Exhibition information synchronization and real-time updating management method and platform based on cloud computing

Abstract

The invention discloses an exhibition information synchronization and real-time updating management method and platform based on cloud computing. The method comprises the following steps: collecting multi-source heterogeneous exhibition information in real time through edge computing nodes; transmitting the collected heterogeneous data to a cloud computing center, carrying out format standardization, and carrying out semantic extraction on an unstructured text to generate a unified exhibition information data pool; periodic synchronization and instant synchronization based on event driving are carried out; when it is detected that multiple versions of the same information are updated, historical operation records of blockchain evidence storage are called, and a final effective version is generated in combination with preset priority rules and timestamp verification; and generating an adaptive information push packet according to the display capability and the interaction permission of the target terminal type. The method is used for solving the technical problems that in traditional exhibition information management, multi-source heterogeneous data integration is difficult, real-time updating response is insufficient, the unstructured text semantic processing capacity is weak, multi-source updating conflict resolution is not credible, and multi-terminal information pushing adaptability is poor.

Description

Technical Field

[0001] The present invention relates to the field of cloud computing application technology, and in particular to a cloud computing-based exhibition information synchronization and real-time update management method and platform. Background Art

[0002] In traditional exhibitions, information management faces multiple technical challenges, mainly in the following aspects: First, integrating heterogeneous data from multiple sources is difficult. Data sources at exhibition sites are complex and diverse, encompassing a wide range of devices, including exhibitor terminals, organizer management systems, and IoT sensors. These data sources employ significantly different transmission protocols (such as RESTful API, SOAP, and MQTT) and data formats (structured fields versus unstructured text), leading to significant data fragmentation. For example, the "booth number" field on an exhibitor's terminal differs from the "booth unique identifier" field in the organizer's system. This also makes it difficult to directly integrate raw sensor data from IoT sensors with structured data from business systems, creating information silos.

[0003] Secondly, real-time requirements are high, but responsiveness is insufficient. During exhibitions, dynamic information such as booth status changes, promotional offers taking effect, and environmental anomalies (such as excessive temperature and humidity, or excessive crowd density) must be captured and synchronized promptly. However, traditional centralized data collection methods rely on long-distance communication transmission, which is prone to data loss or update delays due to network delays or interruptions, making them unable to meet the real-time information requirements.

[0004] Furthermore, the semantic processing capabilities of unstructured text are weak. Unstructured texts such as exhibitor promotional materials and exhibit descriptions contain numerous exhibition-related terms (e.g., "limited-time discounts" and "smart terminals"). However, general language models have a limited understanding of exhibition contexts, often resulting in ambiguous recognition of domain terms, difficulty resolving polysemous words (e.g., "smart terminals" could refer to consumer electronics or industrial equipment), and loss of focus in extracting key elements (e.g., "core selling points" and "applicable scenarios"). This makes it difficult to integrate discrete text information with structured data, preventing the formation of a unified information data pool.

[0005] Furthermore, multi-source update conflicts are difficult to resolve. When exhibitors, organizers' systems, or IoT devices simultaneously modify the same information (such as booth contact information), traditional methods lack a reliable historical operation record traceability mechanism, clear priority rules (such as the authority level of organizer and exhibitor updates), and timestamp verification mechanisms. This can easily lead to "data conflicts" during multi-end information synchronization, making it impossible to generate a single valid version.

[0006] Finally, multi-device information push lacks adaptability. The traditional "one-size-fits-all" push model fails to consider the differences in display capabilities (such as screen resolution and supported media formats) and interaction permissions (such as editing and browsing permissions) across different devices (exhibitor apps, visitor mini-programs, on-site electronic screens, and partner media interfaces). This leads to problems such as information granularity mismatch (such as visitors receiving sensitive business information), display format incompatibility (such as electronic screens unable to adapt to small fonts), an imbalance between real-time performance and resource consumption (such as repeated transmission of routine information), and prominent security risks (such as sensitive data leakage).

[0007] In summary, traditional exhibition information management has significant defects in data integration, real-time response, semantic processing, conflict resolution and multi-terminal adaptation. There is an urgent need for a comprehensive solution based on cloud computing to improve the efficiency and reliability of information synchronization and real-time updates. Summary of the Invention

[0008] In order to overcome the shortcomings of the existing technology, the purpose of the present invention is to provide a cloud computing-based exhibition information synchronization and real-time update management method and platform, which is used to solve the technical problems existing in traditional exhibition information management, such as the difficulty in integrating multi-source heterogeneous data, insufficient real-time update response, weak unstructured text semantic processing capabilities, unreliable multi-source update conflict resolution, and poor adaptability of multi-terminal information push.

[0009] In order to solve the above problems, the technical solutions adopted by the present invention are as follows: A method for synchronizing and updating exhibition information in real time based on cloud computing includes the following steps: By deploying edge computing nodes at the exhibition site, multi-source heterogeneous exhibition information from exhibitors' terminals, organizers' management systems, and IoT sensors can be collected in real time. The collected heterogeneous data is transferred to the cloud computing center, formatted and standardized using pre-configured metadata mapping rules. The exhibition language processing model is used to extract semantics from unstructured text to generate a unified exhibition information data pool. A dual-trigger mechanism is set up in the cloud computing center to perform periodic synchronization based on time thresholds and instant synchronization based on event-driven synchronization; When multiple versions of the same information are detected, the historical operation records stored in the blockchain are called, combined with the preset priority rules and timestamp verification to generate the final valid version; According to the display capabilities and interaction permissions of the target terminal type, an adaptive information push package is dynamically generated through the API gateway of the cloud computing center.

[0010] Preferably, real-time collection of multi-source heterogeneous exhibition information includes: Connect the exhibitor terminal's RESTful API interface and the organizer's management system's SOAP protocol interface through the HTTP / HTTPS interface adapter to achieve standardized access to business system data; Analyze the raw perception data of booth environment sensors and binocular vision people flow counters through the IoT protocol analyzer; The collected heterogeneous data is temporarily stored using the LRU algorithm through the local cache module. When the communication with the cloud computing center is interrupted, the collection records are retained and automatically uploaded to the cloud after the communication is restored.

[0011] Preferably, format standardization and semantic extraction include: Constructing the metadata mapping rules based on the exhibition industry ontology library containing multi-dimensional ontology concepts in the metadata mapping unit; Establish field-level mapping relationships for the RESTful API interface data of exhibitors' terminals, the SOAP protocol data of organizers' systems, and the perception data of IoT sensors, mapping the non-uniform fields in heterogeneous data to the standard data model defined in the exhibition industry ontology library; The exhibition language processing model constructed in the NLP semantic extraction unit is used to perform domain terminology recognition and semantic disambiguation on unstructured text, extract key semantic elements and associate them with the corresponding nodes of the exhibition industry ontology library, and finally generate a unified exhibition information data pool containing structured data and semantic tags.

[0012] Preferably, when mapping to a standard data model, it includes: By formalizing the definition of core concepts, attributes, and inter-conceptual relationships within a domain through multi-dimensional ontology concepts, a unified semantic interpretation framework is provided for the standard data model, and semantic constraints are imposed on the definition, value range, and association rules of each data element in the standard data model. Among them, the abstract semantics of multidimensional ontology concepts need to be operationalized through the specific fields, relationships and constraints of the standard data model.

[0013] Preferably, extracting key semantic elements and associating them with corresponding nodes includes: Build domain dictionaries based on multi-dimensional ontology concepts to identify domain terms in unstructured texts; Combining the hierarchical relationship and attribute constraints of multi-dimensional ontology concepts, it eliminates the ambiguity of polysemous words in unstructured texts; Based on the semantic boundaries of multi-dimensional ontology concepts, key semantic elements that are strongly related to exhibition scenarios are extracted and mapped to corresponding structured fields in the standard data model to achieve semantic and structural conversion of unstructured text.

[0014] Preferably, when constructing the exhibition language processing model, it includes: Based on the BERT pre-trained model, it uses its bidirectional Transformer architecture to capture the contextual semantics of long texts to understand complex semantic relationships in unstructured texts. The BERT pre-trained model is customized and fine-tuned through the exhibition industry corpus, so that the model can learn the special vocabulary, sentence structure and domain logic in the exhibition scenario, and generate the exhibition language processing model.

[0015] Preferably, when setting up a dual trigger mechanism in a cloud computing center, it includes: Flexible configuration of synchronization cycles using Cron expressions through time triggers allows regular synchronization of regular exhibition information that has not changed significantly. The event monitoring component of the rule engine in the event trigger monitors the status changes of the exhibition information data pool in real time, and immediately triggers synchronization when any of the following events are detected: exhibit status change events, promotion event effectiveness events, and environment-related events; The dual trigger mechanism works in coordination with the cloud processing engine module and the conflict management module through the synchronization control module to perform periodic synchronization and event-driven synchronization.

[0016] Preferably, when generating the final valid version, the following is included: Through the blockchain evidence submodule of the conflict management module, all historical operation records of the same information are retrieved based on the alliance chain architecture. Each record contains the operation subject, operation timestamp and hash value of the modified content; Based on the priority rule library built into the version arbitration submodule, the operating subjects of multiple version updates are graded and determined; For multiple version updates with the same priority, extract the operation timestamps of each version and select the version with the latest timestamp as the candidate version; If there are updated versions with different priorities, the version corresponding to the highest priority subject is selected, and the modified content of the version with the latest timestamp is selected and synchronized to the unified exhibition information data pool.

[0017] Preferably, when generating an adapted information push package, the following steps are included: The API gateway dynamically analyzes the display capabilities and interaction permissions of the target terminal by receiving the terminal's request header information and pre-stored terminal type tags; Based on the terminal type, deployment location and authorization level, the rule engine is called to execute differentiated tailoring strategies; The template engine calls the display template that matches the terminal type to convert the exhibition information data pool content in a unified format into a presentation format that is suitable for the terminal; Combined with the trigger mechanism of the synchronization control module, WebSocket long connection is used for instant push of event-driven information; HTTP short connection + cache verification is used for regular information synchronized periodically; Push packages containing sensitive information are encrypted with AES and attached with a terminal-specific dynamic token; watermarks are added to push data from cooperative media interfaces.

[0018] A cloud computing-based exhibition information synchronization and real-time update management platform, including: Multi-source data acquisition module: This module is deployed at the edge computing node at the exhibition site and includes an HTTP / HTTPS interface adapter, an IoT protocol parser, and a local cache module. It is used to collect multi-source heterogeneous exhibition information in real time. Cloud processing engine module: This module runs in the cloud computing center and includes a metadata mapping unit with a built-in exhibition industry ontology library, as well as an NLP semantic extraction unit with a built-in exhibition language processing model and exhibition industry domain corpus. It is used to convert heterogeneous data into a unified exhibition information data pool. Synchronization control module: Integrates time triggers and event triggers to drive periodic synchronization and event-driven synchronization of exhibition information; Conflict management module: includes blockchain evidence storage submodule and version arbitration submodule, used to resolve multi-source update conflicts; Multi-terminal adaptation push module: includes API gateway and template engine, which is used to generate differentiated synchronization information push packages according to terminal types. Compared with the prior art, the present invention has the following beneficial effects: 1. Efficiency and Continuity of Multi-Source Heterogeneous Data Collection By integrating HTTP / HTTPS interface adapters (interfacing with RESTful APIs and SOAP protocols), IoT protocol parsers (supporting sensor protocols such as MQTT and LoRaWAN), and local cache modules (using a LRU algorithm to retain the last 24 hours of data) into edge computing nodes deployed at the exhibition site, we achieve real-time collection of heterogeneous data from multiple sources and ensure data continuity during communication interruptions. This design not only reduces cloud computing pressure but also avoids data loss caused by transmission delays or interruptions in traditional centralized data collection, providing a complete, real-time data source foundation for subsequent processing.

[0019] 2. Structural and Semantic Integration of Heterogeneous Data Based on the exhibition industry ontology, metadata mapping rules are constructed to map heterogeneous data (such as "booth numbers" and "booth unique identifiers") from exhibitor terminals, organizer systems, and IoT sensors into a unified standard data model, addressing the "structural fragmentation" problem inherent in traditional methods. Furthermore, a customized, fine-tuned BERT pre-trained model (integrated with the exhibition industry corpus) identifies domain terms, disambiguates semantics, and extracts key elements (such as "core selling points" and "limited-time discounts") from unstructured text (such as exhibitor promotional copy). These elements are then linked to corresponding nodes in the ontology, achieving the "semantic-to-structural conversion" of unstructured text. These two factors work together to form a unified exhibition information data pool containing both structured data and semantic tags, significantly improving the efficiency of multi-source data integration.

[0020] 3. Balancing the Timeliness and Coverage of Information Synchronization A dual-trigger mechanism (periodic synchronization + event-driven synchronization) balances the coverage of routine information with the real-time response to emergencies. Periodic synchronization (configurable from 10 seconds to 5 minutes) ensures the regular update of routine information that hasn't changed significantly (such as basic booth information). Event-driven synchronization (real-time monitoring of events such as changes in exhibit status, the effectiveness of promotions, and environmental anomalies) ensures the immediate delivery of emergency information (such as warnings of exceeding crowd density thresholds). This design avoids the problems of traditional single-trigger mechanisms, which often result in incomplete coverage or wasted resources, and improves the overall efficiency of information synchronization.

[0021] 4. Trustworthy Resolution of Multi-Source Update Conflicts The blockchain evidence storage submodule (which uses a consortium chain to store the operating entity, timestamp, and hash value of modified content) ensures the credibility of version traceability. Combined with the version arbitration submodule's priority rules (organizer system > exhibitor terminal > IoT device) and timestamp verification, this mechanism clarifies the update authority hierarchy of different entities and handles conflicts of equal priority. This mechanism effectively resolves the "data conflict" problem inherent in traditional methods, ensuring a single valid version when synchronizing across multiple terminals, and guaranteeing the consistency and accuracy of information.

[0022] 5. Adaptability and Security of Multi-Terminal Scenario Push By dynamically parsing terminal display capabilities and interaction permissions through an API gateway, implementing differentiated tailoring strategies in conjunction with a rules engine, and adapting display formats using a template engine, this approach enables precise push notifications for exhibitors, visitors, promotions, guides, and media. Furthermore, WebSocket persistent connections ensure the immediacy of emergency information, HTTP short connections and cache validation reduce the amount of regular information transmitted, and enhance security through AES encryption (for sensitive data), dynamic tokens (for terminal-specific), and watermarking (for media data). This design addresses the information mismatch, display incompatibility, and security risks inherent in the traditional "one-size-fits-all" push model, significantly improving the information acquisition experience and data security for users on multiple devices.

[0023] In summary, the present invention has built an exhibition information management system covering the entire process of data collection, processing, synchronization, conflict resolution and push through the deep integration of cloud computing, edge computing, NLP semantic processing, blockchain and other technologies, providing exhibition activities with efficient, real-time, reliable and adaptive information synchronization and update services.

[0024] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 This is a step diagram of a method for synchronizing and managing exhibition information in real time according to an embodiment of the present invention; Figure 2 This is a flowchart of generating a final valid version according to an embodiment of the present invention; Figure 3 This is a flowchart of generating an adapted information push package according to an embodiment of the present invention; Figure 4 It is a schematic diagram of the structure of the exhibition information synchronization and real-time update management platform according to an embodiment of the present invention.

[0026] Explanation of the accompanying figures: 200, exhibition information synchronization and real-time update management platform; 201, multi-source data acquisition module; 202, cloud processing engine module; 203, synchronization control module; 204, conflict management module; 205, multi-terminal adaptation push module. DETAILED DESCRIPTION

[0027] Exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art.

[0028] At the same time, it should be understood that for the convenience of description, the sizes of the various parts shown in the drawings are not drawn according to the actual proportional relationship.

[0029] The following description of at least one exemplary embodiment is merely illustrative in nature and is in no way intended to limit the present disclosure, its application, or uses.

[0030] Technologies, methods, and equipment known to ordinary technicians in the relevant art may not be discussed in detail, but where appropriate, the technologies, methods, and equipment should be considered part of the specification.

[0031] Example 1, see Figure 1 The present invention provides a method for synchronizing and updating exhibition information in real time. Figure 1 A cloud computing-based exhibition information synchronization and real-time update management method is shown, comprising the following steps: S101. Multi-source heterogeneous data collection: Edge computing nodes deployed at the exhibition site collect multi-source heterogeneous exhibition information in real time from exhibitor terminals, organizer management systems, and IoT sensors (including booth environment sensors and crowd counters). Multi-source heterogeneous exhibition information includes basic booth information (including exhibitor qualifications and exhibit parameters), real-time dynamic information (including exhibit demonstration status and promotional event changes), and environmental information (including booth temperature and humidity, and crowd density). S102, Cloud Platform Standardization Processing: The collected heterogeneous data is transmitted to the cloud computing center, where it is standardized using pre-configured metadata mapping rules. The exhibition language processing model is used to perform semantic extraction on unstructured text (such as exhibitors' promotional copy updated in real time) to generate a unified exhibition information data pool. S103, Dynamic Synchronization Triggering: A dual-trigger mechanism is set up in the cloud computing center to perform periodic synchronization based on time thresholds (the period can be configured from 10 seconds to 5 minutes) and instant synchronization based on event-driven events (events include exhibit status changes, promotions taking effect, and crowd density exceeding thresholds); S104, conflict detection and resolution: When multiple versions of the same information are detected (e.g., the exhibitor's terminal and the organizer's system modify the booth contact information at the same time), the historical operation records stored in the blockchain are called up, combined with the preset priority rules (organizer's system > exhibitor's terminal > IoT device) and timestamp verification, to generate the final valid version; S105. Multi-terminal differentiated push: Based on the display capabilities and interaction permissions of the target terminal types (including exhibitor apps, audience mini-programs, on-site electronic screens, and cooperative media interfaces), adaptive information push packages are dynamically generated through the API gateway of the cloud computing center. The audience end filters sensitive business information, and the on-site electronic screen prioritizes displaying crowd guidance information, achieving scenario-based adaptation of synchronized information.

[0032] Background description: In traditional exhibitions, information management faces the challenges of difficulty in integrating multi-source heterogeneous data and high real-time requirements. On the one hand, the data sources at the exhibition site are complex and diverse, and the transmission protocols and data formats used by different data sources vary significantly, resulting in prominent information island problems; on the other hand, exhibitions have high requirements for the timeliness of information updates. Dynamic information such as booth status changes, promotional activities, and environmental anomalies must be captured and synchronized in a timely manner. Traditional centralized data collection methods are prone to data loss or update delays due to transmission delays or communication interruptions. Based on this: In a possible embodiment, in the above step S101, when collecting multi-source heterogeneous exhibition information in real time, the process includes: Connect the exhibitor terminal's RESTful API interface and the organizer's management system's SOAP protocol interface through the HTTP / HTTPS interface adapter to achieve standardized access to business system data; Analyze the raw perception data of booth environment sensors (including temperature and humidity sensors, CO2 concentration sensors, and light intensity sensors) and binocular vision people flow counters through the Internet of Things protocol parser; The local cache module uses the LRU algorithm to temporarily store the collected heterogeneous data. When the communication with the cloud computing center is interrupted, the collection records of the last 24 hours can be retained and automatically uploaded to the cloud after the communication is restored. Among them, the edge computing node is deployed with HTTP / HTTPS interface adapter, IoT protocol parser and local cache module; the IoT protocol parser supports MQTT, LoRaWAN and ZigBee protocols.

[0033] Further explanation is required regarding this embodiment of the present invention. This embodiment integrates an HTTP / HTTPS interface adapter (for connecting to business systems), an IoT protocol parser (for adapting to sensor protocols), and a local cache module (using a LRU algorithm for temporary data storage) to achieve real-time collection of multi-source, heterogeneous exhibition information. This design not only processes heterogeneous data in different protocols and formats locally, reducing cloud computing pressure, but also ensures data continuity (retaining the last 24 hours of records) during communication interruptions through local caching. This provides a complete, real-time data source foundation for subsequent cloud platform standardization processing and dynamic synchronization.

[0034] Background description: In traditional exhibition information management, the lack of uniformity in the format and semantic ambiguity of multi-source heterogeneous data are the core pain points that hinder information integration and utilization. On the one hand, the raw data from exhibitor terminals (such as RESTful API data), organizer systems (such as SOAP protocol data) and IoT sensors (such as MQTT perception data) use different field names (such as "booth number" and "booth unique identifier"), protocol standards (such as RESTful and SOAP) and data types (structured fields and unstructured text), resulting in "structural fragmentation" of the data and making it impossible to directly integrate them. On the other hand, unstructured text (such as exhibitor promotional copy and exhibit descriptions) contains a large number of domain terms (such as "limited-time discount" and "smart terminal"), but due to the limited understanding of exhibition scenarios by general language models, ambiguity often occurs (such as "smart terminal" may refer to consumer electronics or industrial equipment) or key elements are omitted (such as "core selling points" are not accurately extracted), resulting in "semantic fuzziness". Based on this: In a possible embodiment, in the above step S102, when performing format standardization and semantic extraction, the steps include: Construct metadata mapping rules based on the exhibition industry ontology library in the metadata mapping unit. The exhibition industry ontology library contains multi-dimensional ontology concepts such as exhibit classification system, exhibitor qualification level, and event type label; Field-level mapping relationships are established for the RESTful API interface data of exhibitors' terminals, the SOAP protocol data of the organizer's system, and the perception data of IoT sensors. Non-standard fields such as "booth number", "exhibit model", and "promotion description" in heterogeneous data are mapped to standard data models such as "booth unique identifier", "exhibit technical parameters", and "marketing activity information" defined in the exhibition industry ontology library; The exhibition language processing model in the NLP semantic extraction unit is used to identify domain terms and perform semantic disambiguation on unstructured texts (such as exhibitor promotional copy), extract key semantic elements such as "core selling points" and "applicable scenarios", and associate them with the corresponding nodes in the exhibition industry ontology library, ultimately generating a unified exhibition information data pool containing structured data and semantic tags.

[0035] It is worth further explaining that in this embodiment of the present invention, this embodiment resolves structural inconsistencies through metadata mapping rules (building field-level mapping relationships based on the exhibition industry ontology library), mapping heterogeneous data to standard data models such as "booth unique identifier" and "exhibit technical parameters." Simultaneously, a customized, fine-tuned BERT pre-trained model (combined with a corpus from the exhibition field) is used for semantic extraction, identifying domain terms, resolving ambiguities, and extracting key elements such as "core selling points" and "applicable scenarios," ultimately generating a unified exhibition information data pool. This process transforms heterogeneous data from "fragmentation" to "structuring" and from "fuzzy semantics" to "clear semantics," providing a standardized data source foundation for subsequent dynamic synchronization, conflict resolution, and multi-terminal adaptive push.

[0036] Background: In traditional exhibition information processing, semantic extraction of unstructured text (such as exhibitor promotional materials and exhibit descriptions) faces the following challenges: ambiguous identification of domain terms, difficulty resolving ambiguous polysemous words, and loss of focus in extracting key elements. Furthermore, if the extracted semantic elements are not associated with a standard data model, they cannot be stored in a structured manner, making it difficult to integrate discrete text information with IoT sensor data and structured data from business systems to form a unified information data pool. Based on this: In a possible embodiment, extracting key semantic elements and associating them with corresponding nodes includes: A domain dictionary is constructed based on the multi-dimensional ontology concepts in the exhibition industry ontology library (including exhibit classification system, event type labels, and exhibitor qualification levels) to identify domain terms in unstructured text (such as "limited-time discount", "screen size", and "industry certification"); Combining the hierarchical relationships of multi-dimensional ontology concepts (e.g., the classification hierarchy of "consumer electronics - smart terminals - smartphones") and attribute constraints (e.g., "AAA-rated exhibitors must meet annual revenue ≥ 100 million"), we can eliminate the ambiguity of polysemous words in unstructured text (e.g., clarifying "smart terminals" as "consumer electronics - smart terminals" rather than "smart terminals" in other fields); Based on the semantic boundaries of multi-dimensional ontology concepts (e.g., "promotional activities" corresponds to elements such as "limited-time discounts" and "top 100," and "smartphone" corresponds to attributes such as "screen size" and "battery life"), key semantic elements strongly related to exhibition scenarios (including core selling points, applicable scenarios, and discount strength) are extracted and mapped to corresponding structured fields in the standard data model (e.g., "marketing activity information - activity type" field and "exhibit technical parameters - core selling points" field), achieving semantic and structural conversion of unstructured text; Among them, the exhibition language processing model is based on the BERT (Bidirectional Encoder Representation Transformer) pre-trained model, using its bidirectional Transformer architecture to strongly capture the contextual semantics of long texts, and understand the complex semantic relationships in unstructured texts (such as exhibitor promotional copy and exhibit descriptions); and is customized and fine-tuned through the exhibition industry corpus (including historical exhibitor promotional copy, exhibit technical descriptions, event announcements, audience feedback and other text data), so that the model can learn special vocabulary in the exhibition scenario (such as "limited-time discount", "exhibit life", "industry certification"), sentence structure (such as the promotional statement "50% off if you place an order before 18:00 today") and domain logic (such as the hierarchical association between "smart terminals" and "consumer electronics"), completing the adaptation from a general language model to a special model for the exhibition field.

[0037] It is worth further explaining that in this embodiment of the present invention, by constructing an ontology library for the exhibition industry (including multi-dimensional concepts such as the exhibit classification system and activity type labels), this embodiment provides the model with a "domain dictionary" and "semantic filtering rules": on the one hand, the hierarchical relationships and attribute constraints of the ontology library are used to resolve the ambiguity of polysemous words and clarify the exhibition scene reference of the terms; on the other hand, based on the semantic boundaries of the ontology library, the model is guided to accurately extract key elements such as "core selling points" and "applicable scenarios" and map them to the structured fields of the standard data model. This process realizes the transformation of unstructured text from "fuzzy semantics" to "scenario-based, structured semantics", providing an operational semantic foundation for subsequent information synchronization, conflict detection, and multi-terminal adaptive push.

[0038] In a possible embodiment, multi-dimensional ontology concepts (such as exhibit classification system, exhibitor qualification level, and activity type label) provide a unified semantic interpretation framework for the standard data model by formally defining the core concepts, attributes, and relationships between concepts in the field (for example, in the "exhibit classification system," "consumer electronics" includes "smart terminals," and "smart terminals" include "smart phones" in a hierarchical relationship; in the "exhibitor qualification level," "AAA level" must meet attribute constraints such as "annual revenue ≥ 100 million" and "complete industry certifications." The definition, value range, and association rules of each data element in the standard data model (such as the "exhibit classification" field and the "exhibitor qualification level" field) are directly derived from the semantic constraints of the multi-dimensional ontology concepts. For example, the value of the "exhibit classification" field in the standard data model must strictly correspond to the hierarchical path of the "exhibit classification system" in the ontology library (such as "consumer electronics-smart terminals-smart phones"), rather than any customized classification description; The abstract semantics of multidimensional ontology concepts must be operationalized through the specific fields, relationships, and constraints of a standard data model. For example, concepts such as "Promotional Events - Limited-Time Discounts" and "Interactive Events - Audience Experiences" defined in the "Activity Type Tag" of the exhibition industry ontology library are transformed into enumerated values ​​in the "Activity Type" field of the standard data model (e.g., the value range: ["Promotional Events - Limited-Time Discounts", "Interactive Events - Audience Experiences"]). The progressive hierarchy of "AAA → AA → A" defined in the "Exhibitor Qualification Level" of the ontology library is transformed into the sorting rules of the "Exhibitor Qualification Level" field in the standard data model (e.g., numerical priority identifiers: AAA = 3, AA = 2, A = 1). Through this structured mapping, ontology concepts are transformed from abstract domain knowledge into concrete data models that can be stored and computed.

[0039] What needs to be further explained in the embodiments of the present invention is that the concept of multi-dimensional ontology solves the problem of "semantic inconsistency" of heterogeneous data (for example, the "3C products" of the exhibitor terminal and the "electronic equipment" of the organizer system are uniformly interpreted as "consumer electronics - smart terminals"), while the standard data model solves the problem of "structural inconsistency" of heterogeneous data (for example, the "qualification level A" of the exhibitor terminal and the "first-level certification" of the organizer system are uniformly mapped to the "AA level" field value). The synergy between the two enables heterogeneous data from different sources (exhibitor terminals, organizer systems, IoT sensors) and in different formats (structured fields, unstructured text) to be ultimately converted into a standard data model that follows the same semantic and structural rules, providing a unified data source foundation for subsequent information synchronization, conflict detection, and multi-terminal push. The extraction of key semantic elements (such as "core selling points", "applicable scenarios", "limited-time discounts", etc.) is not aimless text extraction, but domain-oriented extraction based on the semantic constraints of multi-dimensional ontology concepts. For example: The "activity type tags" in the ontology library (such as "promotional activities" and "interactive activities") provide semantic boundaries for the exhibition language processing model, guiding the model to identify key elements of "promotional activities" such as "limited-time discounts" and "top 100" from exhibitor promotional copy, rather than irrelevant information (such as "booth location"); The "exhibit classification system" in the ontology library (e.g., "consumer electronics - smart terminals - smartphones") provides the model with core concepts in the exhibit field, guiding it to extract "screen size" and "battery life" from unstructured descriptions, which are "core selling points" of smartphones, rather than attributes of other types of exhibits (e.g., "material" for furniture exhibits). In short, the multi-dimensional ontology concept provides a "domain dictionary" and "semantic filtering rules" for the extraction of key semantic elements, ensuring that the extracted elements are strongly related to the exhibition scene and conform to the domain specifications; Multidimensional ontology concepts (such as "activity type label" and "exhibit classification system") are abstract domain knowledge frameworks, while key semantic elements are the instantiation of these frameworks in specific texts. For example: The "activity type tag" in the ontology library defines the concept of "promotional activity - limited-time discount." When an exhibitor's copy contains the phrase "50% off for orders placed before 6:00 PM today," the "limited-time discount" and "50% off discount" extracted by the NLP model are specific instances of this ontology concept. The "Exhibit Classification System" in the ontology library defines the attributes of "smart terminal" (such as "interaction mode" and "operating system"). When "supports touch interaction" and "is equipped with Android system" appear in the exhibit description, the extracted "touch interaction" and "Android system" are instantiated elements of the "smart terminal" attribute; Through this instantiation, multidimensional ontological concepts are transformed from abstract semantic frameworks into concrete information units that can be processed by computers; Key semantic elements must be structured and stored using a standard data model so that they can be combined with other structured data (such as temperature and humidity data from IoT sensors and booth numbers from exhibitor terminals) to form a unified exhibition information data pool. For example: Extracted key elements such as "Limited-time discount" and "50% off" are mapped to subfields under the "Marketing Campaign Information" field in the standard data model (for example, "Activity Type" = "Promotion - Limited-time Discount" and "Discount Strength" = "50% off"). The extracted key elements such as "touch interaction" and "Android system" will be mapped to subfields under the "Exhibit Technical Parameters" field in the standard data model (e.g., "Interaction method" = "touch", "Operating system" = "Android"); The standard data model organizes discrete key semantic elements into searchable and syncable structured data by defining field hierarchies, data types, and associations (such as the association between the "marketing campaign information" and "exhibit classification" fields), ultimately supporting subsequent information synchronization and real-time updates.

[0040] Background description: In the traditional exhibition information synchronization process, a single synchronization trigger mechanism is difficult to take into account both the coverage and real-time nature of information updates. Based on this: In a possible embodiment, in the above step S103, when the dual trigger mechanism is set in the cloud computing center, it includes: Periodic synchronization mechanism: Flexibly configure the synchronization period (covering a configurable range of 10 seconds to 5 minutes) through time triggers using Cron expressions. For example, set it to every 30 seconds or every 2 minutes to regularly synchronize routine exhibition information that has not changed significantly (such as basic booth information and stable exhibit parameters). Event-driven synchronization mechanism: The event monitoring component of the rule engine in the event trigger monitors the status changes of the exhibition information data pool in real time, and triggers synchronization immediately when the following events are detected: Exhibit status change events (e.g., the exhibit demonstration status switches from "inactive" to "on demonstration" or "paused"); Promotional event (e.g., the promotion reaches its scheduled start / end time, or the promotion content is modified); Environmental related events (such as crowd density exceeding a preset threshold (such as 20 people / m 2 ), or any of the booth temperature, humidity, CO2 concentration, and light intensity exceeds the set value); Among them, the dual-trigger mechanism works in coordination with the cloud processing engine module 202 and the conflict management module 204 through the synchronization control module 203, periodically synchronizing to ensure the coverage of information updates and event-driven synchronization to handle emergency changes in real time, thereby jointly ensuring the timeliness and accuracy of the synchronization of exhibition information on multiple terminals.

[0041] Further explanation is needed regarding this embodiment of the present invention. This embodiment ensures coverage of routine information (such as basic booth information) that hasn't changed significantly, through periodic synchronization (configurable from 10 seconds to 5 minutes). Event-driven synchronization (real-time monitoring of events such as exhibit status changes, promotional events, and environmental anomalies) handles emergency changes with immediacy, ensuring that emergency information reaches multiple devices quickly. Working together, these two mechanisms not only ensure efficient information coverage but also address the need for real-time response to emergencies, providing a timely and accurate information update foundation for subsequent conflict detection and multi-device push.

[0042] Background description: In traditional exhibition information management, it is difficult to resolve the problem of multi-source update conflicts, mainly because: First, version traceability is unreliable. When the exhibitor terminal, the organizer system or the IoT device simultaneously modifies the same information (such as booth contact information), there is a lack of credible historical operation records, making it difficult to confirm the modification subject, time and content of each version; second, there is a lack of priority rules. Updates from different sources (such as organizers and exhibitors) have no clear authoritative level division, resulting in an inability to objectively determine which version should be adopted; third, there is a lack of timestamp verification. Updates submitted simultaneously from the same source (such as two exhibitor terminals) may mistakenly retain the old version and overwrite the latest modification due to the lack of a timestamp comparison mechanism. These problems lead to the phenomenon of "data conflict" when synchronizing information on multiple terminals. Based on this: In one possible embodiment, see Figure 2 The flowchart of generating the final valid version, in the above step S104, when generating the final valid version, includes: S41. Blockchain evidence tracing: Through the blockchain evidence submodule of the conflict management module 204, all historical operation records of the same information are retrieved based on the alliance chain architecture. Each record includes the operation subject (exhibitor terminal / organizer system / IoT device), the operation timestamp, and the hash value of the modified content; S42, priority classification verification: Based on the priority rule library built into the version arbitration submodule, the operating subjects of multiple version updates are graded and determined. The priority order is organizer system > exhibitor terminal > IoT device; S43, timestamp-assisted arbitration: For multiple versions of updates with the same priority (e.g., two exhibitor terminals modify the same information at the same time), extract the operation timestamps of each version and select the version with the latest timestamp as the candidate version; S44. Valid version generation: If there are updated versions of different priorities, directly select the version corresponding to the highest priority subject; if there are multiple versions of the highest priority subject, select the version with the latest timestamp; finally, synchronize the modified content of this version to the unified exhibition information data pool to complete conflict resolution.

[0043] The present invention's embodiments require further explanation. This embodiment addresses the issue of untrustworthy versions through blockchain evidence storage and traceability (storing complete operation logs based on a consortium chain), ensuring that every update record is traceable and verifiable. Priority tiered verification (organizer system > exhibitor terminal > IoT device) establishes an authority hierarchy, clarifying the update priorities of different entities. Timestamp-assisted arbitration (selecting the most recently timestamped version) handles conflicts of equal priority, ensuring that the latest revision is adopted first. These three elements work together to ultimately generate a single, valid version, resolving conflicts between multiple sources and providing a unified, trusted data source for subsequent multi-terminal synchronization.

[0044] Background description: In the traditional exhibition information push process, the "one-size-fits-all" push model leads to the following problems in multi-terminal application scenarios: information granularity mismatch, display format incompatibility, imbalance between real-time performance and resource consumption, and prominent security risks. Based on this: In one possible embodiment, see Figure 3 The flowchart of generating an adapted information push package, in the above step S105, when generating an adapted information push package, includes: S51. Terminal Type Identification and Capability Analysis: The API Gateway receives the terminal's request header information (including the User-Agent field, screen resolution parameters, and supported media formats) and pre-stored terminal type tags (exhibitor app / visitor mini-program / on-site electronic screen / partner media interface) to dynamically analyze the target terminal's display capabilities (such as maximum supported image resolution, support for animation, and text length limits) and interaction permissions (such as exhibitor editing permissions, visitor browsing permissions, and media reprint authorization scope). S52, scenario-based content tailoring: Based on the terminal type, deployment location, and authorization level, the rule engine is called to execute differentiated tailoring strategies: For exhibitors: Retention of core business information: Exhibitors will retain all sensitive business information related to their own booths (e.g., company contact information, unpublished summaries of letters of intent for cooperation, and unreleased technical specifications of exhibits). Only business information related to unrelated exhibitors (e.g., details of promotional events at competing booths) will be desensitized (displayed as "[Other exhibitor activities]"). Business management data is prioritized: Focus on pushing dynamic information directly related to exhibitors' operations, including: Real-time booth status data (such as display status change records, temperature and humidity abnormality alarms, and the current number of visitors displayed by the crowd counter); Interaction effect statistics (such as the number of inquiries initiated by visitors through the mini program, the number of information form submissions, and the distribution of exhibition page browsing time); Instructions and notifications from the organizer (e.g., "Media coverage is required from 15:00 to 16:00" and "Booth layout is overdue and needs to be rectified"); Open editable fields: To meet exhibitors' self-management rights, editable field modules (such as "Promotion Description" and "Exhibit Parameter Supplementary Instructions") are embedded in the push package. The field format matches the RESTful API interface requirements of the exhibitor's terminal, and supports direct modification and transmission back to the cloud computing center to update the data pool; Irrelevant information filtering: Shield public information irrelevant to the exhibitor's business (such as crowd guidance tips for other exhibition areas and rules for receiving exclusive gifts for visitors), and only retain information with global impact (such as "12:00 closure security inspection reminder" and "sudden weather warning"); Visitor mini-programs: Filter exhibitor contact information and undisclosed business cooperation information, prioritize core selling points of exhibits, real-time promotions (such as "the first 50 visitors can receive gifts"), and crowd flow guidance information in the current area (such as "the queue at Gate 3, Area B, is currently 5 minutes"); On-site electronic screens: Dynamically adjust information priority based on deployment location (entrance / exhibition area / rest area). Entrance screens highlight exhibition overview and popular booth recommendations. Exhibition area screens strengthen current area crowd density warnings (e.g., "Current density 25 people / m 2 , it is recommended to visit during off-peak hours”), and the rest area screen will push catering services and event schedules; For partner media interfaces: information granularity is tailored according to authorization levels (the basic level only provides exhibit names and categories; the advanced level additionally provides exhibit technical parameters and exhibitor qualification levels), and converted into the structured format required by the media (such as the "title-abstract-content" level in JSON); S53. Dynamic format adaptation: The template engine calls the display template that matches the terminal type (such as the card layout template for the audience terminal and the large font carousel template for the electronic screen) to convert the content of the exhibition information data pool in a unified format into a presentation format that is suitable for the terminal. Text information: Create an intelligent summary based on the terminal text length limit (e.g., ≤500 characters per mini program screen), retaining the "core selling points + key data" (e.g., "Smartwatch: 48-hour battery life, 20% off today from 10:00 AM to 12:00 PM"); Multimedia information: Based on the terminal's media support capabilities (e.g., electronic screens support videos, while mini-programs only support images), the cloud media resource library matches the corresponding format of the material (e.g., transcoding 4K videos into 720P images). S54, real-time optimized push: Combined with the trigger mechanism of the synchronization control module 203, event-driven information (such as sudden promotions and warnings of excessive crowd flow) is pushed instantly using WebSocket persistent connections to ensure timely delivery to the terminal; for periodically synchronized routine information (such as basic booth information), HTTP short connections + cache verification (ETag / Last-Modified) are used to reduce the amount of repeated data transmission; S55. Security enhancement processing: Perform AES encryption on push packages containing sensitive information (such as exhibitors' contact information) and attach a terminal-specific dynamic token (generated based on the terminal ID and timestamp) to ensure that only authorized terminals can decrypt and access it; add a watermark to the push data of the cooperative media interface (such as "Data source: XX Exhibition Cloud Platform") to prevent illegal reproduction.

[0045] What needs to be further explained in the embodiments of the present invention is that this embodiment clarifies the boundaries of demand through terminal type identification and capability analysis, and achieves precise push notifications of "exhibitors see business, visitors see discounts, electronic screens see guidance, and media see structure" through scenario-based content tailoring; ensures the display friendliness of information on different terminals through dynamic format adaptation; balances the immediacy of emergency information with the transmission efficiency of regular information through real-time optimized push notifications; and ensures controllable access to sensitive data and traceability of the source of media data through security enhancement processing. This process realizes the transformation of exhibition information from "unified push notifications" to "scenario-based and adaptive push notifications," providing exhibitors, visitors, management parties, and cooperative media with efficient, secure, and user-friendly information synchronization services.

[0046] Example 2, see Figure 4 The present invention provides a schematic diagram of the structure of the exhibition information synchronization and real-time update management platform. Figure 4 The illustrated exhibition information synchronization and real-time update management platform 200 based on cloud computing includes: a multi-source data acquisition module 201 , a cloud processing engine module 202 , a synchronization control module 203 , a conflict management module 204 and a multi-terminal adaptation push module 205 .

[0047] Multi-source data collection module 201: an edge computing node deployed at the exhibition site, including an HTTP / HTTPS interface adapter, an IoT protocol parser, and a local cache module, for real-time collection of multi-source heterogeneous exhibition information; Cloud processing engine module 202: runs in the cloud computing center and includes a metadata mapping unit with a built-in exhibition industry ontology library and an NLP semantic extraction unit with a built-in exhibition language processing model and exhibition industry domain corpus, used to convert heterogeneous data into a unified exhibition information data pool; Synchronization control module 203: integrates time triggers and event triggers to drive periodic synchronization and event-driven synchronization of exhibition information; Conflict management module 204: includes a blockchain evidence storage submodule and a version arbitration submodule, used to resolve multi-source update conflicts; Multi-terminal adaptation push module 205: includes an API gateway and a template engine, and is used to generate differentiated synchronization information push packages according to terminal types.

[0048] In a possible embodiment, the HTTP / HTTPS interface adapter is used to connect to the RESTful API interface of the exhibitor terminal and the SOAP protocol interface of the organizer management system to achieve standardized access to business system data.

[0049] In a possible embodiment, the IoT protocol parser supports MQTT, LoRaWAN, and ZigBee protocols, and is used to parse raw perception data from booth environment sensors (including temperature and humidity sensors, CO2 concentration sensors, and light intensity sensors) and binocular vision people flow counters.

[0050] In a possible embodiment, the local cache module uses an LRU algorithm to temporarily store the collected heterogeneous data, so as to retain the collection records of the last 24 hours when communication with the cloud computing center is interrupted, and automatically upload them to the cloud after communication is restored.

[0051] In a possible embodiment, the exhibition industry ontology library built into the metadata mapping unit is used to construct metadata mapping rules. The exhibition industry ontology library includes multi-dimensional ontology concepts such as exhibit classification system, exhibitor qualification level, and activity type label.

[0052] In a possible embodiment, the metadata mapping unit is used to establish field-level mapping relationships for the RESTful API interface data of the exhibitor terminal, the SOAP protocol data of the organizer system, and the perception data of the Internet of Things sensor, and map the non-uniform fields in the heterogeneous data to the standard data model defined in the exhibition industry ontology library.

[0053] In a possible embodiment, the exhibition language processing model built into the NLP semantic extraction unit is used to perform domain terminology recognition and semantic disambiguation on unstructured text, as well as to extract key semantic elements and associate them with corresponding nodes in the exhibition industry ontology library.

[0054] In one possible embodiment, the exhibition language processing model is based on the BERT pre-trained model, using its bidirectional Transformer architecture to capture the strong semantics of long text contexts to solve the problem of understanding complex semantic relationships in unstructured texts; and customized fine-tuning is performed through the exhibition industry domain corpus, so that the model can learn the specialized vocabulary, sentence structure and domain logic in the exhibition scenario.

[0055] In a possible embodiment, the time trigger uses a Cron expression to flexibly configure the synchronization period, which is used to regularly synchronize regular exhibition information that has not changed significantly.

[0056] In a possible embodiment, the event trigger includes an event monitoring component based on a rule engine. The event monitoring component is used to monitor the status changes of the exhibition information data pool in real time and immediately trigger synchronization when a specific event is detected.

[0057] In one possible embodiment, the blockchain evidence storage submodule uses a consortium chain architecture to retrieve all historical operation records of the same information, and each record contains the operation subject, operation timestamp, and hash value of the modified content.

[0058] In a possible embodiment, the version arbitration submodule has a built-in priority rule library and a timestamp verification algorithm for performing hierarchical determination on operating entities for multi-version updates.

[0059] In a possible embodiment, the API gateway supports RESTful and gRPC protocols, and is used to dynamically parse the display capabilities and interaction permissions of the target terminal by receiving the request header information of the terminal and the pre-stored terminal type tag.

[0060] In a possible embodiment, the multi-terminal adaptation push module 205 calls a rule engine to execute a differentiated tailoring strategy based on the terminal type, deployment location, and authorization level.

[0061] In a possible embodiment, the template engine has built-in display templates for audiences, exhibitors, electronic screens, etc., which are used to call the display template that matches the terminal type and convert the content of the exhibition information data pool in a unified format into a presentation form adapted to the terminal.

[0062] The above embodiments are only preferred embodiments of the present invention and cannot be used to limit the scope of protection of the present invention. Any non-substantial changes and replacements made by technicians in this field on the basis of the present invention fall within the scope of protection required by the present invention.

Claims

1. A method for synchronizing and updating exhibition information in real time based on cloud computing, characterized in that: The following steps are involved: By deploying edge computing nodes at the exhibition site, multi-source heterogeneous exhibition information from exhibitors' terminals, organizers' management systems, and IoT sensors can be collected in real time. The collected heterogeneous data is transferred to the cloud computing center, formatted and standardized using pre-configured metadata mapping rules. The exhibition language processing model is used to extract semantics from unstructured text to generate a unified exhibition information data pool. A dual-trigger mechanism is set up in the cloud computing center to perform periodic synchronization based on time thresholds and instant synchronization based on event-driven synchronization; When multiple versions of the same information are detected, the historical operation records stored in the blockchain are called, combined with the preset priority rules and timestamp verification to generate the final valid version; According to the display capabilities and interaction permissions of the target terminal type, an adaptive information push package is dynamically generated through the API gateway of the cloud computing center.

2. The method according to claim 1, characterized in that Real-time collection of multi-source heterogeneous exhibition information includes: Connect the exhibitor terminal's RESTful API interface and the organizer's management system's SOAP protocol interface through the HTTP / HTTPS interface adapter to achieve standardized access to business system data; Analyze the raw perception data of booth environment sensors and binocular vision people flow counters through the IoT protocol analyzer; The collected heterogeneous data is temporarily stored using the LRU algorithm through the local cache module. When the communication with the cloud computing center is interrupted, the collection records are retained and automatically uploaded to the cloud after the communication is restored.

3. The method according to claim 2, characterized in that When performing format standardization and semantic extraction, this includes: Constructing the metadata mapping rules based on the exhibition industry ontology library containing multi-dimensional ontology concepts in the metadata mapping unit; Establish field-level mapping relationships for the RESTful API interface data of exhibitors' terminals, the SOAP protocol data of organizers' systems, and the perception data of IoT sensors, mapping the non-uniform fields in heterogeneous data to the standard data model defined in the exhibition industry ontology library; The exhibition language processing model constructed in the NLP semantic extraction unit is used to perform domain terminology recognition and semantic disambiguation on unstructured text, extract key semantic elements and associate them with the corresponding nodes of the exhibition industry ontology library, and finally generate a unified exhibition information data pool containing structured data and semantic tags.

4. The method according to claim 3, characterized in that When mapping to a standard data model, this includes: By formalizing the definition of core concepts, attributes, and inter-conceptual relationships within a domain through multi-dimensional ontology concepts, a unified semantic interpretation framework is provided for the standard data model, and semantic constraints are imposed on the definition, value range, and association rules of each data element in the standard data model. Among them, the abstract semantics of multidimensional ontology concepts need to be operationalized through the specific fields, relationships and constraints of the standard data model.

5. The method according to claim 3 or 4, characterized in that Extracting key semantic elements and associating them with corresponding nodes includes: Build domain dictionaries based on multi-dimensional ontology concepts to identify domain terms in unstructured texts; Combining the hierarchical relationship and attribute constraints of multi-dimensional ontology concepts, it eliminates the ambiguity of polysemous words in unstructured texts; Based on the semantic boundaries of multi-dimensional ontology concepts, key semantic elements that are strongly related to exhibition scenarios are extracted and mapped to corresponding structured fields in the standard data model to achieve semantic and structural conversion of unstructured text.

6. The method according to claim 5, characterized in that When building a language processing model for an exhibition, we include: Based on the BERT pre-trained model, it uses its bidirectional Transformer architecture to capture the contextual semantics of long texts to understand complex semantic relationships in unstructured texts. The BERT pre-trained model is customized and fine-tuned through the exhibition industry corpus, so that the model can learn the special vocabulary, sentence structure and domain logic in the exhibition scenario, and generate the exhibition language processing model.

7. The method according to claim 6, characterized in that When setting up a dual trigger mechanism in a cloud computing center, it includes: Flexible configuration of synchronization cycles using Cron expressions through time triggers allows regular synchronization of regular exhibition information that has not changed significantly. The event monitoring component of the rule engine in the event trigger monitors the status changes of the exhibition information data pool in real time, and immediately triggers synchronization when any of the following events are detected: exhibit status change events, promotion event effectiveness events, and environment-related events; The dual trigger mechanism works in coordination with the cloud processing engine module and the conflict management module through the synchronization control module to perform periodic synchronization and event-driven synchronization.

8. The method according to claim 7, characterized in that When generating the final valid version, it includes: Through the blockchain evidence submodule of the conflict management module, all historical operation records of the same information are retrieved based on the alliance chain architecture. Each record contains the operation subject, operation timestamp and hash value of the modified content; Based on the priority rule library built into the version arbitration submodule, the operating subjects of multiple version updates are graded and determined; For multiple version updates with the same priority, extract the operation timestamps of each version and select the version with the latest timestamp as the candidate version; If there are updated versions with different priorities, the version corresponding to the highest priority subject is selected, and the modified content of the version with the latest timestamp is selected and synchronized to the unified exhibition information data pool.

9. The method according to claim 8, characterized in that When generating an adapted information push package, include: The API gateway dynamically analyzes the display capabilities and interaction permissions of the target terminal by receiving the terminal's request header information and pre-stored terminal type tags; Based on the terminal type, deployment location and authorization level, the rule engine is called to execute differentiated tailoring strategies; The template engine calls the display template that matches the terminal type to convert the exhibition information data pool content in a unified format into a presentation format that is suitable for the terminal; Combined with the trigger mechanism of the synchronization control module, WebSocket long connection is used for instant push of event-driven information; HTTP short connection + cache verification is used for regular information synchronized periodically; Push packages containing sensitive information are encrypted with AES and attached with a terminal-specific dynamic token; watermarks are added to push data from cooperative media interfaces.

10. A cloud computing-based exhibition information synchronization and real-time update management platform, characterized by: include: Multi-source data acquisition module: This module is deployed at the edge computing node at the exhibition site and includes an HTTP / HTTPS interface adapter, an IoT protocol parser, and a local cache module. It is used to collect multi-source heterogeneous exhibition information in real time. Cloud processing engine module: This module runs in the cloud computing center and includes a metadata mapping unit with a built-in exhibition industry ontology library, as well as an NLP semantic extraction unit with a built-in exhibition language processing model and exhibition industry domain corpus. It is used to convert heterogeneous data into a unified exhibition information data pool. Synchronization control module: Integrates time triggers and event triggers to drive periodic synchronization and event-driven synchronization of exhibition information; Conflict management module: includes blockchain evidence storage submodule and version arbitration submodule, used to resolve multi-source update conflicts; Multi-terminal adaptation push module: includes API gateway and template engine, which is used to generate differentiated synchronization information push packages according to terminal types.

Images