A News Event Processing Approach Based on Stateful Task Execution and Skill Strategy Reuse
By implementing stateful task execution and reusing skills and strategies, the system addresses the insufficient coverage of news aggregation systems in the long tail and early stages of breaking events, enabling sustainable news event processing and event-level governance, and improving the efficiency and reliability of news event processing.
Patent Information
- Application Number
- CN202611348185.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-09-01
- Publication Date
- 2026-11-17
AI Technical Summary
Existing news aggregation systems lack sufficient coverage of long-tail industries, vertical fields, county and district news, grassroots government announcements, and the early stages of emergencies. They also lack continuous task status management, event-level governance, and version management, resulting in duplicate broadcasts, mis-merging, and information delays.
By using a stateful task execution and skill strategy reuse approach, candidate content is obtained from preset news sources, it is determined whether dynamic search is triggered, a dynamic search task is generated, candidate content is supplemented, and the event processing task trajectory is abstracted into skill strategies and added to the skill strategy library for subsequent news event content search and governance.
It improves the efficiency and reliability of news event handling, ensures coverage of long-tail news and breaking events, enables sustainable task execution and event-level governance, and reduces the cost of repetitive exploration.
Smart Images

Figure CN123115503A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of natural language processing technology, and in particular to a news event processing method based on stateful task execution and skill strategy reuse. Background Technology
[0002] Existing news aggregation systems typically rely on fixed news sources, fixed channels, fixed interfaces, or one-time keyword searches to retrieve news content. While this approach is effective for mainstream trending news, it still has limitations in long-tail industries, vertical sectors, county and district news, grassroots government announcements, and the early stages of emergencies: 1) Insufficient coverage of fixed sources: Fixed news sources and preset interfaces are difficult to cover county and district news, grassroots government affairs, vertical industry announcements and early clues of emergencies in a timely manner, which can easily lead to problems such as insufficient materials, delayed updates, lack of original announcements or only secondary reprints.
[0003] 2) Lack of continuous task status in the search process: Most existing news searches are one-time keyword searches. Even if the system finds that the materials are incomplete or the evidence is insufficient, it is difficult to continue searching, reading and verifying based on the discovered organization name, geographical clues, announcement title, project name or regulation number.
[0004] 3) Difficulty in accumulating and reusing experience: After completing a search or governance, the existing system usually only outputs the results, without accumulating the effective search path, reliable source order, invalid source, verification rules and stopping conditions into reusable strategies for the future, which leads to the repeated starting from scratch for similar long-tail tasks.
[0005] 4) Lack of event-level governance in multi-source news: Different sources may use different headlines to report the same event, and there may also be differences in key facts such as time, location, quantity, responsible parties, and scope of impact. Relying solely on headline deduplication, link deduplication, or similarity merging can easily lead to duplicate reporting, mis-merging, or the synthesis of unverified conflicting facts into definitive conclusions.
[0006] 5) Lack of controlled version management for event evolution: As news events continue to develop, new developments, factual corrections, or source conflicts may occur. Existing systems often lack event-oriented evidence chains, conflict states, and version management mechanisms, making it difficult to control the scope of use of different event states in downstream links such as summaries, broadcasts, Q&A, and recommendations.
[0007] The above problems urgently need to be addressed. Summary of the Invention
[0008] The purpose of this invention is to at least partially solve one of the technical problems existing in the prior art.
[0009] Therefore, one objective of this invention is to provide a news event processing method based on stateful task execution and skill strategy reuse, which improves the efficiency and reliability of news event processing.
[0010] Another objective of this invention is to provide a news event processing device based on stateful task execution and skill strategy reuse.
[0011] To achieve the above-mentioned technical objectives, the technical solutions adopted in the embodiments of the present invention include: On one hand, embodiments of the present invention provide a news event processing method based on stateful task execution and skill strategy reuse, comprising the following steps: Obtain several candidate contents of the first news event from preset fixed news sources, and determine whether to trigger dynamic search based on the source information of each candidate content; When a dynamic search is triggered, a dynamic search task is generated, and the candidate content is obtained by executing the dynamic search task based on the task status. If dynamic search is not triggered, update the event status, credibility profile, and event version of the first news event based on the candidate content; The event handling task trajectory of the first news event is abstracted into a corresponding event handling skill strategy, and the event handling skill strategy is added to a preset skill strategy library; Match the target skill strategy corresponding to the second news event according to the skill strategy library, and perform content search and event management on the second news event according to the target skill strategy.
[0012] Furthermore, in one embodiment of the present invention, the source information includes source type, publication time, news elements, and initial trust marker. The step of determining whether to trigger dynamic search based on the source information of each candidate content specifically includes: Based on the source type, determine whether the news source coverage of the first news event meets the preset coverage requirements; Based on the publication time and the news elements, determine whether there are any missing or conflicting key facts in the first news event; Based on the initial credibility marker, determine whether the first news event has insufficient evidence level; If the coverage of the news source meets the preset coverage requirements, the key facts of the first news event are not missing or conflicting, and the first news event does not have insufficient evidence level, it is determined that dynamic search will not be triggered; otherwise, it is determined that dynamic search will be triggered.
[0013] Furthermore, in one embodiment of the present invention, the step of generating a dynamic search task, which involves executing the dynamic search task based on the task state to obtain the supplemented candidate content, specifically includes: Determine the task objective, triggering cause, seed clues, source constraints, toolset, and execution boundary of the dynamic search task; Multiple candidate actions are generated based on the current task status of the dynamic search task; Determine the expected new facts, verification, conflict resolution, clue expansion, noise risk, duplication risk, and action cost for each candidate action; The priority of each candidate action is determined based on the expected new facts, the expected verification, the expected conflict resolution, the expected clue expansion, the noise risk, the repetition risk, and the action cost. The candidate action with the highest priority is executed to obtain the supplemented candidate content, and the current task status of the dynamic search task is updated.
[0014] Furthermore, in one embodiment of the present invention, the event state includes a candidate content set, a fact atom set, an event fingerprint, and a conflict set. Updating the event state, credibility profile, and event version of the first news event based on the candidate content specifically includes: The candidate content is standardized and fact atoms are extracted to obtain candidate fact atoms; The candidate fact atom is compared with the event fingerprint of the first news event to determine whether the candidate content belongs to the first news event; When the candidate content belongs to the first news event, the candidate fact atom is compared with the current set of fact atoms of the first news event to obtain the conflict field; Update the candidate content set according to the candidate content, update the fact atom set and the event fingerprint according to the candidate fact atom, and update the conflict set according to the conflict field; The trusted profile is updated based on the source type of the candidate content and the initial trusted tag. The event version is updated based on the updated set of fact atoms and the set of conflicts.
[0015] Furthermore, in one embodiment of the present invention, the step of abstracting the event processing task trajectory of the first news event into a corresponding event processing skill strategy specifically includes: The event processing task trajectory is determined based on the content search log and event governance log of the first news event; The event processing task trajectory is segmented to extract continuous action segments, success branches, failure branches, rollback branches, and stop branches. The specific operations and parameters of each continuous action segment and each branch are abstracted into action primitives and reusable slots to obtain the abstraction result. The abstract results are encapsulated into the event handling skill strategy, and the applicable scenarios, input conditions, recommended tool sequences, retrieval templates, verification rules, and conflict handling rules of the event handling skill strategy are determined.
[0016] Furthermore, in one embodiment of the present invention, the step of matching the target skill strategy corresponding to the second news event according to the skill strategy library, and performing content search and event governance on the second news event according to the target skill strategy, specifically includes: Determine the event scenario of the second news event, and match the target skill strategy that is applicable to the event scenario from the skill strategy library; The second news event is searched for content based on the input conditions, recommended tool sequence, and retrieval template of the target skill strategy, and event governance is performed on the second news event according to the verification rules and conflict handling rules of the target skill strategy.
[0017] Furthermore, in one embodiment of the present invention, the news event processing method further includes the following steps: Based on the content search results and event governance results of the second news event, determine the positive benefit items or negative penalty items of the target skill strategy; The skill value of the target skill strategy is updated based on the positive benefit or the negative penalty, and the matching priority of the target skill strategy is adjusted based on the skill value.
[0018] On the other hand, embodiments of the present invention provide a news event processing device based on stateful task execution and skill strategy reuse, including: The preliminary search module is used to obtain several candidate contents of the first news event from preset fixed news sources, and determine whether to trigger dynamic search based on the source information of each candidate content. The dynamic search module is used to generate a dynamic search task when a dynamic search is triggered, and to execute the dynamic search task based on the task status to obtain the supplemented candidate content. The event governance module is used to update the event status, trust profile, and event version of the first news event based on the candidate content when dynamic search is not triggered. The skill strategy abstraction module is used to abstract the event processing task trajectory of the first news event into a corresponding event processing skill strategy, and add the event processing skill strategy to a preset skill strategy library; The skill strategy reuse module is used to match the target skill strategy corresponding to the second news event according to the skill strategy library, and to perform content search and event management for the second news event according to the target skill strategy.
[0019] On the other hand, embodiments of the present invention provide an electronic device, which includes a memory, a processor, a computer program stored in the memory and executable on the processor, and a data bus for enabling communication between the processor and the memory. When the computer program is executed by the processor, it implements the news event processing method based on stateful task execution and skill strategy reuse as described above.
[0020] On the other hand, embodiments of the present invention also provide a storage medium, which is a computer-readable storage medium for computer-readable storage. The storage medium stores one or more computer programs, which can be executed by one or more processors to implement the news event processing method based on stateful task execution and skill strategy reuse as described above.
[0021] On the other hand, embodiments of the present invention also provide a computer program product, including a computer program that, when executed by a processor, implements the news event processing method based on stateful task execution and skill strategy reuse as described above.
[0022] The advantages and beneficial effects of the present invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention: This invention obtains several candidate contents of a first news event from a preset fixed news source. Based on the source information of each candidate content, it determines whether dynamic search should be triggered. If dynamic search is triggered, a dynamic search task is generated, and the task is executed based on the task status to obtain supplementary candidate contents. If dynamic search is not triggered, the event status, credibility profile, and event version of the first news event are updated based on the candidate contents. The event processing task trajectory of the first news event is abstracted into a corresponding event processing skill strategy, and this strategy is added to a preset skill strategy library. A target skill strategy for a second news event is matched against this skill strategy library, and content search and event governance are performed on the second news event based on the target skill strategy. This invention determines whether dynamic search should be triggered based on the source information of the candidate content obtained from the initial search. This allows for the supplementation of candidate content when its availability or credibility is insufficient. Executing dynamic search tasks based on task status models the content collection process as a continuously progressing task state, thereby improving the efficiency and reliability of news content acquisition. Abstracting the event processing task trajectory of a news event into event processing skill strategies for reuse allows subsequent news events to quickly perform content search and event governance based on the skill strategy library, further improving the efficiency and reliability of news event processing. Attached Figure Description
[0023] To more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings used in the embodiments of the present invention are described below. It should be understood that the drawings described below are only for the convenience of clearly describing some embodiments of the technical solutions of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0024] Figure 1 A flowchart illustrating the steps of a news event processing method based on stateful task execution and skill strategy reuse provided in an embodiment of the present invention; Figure 2 This is a schematic diagram illustrating an application scenario of the news event processing method based on stateful task execution and skill strategy reuse provided in an embodiment of the present invention. Figure 3 This is a schematic diagram illustrating the triggering of dynamic search and the execution of dynamic search tasks provided in an embodiment of the present invention; Figure 4 A schematic diagram illustrating event governance and skill strategy generation and reuse provided in an embodiment of the present invention; Figure 5 This is a schematic diagram of the structure of a news event processing device based on stateful task execution and skill strategy reuse provided in an embodiment of the present invention; Figure 6 A schematic diagram of the hardware structure of an electronic device provided in an embodiment of the present invention; Figure 7 This is a schematic diagram of the structure of the storage medium provided in an embodiment of the present invention. Detailed Implementation
[0025] The embodiments of the present invention are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain this application, and should not be construed as limiting this application. It should be noted that although functional modules are divided in the system schematic diagram and a logical order is shown in the flowchart, in some cases, the steps shown or described may be performed in a different order than the module division in the system schematic diagram or the order in the flowchart. The step numbers in the following embodiments are only set for ease of explanation and do not limit the order between steps. The execution order of each step in the embodiments can be adaptively adjusted according to the understanding of those skilled in the art.
[0026] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of this application only and is not intended to limit this application.
[0027] This invention provides a news event processing method based on stateful task execution and skill strategy reuse to achieve the following objectives: 1) Establish a content acquisition mechanism that combines fixed source collection with dynamic search to improve the coverage of long-tail news, vertical industry news and early clues of breaking news.
[0028] 2) Establish a sustainable, stateful task execution mechanism to transform news gathering from a one-off search into a sustainable task process, enabling the system to continuously complete searching, accessing, parsing, extracting, and verifying based on discovered clues and intermediate results.
[0029] 3) Establish an event-level news governance and version control mechanism. Taking news events as the basic object, merge, deduplicate, extract facts, verify sources, identify conflicts, and manage evidence chains of multi-source materials. Control the scope of use of the event in downstream links such as summary, broadcast, Q&A, and recommendation according to the event status.
[0030] 4) Establish a strategy reuse mechanism for news gathering and governance experience, and precipitate effective paths, reliable sources, verification rules and stopping conditions from historical tasks into reusable strategies, which can be called and updated in subsequent similar tasks to reduce the cost of repeated exploration and improve the processing efficiency of similar tasks.
[0031] The following describes some of the terms used in the embodiments of the present invention.
[0032] Stateful task execution: This refers to representing a news collection or governance process as a continuously updated task state, and invoking tools such as search, webpage access, parsing, extraction, comparison, database query, and review based on the state, until the evidence conditions are met, the stopping conditions are met, or the process is transferred to manual review.
[0033] Skill strategy reuse: refers to abstracting effective paths from historical tasks into reusable strategies, including applicable scenarios, triggering conditions, retrieval templates, tool call order, source priority, verification rules, exception handling and stopping conditions, and calling and updating them in subsequent similar tasks.
[0034] Fixed source collection: refers to a basic collection method that periodically obtains news content from pre-configured news sources, government websites, industry sites, RSS, interfaces, databases, or subscribed channels.
[0035] Dynamic search: refers to a collection method in which, when fixed source collection cannot meet the requirements for coverage, facts, evidence, or conflict resolution, the stateful task execution unit actively generates search intent, accesses external pages, and supplements evidence based on the current gap.
[0036] Fact atom: refers to the smallest verifiable fact unit extracted from news headlines, text, announcements, web pages, or structured materials, including subject, behavior, object, time, place, value, conclusion, and corresponding source evidence.
[0037] Event-level governance model: refers to a processing model that takes a single news event as the processing object and continuously maintains candidate materials, fact atoms, source credibility profiles, conflict sets, evidence chains, event versions, and downstream availability status.
[0038] The news event processing method based on stateful task execution and skill strategy reuse provided in this application can be applied to a terminal, a server, or software running on a terminal or server. In some embodiments, the terminal can be a smartphone, tablet, laptop, desktop computer, smart speaker, smartwatch, or in-vehicle terminal, but is not limited to these. The server can be configured as an independent physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms. The server can also be a node server in a blockchain network. The software can be an application implementing a driver state monitoring method based on machine vision, but is not limited to the above forms.
[0039] This application can be used in a wide variety of general-purpose or special-purpose computer system environments or configurations. Examples include: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics, network PCs, minicomputers, mainframe computers, and distributed computing environments including any of the above systems or devices. This application can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc., that perform specific tasks or implement specific abstract data types. This application can also be practiced in distributed computing environments where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program modules can reside in local and remote computer storage media, including storage devices.
[0040] It should be noted that in all specific embodiments of this application, when processing data related to user identity or characteristics, such as user information, user behavior data, user historical data, and user location information, user permission or consent is obtained first. Furthermore, the collection, use, and processing of this data comply with relevant laws, regulations, and standards of the relevant countries and regions. In addition, when embodiments of this application require access to sensitive personal information of users, separate permission or consent from the user is obtained through pop-ups or redirection to confirmation pages. Only after obtaining the user's separate permission or consent is the necessary user-related data for the proper functioning of the embodiments of this application obtained.
[0041] like Figure 1 The diagram shown is a flowchart of a news event processing method based on stateful task execution and skill strategy reuse provided by an embodiment of the present invention. (Refer to...) Figure 1 This invention provides a news event processing method based on stateful task execution and skill strategy reuse, specifically including the following steps: S101. Obtain several candidate contents of the first news event from preset fixed news sources, and determine whether to trigger dynamic search based on the source information of each candidate content. S102. When dynamic search is triggered, a dynamic search task is generated, and the dynamic search task is executed based on the task status to obtain the supplemented candidate content. S103. When dynamic search is not triggered, update the event status, credibility profile, and event version of the first news event based on the candidate content. S104. Abstract the event handling task trajectory of the first news event into a corresponding event handling skill strategy, and add the event handling skill strategy to the preset skill strategy library; S105. Match the target skill strategy corresponding to the second news event according to the skill strategy library, and perform content search and event management for the second news event according to the target skill strategy.
[0042] like Figure 2 The diagram illustrates an application scenario of the news event processing method based on stateful task execution and skill strategy reuse provided in this embodiment of the invention. It comprises a news gathering subsystem, an event governance subsystem, and a skill strategy subsystem. These three subsystems are not linearly connected but rather form a closed loop of continuous feedback around the event state. 1) The news gathering subsystem first obtains basic content materials through fixed sources, and initiates dynamic search when fixed sources are insufficient; 2) The event governance subsystem performs fact extraction, event merging, source verification, conflict control, and version management on candidate materials; 3) The skill strategy subsystem extracts reusable strategies from the collected and managed task trajectories, and initializes search paths, source order, verification rules and stopping conditions in subsequent similar tasks; 4) When the event governance system discovers a factual gap, insufficient evidence, or conflicting sources, it will pass the gap back to the news gathering subsystem, triggering the execution of a new stateful task.
[0043] Through the aforementioned closed loop, the news gathering subsystem no longer generalizes and expands its scope of data collection, but instead supplements evidence around the specific gaps identified in the event governance feedback. The event governance subsystem also no longer passively consumes materials, but is able to identify "what facts are still missing, what sources are needed, and which conflicts need to be verified." The system maintains a gap list, evidence list, conflict list, event version, and task history for each event object, enabling any dynamic search to trace its triggering cause and processing result. The system manages the event status, task status, and strategy status in a coordinated manner, keeping collection, governance, and reuse within the same closed loop, rather than breaking them down into isolated business rules.
[0044] It can be recognized that the embodiments of the present invention determine whether to trigger dynamic search based on the source information of candidate content obtained from the initial search. When the availability or credibility of candidate content is insufficient, candidate content can be supplemented. Dynamic search tasks are executed based on task status, and the content collection process can be modeled as a continuously advancing task status, thereby improving the efficiency and reliability of news content acquisition. The event handling task trajectory of news events is abstracted into event handling skill strategies for reuse, enabling subsequent news events to quickly perform content search and event governance based on the skill strategy library, thereby improving the efficiency and reliability of news event handling.
[0045] As an optional implementation, the source information includes source type, publication time, news elements, and initial trust markers. Based on the source information of each candidate content, it is determined whether to trigger a dynamic search, specifically including: S1011. Determine whether the coverage of the news source of the first news event meets the preset coverage requirements based on the source type; S1012. Determine whether there are missing or conflicting key facts in the first news event based on the release time and news elements; S1013. Determine whether the first news event has insufficient evidence level based on the initial credibility marker; S1014. If the coverage of the news source meets the preset coverage requirements, the key facts of the first news event are not missing or conflicting, and the first news event does not have insufficient evidence level, it is determined that dynamic search will not be triggered; otherwise, it is determined that dynamic search will be triggered.
[0046] like Figure 3 The diagram shown is a schematic diagram of triggering dynamic search and executing dynamic search tasks provided by an embodiment of the present invention. Specifically, fixed source collection is used to cover stable sources, and dynamic search is used to make up for the shortcomings of fixed sources in long-tail, vertical, bursty and insufficient evidence scenarios.
[0047] The system periodically retrieves candidate content from pre-configured news websites, government websites, industry sites, corporate announcements, regulatory announcements, and RSS feeds. For each candidate piece of content, the system schedules it according to channel, region, industry, organization, and historical quality, and records the source, publication time, crawling time, title, body, URL, source type, and initial trust tag for each candidate piece of content.
[0048] Dynamic search is triggered only when a fixed source cannot meet the availability or credibility requirements of an event. When determining whether to trigger a dynamic search, the system first reads the event status, considering coverage, missing key facts, evidence level, conflicting fields, and downstream usage requirements, then makes a comprehensive judgment based on the task objectives. Triggering judgment typically manifests in the following ways: 1. Insufficient coverage of fixed sources, such as targets belonging to county-level news, long-tail industries, vertical organizations, or early clues of breaking events; 2. Key facts are missing, such as the time of publication, location of occurrence, responsible party, quantity, scope of impact, original announcement, or authoritative confirmation; 3. Insufficient level of evidence, for example, existing materials are mainly from reprinted materials, low-credibility websites, or a single source; 4. Conflicts exist regarding key facts, such as different sources giving different accounts of the cause, location, quantity, subject, or state; 5. Downstream applications require more credible evidence, such as for broadcasting, Q&A, topic summaries, or risk warnings.
[0049] In one embodiment of the present invention, the determination of whether dynamic search is triggered is made in the following order: 1. Read the fixed source coverage, key fact missing fields, existing evidence level, conflict fields, and long-tail tags of the current event; 2. Determine in turn whether the current event has insufficient coverage, missing key facts, insufficient level of evidence, conflicting key facts, or long-tail low-coverage topics; 3. If any of the above judgments is true, the corresponding reason will be written into the task status and a dynamic search will be triggered; 4. If none of the above judgments are true, then the fixed source collection will continue, and the dynamic search will not be initiated.
[0050] The triggering of dynamic search is not based solely on the number of news items, but rather on whether the event can be used credibly.
[0051] As an optional implementation, a dynamic search task is generated, and the supplemented candidate content is obtained by executing the dynamic search task based on the task status, specifically including: S1021. Determine the task objective, triggering cause, seed clues, source constraints, toolset, and execution boundary of the dynamic search task; S1022. Generate multiple candidate actions based on the current task status of the dynamic search task; S1023. Determine the expected new facts, verification, conflict resolution, clue expansion, noise risk, duplication risk, and action cost for each candidate action; S1024. Determine the priority of each candidate action based on the expected new facts, expected verification, expected conflict resolution, expected clue expansion, noise risk, repeated risk, and action cost. S1025. Execute the highest priority candidate action to obtain the supplemented candidate content, and update the current task status of the dynamic search task.
[0052] Specifically, after a dynamic search is triggered, the system generates a dynamic search task. This task includes at least: 1. Task objective: Event, topic, or fact fields to be completed; 2. Triggering reasons: Coverage gap, factual gap, evidentiary gap, conflict gap, or long-tail demand; 3. Seed clues: entity name, region, organization, announcement title, industry terms, regulation number, project name or historical events; 4. Source constraints: priority sources, prohibited sources, downgraded sources, and required evidence levels; 5. Toolset: General search, site search, webpage access, text extraction, PDF parsing, structured database query, entity disambiguation, fact extraction, and source comparison tools; 6. Execution boundaries: round budget, time budget, cost budget, stop conditions, and manual review conditions.
[0053] Dynamic search is continuously executed with task status at its core. Task status includes current gaps, visited URLs, extracted facts, confirmed sources, invalid sources, candidate leads, conflicting fields, and remaining budget. In each round of execution, the system generates candidate actions based on the current status, such as searching for an organization's announcement, visiting a local website, re-searching with the title of a newly discovered announcement, querying historical events, comparing company announcements, or parsing PDFs.
[0054] In one embodiment of the present invention, candidate actions are generated and the next action is selected according to the following steps: 1. Read the gap type, visited URLs, extracted facts, confirmed sources, invalid sources, candidate leads, conflict fields, and remaining budget from the current task status; 2. Based on the gap type and triggering reason, load executable action categories from the action template library, including general search, site search, webpage access, PDF parsing, historical event comparison, announcement re-retrieval, structured query, and source comparison; 3. Fill the current seed clues, region, organization name, title keywords, and known facts into the action template to form specific candidate actions; 4. Remove duplicate visits, actions from confirmed invalid sources, actions that exceed source constraints, or actions that exceed budget limits; 5. Sort the remaining actions according to the expected addition of new facts, the expected verification, the expected conflict resolution, the expected expansion of clues, the risk of noise, the risk of repetition, and the cost, and select the optimal action to execute.
[0055] After the action is executed, the task status is updated, and the search is determined to continue or stop based on newly added facts, evidence levels, and changes in conflict. The system stops dynamic searching in the following situations: 1. Key fact fields have been supported by authoritative or independent sources that meet the requirements; 2. Several consecutive rounds of actions have not produced any new facts or valid evidence; 3. The remaining candidate leads are all duplicate reposts or from low-credibility sources; 4. The conflict cannot be resolved further using the current tools and requires manual review; 5. Reach the preset time, round, or cost limit corresponding to the triggering reason.
[0056] In one embodiment, the present invention does not employ a uniform and fixed time, round, and cost cap, but instead generates budget gating parameters based on the triggering conditions of dynamic search tasks. The budget gating parameters include at least the maximum time T_max, maximum rounds R_max, maximum cost C_max, single-action cost cap C_step, no-gain tolerance rounds N_no_gain, and manual review threshold H_manual. The system sets these parameters according to the following steps: 1. Read the current event status and generate an event gap vector. The event gap vector includes coverage gap value, fact gap value, evidence gap value, conflict gap value, and long-tail exploration value. Specifically, the coverage gap value is determined based on the number of candidate sources, the diversity of source types, and the degree of regional or industry coverage; the fact gap value is determined based on the proportion of missing required fact fields; the evidence gap value is determined based on the gap between the existing evidence level and the downstream required evidence level; the conflict gap value is determined based on the number of conflict fields, the importance of conflict fields, and the credibility of conflict sources; and the long-tail exploration value is determined based on the fixed source hit rate, the sparsity of candidate clues, and the coverage of similar historical tasks.
[0057] 2. Determine the primary trigger scenario based on the event gap vector. If the coverage gap value is the highest, it is a coverage gap type; if the fact gap value or evidence gap value is the highest, it is an evidence verification type; if the conflict gap value is the highest, it is a conflict resolution type; if the fixed source hit rate is low and the long-tail exploration value is the highest, it is a long-tail exploration type. In addition to the primary trigger scenario, other gap items exceeding the preset threshold are used as secondary trigger scenarios in the budget adjustment.
[0058] 3. Load the corresponding basic budget parameters according to the main triggering scenario, and adjust T_max, R_max, C_max, C_step, N_no_gain and H_manual in combination with event urgency, regional granularity, source sparsity, downstream trust level and historical cost of similar tasks.
[0059] 4. For tasks with coverage gaps, the system increases the maximum number of rounds and limits the cost per action, enabling the system to expand candidate coverage across multiple geographic, institutional, or industry sources. Dynamic search stops when the coverage gap falls below the available threshold, or when expanding candidate sources yields consistently low returns.
[0060] 5. For evidence verification tasks, the system increases the maximum time and cost per action, limits invalid rounds and repeated access to the same source, and prioritizes access to original announcements, authoritative pages, structured interfaces, or independent trusted sources. Dynamic searching stops when the evidence level meets downstream trust requirements or when no new verifiable evidence is obtained for several consecutive rounds.
[0061] 6. For conflict resolution tasks, the system sets a smaller maximum number of rounds and a higher sensitivity for manual review. If the conflict intensity does not decrease significantly after several rounds of searching, or if both conflicting parties come from highly reliable sources and the current tool cannot further confirm them, the automatic search will stop and the process will proceed to manual review.
[0062] 7. For long-tail exploratory tasks, the system sets lower maximum time, maximum cost, and single-action cost, and allows for more low-cost exploration rounds. Dynamic searching stops when the candidate clue expansion rate, effective fact increment, or source credibility continuously falls below a preset threshold.
[0063] 8. After each round of action execution, the system updates the task revenue and remaining budget based on the number of newly added facts, the improvement of evidence level, the decrease in conflict intensity, the expansion of clues, the proportion of duplicate sources, the proportion of noise sources, and the cost of the action. If any budget parameter reaches its upper limit, or if the task revenue falls below the minimum revenue threshold for N consecutive rounds, the dynamic search stops, and the system outputs the reason for stopping, the facts that have been filled, the remaining gaps, the chain of evidence, and the manual review flag.
[0064] Through the aforementioned budget gating mechanism, the system can control the time, rounds, and cost boundaries of dynamic searches for different scenarios such as coverage completion, evidence verification, conflict resolution, and long-tail exploration, so that the search process can continuously advance around specific fact gaps while avoiding boundless crawling.
[0065] As an optional implementation, the event state includes a candidate content set, a fact atom set, an event fingerprint, and a conflict set. The event state, credibility profile, and event version of the first news event are updated based on the candidate content, specifically including: S1031. Standardize the candidate content and extract fact atoms to obtain candidate fact atoms; S1032. Compare the candidate fact atoms with the event fingerprint of the current first news event to determine whether the candidate content belongs to the first news event; S1033. When the candidate content belongs to the first news event, compare the candidate fact atom with the current fact atom set of the first news event to obtain the conflict field; S1034. Update the candidate content set according to the candidate content, update the fact atom set and event fingerprint according to the candidate fact atom, and update the conflict set according to the conflict field. S1035. Update the trusted profile based on the source type of the candidate content and the initial trusted tag; S1036. Update the event version based on the updated set of fact atoms and the set of conflicts.
[0066] like Figure 4The diagram illustrates event governance and skill strategy generation and reuse provided in an embodiment of the present invention. The event governance in this embodiment is not accomplished through a single scoring formula, but rather by establishing a state-based processing model oriented towards events. For any event task, the system continuously maintains the event state and performs state transitions based on new materials, new facts, and new evidence.
[0067] For each news event, embodiments of the present invention maintain at least the following state information: 1. Candidate material set: including fixed source materials, dynamically searched materials, historical event materials, and artificially supplemented materials; 2. A collection of factual atoms: including subject, action, object, time, place, value, conclusion, and source evidence; 3. Source credibility profile: including source type, domain reliability, geographical reliability, reprint ratio, error correction records, and historical consistency; 4. Event fingerprint: including core entities, core behaviors, time range, geographical or industry tags, key values, and representative sources; 5. Conflict set: includes conflict fields, conflict sources, conflict impact scope, and actions to be verified; 6. Event version and downstream availability: including statuses such as pending certification, available, conflicted, corrected, and archived.
[0068] Whenever fixed-source content or dynamically searched evidence enters the system, the event-level governance module executes one or more state transition actions. State transition actions include: 1. Material Standardization: Convert web pages, announcements, PDFs, interface data, and reprinted content into unified text and metadata; 2. Atomic Fact Extraction: Missing fields are left empty or pending verification, and are not automatically filled in as confirmed facts; 3. Event fingerprint matching: Determine whether candidate materials belong to existing events; 4. Source verification: Distinguish between original announcements, authoritative institutions, independent media, reprinting sites, and low-credibility sources; 5. Conflict identification: Comparing different statements on the same factual field; 6. Version Generation: Update the event version based on new evidence, fact corrections, and conflict states.
[0069] During event merging, this embodiment of the invention employs event fingerprint matching, evidence status judgment, and conflict gating to merge candidate materials, avoiding the erroneous merging of similar but different events. The specific steps are as follows: 1. Generate event fingerprints for candidate materials. Event fingerprints should include at least the core entity, core behavior, time window, geographic label, quantity field, status field, and source type fingerprint. 2. Apply hard constraints to filter candidate materials and existing events. If the core entities are inconsistent, the time windows are too different, or the regional and industry labels are obviously mismatched, they will not be included in the same merged candidate set. 3. Calculate similarity for objects filtered by hard constraints, and comprehensively compare the title, factual atoms in the body text, numerical fields, subject relationships, time relationships, and source consistency; 4. When the similarity reaches the merging threshold and there are no key field conflicts, the candidate materials are grouped into the same event and the evidence chain is updated; 5. When the core entities are the same but there are significant differences in time, place, quantity, responsible party, or event status, they are retained as different versions of the same source or parallel sub-events, and do not directly cover the original facts; 6. When conflicts arise between highly trusted sources, generate conflicting versions and retain evidence of multiple versions until corrections are completed by authoritative sources or manual review.
[0070] Through the above processing, the system can both aggregate multi-source materials on the same event and separate similar but different events for processing, and avoid directly covering the facts based solely on title similarity or human experience.
[0071] The source credibility profile is maintained based on segmented scenarios, rather than using a single global credibility score. For example, local government websites have high credibility for local government announcements, but this does not necessarily support rumors in a particular industry; vertical media can serve as a supplementary source for information in specific industries, but when it comes to regulatory penalties, listed company disclosures, or public safety incidents, confirmation from authoritative institutions or original announcements is still required. The source credibility profile is updated based on the following factors: 1. Source type and source level; 2. Whether it is the original announcement or primary source material; 3. Historical reliability within a specific region, industry, or institutional sector; 4. The proportion of historical facts confirmed; 5. The proportion of reprinted content and the proportion of duplicate content; 6. Error correction records and manual review results; 7. Consistency with other independent and trusted sources.
[0072] This invention manages event versions and downstream availability uniformly within the version status. Event versions include the following types: 1. Version requiring supplementary evidence: Key facts or evidence chains are insufficient, and it is only allowed to enter the internal candidate pool or be displayed under controlled conditions; 2. Available Versions: Key facts meet the evidence requirements and are available for use in summaries, broadcasts, Q&As, and recommendations; 3. Conflicting Version: Key facts remain in conflict and can only be expressed in a controlled manner as "inconsistent statements from multiple parties," or subject to manual review; 4. Revised Version: Generated after authoritative sources or manual review have corrected existing facts, triggering downstream content refresh; 5. Archived Version: The event is outdated or has been processed, but the historical evidence chain is preserved for future reference.
[0073] Based on the updated set of fact atoms and conflict set, determine whether the key facts and evidence of the news event are sufficient and whether there are any unresolved conflicts, thereby updating the event version.
[0074] Through the above event governance model, this embodiment of the invention extends news event governance from one-time deduplication to sustainable event state management.
[0075] As a further optional implementation, the event handling task trajectory of the first news event is abstracted into a corresponding event handling skill strategy, specifically including: S1041. Determine the event processing task trajectory based on the content search log and event governance log of the first news event; S1042. Segment the event processing task trajectory and extract continuous action segments, success branches, failure branches, rollback branches and stop branches; S1043. Abstract the specific operations and parameters of each continuous action segment and each branch into action primitives and reusable slots respectively to obtain the abstract result; S1044. Encapsulate the abstract results into an event handling skill strategy, and determine the applicable scenarios, input conditions, recommended tool sequences, retrieval templates, validation rules, and conflict handling rules for the event handling skill strategy.
[0076] Specifically, embodiments of the present invention record task trajectory logs during content search and event management. The logs include at least: 1. Task type, triggering conditions, and gap fields; 2. Initial seed clues, region, industry, and entity type; 3. The tool retrieves the sequence, search terms, and access source; 4. Page validity, extracted facts, and level of evidence; 5. Conflicting fields, conflict resolution actions, and version changes; 6. Reasons for stopping, reasons for failure, and feedback from manual review.
[0077] When the task outcome exhibits stable reusable value, embodiments of the present invention abstract the task trajectory into a skill strategy. This reusable value is typically reflected in the following aspects: 1. Obtain authoritative sources; 2. Complete the key fact fields; 3. Resolve core conflicts; 4. Reduce duplicate sources; 5. Reduce the number of invalid searches; 6. Confirmed as valid by manual review.
[0078] During the abstraction process, embodiments of the present invention convert task trajectories into skill strategies according to the following steps: 1. Segment the task trajectory and extract continuous action segments, success branches, failure branches, rollback branches, and stop branches; 2. Abstract specific operations into action primitives, such as general search, site search, webpage access, text parsing, PDF parsing, structured query, fact extraction, source comparison, and manual review requests; 3. Abstract specific parameters into reusable slots, such as region, institution type, industry category, keywords, evidence level, source priority, and event status; 4. Retain the sequence of actions, prerequisites, switching conditions, exception recovery actions, and stopping conditions that yield valid results in this task; 5. Remove actions that do not contribute and one-off noise cues, and compress recurring successful paths into standard templates; 6. Encapsulate the abstract results into skill strategies and record the applicable scenarios, input conditions, recommended tool sequences, retrieval templates, validation rules, and conflict handling rules.
[0079] The skill strategies obtained through the above steps include at least the following: 1. Applicable scenarios: such as county and district policies, industry association announcements, enterprise announcements, regulatory penalties, verification of emergencies, or investigation of rumors; 2. Input conditions: such as region, industry, entity type, gap type, source type, and event status; 3. Recommended tool sequence: For example, first use general search, then site search, then PDF parsing, and finally source comparison; 4. Search templates and source priorities: For example, prioritize access to government departments, stock exchange disclosures, corporate announcements, or industry associations; 5. Verification rules and conflict resolution rules: For example, key facts must be supported by an authoritative source or two independent, credible sources; 6. Abnormal handling and stopping conditions: such as demoting the ranking of reprinting sites, skipping pages without content, and stopping continuous low-return actions.
[0080] As an optional implementation, the target skill strategy corresponding to the second news event is matched according to the skill strategy library, and content search and event governance are performed on the second news event according to the target skill strategy, specifically including: S1051. Determine the event scenario of the second news event, and match the target skill strategy that is consistent with the event scenario from the skill strategy library; S1052. Based on the input conditions of the target skill strategy, the recommended tool sequence, and the retrieval template, perform content search on the second news event, and perform event governance on the second news event according to the verification rules and conflict handling rules of the target skill strategy.
[0081] Specifically, when subsequent task assignments related to a second news event arrive, existing skill strategies are matched based on task characteristics and historical effects, and the search templates, source priorities, tool order, and stopping conditions are adjusted accordingly. The matched skill strategies are used to initialize the search templates, source priorities, tool order, and stopping conditions for dynamic search tasks, and can also be used for conflict resolution strategies and version availability determination during the event governance phase. For example, the "County / District Policy Announcement Exploration Strategy" prioritizes searching county / district governments, competent authorities, policy titles, application guidelines, and local media; the "Listed Company Rumor Verification Strategy" prioritizes accessing exchange disclosures, company announcements, regulatory inquiries, and authoritative financial media, while treating forum or self-media content as low-credibility leads; and the "Enterprise Project Progress Verification Strategy" prioritizes searching enterprise announcements, local government responses, bidding platforms, environmental impact assessment announcements, and regulatory documents.
[0082] As an optional implementation, the news event processing method further includes the following steps: S201. Determine the positive benefit items or negative penalty items of the target skill strategy based on the content search results and event governance results of the second news event; S202. Update the skill value of the target skill strategy based on the positive benefit or negative penalty, and adjust the matching priority of the target skill strategy based on the skill value.
[0083] Specifically, the value of a skill strategy can be updated after it has been used. This embodiment of the invention updates the value of the skill strategy according to the following steps: 1. Record whether this call generates valid news or valid events, and the corresponding fact quality, evidence quality, coverage gain, conflict resolution benefit, noise, erroneous merging, incorrect facts, and execution cost; 2. Convert the current results into positive benefit terms and negative penalty terms, and retain historical performance by combining historical retention coefficients; 3. If the current results continue to produce effective output, the skill value will be increased; if noise, misclassification, or excessive costs are found, the skill value will be decreased. 4. Write the updated skill value back to the skill strategy state and use it as the basis for subsequent priority scheduling.
[0084] This invention dynamically adjusts skill priority based on changes in skill strategy value. When the value of a skill strategy falls below a preset threshold, the corresponding skill strategy is downgraded, limited, or disabled. When a skill strategy continues to be effective in a specific scenario, its invocation priority is increased. When errors are discovered during manual review, the relevant search templates, source rules, conflict handling rules, or action paths are corrected.
[0085] The following example, “News Collection and Management of New Energy Vehicle Subsidy Policy in a Certain County,” further illustrates the implementation process of this invention.
[0086] 1) News gathering stage After receiving a news request for "a certain county's new energy vehicle subsidy policy", the system first activates the fixed source collection module to obtain candidate content from mainstream news sources, local media, county government websites, competent department websites, government interfaces and historical databases, and records the source, publication time, title, body text, URL, source type and initial trust mark.
[0087] After preliminary analysis of the candidate content, the event-level governance module found that the fixed source only returned the reprinted title and local media summary, lacking key facts such as the original announcement, issuing department, subsidy recipient, subsidy amount, and application deadline. Therefore, the event was marked as a coverage gap, factual gap, and evidence gap.
[0088] Subsequently, the system triggers a dynamic search task, using county / district names, new energy vehicles, subsidies, applications, announcements, finance, industry and information technology, and development and reform commissions as seed clues, prioritizing access to county / district government websites, competent department websites, government information disclosure platforms, and local integrated media. The system uses tools such as general search, site search, webpage access, and PDF parsing to locate the original policy announcements and extract information such as publication time, issuing department, applicable targets, subsidy amount, application deadline, application materials, and attachment links. For reprinted sites that only duplicate the title without adding new facts, the system records them as duplicate sources and lowers their subsequent access priority.
[0089] 2) Incident Governance Phase The system uses the original announcement as primary evidence and local media outlets and government information disclosure platforms as supplementary evidence. It standardizes the collected multi-source materials, extracts fact atoms, merges events, and verifies the sources.
[0090] In this embodiment, the system maintains fact atoms such as subject, action, object, time, location, value, conclusion, and evidence around the same policy event. For example, the subject is the competent department of a certain county or district, the action is the release of a new energy vehicle subsidy policy, the time includes the announcement date and the application period, the value includes the subsidy amount or subsidy ratio, and the evidence includes the original announcement URL, PDF attachments, and links to supplementary reports.
[0091] If key facts are sufficiently supported by the original announcement or credible sources, the system will update the event status from "Pending Verification" to "Available," allowing it to enter the summary, broadcast, Q&A, and recommendation links. If new materials such as "subsidy amount adjustment" or "application deadline extension" subsequently appear, the system will not directly overwrite the original facts, but will first determine whether it belongs to a new development of the event, a factual correction, or an unverified conflict. For corrected information from announcements by competent authorities, the system generates a corrected version and triggers downstream content updates; for information from low-credibility reprinted sources, the system generates a conflict record to be verified and restricts definitive expressions downstream.
[0092] 3) Skills Accumulation Stage Once the original announcement is successfully located and key facts are supplemented for this task, the system will abstract the effective task trajectory as "County / District Policy Announcement Exploration Skills".
[0093] This skill primarily focuses on the following: applicable scenarios include county and district policies, subsidy policies, application notices, and government announcements; priority sources are county and district government websites, competent department websites, government information disclosure platforms, and local media; the recommended execution path is to first perform a general search, then a site-specific search, and if necessary, access web pages and parse PDFs; the verification rule prioritizes the original announcement or competent department documents for key facts; the exception handling rule is to lower the ranking of sites without reprinted text and skip duplicate title sources; the stopping condition is obtaining the original announcement and supplementing the key facts, or failing to provide new valid evidence for multiple consecutive rounds.
[0094] When the system processes similar county and district policy news again, it can directly reuse this skill to initialize the search template, source priority, tool call order and stopping conditions, thereby reducing invalid search rounds and improving the efficiency of locating original announcements and the quality of event governance.
[0095] The implementation process of the present invention has been described above. It is understood that the embodiments of the present invention determine whether to trigger dynamic search based on the source information of candidate content obtained from the initial search. This allows for the supplementation of candidate content when its availability or credibility is insufficient. Dynamic search tasks are executed based on the task status, and the content collection process can be modeled as a continuously advancing task state, thereby improving the efficiency and reliability of news content acquisition. Furthermore, the event handling task trajectory of news events is abstracted into event handling skill strategies for reuse, enabling subsequent news events to quickly perform content search and event governance based on the skill strategy library, thus improving the efficiency and reliability of news event handling.
[0096] Compared with the prior art, the embodiments of the present invention also have the following advantages: 1) Enhance news coverage: When fixed sources are insufficient, the system can proactively supplement original announcements, authoritative sources, and long-tail leads to enhance the coverage of county and district news, vertical industry news, and early-stage breaking events.
[0097] 2) Improve the controllability of the collection process: Control the dynamic search scope through stateful tasks and stopping conditions to reduce boundless retrieval, duplicate access, and invalid tool calls.
[0098] 3) Reduce the cost of repeated exploration: By reusing skills and strategies, similar news tasks can directly inherit effective search paths, source priorities and verification constraints, reducing the cost of exploring from scratch.
[0099] 4) Improve the quality of event governance: Reduce duplicate content, low credibility content and mis-merging issues through event-level merging, fact extraction, source verification and evidence chain management.
[0100] 5) Reduce the risk of downstream misuse: By controlling the event version and availability, prevent content with insufficient evidence, conflicting sources, or unverified information from directly entering the broadcast, Q&A, and summary links.
[0101] 6) Support for continuous evolution and traceability: The system can record new developments in events, factual corrections, changes in conflicts, and historical evidence chains, which facilitates subsequent traceability, error correction, and content updates.
[0102] like Figure 5 The diagram shown is a structural schematic of a news event processing device based on stateful task execution and skill strategy reuse provided in an embodiment of the present invention. (Refer to...) Figure 5 This invention provides a news event processing device based on stateful task execution and skill strategy reuse, comprising: The preliminary search module is used to obtain several candidate contents of the first news event from preset fixed news sources, and determine whether to trigger dynamic search based on the source information of each candidate content. The dynamic search module is used to generate a dynamic search task when a dynamic search is triggered, and to execute the dynamic search task based on the task status to obtain supplementary candidate content. The event governance module is used to update the event status, trust profile, and event version of the first news event based on the candidate content when dynamic search is not triggered. The skill strategy abstraction module is used to abstract the event handling task trajectory of the first news event into the corresponding event handling skill strategy, and add the event handling skill strategy to the preset skill strategy library; The skill strategy reuse module is used to match the target skill strategy corresponding to the second news event based on the skill strategy library, and to perform content search and event management for the second news event based on the target skill strategy.
[0103] It is understood that the content of the above method embodiments is applicable to the present device embodiments. The specific functions implemented by the present device embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.
[0104] This invention also provides an electronic device, comprising: a memory, a processor, a program stored in the memory and executable on the processor, and a data bus for communication between the processor and the memory. When the program is executed by the processor, it implements the aforementioned news event processing method based on stateful task execution and skill strategy reuse. This electronic device can be any smart terminal, including tablet computers, in-vehicle computers, etc.
[0105] like Figure 6 The diagram shown is a hardware structure schematic of an electronic device provided in an embodiment of the present invention. (Refer to...) Figure 6 This invention provides an electronic device, comprising: The processor 601 can be implemented using a general-purpose CPU (Central Processing Unit), microprocessor, application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of the present invention. The memory 602 can be implemented as a read-only memory (ROM), static storage device, dynamic storage device, or random access memory (RAM). The memory 602 can store the operating system and other applications. When the technical solutions provided in the embodiments of this specification are implemented through software or firmware, the relevant program code is stored in the memory 602 and is invoked by the processor 601 to execute the news event processing method based on stateful task execution and skill strategy reuse of the embodiments of this invention. The input / output interface 603 is used to implement information input and output; The communication interface 604 is used to enable communication and interaction between this device and other devices. Communication can be achieved through wired means (such as USB, network cable, etc.) or wireless means (such as mobile network, WIFI, Bluetooth, etc.). Bus 605 transmits information between various components of the device (e.g., processor 601, memory 602, input / output interface 603, and communication interface 604); The processor 601, memory 602, input / output interface 603, and communication interface 604 are connected to each other within the device via bus 605.
[0106] It is understood that the content of the above method embodiments is applicable to this device embodiment. The specific functions implemented by this device embodiment are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.
[0107] like Figure 7 The diagram shown is a structural schematic of the storage medium provided in an embodiment of the present invention. (Refer to...) Figure 7 The present invention also provides a storage medium, which is a computer-readable storage medium for computer-readable storage. The storage medium stores one or more programs 701, which can be executed by one or more processors to implement the above-described news event processing method based on stateful task execution and skill strategy reuse.
[0108] It is understood that the content of the above method embodiments is applicable to this storage medium embodiment. The specific functions implemented in this storage medium embodiment are the same as those in the above method embodiments, and the beneficial effects achieved are also the same as those achieved in the above method embodiments.
[0109] This invention also discloses a computer program product, including a computer program that, when executed by a processor, implements the aforementioned news event processing method based on stateful task execution and skill strategy reuse.
[0110] It is understood that the content of the above method embodiments is applicable to the embodiments of this program product. The specific functions implemented by the embodiments of this program product are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.
[0111] Memory, as a non-transitory computer-readable storage medium, can be used to store non-transitory software programs and non-transitory computer-executable programs. Furthermore, memory may include high-speed random access memory, and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some embodiments, memory may optionally include memory remotely located relative to the processor, and these remote memories can be connected to the processor via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.
[0112] The embodiments described in this invention are for the purpose of more clearly illustrating the technical solutions of the embodiments of this invention, and do not constitute a limitation on the technical solutions provided by the embodiments of this invention. As those skilled in the art will know, with the evolution of technology and the emergence of new application scenarios, the technical solutions provided by the embodiments of this invention are also applicable to similar technical problems.
[0113] The terms "first," "second," "third," "fourth," etc. (if present) in the specification and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.
[0114] In some alternative embodiments, the functions / operations mentioned in the block diagrams may not occur in the order shown in the operation diagrams. For example, depending on the functions / operations involved, two consecutively shown blocks may actually be executed substantially simultaneously, or the aforementioned blocks may sometimes be executed in reverse order. Furthermore, the embodiments presented and described in the flowcharts of this invention are provided by way of example to provide a more comprehensive understanding of the technology. The disclosed methods are not limited to the operations and logic flows presented herein. Alternative embodiments are contemplated in which the order of various operations is changed and sub-operations described as part of a larger operation are executed independently.
[0115] Furthermore, although the invention has been described in the context of functional modules, it should be understood that, unless otherwise stated, one or more of the aforementioned functions and / or features may be integrated into a single physical device and / or software module, or one or more functions and / or features may be implemented in a separate physical device or software module. It is also understood that a detailed discussion of the actual implementation of each module is unnecessary for understanding the invention. Rather, given the properties, functions, and internal relationships of the various functional modules in the apparatus disclosed herein, the actual implementation of the module will be understood within the scope of conventional skill of an engineer. Therefore, those skilled in the art can implement the invention as set forth in the claims using ordinary techniques without excessive experimentation. It is also understood that the specific concepts disclosed are merely illustrative and not intended to limit the scope of the invention, which is determined by the full scope of the appended claims and their equivalents.
[0116] If the aforementioned functions are implemented as software functional units 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 invention, or the part that contributes to the prior art, or a portion 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, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. 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.
[0117] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-including system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device.
[0118] More specific examples (a non-exhaustive list) of computer-readable media include: electrical connections (electronic devices) having one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which the aforementioned program can be printed, because the aforementioned program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.
[0119] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.
[0120] In the foregoing description of this specification, references to terms such as "one embodiment," "another embodiment," or "some embodiments" indicate that a specific feature, structure, material, or characteristic described in connection with an embodiment or example is included in at least one embodiment or example of the present invention. In this specification, illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.
[0121] Although embodiments of the invention have been shown and described, those skilled in the art will understand that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the claims and their equivalents.
[0122] The above is a detailed description of the preferred embodiments of the present invention. However, the present invention is not limited to the above embodiments. Those skilled in the art can make various equivalent modifications or substitutions without departing from the spirit of the present invention. All such equivalent modifications or substitutions are included within the scope defined by the claims of this application.
Claims
1. A news event processing method based on stateful task execution and skill strategy reuse, characterized in that, Includes the following steps: Obtain several candidate contents of the first news event from preset fixed news sources, and determine whether to trigger dynamic search based on the source information of each candidate content; When a dynamic search is triggered, a dynamic search task is generated, and the candidate content is obtained by executing the dynamic search task based on the task status. If dynamic search is not triggered, update the event status, credibility profile, and event version of the first news event based on the candidate content; The event handling task trajectory of the first news event is abstracted into a corresponding event handling skill strategy, and the event handling skill strategy is added to a preset skill strategy library; Match the target skill strategy corresponding to the second news event according to the skill strategy library, and perform content search and event management on the second news event according to the target skill strategy.
2. The news event processing method based on stateful task execution and skill strategy reuse according to claim 1, characterized in that, The source information includes source type, publication time, news elements, and initial trust markers. The step of determining whether to trigger dynamic search based on the source information of each candidate content specifically includes: Based on the source type, determine whether the news source coverage of the first news event meets the preset coverage requirements; Based on the publication time and the news elements, determine whether there are any missing or conflicting key facts in the first news event; Based on the initial credibility marker, determine whether the first news event has insufficient evidence level; If the coverage of the news source meets the preset coverage requirements, the key facts of the first news event are not missing or conflicting, and the first news event does not have insufficient evidence level, it is determined that dynamic search will not be triggered; otherwise, it is determined that dynamic search will be triggered.
3. The news event processing method based on stateful task execution and skill strategy reuse according to claim 1, characterized in that, The generation of a dynamic search task, which involves executing the dynamic search task based on the task state to obtain the supplemented candidate content, specifically includes: Determine the task objective, triggering cause, seed clues, source constraints, toolset, and execution boundary of the dynamic search task; Multiple candidate actions are generated based on the current task status of the dynamic search task; Determine the expected new facts, verification, conflict resolution, clue expansion, noise risk, duplication risk, and action cost for each candidate action; The priority of each candidate action is determined based on the expected new facts, the expected verification, the expected conflict resolution, the expected clue expansion, the noise risk, the repetition risk, and the action cost. The candidate action with the highest priority is executed to obtain the supplemented candidate content, and the current task status of the dynamic search task is updated.
4. The news event processing method based on stateful task execution and skill strategy reuse according to claim 1, characterized in that, The event state includes a candidate content set, a fact atom set, an event fingerprint, and a conflict set. Updating the event state, credibility profile, and event version of the first news event based on the candidate content specifically includes: The candidate content is standardized and fact atoms are extracted to obtain candidate fact atoms; The candidate fact atom is compared with the event fingerprint of the first news event to determine whether the candidate content belongs to the first news event; When the candidate content belongs to the first news event, the candidate fact atom is compared with the current set of fact atoms of the first news event to obtain the conflict field; Update the candidate content set according to the candidate content, update the fact atom set and the event fingerprint according to the candidate fact atom, and update the conflict set according to the conflict field; The trusted profile is updated based on the source type of the candidate content and the initial trusted tag. The event version is updated based on the updated set of fact atoms and the set of conflicts.
5. A news event processing method based on stateful task execution and skill strategy reuse according to claim 1, characterized in that, The abstraction of the event handling task trajectory of the first news event into a corresponding event handling skill strategy specifically includes: The event processing task trajectory is determined based on the content search log and event governance log of the first news event; The event processing task trajectory is segmented to extract continuous action segments, success branches, failure branches, rollback branches, and stop branches. The specific operations and parameters of each continuous action segment and each branch are abstracted into action primitives and reusable slots to obtain the abstraction result. The abstract results are encapsulated into the event handling skill strategy, and the applicable scenarios, input conditions, recommended tool sequences, retrieval templates, verification rules, and conflict handling rules of the event handling skill strategy are determined.
6. The news event processing method based on stateful task execution and skill strategy reuse according to claim 5, characterized in that, The step of matching the target skill strategy corresponding to the second news event according to the skill strategy library, and performing content search and event management on the second news event according to the target skill strategy, specifically includes: Determine the event scenario of the second news event, and match the target skill strategy that is applicable to the event scenario from the skill strategy library; The second news event is searched for content based on the input conditions, recommended tool sequence, and retrieval template of the target skill strategy, and event governance is performed on the second news event according to the verification rules and conflict handling rules of the target skill strategy.
7. A news event processing method based on stateful task execution and skill strategy reuse according to any one of claims 1 to 6, characterized in that, The news event handling method also includes the following steps: Based on the content search results and event governance results of the second news event, determine the positive benefit items or negative penalty items of the target skill strategy; The skill value of the target skill strategy is updated based on the positive benefit or the negative penalty, and the matching priority of the target skill strategy is adjusted based on the skill value.
8. A news event processing device based on stateful task execution and skill strategy reuse, characterized in that, include: The preliminary search module is used to obtain several candidate contents of the first news event from preset fixed news sources, and determine whether to trigger dynamic search based on the source information of each candidate content. The dynamic search module is used to generate a dynamic search task when a dynamic search is triggered, and to execute the dynamic search task based on the task status to obtain the supplemented candidate content. The event governance module is used to update the event status, trust profile, and event version of the first news event based on the candidate content when dynamic search is not triggered. The skill strategy abstraction module is used to abstract the event processing task trajectory of the first news event into a corresponding event processing skill strategy, and add the event processing skill strategy to a preset skill strategy library; The skill strategy reuse module is used to match the target skill strategy corresponding to the second news event according to the skill strategy library, and to perform content search and event management for the second news event according to the target skill strategy.
9. An electronic device, characterized in that, The electronic device includes a memory, a processor, a computer program stored in the memory and executable on the processor, and a data bus for enabling communication between the processor and the memory. When the computer program is executed by the processor, it implements the news event processing method based on stateful task execution and skill strategy reuse as described in any one of claims 1 to 7.
10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the news event processing method based on stateful task execution and skill strategy reuse as described in any one of claims 1 to 7.
