Next, the system determines at least one structured semantic tree for the online business session to be processed, as well as storage guidance features for each structured semantic tree, based on the business text summary. A structured semantic tree is a structured representation of the session text, clearly reflecting the semantic structure and information hierarchy within the text. Storage guidance features are characteristic information used to guide text storage, such as the importance and urgency of the text.

[0170] Then, for each structured semantic tree, the system inputs the structured semantic tree and its corresponding storage guidance features into the text storage optimization network to obtain the initial structured conversation text updated with the structured semantic tree. The text storage optimization network is a deep learning model that optimizes the conversation text based on the input structured semantic tree and storage guidance features, generating initial structured conversation text that is more suitable for storage and retrieval.

[0171] Subsequently, the system performs text paraphrasing on the corresponding initial structured conversation text based on the structured semantic tree, obtaining the target structured conversation text updated with the structured semantic tree. Text paraphrasing is a natural language processing technique that can restate text while preserving its original meaning. Through text paraphrasing, the system can further optimize the expression of the initial structured conversation text, improving its readability and comprehensibility.

[0172] Finally, the system stores the online business sessions to be processed in a structured manner based on the target structured session text corresponding to each structured semantic tree. Structured storage is a method of storing text data according to a certain structure and format, which can improve the organization and retrieval of data. Through structured storage, the system can save the online business sessions to be processed in a clearer and more organized way, providing convenience for subsequent business processing and decision support.

[0173] In one independent embodiment, the storage guidance features of each structured semantic tree include at least one of the following: logical text for describing the session semantic blocks of the structured semantic tree; and logical text for indicating the structured tags of the initial structured session text corresponding to the structured semantic tree.

[0174] In one standalone embodiment, the system defines specific storage guidance features for each structured semantic tree. These storage guidance features play a crucial role in the structured storage of conversational text, helping the system to more accurately understand and organize textual data. The following is a detailed explanation of these storage guidance features.

[0175] First, the storage guidance features for each structured semantic tree can include logical text used to describe the conversational semantic blocks of that structured semantic tree. A conversational semantic block is the smallest unit with independent meaning in conversational text; it may be a sentence, a phrase, or a word. Logical text is an abstract description of these conversational semantic blocks, summarizing their main content and function in concise language. Through logical text, the system can quickly identify and understand key information in conversational text, facilitating subsequent storage and retrieval.

[0176] For example, in a conversation about government affairs, a semantic block might be "Inquiry about the conditions and procedures for applying for a residence permit." The corresponding logical text could be "Inquiry about the conditions and procedures for applying for a residence permit," which concisely summarizes the topic and content of the semantic block.

[0177] Secondly, the storage guidance feature may also include logical text that indicates the structured tags present in the initial structured session text corresponding to the structured semantic tree. Structured tags are a way of marking different parts of the session text, helping the system better organize and understand the structure and hierarchy of the text. Logical text, on the other hand, is an explanation and description of these structured tags, using easily understandable language to describe their meaning and function.

[0178] Taking a similar government affairs consultation session as an example, suppose the session text is labeled with three structured tags: "Question," "Answer," and "Notes." The corresponding logical text could be "The question raised by the user," "The answer given by the staff," and "Notes to be aware of during the process." These logical texts clearly describe the content and function of each tag, helping the system to more accurately understand and store the structured information of the session text.

[0179] In summary, by defining storage guidance features, including logical text, for each structured semantic tree, the system can more accurately identify, understand, and store key and structured information in session text. This helps improve the organizeability, retrieval, and understandability of session text, providing strong support for subsequent business processing and decision support.

[0180] In one independent embodiment, for each of the structured semantic trees, the step of performing text restatement on the corresponding initial structured conversation text based on the structured semantic tree to obtain the target structured conversation text updated by the structured semantic tree includes: determining the initial structured text features of each semantic tree branch of the structured semantic tree, and the initial structured text features of each semantic tree branch of the corresponding initial structured conversation text; determining a first transitional structured conversation text, wherein the initial structured text features of each semantic tree branch of the first transitional structured conversation text are the difference between the initial structured text features of the structured semantic tree in the same semantic tree branch and the initial structured text features of the corresponding initial structured conversation text in the same semantic tree branch; setting the initial structured text features of the first semantic tree branch in the first transitional structured conversation text as a first feature variable to obtain a second transitional structured conversation text, wherein the initial structured text features of the first semantic tree branch are greater than a preset feature variable; performing dilated convolution on the second transitional structured conversation text to obtain a third transitional structured conversation text; and obtaining the target structured conversation text updated by the structured semantic tree based on the structured semantic tree and the third transitional structured conversation text.

[0181] In one standalone embodiment, the system performs in-depth processing on the structured semantic tree and the corresponding initial structured conversation text to generate updated target structured conversation text. The following is a detailed explanation of the steps.

[0182] First, the system determines the initial structural text features for each semantic tree branch of each structured semantic tree, as well as the initial structural text features for each semantic tree branch of the corresponding initial structured conversation text. Semantic tree branches are different parts of the structured semantic tree, each representing a specific meaning or topic in the conversation text. The initial structural text features are preliminary descriptions and measurements of these semantic tree branches, and can include features related to vocabulary, syntax, and semantics.

[0183] Next, the system determines a first transitional structured conversation text. This transitional text is characterized by the fact that the initial structured text features of each semantic tree branch are the difference between the initial structured text features of the same semantic tree branch and the initial structured conversation text of the corresponding initial semantic tree branch. In other words, the system obtains an intermediate text by calculating the difference; this text is structurally related to both the original structured semantic tree and the initial structured conversation text.

[0184] Then, the system sets the initial structural text features of the first semantic tree branch in the first transitional structured conversation text as the first feature variable, resulting in the second transitional structured conversation text. Here, the first semantic tree branch refers to those branches whose initial structural text features are greater than the preset feature variable. The first feature variable is a preset threshold or standard used to filter and adjust text features. Through this step, the system can further adjust and optimize the structure and features of the transitional text.

[0185] Next, the system performs dilated convolution on the second transitional structured conversation text to obtain the third transitional structured conversation text. Dilated convolution is a special type of convolution operation that can expand the receptive field without increasing model complexity, thereby capturing more contextual information. Here, the system uses dilated convolution to further extract and integrate key information from the transitional text to generate more accurate and coherent text.

[0186] Finally, based on the original structured semantic tree and the third transitional structured conversation text processed by dilated convolution, the system obtains the target structured conversation text after updating the original structured semantic tree. This target text not only retains the key information and structural features of the original conversation text, but also improves the text quality and readability through a series of processing and optimization operations. At the same time, it provides a more valuable data foundation for subsequent business processing and decision support.

[0187] In one independent embodiment, for each of the structured semantic trees, obtaining the updated target structured conversation text based on the structured semantic tree and the third transitional structured conversation text includes: determining a first discrete metric and a first global average metric for the initial structured text features of each semantic tree branch of the structured semantic tree; determining a second discrete metric and a second global average metric for the initial structured text features of each semantic tree branch of the third transitional structured conversation text; for each semantic tree branch of the structured semantic tree, determining the target structured text features of the semantic tree branch based on the first global average metric, the initial structured text features of the third transitional structured conversation text in that semantic tree branch, the second global average metric, the first discrete metric, and the second discrete metric; and generating the updated target structured conversation text based on the target structured text features of each semantic tree branch of the structured semantic tree.

[0188] In one standalone embodiment, the system obtains the updated target structured conversation text through in-depth processing of the structured semantic tree and the third transitional structured conversation text. The following is a detailed explanation of this process.

[0189] First, the system determines a first discrete metric and a first global average metric for the initial structural text features of each branch of the structured semantic tree. The first discrete metric assesses the difference or dispersion among the initial structural text features of each semantic tree branch, helping the system identify which branches have significant variations or uncertainties in their features. The first global average metric, on the other hand, is calculated by averaging the initial structural text features of all semantic tree branches, reflecting the overall level or trend of the entire structured semantic tree in terms of structural text features.

[0190] Next, the system determines a second discrete metric and a second global average metric for the initial structural text features of each semantic tree branch in the third transitional structured conversation text. Similar to the first discrete metric and the first global average metric, the second discrete metric is used to assess the difference or dispersion among the initial structural text features of each semantic tree branch in the third transitional structured conversation text, while the second global average metric reflects the overall level or trend of the structural text features of the entire third transitional structured conversation text.

[0191] Then, for each semantic tree branch in the structured semantic tree, the system determines the target structured text features for that branch based on the first global average metric, the initial structured text features of the third transitional structured conversation text in that semantic tree branch, the second global average metric, the first discrete metric, and the second discrete metric. This step considers the overall feature level of the original structured semantic tree, the features of the transitional text, and the differences and discreteness between them, synthesizing this information to obtain more accurate and comprehensive target structured text features.

[0192] Specifically, the system can use these metrics as input parameters to calculate and process them according to a preset algorithm or model to obtain the target structural text features for each semantic tree branch. This calculation process can be a complex mathematical model or machine learning algorithm, which can be flexibly adjusted and optimized according to actual needs and data characteristics.

[0193] Finally, based on the target structured text features of each branch of the structured semantic tree, the system generates the target structured conversation text updated from the original structured semantic tree. This target text not only retains the key information and structural features of the original structured semantic tree, but also undergoes more in-depth and comprehensive optimization by comprehensively considering information from multiple indicators and transitional texts. Simultaneously, it provides a more accurate, reliable, and valuable data foundation for subsequent business processing and decision support.

[0194] Figure 2A structural block diagram of a smart government affairs platform system is shown, including: a memory 310 for storing program instructions and data; and a processor 320 for coupling with the memory 310 to execute the instructions in the memory 310 to implement the above-mentioned method.

[0195] Furthermore, a computer storage medium is also provided, comprising instructions that, when executed on a processor, implement the above-described method.

[0196] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can also be implemented in other ways. The apparatus and method embodiments described above are merely illustrative. For example, the flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of apparatus, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than those marked in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram and / or flowchart, and combinations of blocks in block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or action, or using a combination of dedicated hardware and computer instructions.

[0197] In addition, the functional modules in the various embodiments of this application can be integrated together to form an independent part, or each module can exist independently, or two or more modules can be integrated to form an independent part.

[0198] If the aforementioned functions are implemented as software functional modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks. It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. In the absence of further restrictions, an element defined by the phrase "comprising a..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0199] The above description is merely a preferred embodiment of this application and is not intended to limit this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A method for generating business text summaries based on an intelligent platform, characterized in that: Applied to the smart government affairs platform system, the method includes: Get the text stream of smart government affairs business conversation; Mining a business interaction state knowledge vector chain from the smart government affairs business conversation text stream according to a business interaction state mining node set; wherein the business interaction state mining node set is obtained by integrating at least one business interaction state mining node based on the target government affairs matter handling task; Based on the business element thermal evaluation data of each business interaction state knowledge vector in the business interaction state knowledge vector chain, determine the state knowledge confidence of each business interaction state knowledge vector; wherein the business element thermal evaluation data is used to reflect the heat map statistics of the business interaction state knowledge vector, and the state knowledge confidence is used to reflect the importance of the corresponding business interaction state knowledge vector; Based on the state knowledge confidence, target key conversation text semantics is determined from the business interaction state knowledge vector chain.

2. The method for generating business text summaries based on a smart platform according to claim 1, characterized in that: The business interaction state mining node set includes at least one or more business interaction state mining nodes among a period linkage processing node, an event linkage processing node, a user linkage processing node, a scenario linkage processing node or an adjustable processing node, and mining a business interaction state knowledge vector chain from the smart government business conversation text flow according to the business interaction state mining node set includes at least one of the following: Mining first business interaction state information continuously generated by the online business session within the target activation period from the smart government business session text flow through the periodic linkage processing node, and determining a business interaction state knowledge vector of the first business interaction state information; Alternatively, the second business interaction state information continuously generated between different online business sessions is mined from the smart government business session text stream through the event linkage processing node, and the business interaction state knowledge vector of the second business interaction state information is determined; Alternatively, mining the third business interaction state information that is continuously and repeatedly generated by the online business session from the text stream of the smart government business session through the user linkage processing node, and determining the business interaction state knowledge vector of the third business interaction state information; Alternatively, mining fourth business interaction state information of the online business conversation realizing question-answering dialogue interaction in the target task process from the smart government business conversation text flow through the scenario linkage processing node, and determining the business interaction state knowledge vector of the fourth business interaction state information; Alternatively, the fifth business interaction state information generated by the online business conversation under the target government affairs matter handling task is mined from the smart government affairs business conversation text stream through an adjustable processing node, and a business interaction state knowledge vector of the fifth business interaction state information is determined; Among them, the first business interaction status information continuously generated by the online business session within the target activation period is mined from the smart government business session text flow through the periodic linkage processing node, and the business interaction status knowledge vector of the first business interaction status information is determined, including at least one of the following: setting the periodic constraint value of the periodic linkage processing node; obtaining the detection result of each business interaction status information continuously generated by the online business session within the target activation period of the periodic constraint value; determining the business interaction status knowledge vector of the first business interaction status information based on the detection result; wherein the detection result includes at least one of the update number of each business interaction status information, the maximum state maintenance stage of each business interaction status information, the minimum state maintenance stage of each business interaction status information, or the generation result of each business interaction status information.

3. The method for generating business text summaries based on a smart platform according to claim 1, characterized in that: The determining of the state knowledge confidence of each business interaction state knowledge vector based on the business element thermal evaluation data of each business interaction state knowledge vector in the business interaction state knowledge vector chain includes: Creating a first reference key text content and a second reference key text content based on the target government affairs matter handling task, wherein the first reference key text content is composed of business interaction state knowledge vectors that do not have hot topics in the target government affairs matter handling task, and the second reference key text content is composed of business interaction state knowledge vectors that have hot topics in the target government affairs matter handling task; Determine business element thermal evaluation data of each business interaction state knowledge vector in the business interaction state knowledge vector chain in the first reference key text content and the second reference key text content; Based on the business element thermal evaluation data of the business interaction state knowledge vector, the state knowledge confidence of each business interaction state knowledge vector is determined.

4. The method for generating business text summaries based on a smart platform according to claim 3 is characterized in that: The determining of the business element thermal evaluation data of each business interaction state knowledge vector in the business interaction state knowledge vector chain in the first reference key text content and the second reference key text content includes: Counting the cumulative value of each business interaction state knowledge vector in the business interaction state knowledge vector chain in the first reference key text content as a first detection variable, and counting the cumulative value of each business interaction state knowledge vector in the business interaction state knowledge vector chain in the second reference key text content as a second detection variable; Determine the total number of business interaction state knowledge vectors contained in the first reference key text content as a third detection variable, and determine the total number of business interaction state knowledge vectors contained in the second reference key text content as a fourth detection variable; Based on the first detection variable and the third detection variable, determining a first operation result between a cumulative value of each business interaction state knowledge vector in the business interaction state knowledge vector chain in the first reference key text content and a total number of business interaction state knowledge vectors contained in the first reference key text content; Based on the second detection variable and the fourth detection variable, determining a second operation result between a cumulative value of each business interaction state knowledge vector in the business interaction state knowledge vector chain in the second reference key text content and a total number of business interaction state knowledge vectors contained in the second reference key text content; Based on the first operation result and the second operation result, the business element thermal evaluation data of each business interaction state knowledge vector in the first reference key text content and the second reference key text content is determined.

5. The method for generating business text summaries based on a smart platform according to claim 3 is characterized in that: The determining of the state knowledge confidence of each business interaction state knowledge vector based on the business element thermal evaluation data of the business interaction state knowledge vector comprises: Performing interval numerical mapping processing on the business element thermal evaluation data of the business interaction state knowledge vector to obtain an interval numerical mapping result of the business element thermal evaluation data; Feature conversion processing is performed based on the interval numerical mapping result of the business element thermal evaluation data, and the obtained feature conversion result is used as the state knowledge confidence of each business interaction state knowledge vector.

6. The method for generating business text summaries based on a smart platform according to claim 1, characterized in that: The determining the target key conversation text semantics from the business interaction state knowledge vector chain based on the state knowledge confidence includes: Based on the target government affairs handling task, determining the semantic feature dimension of the target key conversation text semantics; Determine a confidence interval of the state knowledge confidence of the semantics of the target key conversation text to be selected based on the semantic feature dimension; Based on the confidence interval, target key conversation text semantics are determined from each business interaction state knowledge vector of the business interaction state knowledge vector chain.

7. The method for generating a business text summary based on a smart platform according to any one of claims 1 to 6, characterized in that: The method further comprises: According to the priority of the session output, the priority of each smart government service session text in the smart government service session text stream is updated; Perform joint semantic embedding on each smart government service conversation text after priority update; and clean up the noise service conversation text in the smart government service conversation text stream.

8. The method for generating business text summaries based on a smart platform according to claim 1, characterized in that: The method further comprises: Entering the target key conversation text semantics into a deep structured semantic model to create a business text summary generation network; Entering the text of the smart government affairs business session to be processed of the online business session to be processed into the business text summary generation network to obtain the business text summary of the online business session to be processed; Among them, the method also includes: creating a first reference key text content and a second reference key text content according to the target government affairs matter handling task; determining the derived target key conversation text semantics based on the first reference key text content and the second reference key text content; wherein the derived target key conversation text semantics are used to create a derived business text summary generation network that matches the target government affairs matter handling task, so that the derived business text summary generation network determines the business text summary of the online business session to be processed under the target government affairs matter handling task.

9. A smart government affairs platform system, characterized in that: include: Memory, used to store program instructions and data; A processor, coupled to a memory, and configured to execute instructions in the memory to implement the method according to any one of claims 1 to 8.

10. A computer storage medium, characterized in that: Contains instructions, which, when executed on a processor, implement the method according to any one of claims 1 to 8.

Images