Equipment terminal control method and device for municipal problem abnormal events

Abstract

The embodiment of the invention discloses an equipment terminal control method and device for municipal problem abnormal events. A specific embodiment of the method comprises the steps of preprocessing multi-modal data related to municipal problem anomalous events to generate a structured feature vector set; generating event judgment information; performing text filling and optimization on the called standardized template by utilizing a pre-trained natural language processing large model; performing multi-level sensitive word verification and interception processing on the standardized crime settlement opinion text to generate a target crime settlement opinion; performing association packaging on the structured feature vector set, the target crime settlement opinion and the event judgment information; sending the synchronizable data packet to a target data platform, and receiving synchronization receipt information; and generating a processing instruction, and based on the processing instruction, controlling the corresponding equipment terminal to execute the processing task. According to the embodiment, the response efficiency of municipal problem abnormal event disposal can be improved, and the time cost is saved.

Description

Technical Field

[0001] Embodiments of this disclosure relate to the field of computer technology, and more specifically to device terminal control methods, apparatus, electronic devices, and computer-readable media for abnormal events related to municipal issues. Background Technology

[0002] Currently, the handling process for abnormal municipal issues (such as cases reported through the 12345 citizen hotline) relies heavily on manual processing. This typically includes manual analysis, rule matching, manual drafting of case closure opinions, manual review and cross-departmental communication, and manual on-site operation of relevant equipment to handle the abnormal municipal issues.

[0003] However, when using the above methods to handle abnormal events related to municipal issues, the following technical problems often arise: The long process chain leads to excessively long response and actual handling cycles, failing to meet the timeliness requirements of emergency events. Traditional methods are also unable to fully cover all kinds of sensitive information in massive amounts of text, further resulting in excessively long response times when handling abnormal events in municipal affairs. As a result, a large amount of technical resources and emergency response time are wasted.

[0004] The information disclosed in this background section is only intended to enhance the understanding of the background of the inventive concept, and therefore may contain information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0005] The summary portion of this disclosure is intended to provide a brief overview of the concepts, which will be described in detail in the detailed description portion later. This summary portion is not intended to identify key or essential features of the claimed technical solutions, nor is it intended to limit the scope of the claimed technical solutions.

[0006] Some embodiments of this disclosure provide device terminal control methods, apparatuses, electronic devices, and computer-readable media for municipal problem anomalies, in order to solve one or more of the technical problems mentioned in the background section above.

[0007] In a first aspect, some embodiments of this disclosure provide a device terminal control method for municipal problem anomalies, comprising: responding to receiving a processing request for a municipal problem anomaly, preprocessing multimodal data related to the municipal problem anomaly using a pre-trained natural language processing model to generate a structured feature vector set; generating event determination information based on the structured feature vector set using a pre-trained dynamic determination model; and performing text filling and optimization on a called standardized template using the pre-trained natural language processing model based on the event determination information and the structured feature vector set to generate a standardized template. The process involves: generating a case closure opinion text; performing multi-level sensitive word verification and interception on the standardized case closure opinion text to generate a target case closure opinion; associating and encapsulating the structured feature vector set, the target case closure opinion, and the case judgment information to generate a synchronizeable data packet; sending the synchronizeable data packet to the target data platform using a preset application programming interface (API) and receiving synchronization receipt information returned by the target data platform; responding to the received synchronization receipt information, generating processing instructions for the aforementioned municipal problem anomaly based on the synchronizeable data packet, and controlling the corresponding device terminal to execute processing tasks based on the processing instructions.

[0008] Secondly, some embodiments of this disclosure provide a device terminal control apparatus for municipal problem anomalies, comprising: a preprocessing unit configured to, in response to receiving a processing request for a municipal problem anomaly, preprocess multimodal data related to the municipal problem anomaly using a pre-trained natural language processing model to generate a structured feature vector set; a generation unit configured to, based on the structured feature vector set, generate event determination information using a pre-trained dynamic determination model; and a filling and optimization unit configured to, based on the event determination information and the structured feature vector set, fill and optimize a called standardized template using the pre-trained natural language processing model to generate a standardized case closure opinion text. The interception unit is configured to perform multi-level sensitive word verification and interception processing on the standardized case closure opinion text to generate a target case closure opinion; the encapsulation unit is configured to encapsulate the structured feature vector set, the target case closure opinion, and the case judgment information to generate a synchronizeable data packet; the sending and receiving unit is configured to send the synchronizeable data packet to the target data platform using a preset application programming interface, and to receive the synchronization receipt information returned by the target data platform; the generation and control unit is configured to, in response to receiving the synchronization receipt information, generate a processing instruction for the aforementioned municipal problem abnormal event based on the synchronizeable data packet, and control the corresponding device terminal to execute the processing task based on the processing instruction.

[0009] Thirdly, some embodiments of this disclosure provide an electronic device, including: one or more processors; and a storage device having one or more programs stored thereon, such that when the one or more programs are executed by the one or more processors, the one or more processors implement the method as described in any implementation of the first aspect.

[0010] Fourthly, some embodiments of this disclosure provide a computer-readable medium having a computer program stored thereon, wherein the program, when executed by a processor, implements the method as described in any implementation of the first aspect.

[0011] The various embodiments of this disclosure have the following beneficial effects: The device terminal control method for handling municipal problem / abnormal events, as described in some embodiments of this disclosure, improves the response efficiency for handling such events and saves time. Specifically, the reason for the low response efficiency in handling municipal problem / abnormal events is that traditional processes rely on manual analysis, manual writing, manual review, and on-site manual operation of equipment, with each step sequential and experiencing waiting and transmission delays. Based on this, the device terminal control method for municipal problem / abnormal events, as described in some embodiments of this disclosure, firstly, in response to receiving a processing request for a municipal problem / abnormal event, uses a pre-trained natural language processing (NLP) model to preprocess the multimodal data related to the aforementioned municipal problem / abnormal event to generate a structured feature vector set. By using the pre-trained NLP model to uniformly understand and represent the multimodal data, unstructured citizen demands and information are transformed into a structured feature vector set that can be accurately computed by machines, solving the problems of data heterogeneity and information dispersion. Then, based on the aforementioned structured feature vector set, an event judgment information is generated using a pre-trained dynamic judgment model. A dynamic judgment model is used to analyze structured features, automatically and accurately determining the nature of the case (e.g., type, whether it is a special case) and the handling conclusion (closure, exclusion, or postponement). Then, based on the aforementioned event judgment information and the aforementioned structured feature vector set, the pre-trained natural language processing model is used to fill in and optimize the text of the called standardized template to generate standardized case closure opinion text. By calling the standardized template and using the large model for text filling and optimization, a case closure opinion with standardized expression, complete content, and traceability is automatically generated. This step ensures the standardization of the case closure opinion, avoiding problems such as inconsistent expression and missing key information caused by manual drafting, while also supporting content reuse in cases with multiple plaintiffs, greatly improving the quality and efficiency of opinion generation. Next, multi-level sensitive word verification and interception processing is performed on the aforementioned standardized case closure opinion text to generate the target case closure opinion. By employing a multi-level sensitive word database (basic, government-specific, and departmental personalized) and semantic association analysis, in-depth verification and intelligent interception of case closure opinions are conducted, ensuring the compliance and security of the output content and effectively preventing the leakage of sensitive information or non-standard expressions. This also saves response time in handling abnormal events related to municipal issues. Secondly, the aforementioned structured feature vector set, the target case closure opinions, and the case judgment information are associated and encapsulated to generate a synchronizeable data package. The core features of the case, the final opinion, and the judgment conclusion are associated and encapsulated to form a synchronizeable data package containing a complete chain of evidence. This achieves the integrity and structured packaging of case information, providing standardized data units for secure, efficient data synchronization and subsequent traceability. Thirdly, using a pre-defined application programming interface, the aforementioned synchronizeable data package is sent to the target data platform, and the synchronization receipt information returned by the target data platform is received.By securely sending synchronizeable data packets to the upper-level platform through a pre-defined application programming interface (API) and receiving signed receipts, automated, tamper-proof, and verifiable real-time synchronization of cross-system data is achieved. This solves the problems of low efficiency, error-proneness, and difficulty in traceability associated with traditional manual synchronization, ensuring data consistency. Finally, in response to receiving the synchronization receipt information, processing instructions for the aforementioned municipal problem / anomaly event are generated based on the synchronizeable data packets. Based on these processing instructions, the corresponding equipment terminals are controlled to execute processing tasks. Based on the confirmation of successful synchronization, precise equipment control instructions are automatically generated, driving the corresponding physical equipment (e.g., maintenance vehicles, drones) to perform the handling tasks. This achieves closed-loop control from the information domain to the physical domain, transforming analytical decision-making results into actual actions and significantly shortening the response time from reporting municipal problem / anomalies to on-site handling. Attached Figure Description

[0012] The above and other features, advantages, and aspects of the embodiments of this disclosure will become more apparent from the accompanying drawings and the following detailed description. Throughout the drawings, the same or similar reference numerals denote the same or similar elements. It should be understood that the drawings are schematic, and elements are not necessarily drawn to scale.

[0013] Figure 1 This is a flowchart of some embodiments of the device terminal control method for abnormal events of municipal problems according to the present disclosure; Figure 2 This is a structural schematic diagram of some embodiments of a device terminal control device for abnormal events of municipal problems according to the present disclosure; Figure 3 This is a schematic diagram of the structure of an electronic device suitable for implementing some embodiments of the present disclosure. Detailed Implementation

[0014] Embodiments of this disclosure will now be described in more detail with reference to the accompanying drawings. While some embodiments of this disclosure are shown in the drawings, it should be understood that this disclosure can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided to provide a more thorough and complete understanding of this disclosure. It should be understood that the accompanying drawings and embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of protection of this disclosure.

[0015] It should also be noted that, for ease of description, only the parts relevant to the invention are shown in the accompanying drawings. Unless otherwise specified, the embodiments and features described in this disclosure can be combined with each other.

[0016] It should be noted that the concepts of "first" and "second" mentioned in this disclosure are used only to distinguish different devices, modules or units, and are not used to limit the order of functions performed by these devices, modules or units or their interdependencies.

[0017] It should be noted that the terms "a" and "a plurality of" used in this disclosure are illustrative rather than restrictive, and those skilled in the art should understand that, unless otherwise expressly indicated in the context, they should be understood as "one or more".

[0018] The names of messages or information exchanged between multiple devices in the embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of such messages or information.

[0019] This disclosure will now be described in detail with reference to the accompanying drawings and embodiments.

[0020] refer to Figure 1 The flowchart 100 illustrates some embodiments of a device terminal control method for municipal problem anomalies according to the present disclosure. This device terminal control method for municipal problem anomalies includes the following steps: Step 101: In response to receiving a processing request for an abnormal event related to municipal issues, a pre-trained natural language processing model is used to preprocess the multimodal data related to the abnormal event to generate a structured feature vector set.

[0021] In some embodiments, the executing entity (e.g., an electronic device) of the above-described device terminal control method for abnormal events related to municipal problems can be hardware or software. When the computing device is hardware, it can be implemented as a distributed cluster composed of multiple servers or terminal devices, or as a single server or a single terminal device. When the computing device is software, it can be installed in the hardware devices listed above. It can be implemented as multiple software programs or software modules to provide distributed services, or as a single software program or software module. No specific limitations are made here.

[0022] In other embodiments, the aforementioned execution entity may, in response to receiving a processing request for an abnormal municipal issue event, utilize a pre-trained natural language processing (NLP) model to preprocess the multimodal data related to the abnormal municipal issue event to generate a structured feature vector set. The aforementioned abnormal municipal issue event can refer to an abnormal situation occurring in city operations that requires intervention from government functional departments. For example, the aforementioned abnormal municipal issue event could be a complaint request from a citizen reporting an elevator malfunction in a residential community through channels such as the 12345 hotline. The aforementioned processing request can be an instruction or data input that triggers the system to initiate the processing procedure. For example, the aforementioned processing request could be a complaint form submitted by a citizen through the 12345 hotline. The aforementioned pre-trained NLP model can refer to a NLP model pre-trained on massive amounts of general text, possessing strong language understanding capabilities, and optimized for specific tasks (e.g., a domain-optimized model with a Transformer+Mamba hybrid architecture). The input to the aforementioned pre-trained NLP model can be a standard dataset of multimodal related data after format unification and cleaning. The output of the aforementioned pre-trained large-scale natural language processing model can be a set of structured feature vectors. This pre-trained model may include: an input encoding layer (input is a standard dataset, output is a sequence of word vectors), a hybrid feature extraction layer (input is the sequence of word vectors output by the encoding layer, output is an enhanced feature representation fusing global dependencies and long sequence features), a feature decoding layer (input is the enhanced feature representation, output is the initial prediction result (including confidence) of structured feature labels), and a post-processing optimization layer (input is the initial prediction result of the feature decoding layer, output is a set of structured feature vectors). The aforementioned multimodal data can refer to various data in different forms related to the same abnormal event in municipal affairs. For example, the aforementioned multimodal data could be the speech-to-text of a complaint call (text), photos taken at the scene (images), the geographical location of the incident (coordinates), or the case submission time (time-series data). The aforementioned structured feature vector set can refer to a set of numerical vectors with clear semantics and mathematical form that transforms multimodal data into directly processable data.

[0023] In some optional implementations of certain embodiments, the aforementioned execution entity may, in response to receiving a processing request for an abnormal municipal problem event, utilize a pre-trained natural language processing model to preprocess the multimodal data related to the abnormal municipal problem event to generate a structured feature vector set, which may include the following steps: The first step is to standardize and clean the aforementioned multimodal data to generate a standard dataset. This standard dataset can be a structured collection of data that has undergone standardized formatting and data cleaning. In practice, firstly, the multimodal data (text, images, locations) undergoes format conversion (standardized encoding to UTF-8, coordinate conversion to WGS84 standard) and data cleaning (removing garbled characters, completing missing fields, and standardizing time format). Then, it is encapsulated into a standardized JSON structure to generate the standard dataset.

[0024] The second step involves using the pre-trained large-scale natural language processing model to extract and fuse features from the standard dataset, generating an intermediate feature vector set. This intermediate feature vector set can be the original, unoptimized high-dimensional semantic feature representation extracted by the model. In practice, first, the standard dataset is input into a hybrid architecture (Transformer + Mamba layer) of the pre-trained large-scale natural language processing model. Then, core features (e.g., focus of the appeal, urgency) are extracted through semantic encoding. Finally, these features are fused into an intermediate feature vector set.

[0025] The third step involves performing structured encoding and dimensionality reduction on the intermediate feature vector set to generate a structured feature vector set. In practice, this is first done using PCA to reduce the dimensionality of the intermediate feature vector set. Then, structured encoding (e.g., one-hot encoding of case types) is applied. Finally, the structured feature vector set is output.

[0026] Step 102: Based on the structured feature vector set, generate event determination information using a pre-trained dynamic determination model.

[0027] In some embodiments, the aforementioned executing entity can generate event determination information based on the aforementioned structured feature vector set and using a pre-trained dynamic determination model. The pre-trained dynamic determination model can be a model trained on historical case closure / removal data and integrating a rule engine and a machine learning module. The inputs to the pre-trained dynamic determination model include: rule-based data (preliminary rule matching results), the structured feature vector set, and historical data on similar events (case closure records of similar cases, historical removal rates of departments), and the output is event determination information. The aforementioned pre-trained dynamic judgment model may include: a core input layer (a neural network layer (an encoder network consisting of an embedding layer, a fully connected layer, and a normalization layer connected sequentially), with inputs of a pre-defined rule base and a set of structured feature vectors, and outputs standardized and quantized data in a unified format), a rule engine layer (a feedforward network based on a decision tree, with inputs of the unified format data output from the core input layer and outputs preliminary rule matching results), a machine learning judgment layer (using a gradient boosting tree (XGBoost architecture), with inputs of preliminary rule matching results, a set of structured feature vectors, and historical data of similar events, and outputs preliminary judgment information), and a dynamic weight adjustment layer (an attention mechanism and a gating fusion network, with inputs of preliminary judgment information output from the machine learning judgment layer, dynamic parameters of assessment requirements, and the remaining time factor of the current assessment cycle, and outputs event judgment information). The aforementioned event judgment information may be the result of handling abnormal events related to municipal issues, and may include: conclusions, basis, and time limits. For example, the aforementioned event judgment information may be {"Conclusion": "Case closed with permission", "Basis": "Three rates met", "Time limit": "24 hours"}.

[0028] In some optional implementations of certain embodiments, the execution entity can generate event determination information based on the structured feature vector set and using a pre-trained dynamic determination model, which may include the following steps: The first step involves performing a matching mapping operation based on the aforementioned structured feature vector set and the preset rule base to generate preliminary rule matching results. The preset rule base can be a predefined set of business rules, which may include basic and personalized rules. For example, the preset rule base may include: {"Rule 1": "Elevator malfunction requires a response within 24 hours", "Rule 2": "Repeated complaints require secondary verification"}. In practice, firstly, the structured feature vector (e.g., [Case Type: Property Management: 0.9]) is matched against the preset rule base. Then, it is checked whether a mandatory rule (e.g., "Safety-related issues require a 2-hour response") is triggered. Finally, the preliminary rule matching results are output.

[0029] The second step involves generating preliminary judgment information based on the preliminary rule matching results, the structured feature vector set, and historical similar event data, using the pre-trained dynamic judgment model. This preliminary judgment information can serve as the initial decision suggestion output by the dynamic judgment model. The historical similar event data can be the processing records and results of previous similar cases. For example, the average processing time for three "streetlight malfunction" cases on the same road segment last year was 2.5 hours. First, the preliminary rule matching results, the structured feature vector set, and historical similar event data (e.g., elevator malfunction processing records from the past six months) are input into the machine learning judgment layer of the dynamic judgment model. Then, the model calculates multi-dimensional feature weights. Finally, it outputs preliminary judgment information with confidence levels.

[0030] The third step involves adjusting dynamic thresholds and redistributing priorities based on the preliminary assessment information, pre-acquired dynamic parameters of the assessment requirements, and the remaining time factor of the current assessment period. This generates optimized event handling information. The dynamic parameters of the assessment requirements can be quantifiable indicators for the current assessment period, changing over time. For example, the current assessment requirement could be an increase in the "completion rate" requirement from 90% to 95% this month. The remaining time factor of the current assessment period can be the coefficient of influence of the remaining time in the assessment period on the urgency of the assessment. For example, the remaining time factor of the current assessment period could be the last week of the quarter, with the time factor weight increasing from 0.3 to 0.6. The optimized event handling information can be a refined handling plan after dynamic adjustment. For example, the optimized event handling information could be {"Handling Level": "Urgent", "Priority": "P1", "Adjusted Time Limit": "12 hours"}. In practice, first, the dynamic parameters of the assessment requirements are read (e.g., a satisfaction rate requirement of 95% this month). Then, the judgment threshold is adjusted based on the remaining time factor of the current assessment period (e.g., the last 3 days of the month). Finally, the priorities are recalculated to generate optimized event handling information. For example, the time limit for "routine maintenance" will be reduced from 48 hours to 36 hours at the end of the quarter.

[0031] The fourth step involves matching corresponding configuration information from a pre-defined resource matching library based on the optimized incident handling information to generate a structured contingency plan. This pre-defined resource matching library can be a directory of available departments, personnel, and equipment resources. For example, it could be {"Streetlight Maintenance Team": "3 teams available", "Engineering Vehicles": "2 vehicles available"}. The structured contingency plan can be a complete action plan including specific measures, resources, and procedures.

[0032] The fifth step involves cross-departmental verification of the structured contingency plan to generate event determination information. In practice, the structured contingency plan is first sent to relevant departments (e.g., power companies, traffic police departments) for review. Then, feedback is received and conflicts are corrected, and finally, final determination information confirmed by all parties is generated.

[0033] Step 103: Based on the event determination information and the structured feature vector set, use a pre-trained natural language processing model to fill in and optimize the text of the called standardized template to generate a standardized case closure opinion text.

[0034] In some embodiments, the aforementioned executing entity may, based on the aforementioned event determination information and the aforementioned structured feature vector set, utilize the aforementioned pre-trained natural language processing model to perform text filling and optimization on the invoked standardized template, thereby generating a standardized case closure opinion text. The aforementioned standardized template may be a pre-designed structured text framework, which may include fixed text and variable placeholders. The aforementioned standardized case closure opinion text may be a response text that conforms to government document standards, is complete in elements, and has a consistent expression.

[0035] In some optional implementations of certain embodiments, the execution entity may, based on the event determination information and the structured feature vector set, utilize the pre-trained natural language processing model to perform text filling and optimization on the invoked standardized template to generate standardized case closure opinion text, which may include the following steps: The first step involves retrieving the event type identifier and processing result from the aforementioned event determination results, and then calling the matching target template from a pre-defined standardized template library. The event type identifier can be a unique code identifier for the event category. The processing result can be the specific handling type of the determination conclusion (e.g., "removed" or "transferred for verification"). The target template can be a specific template selected based on the event type and processing result. In practice, the event determination information is first parsed to extract the event type identifier (e.g., "PS003 - Road Damage") and processing result (e.g., "Case Closed"). Then, it is matched against the pre-defined standardized template library, and the corresponding template is called. Finally, the target template is returned.

[0036] The second step involves extracting padding data corresponding to the preset target template based on the aforementioned structured feature vector set, thereby generating template padding data. This padding data can be specific information extracted from the feature vectors used to replace template variables. The template padding data can also be a complete dataset with completed variable mapping. In practice, this can be achieved by determining the variable placeholders of the target template (e.g., {location}), extracting corresponding data from the structured feature vector set, and assembling them into key-value pair padding data to obtain the template padding data.

[0037] The third step involves inputting the target template and the template-filled data into the pre-trained natural language processing (NLP) model to generate a preliminary opinion text. This preliminary opinion text can be a preliminary closing opinion generated based on the template and data. In practice, the template text and the filled data are first concatenated. Then, it is input into the NLP model for semantic coherence optimization. Finally, a preliminary opinion text is generated (e.g., "After verification by the street office, the elevator malfunction you reported in XX residential area has been repaired.").

[0038] The fourth step involves performing a content reuse detection test on multiple plaintiff cases based on the event determination information mentioned above, generating a detection result. The multiple plaintiff case identifier can be a marker indicating that multiple claims point to the same event. The detection result can be an analytical conclusion determining whether the current case belongs to the category of multiple plaintiff cases. In practice, first, it checks whether the event carries a plaintiff case identifier. Then, it queries the related case database. Finally, it analyzes the similarity of claims to generate the detection result.

[0039] The fifth step involves extracting the case closure opinion content in response to the aforementioned detection results meeting preset conditions, to obtain the content-reused opinion text. The preset conditions can be thresholds for triggering content reuse. The content-reused opinion text can be a batch of text generated by reusing the core content of existing case closure opinions. In practice, firstly, the main case opinion text is extracted based on the detection results. Then, the differentiating fields are replaced. Finally, the content-reused opinion text is generated.

[0040] The sixth step involves performing standardization checks and logical corrections on the reused opinion text to generate a standardized closing opinion text. In practice, this involves first checking the text's standardization (e.g., policy citation format), then verifying logical consistency (e.g., chronological order), and finally outputting the standardized closing opinion.

[0041] Step 104: Perform multi-level sensitive word verification and interception processing on the standardized case closure opinion text to generate the target case closure opinion.

[0042] In addressing the aforementioned technical challenges by employing technical solutions, the application scenario—reviewing the closing opinions of large-scale, high-frequency municipal hotline calls—often presents the following technical problems: traditional review methods (e.g., single keyword matching technology) struggle to comprehensively cover all explicit and implicit sensitive information within massive amounts of text, leading to a high risk of non-compliant content leaks, low review efficiency, wasted response time, and additional costs associated with handling public opinion due to misjudgments. Considering the following requirements for this application scenario—high accuracy, real-time performance, scalability, and automated processing capabilities—we have decided to adopt the following solution: In some embodiments, the aforementioned implementing entity may perform multi-level sensitive word verification and interception processing on the standardized case closure opinion text to generate a target case closure opinion. The aforementioned multi-level sensitive words may refer to a sensitive word classification system based on risk level (e.g., high risk (political), medium risk (non-standard terminology), low risk (colloquial expression)). The aforementioned target case closure opinion may refer to the compliant and publishable final case closure text generated after sensitive word verification and interception.

[0043] In some optional implementations of certain embodiments, the aforementioned executing entity may perform multi-level sensitive word verification and interception processing on the standardized case closure opinion text to generate the target case closure opinion, which may include the following steps: The first step is to construct a multi-level distributed sensitive word index based on pre-defined multi-type thesaurus to build a real-time sensitive word detection engine. The multi-type thesaurus can be collections of sensitive words categorized by application scenario and sensitivity level, including: a basic thesaurus (storing nationally common sensitive words, such as illegal / irregular, politically sensitive, vulgar / abusive terms), a government-specific thesaurus (e.g., customized for citizen hotline scenarios, including unpublished policy numbers, non-standard government terminology, and terms related to illegal promises), and a department-specific thesaurus (supporting each responsible unit to add its own exclusive sensitive words). The multi-level distributed sensitive word index can refer to a multi-level thesaurus index stored using a prefix tree structure, supporting fast matching. The real-time sensitive word detection engine can refer to a distributed index built on the multi-type thesaurus (basic thesaurus, government-specific thesaurus, department-specific thesaurus), utilizing an efficient data structure (e.g., a prefix tree Trie) to support millisecond-level retrieval. In practice, firstly, a distributed index is built based on the multi-type thesaurus using a prefix tree structure to construct an engine supporting real-time detection.

[0044] The second step involves inputting the standardized case closure opinion text into the real-time sensitive word detection engine. A precise word comparison is then performed using a target algorithm to generate a list of precisely matched sensitive words. The target algorithm can refer to the core algorithm used for sensitive word matching, such as the AC automaton. The fuzzy sensitive word list can be a list of sensitive words identified through literal matching. For example, the sensitive word list could include "contraband". In practice, first, the standardized case closure opinion text is input into the real-time sensitive word detection engine. Then, literal matching is performed using the AC automaton algorithm. Finally, a list of precisely matched sensitive words is generated.

[0045] Step 3: Based on the preset homophone variant rule library and the mapping table of similar-shaped characters, perform fuzzy matching detection on the above standardized case closing opinion text to generate a list of fuzzy sensitive words. Among them, the above homophone variant rule library can be a library storing the mapping relationships of homophone variants of sensitive words. For example, the above homophone variant rule library can include: mapping "氵查" to "查" and "无馬" to "无码". The above mapping table of similar-shaped characters can be a mapping table storing the substitution relationships of similar-shaped characters. The above list of fuzzy sensitive words can be a list of sensitive words identified through fuzzy matching such as homophones and similar-shaped characters. In practice, first, perform fuzzy matching detection on the text based on the homophone rule library and the mapping table of similar-shaped characters. Then, generate a list of sensitive words for fuzzy matching.

[0046] Step 4: Perform semantic association on the above standardized case closing opinion text to generate semantically implicit sensitive content. Among them, the above semantically implicit sensitive content can be implicitly illegal expressions identified through semantic analysis. In practice, first, analyze the semantics of the standardized case closing opinion text. Then, identify the implicitly illegal expressions. Finally, generate sensitive content at the semantic level.

[0047] Step 5: Based on the above context scenario information of the standardized case closing opinion text, the above semantically implicit sensitive content, the above list of fuzzy sensitive words, and the above list of accurately matched sensitive words, determine the risk classification result. Among them, the above context scenario information can refer to context information such as event type and processing scenario. For example, the above context scenario information can be the specific scenario of property management cases. The above risk classification result can be a classification based on the risk level of sensitive content. For example, high risk (illegal), medium risk (non-standard terms), low risk (colloquial). In practice, first, combine the context scenario information. Then, comprehensively consider the list of accurately matched sensitive words, the semantically implicit sensitive content, and the list of fuzzy sensitive words. Finally, classify according to the risk level to determine the risk classification result.

[0048] Step 6: According to the above risk classification result, perform hierarchical interception processing to generate an interception result. Among them, the above interception result can be the result of taking corresponding processing measures according to the risk level. In practice, first, perform hierarchical processing according to the risk classification result, then directly intercept high-risk content, suggest replacement for medium-risk content, and give a prompt for low-risk content. Finally, generate an interception result. For example, directly intercept and give an early warning for high-risk words, and suggest replacing medium-risk words with standard expressions.

[0049] Step 7: Verify the above interception result to generate the target case closing opinion. In practice, first, verify the interception result. Then, ensure that all sensitive content has been processed. Finally, generate a compliant target case closing opinion.

[0050] The above-described operational steps, as an inventive point of this disclosure, solve the technical problem mentioned in the background section: "Traditional review methods are unable to comprehensively cover various explicit and implicit sensitive information in massive amounts of text, leading to a high risk of non-compliant content leakage, low review efficiency, wasted valuable response time, and additional handling costs for public opinion caused by misjudgments." The reasons for these technical problems are as follows: single keyword matching technology cannot effectively handle complex variations of sensitive information such as homophones, similar-looking characters, and semantic implicatures, and lacks understanding of the business scenario context, resulting in high rates of misjudgment and missed judgment. This invention, by constructing a three-level intelligent detection and hierarchical interception mechanism of "precise matching + fuzzy matching + semantic analysis," achieves millisecond-level, comprehensive, adaptive security filtering and standardization of case closure opinions, reducing time costs and lowering additional handling costs caused by the leakage of sensitive information.

[0051] Step 105: The structured feature vector set, target case closure opinions, and case judgment information are associated and encapsulated to generate a synchronizeable data packet.

[0052] In some embodiments, the aforementioned executing entity may associate and encapsulate the aforementioned structured feature vector set, the aforementioned target case closure opinion, and the aforementioned case judgment information to generate a synchronizeable data packet. This synchronizeable data packet may be a structured data unit that encapsulates key data from the entire process of municipal problem anomalies according to platform specifications for cross-system synchronization.

[0053] As an example: First, the structured feature vector set (core case features), the target case closure opinion (final review text), and the case judgment information (case closure / removal conclusion) are mapped and assembled according to the city-level platform interface specifications (e.g., JSON Schema). Then, timestamps and data signatures are injected. Finally, the data packets are encrypted and compressed to generate standard, synchronizeable data packets.

[0054] Step 106: Using the preset application programming interface, send the synchronizeable data packet to the target data platform and receive the synchronization receipt information returned by the target data platform.

[0055] In some embodiments, the executing entity can utilize a preset application programming interface (API) to send the synchronizeable data packets to the target data platform and receive synchronization receipt information returned by the target data platform. The API can be a programming interface for interacting with the target data platform. The target data platform can be a superior or external system receiving the synchronized data, such as a municipal hotline management platform. The synchronization receipt information can be a receipt confirmation credential returned by the target data platform.

[0056] In addressing the technical challenges mentioned above, the application scenario of cross-level synchronization of government data often presents the following challenges: difficulty in ensuring data security, low transmission reliability, and untimely status confirmation, leading to a significant waste of operational resources. Considering the specific requirements of this application scenario—high security, strong consistency, and real-time feedback—we have decided to adopt the following solution: In some optional implementations of certain embodiments, the execution entity may utilize a preset application programming interface to send the synchronizeable data packet to the target data platform and receive synchronization receipt information returned by the target data platform, which may include the following steps: The first step involves performing interface adaptation and conversion based on the aforementioned synchronizeable data packets and the interface specifications of the target data platform to generate platform-compatible data packets. The interface specifications can be the constraints imposed by the target data platform on data formats and protocols. The platform-compatible data packets can be data packets converted according to the interface specifications. In practice, the field names and data types of the synchronizeable data packets can be mapped and converted according to the interface specifications of the target data platform. Then, platform-compatible data packets that meet the platform requirements are generated.

[0057] The second step involves identifying privacy fields in the platform-compatible data packets to generate a privacy field list. This privacy field list can be a list of sensitive fields in the data packets that need to be de-identified. In practice, this can be achieved by first scanning all fields in the data packet, then matching them against a pre-defined privacy field rule base, and finally generating the privacy field list.

[0058] The third step involves performing differentiated de-identification transformation based on the aforementioned privacy field list to generate intermediate de-identified data. This intermediate de-identified data can be transitional data with some fields already de-identified. In practice, first, the privacy field list is used. Then, different fields are processed using corresponding de-identification algorithms. Finally, partially de-identified intermediate data is generated.

[0059] The fourth step involves injecting a reversible mapping identifier into the intermediate de-identified data to generate a de-identified data packet. This reversible mapping identifier can be a unique identifier that supports data restoration under authorized scenarios. The de-identified data packet can be a complete data packet that has completed privacy protection processing. In practice, firstly, a reversible mapping identifier is injected into each de-identified field. Then, the mapping relationship between each identifier and the original value is encrypted and stored. Finally, a complete de-identified data packet is generated.

[0060] The fifth step involves encrypting the de-identified data packets using a preset encryption algorithm to generate encrypted transmission data packets. The preset encryption algorithm can be a pre-defined data encryption standard (e.g., the Chinese national standard SM4 algorithm). The encrypted transmission data packets can be ciphertext data packets processed by the encryption algorithm.

[0061] The sixth step involves sending the encrypted data packet to the target data platform via the aforementioned application programming interface (API) and receiving preliminary response data. This preliminary response data can be the initial response from the target data platform after its first receipt of the request. In practice, the encrypted data packet is first sent to the target platform's API via HTTPS. Then, the platform's preliminary response data is received to record the response status.

[0062] Step 7: Perform status verification on the preliminary response data to generate a status verification report. This status verification report can be a structured report generated after parsing and verifying the preliminary response. In practice, first, the status code and header information of the preliminary response are parsed; then, the validity of the digital signature is verified; and finally, a status verification report is generated.

[0063] Step 8: Based on the aforementioned status verification report, initiate a confirmation request to the target data platform and determine the complete synchronization confirmation information returned by the target data platform. This complete synchronization confirmation information can be the final processing result returned by the target platform. In practice, first, according to the instructions in the status report, then, initiate a confirmation request to the target platform to query the final status, and finally, obtain the complete confirmation information.

[0064] The ninth step is to extract the digital signature and timestamp from the complete synchronization confirmation information to generate a synchronization receipt. In practice, first, key elements are extracted from the complete confirmation information, and then a local timestamp is added to generate a standardized receipt as the synchronization receipt information.

[0065] The above-described steps, as an inventive point of this disclosure, solve the technical problems mentioned in the background art: "difficulty in ensuring data security, low transmission reliability, and untimely status confirmation, leading to a waste of a large amount of operation and maintenance resources." The reasons for these technical problems are as follows: traditional synchronization solutions lack a complete privacy protection mechanism, encrypted transmission guarantee, and reliable status confirmation process, resulting in a high risk of data leakage, unreliable transmission process, and difficulty in verifying synchronization results. This invention, by establishing a complete secure synchronization chain including privacy desensitization, encrypted transmission, and dual confirmation, achieves secure, reliable, and auditable cross-platform data synchronization, saving operation and maintenance resources caused by data leakage.

[0066] Step 107: In response to receiving the synchronization receipt information, generate a processing instruction for the abnormal event of the municipal problem based on the synchronizeable data packet, and control the corresponding device terminal to execute the processing task based on the processing instruction.

[0067] In some embodiments, the execution entity may, in response to receiving the synchronization receipt information, generate a processing instruction for the aforementioned municipal problem anomaly based on the synchronizeable data packet, and control the corresponding device terminal to execute the processing task based on the processing instruction. The processing instruction may be a sequence of commands to control the device to perform specific operations. The device terminal may be a physical device performing specific municipal processing. For example, the device terminal may be a drone, a maintenance robot, or a sensor. The processing task may be an execution unit containing specific operation instructions generated by the system and issued to the device terminal in response to the municipal problem anomaly. For example, the processing task may be an instruction to a maintenance robot to "go to Building 3 of XX Community to repair the elevator."

[0068] In addressing the technical challenges of the aforementioned background technologies through technological solutions, the application scenario—intelligent emergency repair and automated operation of municipal facilities (e.g., street light repair, road flooding drainage, temporary fencing for missing manhole covers)—often presents the following technical problems: long on-site response delays and insufficient operational accuracy in complex environments, leading to increased safety hazards, prolonged public service interruptions, wasted valuable emergency response windows, and redundant maintenance resources. Considering the following requirements for this application scenario: rapid response, precise control, and adaptive adjustment, we have decided to adopt the following solution: In some optional implementations of certain embodiments, the execution entity may, in response to receiving the synchronization receipt information, generate a processing instruction for the aforementioned municipal problem anomaly based on the synchronizeable data packet, and control the corresponding device terminal to execute the processing task based on the processing instruction, which may include the following steps: The first step is to generate a transaction confirmation result based on the transaction status code and platform confirmation identifier from the aforementioned synchronization receipt information. The transaction status code can be the status identifier code from the synchronization receipt. For example, the transaction status code could be HTTP 200 for success and 500 for failure. The platform confirmation identifier can be a unique identifier for the synchronization transaction returned by the target data platform. The transaction confirmation result can be a structured judgment result after parsing the synchronization status. In practice, first, the status code and confirmation identifier in the synchronization receipt are parsed. Then, their validity is verified to generate the transaction confirmation result.

[0069] The second step involves generating an event handling feature set by extracting the event type identifier and handling priority parameters from the synchronizeable data packets. The event type identifier can be an event classification label, such as "elevator malfunction" or "road damage." The handling priority parameters can be quantitative indicators of the handling order set based on factors such as the urgency and scope of impact of the event; for example, cases involving personal safety are set to "highest priority," while ordinary consultation cases are set to "ordinary." The event handling feature set can be a set of features extracted from the data packets (e.g., {Type: "road flooding", Depth: "15cm", Area: "100m²"}). In practice, first, the event type identifier and priority parameters are extracted from the synchronizeable data packets. Then, combined with data such as the event location, the event handling feature set is generated.

[0070] The third step involves matching the aforementioned event handling feature set with a pre-defined equipment instruction rule base to generate a preliminary equipment control instruction sequence. This pre-defined rule base can be a library of equipment control instruction templates, such as a template for an unmanned aerial vehicle (UAV) patrolling a pre-defined route. The preliminary equipment control instruction sequence can be an initial instruction generated by matching the rule base. For example, the preliminary equipment control instruction sequence could be "take off to 50 meters," "cruise to point A," or "take a picture." In practice, first, the event handling feature set is matched with the pre-defined equipment instruction rule base. Then, the corresponding instruction template is invoked. Finally, the initial instruction sequence is generated.

[0071] The fourth step involves optimizing the initial equipment control command sequence based on real-time environmental sensor data and equipment status feedback information to generate processing commands. The real-time environmental sensor data can be real-time monitoring data of the equipment's operating environment, including wind speed, temperature, and visibility. The equipment status feedback information can be the equipment's own status report, such as 80% battery level or good GPS signal. In practice, first, real-time environmental data (e.g., ambient temperature) and equipment status (e.g., robot battery level) are acquired. Then, the initial command sequence is optimized to generate processing commands.

[0072] The fifth step involves determining the corresponding action parameters, execution sequence, and device identifier based on the aforementioned processing instructions to generate a device control task package. The action parameters can be specific execution parameters of the instructions, such as a flight speed of 5 m / s or a camera resolution of 4K. The execution sequence can be the execution order and time interval of multiple instructions, for example, executing instruction 1 followed by instruction 2 after a 10-second interval. The device identifier can be a unique identifier for the device. In practice, the specific actions in the processing instructions are first parsed. Then, the parameters, execution sequence, and device identifier for each action are determined. Finally, this is encapsulated into a device control task package.

[0073] Step 6: Based on the aforementioned device control task package, establish a device control channel. This device control channel can be a secure data link for communication with the device. In practice, first, a communication connection is established based on the device identifier. Then, two-way authentication is performed. Finally, a secure control channel is established to serve as the device control channel.

[0074] Step 7: Distribute the aforementioned device control task package to the corresponding device terminal, and receive real-time execution feedback information through the aforementioned device control channel to generate an instruction execution state stream. The aforementioned real-time execution feedback information can be the immediate status returned by the device during execution. The aforementioned instruction execution state stream can be a continuously updated sequence of device execution state data. In practice, first, the device control task package is sent through the control channel. Then, the real-time execution feedback information returned by the device is continuously received. Finally, the instruction execution state stream is generated.

[0075] Step 8: Based on the aforementioned instruction execution state flow, the control parameters of the corresponding terminal device are dynamically adjusted based on a preset physical anomaly detection model to control the corresponding device terminal to execute the processing task. The aforementioned physical anomaly detection model can be an AI model that identifies device execution anomalies (e.g., a model based on vibration and current waveform analysis, or a fault prediction model based on detecting jamming of a maintenance robot's robotic arm through current waveform detection (e.g., a physical anomaly detection model built using TensorFlow or PyTorch)). The aforementioned control parameters can be dynamically adjustable device operating parameters. In practice, firstly, the execution process is monitored based on the state flow. Then, problems (e.g., obstacles) are identified through the physical anomaly detection model. Finally, the control parameters are dynamically adjusted. For example, if the physical anomaly detection model detects an obstacle ahead, the system adjusts the detour mode in real time to control the successful execution of the maintenance task.

[0076] The above-described operational steps, as an inventive point of this disclosure, solve the technical problems mentioned in the background art: "long delays in on-site response and insufficient operational accuracy in complex environments lead to increased safety hazards, prolonged public service interruptions, and wasted valuable emergency response windows and redundant maintenance resources." The reasons for these technical problems are as follows: traditional handling methods rely on manual personnel carrying equipment to the site before making operational decisions, resulting in a long chain from receiving instructions to physical execution; in the face of dynamic environments (such as weather and obstacles), manual operation is difficult to adjust accurately in real time; and the lack of an intelligent control closed loop based on real-time feedback makes it impossible to cope with sudden on-site situations. This invention establishes a complete automated control chain from intelligent instruction generation and environmental adaptive optimization to dynamic adjustment based on real-time feedback, achieving remote, precise, and adaptive operations for municipal emergency maintenance, saving the social costs caused by public service interruptions and the equipment damage and redundant operation resources caused by inaccurate operation.

[0077] In some optional implementations of certain embodiments, after the execution entity sends the synchronizeable data packet to the target data platform using the preset application programming interface and receives the synchronization receipt information returned by the target data platform, it may include the following steps: The first step, in response to the received synchronization receipt information, is to generate a status confirmation code and a version identifier for the target data platform based on the synchronization receipt information. The status confirmation code can be a processing status code of the target platform after receiving the data (e.g., "RECEIVED" or "PROCESSED"). The version identifier can be a version marker of the synchronized data on the target platform. In practice, first, the synchronization receipt information is parsed to extract fields related to the platform's processing status, and a status confirmation code is generated. Then, the data version identifier field is extracted to generate the version identifier.

[0078] The second step involves performing a data difference comparison between the local case database and the target data platform based on the aforementioned status confirmation code, version identifier, and preset timing strategy to generate an automatic repair script. The preset timing strategy can be a pre-configured data comparison trigger time and frequency rule. For example, the preset timing strategy could be to perform a full comparison at 2 AM daily, or an incremental comparison every 30 minutes. The automatic repair script can be an automatically generated set of data synchronization difference repair instructions. In practice, firstly, when the preset timing strategy is triggered (e.g., 2 AM), the status confirmation code and version identifier are used as query conditions to retrieve corresponding data from the target platform. Then, the retrieved data is compared field-by-field with the corresponding records in the local case database. Finally, the differences are analyzed, and an executable automatic repair script is generated.

[0079] The third step involves simultaneously executing atomic update operations on both the target data platform and the local case database, based on the aforementioned automatic repair script. In practice, this is done within a single database transaction: first, update operations are initiated simultaneously on both the target data platform and the local case database; then, the automatic repair script is executed; finally, it is ensured that both updates either succeed simultaneously or fail simultaneously (atomicity); and finally, the repair operation log is recorded.

[0080] The various embodiments of this disclosure have the following beneficial effects: The device terminal control method for handling municipal problem / abnormal events, as described in some embodiments of this disclosure, improves the response efficiency for handling such events and saves time. Specifically, the reason for the low response efficiency in handling municipal problem / abnormal events is that traditional processes rely on manual analysis, manual writing, manual review, and on-site manual operation of equipment, with each step sequential and experiencing waiting and transmission delays. Based on this, the device terminal control method for municipal problem / abnormal events, as described in some embodiments of this disclosure, firstly, in response to receiving a processing request for a municipal problem / abnormal event, uses a pre-trained natural language processing (NLP) model to preprocess the multimodal data related to the aforementioned municipal problem / abnormal event to generate a structured feature vector set. By using the pre-trained NLP model to uniformly understand and represent the multimodal data, unstructured citizen demands and information are transformed into a structured feature vector set that can be accurately computed by machines, solving the problems of data heterogeneity and information dispersion. Then, based on the aforementioned structured feature vector set, an event judgment information is generated using a pre-trained dynamic judgment model. A dynamic judgment model is used to analyze structured features, automatically and accurately determining the nature of the case (e.g., type, whether it is a special case) and the handling conclusion (closure, exclusion, or postponement). Then, based on the aforementioned event judgment information and the aforementioned structured feature vector set, the pre-trained natural language processing model is used to fill in and optimize the text of the called standardized template to generate standardized case closure opinion text. By calling the standardized template and using the large model for text filling and optimization, a case closure opinion with standardized expression, complete content, and traceability is automatically generated. This step ensures the standardization of the case closure opinion, avoiding problems such as inconsistent expression and missing key information caused by manual drafting, while also supporting content reuse in cases with multiple plaintiffs, greatly improving the quality and efficiency of opinion generation. Next, multi-level sensitive word verification and interception processing is performed on the aforementioned standardized case closure opinion text to generate the target case closure opinion. By employing a multi-level sensitive word database (basic, government-specific, and departmental personalized) and semantic association analysis, in-depth verification and intelligent interception of case closure opinions are conducted, ensuring the compliance and security of the output content and effectively preventing the leakage of sensitive information or non-standard expressions. This also saves response time in handling abnormal events related to municipal issues. Secondly, the aforementioned structured feature vector set, the target case closure opinions, and the case judgment information are associated and encapsulated to generate a synchronizeable data package. The core features of the case, the final opinion, and the judgment conclusion are associated and encapsulated to form a synchronizeable data package containing a complete chain of evidence. This achieves the integrity and structured packaging of case information, providing standardized data units for secure, efficient data synchronization and subsequent traceability. Thirdly, using a pre-defined application programming interface, the aforementioned synchronizeable data package is sent to the target data platform, and the synchronization receipt information returned by the target data platform is received.By securely sending synchronizeable data packets to the upper-level platform through a pre-defined application programming interface (API) and receiving signed receipts, automated, tamper-proof, and verifiable real-time synchronization of cross-system data is achieved. This solves the problems of low efficiency, error-proneness, and difficulty in traceability associated with traditional manual synchronization, ensuring data consistency. Finally, in response to receiving the synchronization receipt information, processing instructions for the aforementioned municipal problem / anomaly event are generated based on the synchronizeable data packets. Based on these processing instructions, the corresponding equipment terminals are controlled to execute processing tasks. Based on the confirmation of successful synchronization, precise equipment control instructions are automatically generated, driving the corresponding physical equipment (e.g., maintenance vehicles, drones) to perform the handling tasks. This achieves closed-loop control from the information domain to the physical domain, transforming analytical decision-making results into actual actions and significantly shortening the response time from reporting municipal problem / anomalies to on-site handling.

[0081] Further reference Figure 2 As an implementation of the methods shown in the above figures, this disclosure provides some embodiments of a device terminal control device for abnormal events in municipal affairs. These device embodiments are similar to... Figure 1 Corresponding to the method embodiments shown, this device terminal control device for municipal problem anomalies can be specifically applied to various electronic devices.

[0082] like Figure 2As shown, a device terminal control device 200 for municipal problem anomalies includes: a preprocessing unit 201, a generation unit 202, a filling and optimization unit 203, an interception unit 204, an encapsulation unit 205, a sending and receiving unit 206, and a generation and control unit 207. The preprocessing unit 201 is configured to: in response to receiving a processing request for a municipal problem anomaly, preprocess the multimodal data related to the municipal problem anomaly using a pre-trained natural language processing model to generate a structured feature vector set. The generation unit 202 is configured to: generate event judgment information based on the structured feature vector set using a pre-trained dynamic judgment model. The filling and optimization unit 203 is configured to: fill and optimize the called standardized template using the pre-trained natural language processing model based on the event judgment information and the structured feature vector set to generate a standardized case closure opinion text. The interception unit 204 is configured to: perform multi-level sensitive word verification and interception processing on the standardized case closure opinion text to generate a target case closure opinion. The encapsulation unit 205 is configured to encapsulate the structured feature vector set, the target case closure opinion, and the case judgment information to generate a synchronizeable data packet. The sending and receiving unit 206 is configured to send the synchronizeable data packet to the target data platform using a preset application programming interface, and to receive synchronization receipt information returned by the target data platform. The generation and control unit 207 is configured to: in response to receiving the synchronization receipt information, generate a processing instruction for the aforementioned municipal problem anomaly based on the synchronizeable data packet, and control the corresponding device terminal to execute the processing task based on the processing instruction.

[0083] It is understandable that the units described in the equipment terminal control device 200 for abnormal events related to municipal problems are similar to those in the reference device. Figure 1 The steps in the described method correspond accordingly. Therefore, the operations, features, and beneficial effects described above for the method also apply to the equipment terminal control device 200 and the units contained therein for handling abnormal events related to municipal problems, and will not be repeated here.

[0084] The following is for reference. Figure 3 It shows a schematic diagram of the structure of an electronic device (e.g., an electronic device) 300 suitable for implementing some embodiments of the present disclosure. Figure 3 The electronic device shown is merely an example and should not be construed as limiting the functionality and scope of the embodiments of this disclosure.

[0085] like Figure 3As shown, the electronic device 300 may include a processing unit (e.g., a central processing unit, a graphics processing unit, etc.) 301, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 302 or a program loaded from a storage device 308 into a random access memory (RAM) 303. The RAM 303 also stores various programs and data required for the operation of the electronic device 300. The processing unit 301, ROM 302, and RAM 303 are interconnected via a bus 304. An input / output (I / O) interface 305 is also connected to the bus 304.

[0086] Typically, the following devices can be connected to I / O interface 305: input devices 306 including, for example, touchscreens, touchpads, keyboards, mice, cameras, microphones, accelerometers, gyroscopes, etc.; output devices 307 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; storage devices 308 including, for example, magnetic tapes, hard disks, etc.; and communication devices 309. Communication device 309 allows electronic device 300 to communicate wirelessly or wiredly with other devices to exchange data. Although Figure 3 An electronic device 300 with various devices is shown; however, it should be understood that it is not required to implement or possess all of the devices shown. More or fewer devices may be implemented or possessed alternatively. Figure 3 Each box shown can represent a device or multiple devices as needed.

[0087] In particular, according to some embodiments of this disclosure, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, some embodiments of this disclosure include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via communication device 309, or installed from storage device 308, or installed from ROM 302. When the computer program is executed by processing device 301, it performs the functions defined in the methods of some embodiments of this disclosure.

[0088] It should be noted that, in some embodiments of this disclosure, the computer-readable medium described above may be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. A computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In some embodiments of this disclosure, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In some embodiments of this disclosure, a computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium can be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wires, optical fibers, RF (radio frequency), etc., or any suitable combination thereof.

[0089] In some implementations, clients and servers can communicate using any currently known or future-developed network protocol such as HTTP (Hypertext Transfer Protocol) and can interconnect with digital data communication (e.g., communication networks) of any form or medium. Examples of communication networks include local area networks (“LANs”), wide area networks (“WANs”), the Internet (e.g., the Internet of Things), and peer-to-peer networks (e.g., ad hoc peer-to-peer networks), as well as any currently known or future-developed networks.

[0090] The aforementioned computer-readable medium may be included within the aforementioned electronic device; or it may exist independently without being assembled into the electronic device. The aforementioned computer-readable medium carries one or more programs, which, when executed by the electronic device, cause the electronic device to: respond to receiving a processing request for an abnormal event related to a municipal issue; preprocess the multimodal data related to the abnormal event using a pre-trained natural language processing model to generate a structured feature vector set; based on the structured feature vector set, generate event determination information using a pre-trained dynamic judgment model; and, according to the event determination information and the structured feature vector set, perform text filling and optimization on the invoked standardized template using the pre-trained natural language processing model to generate… The system generates a standardized case closure opinion text; performs multi-level sensitive word verification and interception processing on the standardized case closure opinion text to generate a target case closure opinion; associates and encapsulates the structured feature vector set, the target case closure opinion, and the case judgment information to generate a synchronizeable data packet; uses a preset application programming interface to send the synchronizeable data packet to the target data platform and receives the synchronization receipt information returned by the target data platform; in response to receiving the synchronization receipt information, it generates a processing instruction for the aforementioned municipal problem abnormal event based on the synchronizeable data packet, and controls the corresponding device terminal to execute the processing task based on the processing instruction.

[0091] Computer program code for performing operations of some embodiments of this disclosure can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0092] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this disclosure. 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 indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated 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 the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0093] The units described in some embodiments of this disclosure can be implemented in software or hardware. The described units can also be housed in a processor; for example, a processor may be described as including a preprocessing unit, a generation unit, a filling and optimization unit, an interception unit, a packaging unit, a sending and receiving unit, and a generation and control unit. The names of these units do not necessarily limit the specific unit itself; for example, a preprocessing unit may be described as "a unit that, in response to receiving a processing request for an abnormal municipal problem event, preprocesses the multimodal data related to the abnormal municipal problem event using a pre-trained natural language processing model to generate a structured feature vector set."

[0094] The functions described above in this document can be performed at least in part by one or more hardware logic components. For example, exemplary types of hardware logic components that can be used, without limitation, include: field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), system-on-a-chip (SoCs), complex programmable logic devices (CPLDs), and so on.

[0095] The above description is merely a selection of preferred embodiments of this disclosure and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of the invention involved in the embodiments of this disclosure is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the above-described inventive concept. For example, technical solutions formed by substituting the above-described features with (but not limited to) technical features with similar functions disclosed in the embodiments of this disclosure.

Claims

1. A device terminal control method for abnormal events related to municipal problems, comprising: In response to receiving a processing request for an abnormal event related to a municipal problem, a pre-trained natural language processing model is used to preprocess the multimodal data related to the abnormal event to generate a structured feature vector set. Based on the structured feature vector set, event determination information is generated using a pre-trained dynamic determination model. Based on the event determination information and the structured feature vector set, the pre-trained natural language processing model is used to fill in and optimize the called standardized template to generate standardized case closure opinion text. Multi-level sensitive word verification and interception processing are performed on the standardized case closure opinion text to generate the target case closure opinion; The structured feature vector set, the target case closure opinion, and the case judgment information are associated and encapsulated to generate a synchronizeable data packet; Using a pre-defined application programming interface, the synchronizeable data packet is sent to the target data platform, and the synchronization receipt information returned by the target data platform is received. In response to receiving the synchronization receipt information, based on the synchronizeable data packet, a processing instruction for the abnormal municipal problem event is generated, and based on the processing instruction, the corresponding device terminal is controlled to execute the processing task.

2. The method according to claim 1, wherein, After sending the synchronizeable data packet to the target data platform using a preset application programming interface and receiving the synchronization receipt information returned by the target data platform, the method further includes: In response to receiving the synchronization receipt information, a status confirmation code and version identifier for the target data platform are generated based on the synchronization receipt information; Based on the status confirmation code, the version identifier, and the preset timing strategy, a data difference comparison is performed between the local case database and the target data platform to generate an automatic repair script; Based on the automatic repair script, atomic update operations are performed synchronously on the target data platform and the local case database.

3. The method according to claim 1, wherein, In response to receiving a processing request for an abnormal event related to a municipal issue, the system utilizes a pre-trained natural language processing model to preprocess the multimodal data related to the abnormal event to generate a structured feature vector set, including: The multimodal related data are formatted and cleaned to generate a standard dataset; Using the pre-trained large-scale natural language processing model, feature extraction and fusion are performed on the standard dataset to generate an intermediate feature vector set; The intermediate feature vector set is subjected to structured encoding and dimensionality reduction processing to generate a structured feature vector set.

4. The method according to claim 1, wherein, The step of generating event determination information based on the structured feature vector set and using a pre-trained dynamic determination model includes: Based on the structured feature vector set and the preset rule base, a matching mapping operation is performed to generate preliminary rule matching results; Based on the preliminary rule matching results, the structured feature vector set, and historical data of similar events, preliminary judgment information is generated using the pre-trained dynamic judgment model. Based on the preliminary judgment information, the pre-acquired dynamic parameters of the assessment requirements, and the remaining time factor of the current assessment cycle, dynamic threshold adjustment and priority reallocation are performed to generate optimized event handling information. Based on the optimized event handling information, the corresponding configuration information is matched from the preset resource adaptation library to generate a structured handling plan; The structured response plan is subjected to cross-departmental verification to generate event determination information.

5. The method according to claim 1, wherein, The step of using the pre-trained natural language processing model to fill in and optimize the called standardized template based on the event determination information and the structured feature vector set to generate standardized case closure opinion text includes: Based on the event type identifier and processing result of the event determination result, a matching target template is called from the preset standardized template library; Based on the structured feature vector set, fill data corresponding to the preset target template is extracted to generate template fill data; The target template and the template filling data are input into the pre-trained natural language processing large model to generate preliminary opinion text; Based on the multi-person co-prosecution case identifier in the event determination information, a content reuse detection of co-prosecution cases is performed to generate detection results; In response to the detection results meeting preset conditions, the content of the case closure opinion is extracted to obtain the opinion text after content reuse; The reused opinion text undergoes standard checks and logical corrections to generate a standardized case closure opinion text.

6. A device terminal control device for abnormal events in municipal affairs, comprising: The preprocessing unit is configured to, in response to receiving a processing request for an abnormal event related to a municipal problem, use a pre-trained natural language processing model to preprocess the multimodal data related to the abnormal event to generate a structured feature vector set. The generation unit is configured to generate event determination information based on the structured feature vector set and using a pre-trained dynamic determination model. The filling and optimization unit is configured to fill and optimize the called standardized template using the pre-trained natural language processing big model based on the event determination information and the structured feature vector set, so as to generate standardized case closure opinion text. The interception unit is configured to perform multi-level sensitive word verification and interception processing on the standardized case closure opinion text to generate the target case closure opinion; The encapsulation unit is configured to encapsulate the structured feature vector set, the target case closure opinion, and the case determination information in association to generate a synchronizeable data packet; The sending and receiving unit is configured to use a preset application programming interface to send the synchronizeable data packet to the target data platform and receive the synchronization receipt information returned by the target data platform. The generation and control unit is configured to, in response to receiving the synchronization receipt information, generate a processing instruction for the municipal problem anomaly based on the synchronizeable data packet, and control the corresponding device terminal to execute the processing task based on the processing instruction.

7. An electronic device, comprising: One or more processors; Storage device, on which one or more programs are stored, When the one or more programs are executed by the one or more processors, the one or more processors implement the method as described in any one of claims 1-7.

8. A computer-readable medium having a computer program stored thereon, wherein, When the program is executed by the processor, it implements the method as described in any one of claims 1-7.

Images