A power failure information sensitive word intelligent management method and system

Abstract

This invention discloses an intelligent management method and system for sensitive words in power outage information. The method includes: acquiring and preprocessing the original power outage information text to be published; performing text analysis to identify sensitive words related to power outages in the original power outage information text; determining the multidimensional features of the sensitive words and evaluating their sensitivity to the sensitive words to obtain a sensitivity evaluation value; determining the sensitivity level of the sensitive words based on the sensitivity evaluation value and determining an intelligent management strategy for the sensitive words based on the sensitivity level; and processing the sensitive words in the original power outage information text based on the intelligent management strategy to obtain the processed power outage information text. This invention proactively discovers risks through multidimensional data mining, uses a quantitative model to transform subjective sensitivity into objective values, and matches adaptive processing strategies according to precise levels, improving the efficiency and consistency of information processing, achieving refined and precise risk control, and ensuring information security.

Description

Technical Field

[0001] This invention relates to the field of electrical digital data processing technology, and in particular to an intelligent management method and system for sensitive words in power outage information. Background Technology

[0002] In today's era of highly digitized energy infrastructure and instantaneous social information dissemination, the release of power outage information is not only a basic public service, but also a matter of national security, social stability, and public trust.

[0003] However, traditional power outage information management mainly relies on manual review and experience-based judgment, facing three major dilemmas: First, efficiency and accuracy are difficult to balance. The massive amount of information leads to review delays, and inconsistent human judgment standards easily result in omissions and misjudgments. Second, risk assessment lacks scientific basis, the definition of "sensitive" is vague, and the handling scale often varies from person to person, making it difficult to achieve standardization and compliance auditing. Third, passive response cannot adapt to dynamic risks. Static keyword databases are difficult to cover new risk points and cannot cope with the ever-changing public opinion and security situation. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention provides a method and system for intelligent management of sensitive words related to power outage information, comprising: Obtain the original power outage information text to be published, and preprocess the original power outage information text; Text analysis was performed on the preprocessed original power outage information text to identify sensitive words related to power outages in the original power outage information text; The multidimensional features of sensitive words are determined, and the sensitivity of sensitive words is evaluated based on the multidimensional features to obtain the sensitivity evaluation value of sensitive words; Sensitivity levels of sensitive words are determined based on sensitivity assessment values, and intelligent management strategies for sensitive words are determined based on sensitivity levels. Based on intelligent management strategies, sensitive words in the original power outage information text are processed to obtain the processed power outage information text.

[0005] Furthermore, the step of obtaining the original power outage information text to be published and preprocessing the original power outage information text includes: Obtain the original power outage information text to be published, and preprocess the original power outage information text, including text cleaning and standardization.

[0006] Furthermore, the preprocessed original power outage information text is subjected to text analysis to identify sensitive words related to power outages in the original power outage information text, including: Determine the boundaries of sentences in the preprocessed original power outage information text, and divide the preprocessed original power outage information text into multiple sentences based on the sentence boundaries; Semantic analysis is performed on each sentence to identify entities in each sentence, and the entities are compared and matched with a preset sensitive word lexicon. Entities that match successfully are identified as candidate sensitive words. The semantic meaning of candidate sensitive words is determined based on context analysis, and candidate sensitive words containing information related to power outages in their semantic meanings are identified and determined as sensitive words related to power outages in the original power outage information text.

[0007] Furthermore, the step of comparing and matching entities with a preset sensitive word lexicon, and determining successfully matched entities as candidate sensitive words, includes: Identify each preset sensitive word in the preset sensitive word lexicon and calculate the semantic similarity between each entity and each preset sensitive word; Entities with semantic similarity greater than a preset threshold are selected as successfully matched entities and then identified as candidate sensitive words.

[0008] Furthermore, the multidimensional features for determining sensitive words include: The semantic meaning of sensitive words, their position in sentences, and their frequency in the original power outage information text are determined. Based on the semantic meaning, position, and frequency, the semantic features, positional features, and frequency features of the sensitive words are determined respectively. Sensitive words are constructed based on semantic features, positional features, and frequency features.

[0009] Furthermore, the evaluation of the sensitivity of sensitive words based on multidimensional features to obtain a sensitivity evaluation value for sensitive words includes: Sensitivity analysis is performed on the multidimensional features of sensitive words, and the sensitivity of each multidimensional feature is evaluated based on the analysis results to obtain the sensitivity evaluation value of each multidimensional feature. Determine the preset sensitivity evaluation value for each multidimensional feature, and calculate the difference between the sensitivity evaluation value of each multidimensional feature and the corresponding preset sensitivity evaluation value to obtain the sensitivity difference value of each multidimensional feature. Determine the preset weights of each multidimensional feature, and calculate the sensitivity assessment value of sensitive words based on the preset weights and sensitivity difference values ​​of each multidimensional feature.

[0010] Furthermore, the formula for calculating the sensitivity assessment value of the sensitive words is as follows: , Where P is the sensitivity assessment value of the sensitive word, αi is the preset weight of the i-th multidimensional feature, Di is the sensitivity difference value of the i-th multidimensional feature, and n is the number of multidimensional features.

[0011] Furthermore, determining the sensitivity level of sensitive words based on sensitivity assessment values ​​includes: A preset sensitivity level-sensitivity assessment value range correspondence is set in advance. For each sensitivity assessment value range, a corresponding preset sensitivity level is associated with it. Determine the sensitivity assessment value of the sensitive words, and based on the mapping relationship between the sensitivity assessment value interval to which the sensitivity assessment value belongs and the preset sensitivity level-sensitivity assessment value interval correspondence, select the preset sensitivity level corresponding to the sensitivity assessment value interval as the sensitivity level of the sensitive words.

[0012] Furthermore, the intelligent management strategy for determining sensitive words based on sensitivity levels includes: Determine a first preset threshold and a second preset threshold, and determine the relationship between the sensitivity level of the sensitive word and the first preset threshold and the second preset threshold; If the sensitivity level of a sensitive word is less than or equal to the first preset threshold, the intelligent management strategy for that sensitive word is to semantically fuzzy rewrite the sensitive word in the original power outage information text. If the sensitivity level of a sensitive word is greater than the first preset threshold and less than the second preset threshold, the intelligent management strategy for the sensitive word is to replace the sensitive word in the original power outage information text with semantically standardized text. If the sensitivity level of a sensitive word is greater than or equal to the second preset threshold, the intelligent management strategy for that sensitive word is to cover or delete the sensitive word in the original power outage information text.

[0013] This invention also provides an intelligent management system for sensitive words related to power outage information, comprising: The acquisition module is used to acquire the original power outage information text to be published and to preprocess the original power outage information text. The analysis module is used to perform text analysis on the preprocessed original power outage information text and identify sensitive words related to power outages in the original power outage information text. The evaluation module is used to determine the multidimensional features of sensitive words and evaluate the sensitivity of sensitive words based on the multidimensional features to obtain the sensitivity evaluation value of sensitive words; The determination module is used to determine the sensitivity level of sensitive words based on the sensitivity assessment value, and to determine the intelligent management strategy for sensitive words based on the sensitivity level; The processing module is used to process sensitive words in the original power outage information text based on intelligent management strategies, and obtain the processed power outage information text.

[0014] Compared with existing technologies, the intelligent management method and system for sensitive words in power outage information according to embodiments of the present invention have the following advantages: This invention proactively identifies risks through multi-dimensional data mining, transforms vague, sensitive feelings into objective numerical values ​​using data models, and achieves standardization and objectification of risk assessment, eliminating individual subjective differences. Based on the quantified risk level, it automatically and accurately invokes pre-set optimal processing strategies. Ultimately, while ensuring the inherent security of information, it achieves efficient, accurate, and compliant information dissemination and communication, ensuring the consistency and scientific nature of control measures. Under the premise of safeguarding state secrets, public interests, and social stability, it significantly improves the efficiency and consistency of information processing, achieving refined and precise risk control. At the same time, through full-process data recording, it meets the requirements of compliance auditing and accountability traceability, building a robust, intelligent, and evolving information dissemination security firewall for power companies. Attached Figure Description

[0015] Figure 1 This is a schematic diagram of the process structure of the intelligent management method for sensitive words related to power outage information in an embodiment of the present invention; Figure 2 This is a schematic diagram of the composition of the intelligent management system for sensitive words related to power outage information in an embodiment of the present invention. Detailed Implementation

[0016] The specific embodiments of this application will be described in further detail below with reference to the accompanying drawings and examples. The following examples are for illustrative purposes only and are not intended to limit the scope of the invention.

[0017] like Figure 1 As shown in the embodiments of this application, a method for intelligent management of sensitive words in power outage information is provided, including: S100: obtaining the original power outage information text to be published and preprocessing the original power outage information text; S200: performing text analysis on the preprocessed original power outage information text to identify sensitive words related to power outage in the original power outage information text; S300: determining the multi-dimensional features of the sensitive words and evaluating the sensitivity of the sensitive words based on the multi-dimensional features to obtain a sensitivity evaluation value of the sensitive words; S400: determining the sensitivity level of the sensitive words based on the sensitivity evaluation value and determining an intelligent management strategy for the sensitive words based on the sensitivity level; S500: processing the sensitive words in the original power outage information text based on the intelligent management strategy to obtain the processed power outage information text.

[0018] Furthermore, this invention proactively identifies risks through multi-dimensional data mining, transforms vague perceptions into objective numerical values ​​using data models, and achieves standardization and objectification of risk assessment, eliminating subjective differences. Based on the quantified risk level, it automatically and accurately invokes pre-set optimal processing strategies. Ultimately, while ensuring the inherent security of information, it achieves efficient, accurate, and compliant information dissemination and communication, ensuring the consistency and scientific nature of control measures. Under the premise of safeguarding state secrets, public interests, and social stability, it significantly improves the efficiency and consistency of information processing, achieving refined and precise risk control. At the same time, through full-process data recording, it meets the requirements of compliance auditing and accountability traceability, building a robust, intelligent, and evolving information dissemination security firewall for power companies.

[0019] In the embodiments of this application, a method for intelligent management of sensitive words in power outage information is provided. The step of obtaining the original power outage information text to be published and preprocessing the original power outage information text includes: obtaining the original power outage information text to be published and preprocessing the original power outage information text, the preprocessing including text cleaning and standardization.

[0020] Specifically, irrelevant characters, garbled text, and formatting noise are removed using rules and regular expressions to ensure text purity. Next, formatting is standardized, converting full-width and half-width characters, diverse date and time expressions, and numerical units into machine-readable standard formats. Simultaneously, spell correction and terminology normalization are performed to correct human input errors and force different expressions of the same concept within the industry to be mapped to a single standardized term. This step transforms the chaotic raw information flow into highly standardized and consistent high-quality text data, fundamentally solving the machine understanding ambiguity problem caused by data inconsistency. It provides a reliable and accurate input foundation for subsequent named entity recognition, sensitive word discovery, and semantic analysis, serving as the cornerstone for ensuring the accuracy, consistency, and efficiency of the entire intelligent management system. It achieves the crucial transformation from raw data to analyzable knowledge.

[0021] In embodiments of this application, a method for intelligent management of sensitive words in power outage information is provided. The method involves performing text analysis on the preprocessed original power outage information text to identify sensitive words related to power outages. This includes: determining the boundaries of sentences in the preprocessed original power outage information text and dividing the preprocessed original power outage information text into multiple sentences based on the sentence boundaries; performing semantic analysis on each sentence to identify entities in each sentence and comparing and matching the entities with a preset sensitive word lexicon, identifying successfully matched entities as candidate sensitive words; determining the semantic meaning of the candidate sensitive words based on context analysis, and identifying candidate sensitive words whose semantic meaning contains information related to power outages, and determining them as sensitive words related to power outages in the original power outage information text.

[0022] Specifically, sentence boundary recognition precisely segments coherent text paragraphs into independent semantic units (sentences), establishing a foundational analytical granularity for subsequent refined analysis and avoiding semantic confusion caused by cross-sentence comprehension. Semantic analysis and entity recognition are performed on each sentence, using named entity recognition technology to accurately locate key objects within the sentence and quickly compare them with a pre-set sensitive word library to initially screen candidate sensitive words. By analyzing the grammatical role, semantic relationship, and emotional tone of candidate words in specific sentences, their true meaning is determined. Only entities that genuinely carry risks, impacts, or sensitive attributes related to the power outage event in the context are ultimately identified as sensitive words related to the power outage. This step fundamentally overcomes the shortcomings of traditional keyword matching, which often relies on taking things out of context. It effectively distinguishes between literal and contextual sensitivity, significantly reducing the false positive rate. Furthermore, through contextual understanding, it can discover and capture sensitive expressions not in the basic word library but newly generated through semantic combination, thereby significantly improving recall and coverage. This provides high-quality, highly relevant input for subsequent quantitative risk assessment based on multi-dimensional features, a core prerequisite for ensuring the scientific and effective implementation of the entire management strategy.

[0023] In the embodiments of this application, a method for intelligent management of sensitive words for power outage information is provided. The step of comparing and matching entities with a preset sensitive word library and determining the successfully matched entities as candidate sensitive words includes: determining each preset sensitive word in the preset sensitive word library and calculating the semantic similarity between each entity and each preset sensitive word; selecting entities with semantic similarity greater than a preset threshold and determining them as successfully matched entities, and determining the successfully matched entities as candidate sensitive words.

[0024] Specifically, each word in the pre-defined sensitive word library and various entities identified from the text are transformed into high-dimensional semantic vectors using a natural language processing model. The cosine similarity between each entity vector and all pre-defined sensitive word vectors is calculated as an equidistant measure, thus obtaining a quantified semantic association score. An optimized pre-defined threshold is set; any entity whose semantic similarity to any pre-defined sensitive word exceeds this threshold, even if its literal form does not directly appear in the word library, will be captured and marked as a successfully matched candidate sensitive word. This step greatly enhances the system's ability to discover out-of-vocabulary words, variant expressions, and synonym substitutions. When sensitive information appears in the form of unregistered aliases, abbreviations, obscure references, or even partial spelling errors, it can still be accurately located through deep semantic association, significantly improving the system's recall and generalization capabilities. It can cope with flexible and varied information expressions, greatly reducing the risk of missed detections due to expression changes, and providing key technical support for building a more robust intelligent security barrier.

[0025] In the embodiments of this application, a method for intelligent management of sensitive words in power outage information is provided. The method for determining the multidimensional features of sensitive words includes: determining the semantic meaning of the sensitive words, the position of the sensitive words in the sentence, and the frequency of the sensitive words in the original power outage information text; and determining the semantic features, position features, and frequency features of the sensitive words based on the semantic meaning, position, and frequency, respectively; and constructing the multidimensional features of the sensitive words based on the semantic features, position features, and frequency features.

[0026] Specifically, the semantic meaning of sensitive words is determined by analyzing their context to identify their specific references, thus forming semantic features that characterize their essential attributes. The position of sensitive words in sentences is analyzed, such as whether they appear in the subject / object position stating a core fact or in a modifying adverbial clause, generating positional features reflecting their information-carrying weight. The frequency of sensitive words throughout the text is statistically analyzed; high frequency often indicates that the concept is one of the core focuses of the text description, thus forming frequency features representing their degree of emphasis. These three dimensions are integrated into a multi-dimensional feature vector describing the sensitive word. This step no longer relies solely on a binary judgment based on whether the word is on the sensitive list, but comprehensively considers three key dimensions: its meaning (semantics), its importance in the text (position), and how many times it is emphasized (frequency). This constructs a three-dimensional, quantitative risk profile, greatly improving the accuracy and rationality of risk classification and subsequent strategy matching. It endows the entire management method with a comprehensive judgment capability similar to a human reviewer, while ensuring the transparency and traceability of decision-making basis.

[0027] In embodiments of this application, a method for intelligent management of sensitive words in power outage information is provided. The method for evaluating the sensitivity of sensitive words based on multi-dimensional features to obtain a sensitivity evaluation value for the sensitive words includes: performing sensitivity analysis on the multi-dimensional features of the sensitive words, and evaluating the sensitivity of each multi-dimensional feature based on the analysis results to obtain a sensitivity evaluation value for each multi-dimensional feature; determining a preset sensitivity evaluation value for each multi-dimensional feature, and calculating the difference between the sensitivity evaluation value of each multi-dimensional feature and the corresponding preset sensitivity evaluation value to obtain a sensitivity difference value for each multi-dimensional feature; determining a preset weight for each multi-dimensional feature, and calculating the sensitivity evaluation value of the sensitive words based on the preset weight and the sensitivity difference value for each multi-dimensional feature.

[0028] Specifically, sensitivity analysis is performed on each multidimensional feature (such as semantics, location, and frequency). Based on its specific performance (such as the negativeness of semantics, the coreness of location, and the level of frequency), it is independently evaluated and assigned an initial sensitivity assessment value. An optimized preset sensitivity assessment value is introduced as a baseline, and the difference between the actual assessment value of each feature and the baseline value (sensitivity difference value) is calculated. This difference value directly reflects the degree to which the current feature performance deviates from the normal or expected risk level. According to the predefined preset weights, the difference values ​​of each feature are weighted and comprehensively calculated, and finally a single and accurate sensitivity assessment value is output. This step, by introducing a comparison with a preset benchmark value, allows the system to sensitively capture situations where features exhibit abnormal behavior. This enables the evaluation results to dynamically adapt to the unique context of the specific text, rather than mechanically applying fixed formulas. The weighted summation calculation method ensures that the final evaluation value can scientifically reflect the differentiated contribution of different features to the overall risk. Furthermore, the entire calculation process is fully data-driven and traceable, providing a solid and objective numerical basis for subsequent sensitivity level classification and management strategy matching. This transforms the entire system into a quantitative decision support system with clear decision-making logic, adjustable parameters, and auditable results.

[0029] In an embodiment of this application, a method for intelligent management of sensitive words in power outage information is provided, wherein the formula for calculating the sensitivity assessment value of the sensitive words is as follows: , Where P is the sensitivity assessment value of the sensitive word, αi is the preset weight of the i-th multidimensional feature, Di is the sensitivity difference value of the i-th multidimensional feature, and n is the number of multidimensional features.

[0030] In an embodiment of this application, a method for intelligent management of sensitive words in power outage information is provided. The step of determining the sensitivity level of a sensitive word based on a sensitivity assessment value includes: pre-setting a preset sensitivity level-sensitivity assessment value interval correspondence relationship, wherein each sensitivity assessment value interval is associated with a corresponding preset sensitivity level; determining the sensitivity assessment value of a sensitive word, and selecting the preset sensitivity level corresponding to the sensitivity assessment value interval as the sensitivity level of the sensitive word based on the mapping relationship between the sensitivity assessment value interval to which the sensitivity assessment value belongs and the preset sensitivity level-sensitivity assessment value interval correspondence relationship.

[0031] Specifically, the system transforms complex numerical calculations into simple and intuitive level labels, enabling subsequent strategy matching and manual review to quickly and consistently understand risk levels. The preset interval boundaries serve as flexible policy levers, allowing managers to dynamically adjust the score thresholds for each level based on different security situations, public opinion tolerance, or policy requirements. This allows for a global tightening or loosening of control without altering the underlying algorithm, giving the system strong policy adaptability and agile response capabilities. This grading method based on clearly defined numerical intervals makes the entire grading decision-making process completely transparent, auditable, and repeatable, thoroughly eliminating subjective arbitrariness and providing a clear and solid basis for standardized management and compliance review.

[0032] In embodiments of this application, a method for intelligent management of sensitive words in power outage information is provided. The intelligent management strategy for determining sensitive words based on sensitivity levels includes: determining a pre-set first preset threshold and a second preset threshold, and judging the relationship between the sensitivity level of the sensitive word and the first preset threshold and the second preset threshold; if the sensitivity level of the sensitive word is less than or equal to the first preset threshold, the intelligent management strategy for the sensitive word is to semantically fuzzily rewrite the sensitive word in the original power outage information text; if the sensitivity level of the sensitive word is greater than the first preset threshold and less than the second preset threshold, the intelligent management strategy for the sensitive word is to semantically standardize and replace the sensitive word in the original power outage information text; if the sensitivity level of the sensitive word is greater than or equal to the second preset threshold, the intelligent management strategy for the sensitive word is to cover or delete the sensitive word in the original power outage information text.

[0033] Specifically, two key strategic decision thresholds (a first preset threshold and a second preset threshold) are pre-set to divide the continuous sensitivity levels into three decision intervals with clear boundaries. The sensitivity level of each sensitive word is quickly and logically judged against these two thresholds: if its level is less than or equal to the first threshold, it is judged as low risk, triggering a semantic fuzzy rewriting strategy (e.g., changing "XX Precision Instrument Factory" to "a certain industrial user"); if its level is between the two thresholds, it is judged as medium risk, triggering a semantic standardization replacement strategy (e.g., replacing "deliberately damaged by external forces" with "failed due to external causes"); if its level is greater than or equal to the second threshold, it is judged as high risk, triggering the strictest masking or deletion strategy (e.g., directly concealing the name of a key facility). This step, through clear threshold logic, instantly transforms abstract risk levels into concrete, actionable instructions, achieving full automation and millisecond-level response in handling strategies, greatly improving management efficiency. The dual threshold setting constitutes a gradual "buffer zone," avoiding a crude dichotomy of either high or low risk. This allows for a balanced approach of "optimization rather than elimination" for large amounts of medium-risk information, thereby maximizing the preservation of information effectiveness and public service value while eliminating risk. By adjusting the values ​​of the two thresholds, managers can flexibly and globally control the overall level of management, much like adjusting a "valve," dynamically adapting to the security and communication needs of different periods and events. This gives the entire system both principled rigidity and execution-level flexibility. like Figure 2 As shown in the embodiments of this application, a power outage information sensitive word intelligent management system is provided, including: an acquisition module, used to acquire the original power outage information text to be published and preprocess the original power outage information text; an analysis module, used to perform text analysis on the preprocessed original power outage information text and identify sensitive words related to power outage in the original power outage information text; an evaluation module, used to determine the multi-dimensional features of the sensitive words and evaluate the sensitivity of the sensitive words based on the multi-dimensional features to obtain a sensitivity evaluation value of the sensitive words; a determination module, used to determine the sensitivity level of the sensitive words based on the sensitivity evaluation value and determine the intelligent management strategy of the sensitive words based on the sensitivity level; and a processing module, used to process the sensitive words in the original power outage information text based on the intelligent management strategy to obtain the processed power outage information text.

[0034] In summary, this invention provides an intelligent management method and system for sensitive words in power outage information, comprising: acquiring and preprocessing the original power outage information text to be published; performing text analysis on the text to identify sensitive words related to power outages; determining the multidimensional features of the sensitive words and evaluating their sensitivity to the sensitive words to obtain a sensitivity evaluation value; determining the sensitivity level of the sensitive words based on the sensitivity evaluation value and determining an intelligent management strategy for the sensitive words based on the sensitivity level; and processing the sensitive words in the original power outage information text based on the intelligent management strategy to obtain the processed power outage information text. This invention proactively discovers risks through multidimensional data mining, uses a quantitative model to transform subjective sensitivity into objective values, and matches adaptive processing strategies according to precise levels, thereby improving the efficiency and consistency of information processing, achieving refined and precise risk control, and ensuring information security.

[0035] Finally, it should be noted that those skilled in the art can obviously make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for intelligent management of sensitive words in power outage information, characterized in that, include: Obtain the original power outage information text to be published, and preprocess the original power outage information text; Text analysis was performed on the preprocessed original power outage information text to identify sensitive words related to power outages in the original power outage information text; The multidimensional features of sensitive words are determined, and the sensitivity of sensitive words is evaluated based on the multidimensional features to obtain the sensitivity evaluation value of sensitive words; Sensitivity levels of sensitive words are determined based on sensitivity assessment values, and intelligent management strategies for sensitive words are determined based on sensitivity levels. Based on intelligent management strategies, sensitive words in the original power outage information text are processed to obtain the processed power outage information text.

2. The intelligent management method for sensitive words in power outage information according to claim 1, characterized in that, The process of obtaining the original power outage information text to be published and preprocessing the original power outage information text includes: Obtain the original power outage information text to be published, and preprocess the original power outage information text, including text cleaning and standardization.

3. The intelligent management method for sensitive words in power outage information according to claim 2, characterized in that, The text analysis of the preprocessed original power outage information text identifies sensitive words related to power outages, including: Determine the boundaries of sentences in the preprocessed original power outage information text, and divide the preprocessed original power outage information text into multiple sentences based on the sentence boundaries; Semantic analysis is performed on each sentence to identify entities in each sentence, and the entities are compared and matched with a preset sensitive word lexicon. Entities that match successfully are identified as candidate sensitive words. The semantic meaning of candidate sensitive words is determined based on context analysis, and candidate sensitive words containing information related to power outages in their semantic meanings are identified and determined as sensitive words related to power outages in the original power outage information text.

4. The intelligent management method for sensitive words in power outage information according to claim 3, characterized in that, The step of comparing and matching entities with a preset sensitive word lexicon, and determining successfully matched entities as candidate sensitive words, includes: Identify each preset sensitive word in the preset sensitive word lexicon and calculate the semantic similarity between each entity and each preset sensitive word; Entities with semantic similarity greater than a preset threshold are selected as successfully matched entities and then identified as candidate sensitive words.

5. The intelligent management method for sensitive words in power outage information according to claim 3, characterized in that, The multidimensional features for determining sensitive words include: The semantic meaning of sensitive words, their position in sentences, and their frequency in the original power outage information text are determined. Based on the semantic meaning, position, and frequency, the semantic features, positional features, and frequency features of the sensitive words are determined respectively. Sensitive words are constructed based on semantic features, positional features, and frequency features.

6. The intelligent management method for sensitive words in power outage information according to claim 5, characterized in that, The method of evaluating the sensitivity of sensitive words based on multidimensional features to obtain a sensitivity evaluation value for sensitive words includes: Sensitivity analysis is performed on the multidimensional features of sensitive words, and the sensitivity of each multidimensional feature is evaluated based on the analysis results to obtain the sensitivity evaluation value of each multidimensional feature. Determine the preset sensitivity evaluation value for each multidimensional feature, and calculate the difference between the sensitivity evaluation value of each multidimensional feature and the corresponding preset sensitivity evaluation value to obtain the sensitivity difference value of each multidimensional feature. Determine the preset weights of each multidimensional feature, and calculate the sensitivity assessment value of sensitive words based on the preset weights and sensitivity difference values ​​of each multidimensional feature.

7. The intelligent management method for sensitive words in power outage information according to claim 6, characterized in that, The formula for calculating the sensitivity assessment value of the sensitive words is as follows: ， Where P is the sensitivity assessment value of the sensitive word, αi is the preset weight of the i-th multidimensional feature, Di is the sensitivity difference value of the i-th multidimensional feature, and n is the number of multidimensional features.

8. The intelligent management method for sensitive words in power outage information according to claim 6, characterized in that, The determination of the sensitivity level of sensitive words based on sensitivity assessment values ​​includes: A preset sensitivity level-sensitivity assessment value range correspondence is set in advance. For each sensitivity assessment value range, a corresponding preset sensitivity level is associated with it. Determine the sensitivity assessment value of the sensitive words, and based on the mapping relationship between the sensitivity assessment value interval to which the sensitivity assessment value belongs and the preset sensitivity level-sensitivity assessment value interval correspondence, select the preset sensitivity level corresponding to the sensitivity assessment value interval as the sensitivity level of the sensitive words.

9. The intelligent management method for sensitive words in power outage information according to claim 8, characterized in that, The intelligent management strategy for determining sensitive words based on sensitivity levels includes: Determine a first preset threshold and a second preset threshold, and determine the relationship between the sensitivity level of the sensitive word and the first preset threshold and the second preset threshold; If the sensitivity level of a sensitive word is less than or equal to the first preset threshold, the intelligent management strategy for that sensitive word is to semantically fuzzy rewrite the sensitive word in the original power outage information text. If the sensitivity level of a sensitive word is greater than the first preset threshold and less than the second preset threshold, the intelligent management strategy for the sensitive word is to replace the sensitive word in the original power outage information text with semantically standardized text. If the sensitivity level of a sensitive word is greater than or equal to the second preset threshold, the intelligent management strategy for that sensitive word is to cover or delete the sensitive word in the original power outage information text.

10. A smart management system for sensitive words related to power outage information, characterized in that, include: The acquisition module is used to acquire the original power outage information text to be published and to preprocess the original power outage information text. The analysis module is used to perform text analysis on the preprocessed original power outage information text and identify sensitive words related to power outages in the original power outage information text. The evaluation module is used to determine the multidimensional features of sensitive words and evaluate the sensitivity of sensitive words based on the multidimensional features to obtain the sensitivity evaluation value of sensitive words; The determination module is used to determine the sensitivity level of sensitive words based on the sensitivity assessment value, and to determine the intelligent management strategy for sensitive words based on the sensitivity level; The processing module is used to process sensitive words in the original power outage information text based on intelligent management strategies, and obtain the processed power outage information text.

Images