A global full-factor information processing method and system based on multi-source data fusion

Abstract

The application relates to the technical field of information processing, and discloses a global full-element information processing method and system based on multi-source data fusion. The method collects city operation information in parallel from multiple distributed data sources such as Internet of Things sensors, video monitoring equipment and social media platforms, and performs format normalization processing, effectively solving the problem of non-uniform data format faced by traditional systems when processing multi-source heterogeneous data. Through entity recognition and event element extraction on the normalized information, and preliminary association according to geographical location tags and time stamps, the application can effectively integrate information segments describing the same event, laying a foundation for subsequent conflict processing.

Description

Technical Field

[0001] This invention relates to the field of information processing technology, and more specifically, to a method and system for processing information across all domains and elements based on multi-source data fusion. Background Technology

[0002] In the field of urban management and planning, the deepening of the smart city concept has placed higher demands on the comprehensive perception of urban operational status and accurate decision-making. Traditional urban information processing systems perform well when processing structured, clearly sourced data, effectively responding to emergencies and allocating resources. However, when these systems need to integrate unconventional, unstructured, or semi-structured information from a wider range of channels, such as citizen reports, social media content, and third-party commercial data, they face significant challenges. While these new information sources provide unprecedented urban detail and potential early warning value, their information expression methods, semantic definitions, and inherent reliability differ significantly from traditional official data sources. This makes it difficult for the system to effectively identify and handle semantic inconsistencies and potential conflicts during information fusion, thereby affecting the accuracy of decision-making. Summary of the Invention

[0003] This invention provides a method and system for processing information across the entire domain and all elements based on multi-source data fusion. It aims to solve the problem that traditional urban information processing systems have difficulty effectively identifying and processing semantic inconsistencies and potential conflicts when integrating non-traditional, unstructured, or semi-structured information, thereby affecting the accuracy of decision-making.

[0004] The technical solution of this application is as follows: Firstly, this application discloses a method for processing full-domain, full-element information based on multi-source data fusion, specifically including: Urban operation information is collected in parallel from multiple distributed data sources, including IoT sensors, video surveillance equipment, and social media platforms. The collected urban operation information is normalized and uniformly converted into a key-value pair format containing geographic location tags and timestamps. Entity identification and event element extraction are performed on the normalized urban operation information. Information fragments describing the same event are initially associated based on geographic location tags and timestamps to form at least one set of information fragments. For each information fragment in each information fragment set, a weighted sum is calculated based on the information source type to which the information fragment belongs, the textual subjectivity score of the information fragment, and the historical accuracy of the information source, to calculate the reliability score of each information fragment; By comparing the description values ​​of the same event attribute from different information fragments in the same information fragment set, and calculating the degree of difference between the description values ​​according to the data type of the event attribute, the conflict quantification result is obtained. Based on the reliability score and conflict quantification results, a multi-stage conflict reconciliation is performed: when the conflict quantification result is greater than the first conflict threshold and the range of the reliability scores is greater than the score threshold, the information segment with the highest reliability score is selected as the reconciliation result; when the conflict quantification result is less than or equal to the first conflict threshold and greater than the second conflict threshold, or when the range of the reliability scores is less than or equal to the score threshold, auxiliary information is retrieved from external data sources for cross-validation, and a reconciled semantic description is generated based on the validation results. The harmonized semantic description is used as the unified semantic description of the event, and the geographic location label and timestamp are used as the joint primary key to write it into the city situation awareness database.

[0005] This technical solution can effectively integrate multi-source heterogeneous urban operation information, and resolve information conflicts and semantic inconsistencies through reliability assessment and multi-stage conflict reconciliation mechanisms, thereby improving the accuracy of urban situational awareness and the reliability of decision-making, and overcoming the shortcomings of existing technologies that cause misjudgments due to information conflicts.

[0006] Furthermore, this application also proposes to calculate the reliability score of each information segment, specifically including: Map the information source type to a preset type weight value; Perform sentiment analysis on the information fragments, calculate the proportion of subjective words in all words, and obtain a text subjectivity score; The historical accuracy rate is calculated by retrieving the accuracy statistics of information sources in past events. The type weight value, text subjectivity score, and historical accuracy are weighted and summed according to a preset weighting coefficient, and the sum is used as the reliability score.

[0007] This technical solution enables a more comprehensive and objective assessment of the reliability of information fragments by taking into account the type of information source, text subjectivity, and historical accuracy. This results in more accurate reliability assessments and provides a more reliable basis for subsequent conflict resolution.

[0008] Based on the above, this application further proposes to calculate the degree of difference between descriptive values, specifically including: When the event attribute is a numerical attribute, calculate the ratio of the difference between the maximum and minimum values ​​of the numerical attribute in each information fragment to the average value, and use the ratio as the degree of difference; When the event attribute is a categorical attribute, calculate the proportion of inconsistent samples with categorical attributes in each information fragment to the total number of samples, and use the proportion as the degree of difference; When the event attribute is a text attribute, calculate the semantic edit distance between the text descriptions in each information fragment, and use the maximum semantic edit distance as the difference.

[0009] This technical solution enables the use of targeted difference calculation methods based on the event attributes of different data types, making the conflict quantification results more accurate and reasonable, and effectively avoiding the limitations of a single quantification method in processing different types of data.

[0010] In some preferred implementations, auxiliary information is retrieved from external data sources for cross-validation, specifically including: Based on the geographic location labels in the information fragment set, obtain the physical environment data of the corresponding location through the geographic information system interface; Retrieve user-generated content corresponding to geotags from social media platforms using application programming interfaces; Using physical environment data and user-generated content as auxiliary information, the auxiliary information is spatiotemporally aligned with the information fragments in the information fragment set based on timestamps.

[0011] This technical solution allows for the introduction of external auxiliary information for cross-validation, effectively compensating for the shortcomings of a single information source and improving the accuracy and reliability of conflict reconciliation. It provides stronger decision support, especially when information conflicts are complex or reliability assessments are unclear. Furthermore, based on the validation results, a reconciled semantic description is generated, specifically including: The spatiotemporally aligned auxiliary information is compared one by one with each information fragment in the information fragment set, and the semantic matching degree between the auxiliary information and each information fragment is calculated. Select the information fragment with the highest semantic matching degree as the benchmark fragment; The system identifies portions of auxiliary information that have semantic differences from the baseline segment and whose semantic matching degree is higher than a preset cross-validation threshold. It then uses the identified portions to correct the corresponding content in the baseline segment, generating a harmonized semantic description.

[0012] This technical solution enables the use of auxiliary information to refine conflict information, ensuring that the reconciliation result not only resolves the conflict but also is semantically closer to the real situation, thereby generating a more accurate and persuasive unified semantic description.

[0013] Based on the above, after implementing multi-stage conflict reconciliation, the following is also included: Record the identifiers of the information fragments on which the reconciled semantic description is based, as well as the types of information sources used; The reconciled semantic descriptions, identifiers, and information source types are associated and stored in the reconciliation log database, which is used to track the conflict reconciliation results at different time points under the same geographic location label.

[0014] This technical solution enables the establishment of conflict reconciliation logs, providing valuable data support for subsequent system optimization, problem tracing, and decision analysis, and helping to continuously improve the system's performance and reliability.

[0015] Preferably, the harmonized semantic description is written into the urban situational awareness database, specifically including: Extract geolocation tags and timestamps from the harmonized semantic description; Using geographic location tags and timestamps as a combined primary key, query whether a corresponding record exists in the city situational awareness database; When a corresponding record is found, the corresponding record is overwritten and updated with the harmonized semantic description. When no corresponding record is found, the harmonized semantic description is inserted as a new record into the city situation awareness database.

[0016] This technical solution ensures that the information in the urban situational awareness database is always up-to-date and accurate. Through intelligent update and insertion mechanisms, it avoids data redundancy and inconsistency, thereby improving the database maintenance efficiency.

[0017] In some implementations, the method also includes periodically re-evaluating the set of information fragments: At the end of each preset evaluation period, acquire the newly added urban operation information during the evaluation period; After the newly added city operation information is formatted and normalized, it is re-associated with the information fragments in the existing information fragment set according to the geographic location label and timestamp to form an updated information fragment set; The reliability score calculation and conflict quantification are re-performed on the updated set of information fragments to obtain the updated reconciled semantic description; The updated harmonized semantic description is compared with the existing records in the urban situation awareness database. When the difference exceeds the preset update threshold, the corresponding record in the urban situation awareness database is updated.

[0018] This technical solution enables dynamic updates and continuous optimization of urban situation information, ensuring that the system can respond promptly to changes in the city's operational status and maintain the timeliness and accuracy of situational awareness.

[0019] As a technical improvement, the method also includes adaptive adjustments to the calculation parameters of the reliability score: Record the feedback deviation between the semantic description after each conflict resolution and the actual outcome of the event; When the accumulated feedback deviation exceeds the preset accumulated deviation threshold, the weighting coefficients used in the weighted summation are adjusted. The adjustment method is to increase the weighting coefficient corresponding to historical accuracy and decrease the weighting coefficient corresponding to text subjectivity scores.

[0020] This technical solution enables the reliability assessment model to have self-learning and adaptive capabilities, continuously optimizing parameters through a feedback mechanism, thereby improving the robustness and accuracy of the system in complex and ever-changing environments.

[0021] Secondly, this application also discloses a comprehensive information processing system based on multi-source data fusion, specifically including: The multi-source acquisition and normalization module is used to collect urban operation information in parallel from multiple distributed data sources, including IoT sensors, video surveillance equipment and social media platforms. The module performs format normalization processing on the collected urban operation information and converts it into a key-value pair format containing geographic location tags and timestamps. The information association module is used to perform entity recognition and event element extraction on the normalized urban operation information. It preliminarily associates information fragments describing the same event based on geographic location tags and timestamps to form at least one set of information fragments. The reliability assessment module is used to calculate the reliability score of each information fragment in each information fragment set by weighting and summing the information fragments according to the information source type to which the information fragment belongs, the subjective score of the information fragment text, and the historical accuracy of the information source. The conflict quantification module is used to compare the description values ​​of different information fragments in the same information fragment set for the same event attribute, calculate the degree of difference between the description values ​​according to the data type of the event attribute, and obtain the conflict quantification result. The multi-stage reconciliation module performs multi-stage conflict reconciliation based on the reliability score and conflict quantification result: when the conflict quantification result is greater than the first conflict threshold and the range of the reliability scores is greater than the score threshold, the information segment with the highest reliability score is selected as the reconciliation result; when the conflict quantification result is less than or equal to the first conflict threshold and greater than the second conflict threshold, or when the range of the reliability scores is less than or equal to the score threshold, auxiliary information is retrieved from external data sources for cross-validation, and a reconciled semantic description is generated based on the validation results. The database writing module is used to write the harmonized semantic description as a unified semantic description of the event, using geographic location tags and timestamps as a combined primary key, into the city situational awareness database.

[0022] This technical solution provides an integrated system solution. Through modular design, it enables efficient fusion, intelligent assessment, and conflict resolution of multi-source data, thereby building an intelligent platform that can accurately perceive urban conditions and support precise decision-making. Beneficial effects

[0023] This application discloses a method and system for processing comprehensive, multi-element information based on multi-source data fusion. By collecting urban operation information in parallel from multiple distributed data sources such as IoT sensors, video surveillance equipment, and social media platforms, and performing format normalization processing, it effectively solves the problem of inconsistent data formats faced by traditional systems when processing multi-source heterogeneous data. By performing entity recognition and event element extraction on the normalized information, and making preliminary associations based on geographic location tags and timestamps, this application can effectively integrate information fragments describing the same event, laying the foundation for subsequent conflict resolution. Attached Figure Description

[0024] Figure 1 This is a flowchart illustrating a method for processing full-domain, full-element information based on multi-source data fusion, provided in an embodiment of the present invention. Figure 2 This is a flowchart of a method for calculating the reliability score of each information segment provided in an embodiment of the present invention; Figure 3 This is a schematic diagram of the structure of a full-domain, full-element information processing system based on multi-source data fusion provided in an embodiment of the present invention. Detailed Implementation

[0025] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0026] Reference Figure 1 , Figure 1 This is a flowchart illustrating a method for processing full-domain, full-element information based on multi-source data fusion, provided by an embodiment of the present invention, including: S11, collect city operation information in parallel from multiple distributed data sources, including IoT sensors, video surveillance equipment and social media platforms, and perform format normalization processing on the collected city operation information, uniformly converting it into a key-value pair format containing geographic location tags and timestamps; S12, perform entity recognition and event element extraction on the normalized city operation information, and preliminarily associate information fragments describing the same event according to the geographic location label and the timestamp to form at least one set of information fragments; S13, For each information fragment in each set of information fragments, a weighted sum is calculated based on the information source type to which the information fragment belongs, the text subjectivity score of the information fragment, and the historical accuracy of the information source, to calculate the reliability score of each information fragment; S14, compare the description values ​​of different information fragments in the same set of information fragments for the same event attribute, calculate the degree of difference between the description values ​​according to the data type of the event attribute, and obtain the conflict quantification result; S15, based on the reliability score and the conflict quantification result, perform multi-stage conflict reconciliation: when the conflict quantification result is greater than the first conflict threshold and the range of the reliability scores is greater than the score threshold, select the information segment with the highest reliability score as the reconciliation result; when the conflict quantification result is less than or equal to the first conflict threshold and greater than the second conflict threshold, or when the range of the reliability scores is less than or equal to the score threshold, retrieve auxiliary information from an external data source for cross-validation, and generate a reconciled semantic description based on the validation results; S16, the harmonized semantic description is used as the unified semantic description of the event, and the geographic location tag and the timestamp are used as the joint primary key to write it into the city situation awareness database.

[0027] This application aims to effectively address the semantic inconsistencies and conflicts in the fusion of multi-source heterogeneous information by introducing multi-source data fusion, reliability assessment, conflict quantification, and a multi-stage conflict reconciliation mechanism, thereby improving the accuracy of urban situational awareness and the scientific nature of decision-making.

[0028] To better understand the method proposed in this application, some key terms are explained first. Urban operation information refers to data reflecting the functional status and activities of a city, and its sources are wide-ranging, including but not limited to IoT sensors, video surveillance equipment, and social media platforms. IoT sensors can provide structured data such as environmental parameters (e.g., temperature, humidity, air quality), traffic flow, and equipment status. Video surveillance equipment provides images and video streams, and through visual analysis, information such as pedestrian density and abnormal behavior can be extracted. Social media platforms contain a large amount of user-generated content, such as text, images, and videos, reflecting public sentiment, event discussions, and emergencies. Geographic location tags are used to identify the specific spatial location where information occurs; these can be latitude and longitude coordinates, administrative division codes, or specific location names. Timestamps are used to record the specific time when information is generated or collected, ensuring the chronological order of events. Key-value pair format is a simple data representation method where each data item consists of a key and a value, facilitating data storage and retrieval. An information fragment set refers to a logical grouping formed after initially associating information describing the same event from different data sources. A reliability score is a quantitative indicator measuring the credibility of information fragments. Conflict quantification results indicate the degree of difference between the descriptive values ​​of the same event attribute from different information fragments. Multi-stage conflict reconciliation is a hierarchical conflict handling strategy, aiming to adopt different reconciliation mechanisms based on the severity of the conflict and the reliability of the information. Unified semantic description refers to a standardized and unambiguous description of an event after reconciliation. The urban situational awareness database is a centralized database used to store and manage unified semantic descriptions of the city's operational status.

[0029] The method proposed in this application first collects urban operation information in parallel from multiple distributed data sources. For example, a data collection agent can be deployed that can simultaneously connect to IoT sensor networks, public safety video surveillance systems, and open APIs of mainstream social media platforms throughout the city. These agents can be configured to periodically pull data or receive data streams in real time when specific events are triggered. The raw data collected may vary in format; for example, IoT sensor data may be numerical values ​​in JSON format, video surveillance data may be H.264 encoded video streams, and social media data may be XML or HTML content containing text, images, and links. To ensure consistency in subsequent processing, this collected urban operation information needs to be format-normalized.

[0030] One approach is to develop a series of data parsers and converters to parse the raw format of each data source and uniformly convert it into a key-value pair format containing geographic location tags and timestamps. For example, temperature data reported by a sensor can be converted to `{"location": "Park A", "timestamp": "2023-10-27T10:00:00Z", "type": "temperature", "value": "25°C"}`. A video analysis result can be converted to `{"location": "Park A", "timestamp": "2023-10-27T10:05:00Z", "type": "event", "description": "crowd gathering", "confidence": "0.8"}`. A social media post can be converted to `{"location": "Park A", "timestamp": "2023-10-27T10:10:00Z", "type": "social_post", "content": "The park is bustling with activity, there's a concert!"}`.

[0031] Next, entity recognition and event element extraction are performed on the normalized urban operation information. This can be achieved through Natural Language Processing (NLP) techniques and machine learning models. For example, Named Entity Recognition (NER) models can be used to identify entities such as names of people, places, organizations, and times from text descriptions. Simultaneously, event extraction models can identify event types (such as "gathering," "conflict," and "performance") and their participants, time, and location. After extracting entity and event elements, information fragments describing the same event are initially associated based on geographic location tags and timestamps, forming at least one set of information fragments. For example, if multiple information fragments all point to a "crowd gathering" event that occurred at "Park A" between "2023-10-27T10:00:00Z" and "2023-10-27T10:30:00Z," these information fragments will be grouped into the same set. This initial association can be achieved by setting a spatiotemporal proximity threshold, meaning that information occurring within a specific geographic area and time window is considered relevant.

[0032] Subsequently, a reliability score is calculated for each information fragment in each set of information fragments. This requires considering the type of information source to which the information fragment belongs, the textual subjectivity score of the information fragment, and the historical accuracy of the information source. For example, weights can be preset for different information source types, such as official sensor data having a higher weight than social media user reports. The textual subjectivity score can be calculated using a sentiment analysis model; for example, a text containing many subjective adjectives (such as "horrible" or "wonderful") will have a higher subjectivity score. The historical accuracy of the information source can be obtained by tracking and evaluating the performance of each data source in past events over a long period. For example, if a social media account frequently posts false information, its historical accuracy will be lower. By weighting and summing these factors according to preset weighting coefficients, the reliability score for each information fragment can be obtained.

[0033] While calculating the reliability score, it's also necessary to compare the descriptive values ​​of the same event attribute from different information fragments within the same set of information fragments to obtain the conflict quantification result. For example, if one information fragment describes "Number of people: 100" and another describes "Number of people: 500," then there is a conflict between these two descriptive values. The calculation method for the conflict quantification result depends on the data type of the event attribute. For numerical attributes (such as number of people, temperature), the ratio of the difference between the maximum and minimum values ​​to the average value can be calculated as the degree of discrepancy. For categorical attributes (such as event type: conflict / performance), the proportion of inconsistent samples to the total number of samples can be calculated. For textual attributes (such as event description), the semantic edit distance between text descriptions can be calculated.

[0034] Finally, based on the reliability score and conflict quantification results, a multi-stage conflict reconciliation is performed. In the first stage, when the conflict quantification result is greater than the first conflict threshold (indicating severe conflict) and the range of reliability scores is greater than the score threshold (indicating significant differences in the reliability of information sources), the information fragment with the highest reliability score is selected as the reconciliation result. For example, if sensor data shows a temperature of 25°C, while a low-reliability social media post claims a temperature of 35°C, and the sensor data's reliability is much higher than the social media post's, then the sensor data is selected as the reconciliation result. In the second stage, when the conflict quantification result is less than or equal to the first conflict threshold but greater than the second conflict threshold (indicating moderate conflict), or when the range of reliability scores is less than or equal to the score threshold (indicating similar information source reliability), auxiliary information is retrieved from external data sources for cross-validation. For example, if two information sources with similar reliability describe the nature of an event inconsistently (one says "conflict," the other says "performance"), physical environment data of the geographical location (such as whether there is a stage setup or lighting equipment) or user-generated content of the area (such as other social media posts, images, and videos) can be retrieved for cross-validation. Based on the validation results, a reconciled semantic description is generated. For example, if the supporting information shows a stage and lighting, the event is likely to be reconciled as a "performance".

[0035] The harmonized semantic description is used as the unified semantic description of the event, with the geographic location tag and timestamp as the joint primary key, and written into the city situational awareness database. For example, after harmonization, the final unified semantic description might be "Park A held a concert from 2023-10-27T10:00:00Z to 2023-10:30:00Z, with crowd gathering and no conflict." This unified semantic description will be stored in the database for subsequent use by the city situational awareness and decision support system.

[0036] The method proposed in this application effectively solves the semantic inconsistency and conflict problems faced by traditional urban information processing systems when processing multi-source heterogeneous information by introducing a series of technical means such as parallel acquisition and normalization of multi-source data, entity recognition and event element extraction, reliability assessment, conflict quantification, and multi-stage conflict reconciliation.

[0037] refer to Figure 2 , Figure 2 This is a flowchart of a method for calculating the reliability score of each information segment according to an embodiment of the present invention, S13: S131, map the information source type to a preset type weight value; S132, Perform text sentiment analysis on the information fragment, calculate the proportion of subjective words to all words, and obtain a text subjectivity score; S133, query the historical accuracy statistics of information sources in past events, as the historical accuracy; S134: The type weight value, text subjectivity score and historical accuracy are weighted and summed according to the preset weighting coefficient, and the sum is used as the reliability score.

[0038] Mapping information source types to preset type weight values ​​involves assigning a numerical weight to each source based on its inherent characteristics and credibility. For example, data from official IoT sensors is generally considered more credible than user-generated content on social media platforms and can therefore be assigned a higher type weight. This mapping can be pre-configured in the system and adjusted based on actual application scenarios and experience to reflect the relative reliability of different information sources in specific contexts.

[0039] Furthermore, sentiment analysis is performed on the information fragment to calculate the proportion of subjective words among all words, resulting in a text subjectivity score. Sentiment analysis technology can be used to identify the emotional tendency and subjective expressions contained in information fragments. A higher proportion of subjective words generally indicates lower objectivity in the information fragment, and its reliability may be correspondingly reduced. This score aims to quantify the objectivity of the information fragment, thus serving as an important dimension for assessing its reliability.

[0040] Furthermore, the accuracy statistics of information sources in past events are retrieved from the query history and used as historical accuracy. This means the system maintains a database of the historical performance of each information source. The more times an information source has been verified as accurate in past reported events, the higher its historical accuracy. This historical accuracy reflects the long-term reliability of the information source and provides empirical evidence for assessing the reliability of current information fragments.

[0041] Finally, the type weight value, text subjectivity score, and historical accuracy are weighted and summed using preset weighting coefficients, and the sum is used as the reliability score. This weighted summation method comprehensively considers the inherent credibility of the information source, the objectivity of the information content, and the historical performance of the information source, thus obtaining a comprehensive and quantitative reliability score. The preset weighting coefficients can be adjusted according to actual needs and the importance of different factors to ensure that the calculated reliability score accurately reflects the true reliability of the information fragment.

[0042] This application's solution quantifies the reliability of each information fragment by comprehensively considering three key dimensions: information source type, text subjectivity, and historical accuracy. Through this technical solution, the application overcomes the limitations of a single evaluation dimension, providing a more refined and comprehensive mechanism for assessing the reliability of information fragments. This multi-factor weighted summation calculation method enables the system to more accurately determine the authenticity and credibility of each information fragment when faced with complex, ever-changing, and often unreliable urban operational information. This provides a solid foundation for subsequent multi-stage conflict resolution, significantly improving the effectiveness of conflict resolution and the accuracy and reliability of the unified semantic description in the final urban situational awareness database, thereby enhancing the overall performance of comprehensive, all-element information processing.

[0043] Specifically, the calculation of the difference between the described values ​​includes the following steps: When the event attribute is a numerical attribute, such as temperature, humidity, or traffic flow, the dissimilarity is calculated as follows: the ratio of the difference between the maximum and minimum values ​​of the numerical attribute in each information segment to the average value is used as the dissimilarity. This method can effectively measure the degree of dispersion of numerical data across different information sources.

[0044] When the event attribute is a categorical attribute, such as event type (fire, traffic accident) or equipment status (normal, malfunction), the dissimilarity is calculated as follows: the proportion of inconsistent samples with the categorical attribute in each information segment to the total number of samples is calculated, and this proportion is used as the dissimilarity. This method is suitable for evaluating the consistency of discrete categorical data.

[0045] When the event attribute is a text-based attribute, such as an event description or user comments, the difference is calculated as follows: the semantic edit distance between the text descriptions in each information fragment is calculated, and the maximum semantic edit distance is used as the difference. Semantic edit distance reflects the degree of similarity or difference between text content at the semantic level, thus quantifying the conflict in text descriptions.

[0046] This application's solution employs a customized difference calculation method based on different data types (numerical, categorical, and textual) of event attributes, enabling more precise quantification of the degree of conflict between different information fragments describing the same event attribute. Through this technical solution, the application can use the most suitable quantification method to calculate the difference between descriptive values ​​for different types of event attributes, thereby significantly improving the accuracy and precision of conflict quantification results. This refined difference calculation avoids errors that may result from a "one-size-fits-all" quantification approach, allowing subsequent multi-stage conflict reconciliation processes to make decisions based on more reliable conflict information. Specifically, for numerical data, it can more sensitively capture actual numerical fluctuations; for categorical data, it can clearly identify disagreements in classification judgments; and for textual data, it can delve into the semantic level, more accurately assessing conflicts in textual descriptions, thus providing a more realistic and reliable unified semantic description of events for the urban situational awareness database.

[0047] Specifically, in the aforementioned multi-stage conflict reconciliation process, when it is necessary to retrieve auxiliary information from external data sources for cross-validation, this application proposes a specific method for retrieving auxiliary information for cross-validation.

[0048] The steps described above for retrieving auxiliary information from external data sources for cross-validation include: Based on the geographic location labels in the information fragment set, obtain the physical environment data of the corresponding location through the geographic information system interface; Retrieve user-generated content corresponding to geotags from social media platforms using application programming interfaces; Using physical environment data and user-generated content as auxiliary information, the auxiliary information is spatiotemporally aligned with the information fragments in the information fragment set based on timestamps.

[0049] Geographic Information System (GIS) interfaces can be understood as a standard way for systems to access and process geospatial data. For example, they can obtain environmental information such as weather, terrain, and traffic conditions for specific geographic coordinates by calling map service APIs. The purpose is to provide objective physical environmental background information for the location of an event.

[0050] Application Programming Interfaces (APIs) are interfaces provided by social media platforms to external developers for accessing their data and functions. For example, calling the APIs of platforms like Weibo and WeChat allows developers to access text, images, videos, and other content posted by users within a specific region. Their purpose is to obtain subjective or objective information related to events and generated by the public.

[0051] Spatiotemporal alignment refers to the process of matching and associating data from different sources based on their geographic location and time information. Specifically, auxiliary information (physical environment data and user-generated content) is precisely matched with existing information fragments in the information fragment set based on its recorded geographic location tags and timestamps. This ensures that all relevant information points to the same time period and the same geographic region, facilitating effective cross-validation. The purpose is to ensure the validity and relevance of the auxiliary information, providing accurate references for subsequent conflict resolution.

[0052] This application's solution utilizes physical environment data and user-generated content corresponding to the location and time of the event as supplementary information, providing multi-dimensional and multi-perspective external references for conflict reconciliation. The physical environment data acquired through the Geographic Information System (GIS) interface provides objective, contextual information, helping to verify the authenticity of the event description. User-generated content acquired from social media platforms provides subjective or objective perspectives from on-site or relevant personnel, helping to supplement and corroborate event details. By aligning this supplementary information with existing information fragments in time and space, the temporal and spatial relevance of all data is ensured, thus laying a solid foundation for subsequent cross-validation and semantic description generation, improving the accuracy and reliability of the reconciliation results.

[0053] Through the aforementioned technical solution, this application can effectively acquire auxiliary information highly relevant to the event to be reconciled from external data sources and perform precise spatiotemporal alignment. This not only enriches the information dimensions in the conflict reconciliation process but also provides an effective cross-validation method when the reliability score difference of information fragments is small or the conflict quantification results are at a moderate level. Therefore, it avoids the limitations of subjective judgment based solely on internal information, significantly improving the objectivity, accuracy, and credibility of the conflict reconciliation results, making the event descriptions in the urban situational awareness database closer to the real situation.

[0054] The above-mentioned reconciled semantic description generated based on the verification results includes: The spatiotemporally aligned auxiliary information is compared one by one with each information fragment in the information fragment set, and the semantic matching degree between the auxiliary information and each information fragment is calculated. Select the information fragment with the highest semantic matching degree as the benchmark fragment; Identify the portion of the auxiliary information that has semantic differences from the baseline segment and whose semantic matching degree is higher than a preset cross-validation threshold, and use the identified portion to correct the corresponding content in the baseline segment to generate the harmonized semantic description.

[0055] Specifically, the spatiotemporally aligned auxiliary information is compared one by one with each information fragment in the information fragment set. The purpose is to quantify the semantic relevance between the auxiliary information and each original information fragment. Semantic matching degree can be understood as measuring the degree of similarity between two texts or information fragments at the semantic level. This can be calculated using natural language processing techniques, such as word vector models (e.g., Word2Vec, GloVe), sentence vector models (e.g., Sentence-BERT), or deep learning-based semantic similarity calculation methods. For example, the cosine similarity between the auxiliary information and the text embedding vector of each information fragment can be calculated, and this similarity value can be used as the semantic matching degree.

[0056] Furthermore, after calculating the semantic matching degree between all information fragments and auxiliary information, the information fragment with the highest semantic matching degree is selected as the benchmark fragment. This benchmark fragment is considered to be the original information description that is most consistent with the auxiliary information, providing a reliable starting point for subsequent corrections.

[0057] Based on this, the auxiliary information is identified as having semantic differences from the baseline segment and a semantic matching degree higher than a preset cross-validation threshold. Semantic differences refer to inconsistencies or supplementary information between the auxiliary information and the baseline segment in describing the same event attribute. The preset cross-validation threshold is an empirical value used to filter out valid information that has high consistency with the auxiliary information but also exhibits differences. For example, by comparing the keywords, entities, or key descriptive sentences of the auxiliary information with the baseline segment, parts that are semantically inconsistent or where the auxiliary information provides a more detailed and accurate description can be identified.

[0058] Finally, the identified corrections to the corresponding content in the baseline fragment are used to generate a harmonized semantic description. The correction process may include replacing inaccurate or missing information in the baseline fragment, supplementing important details not mentioned in the baseline fragment but present in the auxiliary information, or clarifying ambiguous descriptions in the baseline fragment. For example, if the baseline fragment describes "road congestion," while auxiliary information (such as real-time traffic data) shows "the road is severely congested due to a traffic accident, expected to last 2 hours," the baseline fragment can be corrected to "the road is severely congested due to a traffic accident, expected to last 2 hours."

[0059] This application's solution effectively addresses the issue of insufficient accuracy in reconciliation results when relying solely on external auxiliary information for cross-validation by introducing refined semantic comparison, benchmark selection, and difference correction mechanisms. Through this technical solution, the application overcomes the inaccuracies and low information utilization efficiency that may exist in traditional methods when generating semantic descriptions based on external validation results. Specifically, by introducing steps such as semantic matching degree calculation, benchmark fragment selection, and difference identification and correction, it ensures that auxiliary information can be systematically and effectively used to optimize the original information fragments. As a result, the generated reconciled semantic description has higher accuracy, consistency, and reliability, more realistically reflecting the situation of urban operational events and providing city managers with more accurate decision-making basis. Furthermore, this structured correction process reduces the need for manual intervention and improves the automation level and efficiency of information processing.

[0060] This application further proposes that, following the implementation of the aforementioned multi-stage conflict reconciliation, the following additional steps are included: Record the identifiers of the information fragments on which the harmonized semantic description is based, as well as the types of information sources used; The reconciled semantic description, the identifier, and the information source type are associated and stored in a reconciliation log database, which is used to track the conflict reconciliation results at different time points under the same geographic location label.

[0061] Specifically, the phrase "recording the identifiers of the information fragments upon which the reconciled semantic description is based and the types of information sources used" means that after the multi-stage conflict reconciliation process is completed, the system identifies and records the unique identifiers of the original information fragments that are ultimately selected or used to generate the reconciliation result. Simultaneously, it also records the specific information source types from which these original information fragments originate, such as IoT sensors, video surveillance equipment, or social media platforms. These identifiers and information source types are key metadata for understanding the formation process of the reconciliation result. The phrase "associating and storing the reconciled semantic description, the identifiers, and the information source types in the reconciliation log database" can be understood as logically binding the finally generated reconciled semantic description with the recorded original information fragment identifiers and information source types, and persistently storing them together in a dedicated database, namely the reconciliation log database. This associated storage ensures the integrity and traceability between the reconciliation result and its generation basis.

[0062] In practical applications, the phrase "the reconciliation log database is used to track conflict reconciliation results at different time points under the same geographic location label" means that this database not only stores the results of a single reconciliation and its metadata, but more importantly, it can record detailed information about different conflict reconciliation events over time at a specific geographic location. For example, for traffic congestion events occurring at the same intersection at different times, the reconciliation log database can record the reconciliation results, basis, and source of each event, thus forming a historical record that facilitates subsequent analysis and auditing.

[0063] This application's solution effectively addresses the issues of insufficient transparency and poor traceability of reconciliation results that may exist in the basic solution by adding the recording and storage of key information from the reconciliation process after multi-stage conflict resolution. Through the above technical solution, this application significantly improves the transparency and traceability of conflict resolution results in the urban operation information processing process. Specifically, by recording and associating the source information of reconciliation results, the system can provide a clear audit path, allowing managers and analysts to review the formation process of any reconciliation result at any time, thereby enhancing trust in the system's output. Furthermore, the reconciliation log database's ability to track conflict resolution results at different time points under the same geographic location tag provides historical data support for the long-term evolution analysis of urban situations, helping to identify potential data source problems, optimize reconciliation strategies, and support more precise urban management decisions. This mechanism effectively compensates for the deficiencies of the basic solution in terms of result interpretability and process controllability, making the entire information processing system more robust and reliable.

[0064] In some embodiments of this application, the harmonized semantic description is used as a unified semantic description of the event, and written into the city situation awareness database using geographic location tags and timestamps as a joint primary key. Specifically, the steps of writing the harmonized semantic description into the city situation awareness database include: Extract the geolocation tags and timestamps from the harmonized semantic description; Using the geographic location tag and the timestamp as a joint primary key, query whether a corresponding record exists in the city situation awareness database; When the corresponding record is found, the corresponding record is overwritten and updated with the harmonized semantic description. When no corresponding record is found, the harmonized semantic description is inserted as a new record into the city situation awareness database.

[0065] Specifically, before writing the harmonized semantic description into the city situational awareness database, it is first necessary to accurately extract the geographic location tags and timestamps contained within it. These tags and timestamps are unique identifiers of the location and time of the event, crucial for subsequent data management and querying. Furthermore, the extracted geographic location tags and timestamps are combined to form a composite primary key. Using this composite primary key, the system will query the city situational awareness database to determine if a record corresponding to this composite primary key already exists. This query aims to avoid data redundancy and ensure data consistency.

[0066] When the query result indicates that a record corresponding to the composite primary key already exists in the database, it means that the event already has a preliminary or older record in the database. In this case, to maintain the up-to-dateness and accuracy of the data, the newly generated reconciled semantic description will be used to overwrite and update the existing record in the database. This overwrite and update mechanism ensures that the database always stores the most reliable event semantic descriptions that have undergone the latest conflict reconciliation. Conversely, when the query result shows that a record corresponding to the composite primary key does not exist in the database, it indicates that this is a completely new event or an event that has been identified and reconciled by the system for the first time. In this case, the reconciled semantic description will be inserted as a completely new record into the city situational awareness database. This ensures that all processed event information can be effectively stored and managed.

[0067] This application's solution achieves fine-grained control over data writing operations by first extracting geographic location tags and timestamps as a composite primary key before writing to the urban situation perception database, and then performing database queries based on this primary key. This mechanism can intelligently determine whether the event information to be written is entirely new data or an update to existing data. By distinguishing between these two cases and performing the operation of inserting a new record or overwriting an existing record accordingly, the uniqueness, real-time nature, and accuracy of the data in the database are ensured. This avoids data redundancy and inconsistency problems that may result from simple repeated insertions, while ensuring that the perception of urban situation is based on the latest and most reliable information.

[0068] Through the above technical solution, this application can effectively manage event information in the urban situation awareness database. Specifically, by introducing a joint primary key query mechanism based on geographic location tags and timestamps, it can avoid repeatedly storing different descriptions of the same event, thereby significantly reducing data redundancy in the database. Furthermore, when event information changes and is readjusted, the corresponding records in the database can be updated promptly and accurately, ensuring the real-time nature and consistency of data in the urban situation awareness database. This intelligent writing strategy not only improves the efficiency of data management but also provides a more reliable and accurate data foundation for subsequent situation analysis and decision-making.

[0069] In some of the embodiments described above in this application, a method for processing comprehensive, all-element information based on multi-source data fusion is proposed. This method can perform reliability assessment and conflict reconciliation of collected urban operation information and write it into the urban situation awareness database. However, urban operation information is dynamically changing; for example, traffic conditions, environmental indicators, or the progress of emergencies are constantly updated over time. If the processed and stored information in the database is not periodically updated and reassessed, the information in the database may deviate from the actual urban situation, thereby affecting the real-time performance, accuracy, and effectiveness of decision support for urban situation awareness.

[0070] In response, this application further proposes a method for periodically re-evaluating the aforementioned set of information fragments, specifically including: At the end of each preset evaluation period, acquire the newly added urban operation information within the evaluation period; After the newly added city operation information is formatted and normalized, it is re-associated with the information fragments in the existing information fragment set according to the geographic location tag and the timestamp to form an updated information fragment set; The reliability score calculation and conflict quantification are re-performed on the updated set of information fragments to obtain an updated reconciled semantic description; The updated harmonic semantic description is compared with the existing records in the urban situation awareness database. When the difference exceeds a preset update threshold, the corresponding record in the urban situation awareness database is updated.

[0071] Specifically, "periodic reassessment" refers to the system reassessing and updating processed and stored information at predetermined time intervals (e.g., hourly, daily, or weekly). This mechanism aims to ensure that the data in the urban situational awareness database remains up-to-date and highly accurate. The "preset assessment cycle" can be flexibly configured according to the needs of actual application scenarios. For example, for information with high real-time requirements, such as traffic flow, the assessment cycle can be set to a shorter interval, while for information with relatively slow changes, such as environmental quality, the assessment cycle can be appropriately extended.

[0072] At the end of each evaluation period, the system automatically acquires newly generated city operation information during that period. This new information may come from existing distributed data sources such as IoT sensors, video surveillance equipment, and social media platforms, or from other newly connected data sources. The acquired new city operation information will first undergo format normalization processing to ensure that it is uniformly converted into a key-value pair format containing geographic location tags and timestamps, so as to facilitate subsequent processing and correlation.

[0073] Subsequently, these normalized new information pieces are re-associated with existing information pieces in the current system. The re-association is based on geographic location tags and timestamps. That is, if the new information and existing information describe events that occurred at the same geographic location and at similar times, they will be associated to form an "updated set of information pieces" that includes both new and old information.

[0074] For this updated set of information fragments, the system will again perform reliability score calculation and conflict quantification. This means that even previously processed information will have its reliability and conflict with other information reassessed after new data is added. Through this process, an "updated and reconciled semantic description" reflecting the latest situation can be obtained.

[0075] Finally, the system compares this updated and harmonized semantic description with the corresponding existing record in the city situation awareness database. The purpose of the comparison is to determine whether there are significant differences between the old and new information. If the comparison results show that the difference exceeds a preset "update threshold" (for example, significant changes in event type, severity, key parameters, etc.), it will trigger an update operation on the corresponding record in the city situation awareness database, overwriting the old record with the new semantic description.

[0076] This application's solution effectively addresses the problem of outdated or inaccurate database information caused by dynamic changes in urban operational information by introducing a periodic reassessment mechanism. Through this technical solution, the application enables dynamic maintenance and updating of information in the urban situational awareness database, significantly improving the real-time performance and accuracy of urban situational awareness. This periodic reassessment mechanism ensures that even in an environment of constantly changing information, the system can continuously acquire, process, and integrate the latest data, thus ensuring that the event descriptions in the database remain highly consistent with the actual situation. Furthermore, by setting update thresholds, frequent and unnecessary database updates are avoided, optimizing the efficiency of system resource utilization while ensuring information accuracy. This mechanism enables urban managers to make decisions based on the latest and most reliable information, thereby improving the intelligence level of urban governance and emergency response capabilities.

[0077] In some embodiments described above, the reliability score of an information fragment is obtained by weighting and summing the information source type, text subjectivity score, and historical accuracy of the information source using preset weighting coefficients. However, these preset weighting coefficients may not fully adapt to the dynamic changes in urban operational information at different times and in different scenarios, affecting the accuracy of the reliability score calculation and potentially impacting the accuracy of conflict reconciliation. If these problems are not addressed, the system may experience a decrease in its ability to perceive complex and ever-changing urban events over long-term operation due to parameter fixation. Therefore, this application proposes a method for adaptively adjusting the calculation parameters of the reliability score. This method aims to improve the accuracy of the reliability score calculation and the robustness of the system by introducing a feedback mechanism to dynamically optimize the weighting coefficients.

[0078] The above method also includes adaptive adjustment of the calculation parameters for the reliability score. Specifically, this adjustment process includes: Record the feedback deviation between the reconciled semantic description and the actual event outcome after each conflict resolution. The feedback deviation can be understood as the degree of difference between the unified semantic description output by the system and the actual event truth or authoritative verification results. For example, the actual event outcome can be obtained through manual verification, comparison with authoritative data sources, or verification through subsequent event developments, and then compared with the reconciled semantic description generated by the system to quantify the inconsistency.

[0079] When the accumulated feedback deviation exceeds a preset deviation accumulation threshold, the system will automatically adjust the weighting coefficients used in the weighted summation. The deviation accumulation threshold is a preset value used to determine whether there is a significant decline in system performance, thereby triggering the parameter adjustment mechanism.

[0080] The adjustment method is to increase the weighting coefficient corresponding to historical accuracy and decrease the weighting coefficient corresponding to textual subjectivity scores. This means that when adjusting parameters, the system will place more emphasis on the objective historical performance of the information source and relatively reduce its reliance on subjective descriptions of information fragments.

[0081] This application's solution, by introducing a mechanism for recording and accumulating feedback deviations, enables real-time monitoring of the system's performance in conflict reconciliation. When a persistent or significant deviation exists between the system's output of the reconciled semantic description and the actual event outcome, it indicates that the current reliability score calculation parameters may no longer be optimal. By comparing the accumulated feedback deviations with a preset deviation accumulation threshold, the system can intelligently determine when parameter adjustments are needed.

[0082] Furthermore, when parameter adjustments are triggered, increasing the weighting coefficient corresponding to historical accuracy allows information sources that have demonstrated high accuracy in past events to receive greater weight in subsequent reliability score calculations, thereby enhancing the system's adoption of objective and credible information. Simultaneously, decreasing the weighting coefficient corresponding to text subjectivity scores helps reduce the impact of subjective descriptions in information fragments on reliability scores, avoiding misjudgments due to excessive subjectivity. This adjustment is particularly effective in improving the objectivity and accuracy of reliability assessments when dealing with highly subjective data sources such as social media. It is precisely this adaptive parameter adjustment mechanism that enables the system to self-optimize based on actual operational performance, thereby continuously improving its processing accuracy and decision support capabilities for urban operational information.

[0083] Through the above technical solution, this application enables adaptive adjustment of reliability score calculation parameters, effectively solving the problem of decreased accuracy that may occur when traditional fixed weighting coefficients are used to deal with dynamically changing urban operation information. By introducing a feedback mechanism, the system can learn and optimize itself based on actual conflict reconciliation effects, thereby significantly improving the accuracy and robustness of reliability score calculation. Consequently, the system can generate a more accurate and reliable unified semantic description, further enhancing the accuracy of urban situational awareness and the effectiveness of decision support, making the entire information processing method more adaptable and intelligent.

[0084] refer to Figure 3 , Figure 3 This is a schematic diagram of the structure of a full-domain, full-element information processing system based on multi-source data fusion provided in an embodiment of the present invention, including: The multi-source acquisition and normalization module is used to collect urban operation information in parallel from multiple distributed data sources, including IoT sensors, video surveillance equipment and social media platforms. The module performs format normalization processing on the collected urban operation information and converts it into a key-value pair format containing geographic location tags and timestamps. The information association module is used to perform entity recognition and event element extraction on the normalized city operation information, and to initially associate information fragments describing the same event based on the geographic location tag and the timestamp to form at least one set of information fragments. The reliability assessment module is used to calculate the reliability score of each information fragment in each set of information fragments by performing a weighted sum based on the information source type to which the information fragment belongs, the text subjectivity score of the information fragment, and the historical accuracy of the information source. The conflict quantification module is used to compare the description values ​​of different information fragments in the same set of information fragments for the same event attribute, calculate the degree of difference between the description values ​​according to the data type of the event attribute, and obtain the conflict quantification result. A multi-stage reconciliation module is used to perform multi-stage conflict reconciliation based on the reliability score and the conflict quantification result: when the conflict quantification result is greater than a first conflict threshold and the range of the reliability scores is greater than a score threshold, the information segment with the highest reliability score is selected as the reconciliation result; when the conflict quantification result is less than or equal to the first conflict threshold and greater than a second conflict threshold, or when the range of the reliability scores is less than or equal to the score threshold, auxiliary information is retrieved from an external data source for cross-validation, and a reconciled semantic description is generated based on the validation results; The database writing module is used to write the harmonized semantic description as the unified semantic description of the event, using the geographic location tag and the timestamp as a joint primary key, into the city situation awareness database.

[0085] The system proposed in this application integrates multi-source acquisition and normalization modules, information association modules, reliability assessment modules, conflict quantification modules, multi-stage reconciliation modules, and database writing modules, forming a complete and intelligent information processing flow. This system can collect urban operation information in parallel from diverse data sources such as IoT sensors, video surveillance equipment, and social media platforms, and perform format normalization processing on it. Subsequently, through entity recognition and event element extraction, information fragments describing the same event are initially associated. The system further performs reliability assessment and conflict quantification on the information fragments, and performs multi-stage conflict reconciliation based on the assessment and quantification results. Finally, the reconciled unified semantic description is written into the urban situational awareness database. This modular design ensures the collaborative work of each functional unit, jointly achieving a comprehensive and accurate perception of the urban operation status. It aims to effectively solve the semantic inconsistency and conflict problems in the fusion of multi-source heterogeneous information, thereby improving the accuracy of urban situational awareness and the scientific nature of decision-making.

[0086] The above description is merely an embodiment of this application and is not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A method for processing comprehensive, all-element information based on multi-source data fusion, characterized in that, include: Urban operation information is collected in parallel from multiple distributed data sources, including IoT sensors, video surveillance equipment, and social media platforms. The collected urban operation information is then normalized and uniformly converted into a key-value pair format containing geographic location tags and timestamps. Entity recognition and event element extraction are performed on the normalized city operation information. Based on the geographic location label and the timestamp, information fragments describing the same event are initially associated to form at least one set of information fragments. For each information fragment in each set of information fragments, a reliability score for each information fragment is calculated by weighting and summing the information fragments according to the information source type to which the information fragments belong, the text subjectivity score of the information fragments, and the historical accuracy of the information source. By comparing the description values ​​of the same event attribute for different information fragments in the same set of information fragments, and calculating the degree of difference between the description values ​​according to the data type of the event attribute, the conflict quantification result is obtained. Based on the reliability score and the conflict quantification result, a multi-stage conflict reconciliation is performed: when the conflict quantification result is greater than a first conflict threshold and the range of the reliability scores is greater than a score threshold, the information segment with the highest reliability score is selected as the reconciliation result; when the conflict quantification result is less than or equal to the first conflict threshold and greater than a second conflict threshold, or when the range of the reliability scores is less than or equal to the score threshold, auxiliary information is retrieved from an external data source for cross-validation, and a reconciled semantic description is generated based on the validation results. The harmonized semantic description is used as the unified semantic description of the event, and the geographic location tag and the timestamp are used as the joint primary key to write it into the city situation awareness database.

2. The method according to claim 1, characterized in that, The calculation of the reliability score for each information segment includes: Map the information source type to a preset type weight value; Perform sentiment analysis on the information fragment, calculate the proportion of subjective words to all words, and obtain the text subjectivity score; The historical accuracy rate is calculated by retrieving the accuracy statistics of the information sources mentioned in the query history in past events. The type weight value, the text subjectivity score, and the historical accuracy are weighted and summed according to a preset weighting coefficient, and the sum is used as the reliability score.

3. The method according to claim 1, characterized in that, The calculation of the degree of difference between the descriptive values ​​includes: When the event attribute is a numerical attribute, the ratio of the difference between the maximum and minimum values ​​of the numerical attribute in each information segment to the average value is calculated, and the ratio is used as the difference degree. When the event attribute is a categorical attribute, the proportion of inconsistent samples of the categorical attribute in each information fragment to the total number of samples is calculated, and the proportion is used as the degree of difference. When the event attribute is a text attribute, the semantic edit distance between the text descriptions in each information fragment is calculated, and the maximum semantic edit distance is used as the difference.

4. The method according to claim 1, characterized in that, The step of retrieving auxiliary information from the external data source for cross-validation includes: Based on the geographic location tags in the information fragment set, the physical environment data of the corresponding location is obtained through the geographic information system interface; User-generated content corresponding to the geolocation tag is obtained from the social media platform via an application programming interface; The physical environment data and the user-generated content are used as auxiliary information, and the auxiliary information is spatiotemporally aligned with the information fragments in the information fragment set according to the timestamp.

5. The method according to claim 4, characterized in that, The generation of the harmonized semantic description based on the verification results includes: The spatiotemporally aligned auxiliary information is compared one by one with each information fragment in the information fragment set, and the semantic matching degree between the auxiliary information and each information fragment is calculated. Select the information fragment with the highest semantic matching degree as the benchmark fragment; Identify the portion of the auxiliary information that has semantic differences from the baseline segment and whose semantic matching degree is higher than a preset cross-validation threshold, and use the identified portion to correct the corresponding content in the baseline segment to generate the harmonized semantic description.

6. The method according to claim 1, characterized in that, After performing the multi-stage conflict reconciliation, the process also includes: Record the identifiers of the information fragments on which the harmonized semantic description is based, as well as the types of information sources used; The reconciled semantic description, the identifier, and the information source type are associated and stored in a reconciliation log database, which is used to track the conflict reconciliation results at different time points under the same geographic location label.

7. The method according to claim 1, characterized in that, The step of writing the harmonized semantic description into the urban situation awareness database includes: Extract the geolocation tags and timestamps from the harmonized semantic description; Using the geographic location tag and the timestamp as a joint primary key, query whether a corresponding record exists in the city situation awareness database; When the corresponding record is found, the corresponding record is overwritten and updated with the harmonized semantic description. When no corresponding record is found, the harmonized semantic description is inserted as a new record into the city situation awareness database.

8. The method according to claim 1, characterized in that, The method further includes periodically re-evaluating the set of information fragments: At the end of each preset evaluation period, acquire the newly added urban operation information within the evaluation period; After the newly added city operation information is formatted and normalized, it is re-associated with the information fragments in the existing information fragment set according to the geographic location tag and the timestamp to form an updated information fragment set; The reliability score calculation and conflict quantification are re-performed on the updated set of information fragments to obtain an updated reconciled semantic description; The updated harmonic semantic description is compared with the existing records in the urban situation awareness database. When the difference exceeds a preset update threshold, the corresponding record in the urban situation awareness database is updated.

9. A comprehensive, all-element information processing system based on multi-source data fusion, characterized in that, include: The multi-source acquisition and normalization module is used to collect urban operation information in parallel from multiple distributed data sources, including IoT sensors, video surveillance equipment and social media platforms. The module performs format normalization processing on the collected urban operation information and converts it into a key-value pair format containing geographic location tags and timestamps. The information association module is used to perform entity recognition and event element extraction on the normalized city operation information, and to initially associate information fragments describing the same event based on the geographic location tag and the timestamp to form at least one set of information fragments. The reliability assessment module is used to calculate the reliability score of each information fragment in each set of information fragments by performing a weighted sum based on the information source type to which the information fragment belongs, the text subjectivity score of the information fragment, and the historical accuracy of the information source. The conflict quantification module is used to compare the description values ​​of different information fragments in the same set of information fragments for the same event attribute, calculate the degree of difference between the description values ​​according to the data type of the event attribute, and obtain the conflict quantification result. A multi-stage reconciliation module is used to perform multi-stage conflict reconciliation based on the reliability score and the conflict quantification result: when the conflict quantification result is greater than a first conflict threshold and the range of the reliability scores is greater than a score threshold, the information segment with the highest reliability score is selected as the reconciliation result; when the conflict quantification result is less than or equal to the first conflict threshold and greater than a second conflict threshold, or when the range of the reliability scores is less than or equal to the score threshold, auxiliary information is retrieved from an external data source for cross-validation, and a reconciled semantic description is generated based on the validation results; The database writing module is used to write the harmonized semantic description as the unified semantic description of the event, using the geographic location tag and the timestamp as a joint primary key, into the city situation awareness database.

Images