Abnormal conference rental monitoring method, device, equipment, medium and program product

Abstract

The present application relates to the field of electronic technology, and discloses an abnormal conference rental monitoring method, device, equipment, medium and program product, the method comprises: obtaining multi-modal conference organization rental data;Multi-modal conference organization rental data is carried out feature fusion processing, and multi-modal fusion feature is obtained;The multi-modal fusion feature and the conference rental compliance knowledge graph are input into the pre-constructed semantic understanding big model, so that the semantic understanding model outputs the preliminary abnormal list.The method provided by the present application inputs the multi-modal fusion feature and the conference rental compliance knowledge graph into the pre-constructed semantic understanding big model, and the model automatically completes compliance verification and abnormal determination and outputs the standardized preliminary abnormal list, without staff checking a large amount of multi-type data one by one, which greatly reduces the time consumption of abnormal investigation, reduces the labor cost, and can stably output the unified standard determination result.

Description

Technical Field

[0001] This invention relates to the field of electronic technology, specifically to methods, devices, equipment, media, and program products for monitoring abnormal meeting rentals. Background Technology

[0002] In corporate management, conference car rental is a core component for business travel, large conferences, and foreign affairs receptions. Its compliance directly impacts cost control, legal risk avoidance, and the implementation of official vehicle management regulations. Inspection dimensions cover multiple aspects, including car rental qualification compliance, vehicle type and passenger matching, fee standard compliance, consistency between actual vehicle service and contractual agreement, and driver qualifications and service standards. Currently, conference car rental compliance checks often employ a combination of manual labor and simple tools. The core process involves: collecting basic data via API, comparing against preset rules, manually identifying suspected anomalies, and having a dedicated person compile and write a report. However, this approach has a significant weakness in anomaly detection. Simple tools, based on preset rules, struggle to handle the complex and ever-changing needs of conference car rental scenarios. Complex anomalies outside the rule coverage require staff to meticulously review rental contracts, itinerary records, and expense receipts. This process is not only time-consuming and labor-intensive with low efficiency, but the results are also directly influenced by the staff's professional level and experience, leading to inconsistent anomaly detection accuracy and a high risk of misjudgments and missed detections. Summary of the Invention

[0003] This invention provides a method, device, equipment, medium, and program product for monitoring abnormal conference vehicle rentals, in order to solve the problems of overall time-consuming and labor-intensive, low identification efficiency, and inconsistent accuracy of abnormal identification in related technologies that use a combination of manual and simple tools for compliance checks of conference vehicle rentals.

[0004] In a first aspect, the present invention provides a method for monitoring abnormal conference car rentals. The method includes: acquiring multimodal conference organization car rental data; performing feature fusion processing on the multimodal conference organization car rental data to obtain multimodal fusion features; and inputting the multimodal fusion features and a conference car rental compliance knowledge graph into a pre-constructed semantic understanding model so that the semantic understanding model outputs a preliminary list of anomalies.

[0005] The abnormal meeting car rental monitoring method provided by this invention acquires multimodal meeting organization car rental data, providing a complete data foundation for full-dimensional compliance verification and avoiding blind spots caused by single data dimensions. By performing feature fusion processing on multimodal data, dynamic weight adaptation of different types of data is achieved, which can flexibly adapt to various complex business scenarios of meeting car rental. It also has the ability to intelligently identify implicit and complex anomalies that traditional rules cannot cover. The multimodal fusion features and the meeting car rental compliance knowledge graph are jointly input into a pre-built semantic understanding model. The model automatically completes compliance verification and anomaly judgment and outputs a standardized preliminary anomaly list. This eliminates the need for staff to manually review and verify massive amounts of various types of data, which not only greatly reduces the time spent on anomaly investigation and lowers the cost of manpower, but also eliminates the dependence of the verification results on the professional level and experience of the staff. It can stably output unified standard judgment results, effectively reducing the misjudgment and omission problems that are prone to occur in manual verification, and improving the professionalism, stability and reliability of meeting car rental compliance inspection.

[0006] In an optional implementation, the method further includes: standardizing and annotating each abnormal information in the preliminary abnormality list to obtain a standardized evidence package classified by abnormality; inputting the standardized evidence package and a pre-built meeting car rental compliance report template into a pre-built report generation big language model so that the report generation big language model outputs monitoring result report text data.

[0007] The method provided by this optional implementation can automatically classify, collect, and standardize abnormal evidence, replacing the tedious work of manually organizing multi-source evidence and ensuring the integrity and standardization of the evidence chain. At the same time, it can automatically output monitoring result reports with the help of a large language model for report generation and preset report templates, eliminating the need for manual writing and formatting by dedicated personnel, greatly shortening the report generation cycle, reducing labor costs, and effectively avoiding problems such as data omissions and non-standard expressions that are prone to occur in manual operation, thus ensuring the professionalism and consistency of the reports.

[0008] In one alternative implementation, the conference car rental compliance knowledge graph is constructed through the following steps: obtaining a conference car rental rule dataset; performing structured processing on the conference car rental rule dataset to obtain triple data; and constructing the conference car rental compliance knowledge graph based on the triple data.

[0009] The method provided in this optional implementation converts multi-source meeting car rental rule data into triples through structured processing and constructs a compliance knowledge graph. This integrates scattered internal corporate policies, official vehicle regulations, industry compliance standards, and other rules into a unified, standardized, and machine-accessible structured rule system. This provides a clear and unified basis for anomaly identification, ensuring the consistency of compliance judgment standards and the traceability of results. It also facilitates the dynamic updating and maintenance of subsequent rules, effectively avoiding problems such as rule reference bias and inconsistent standard implementation that are prone to occur under manual verification mode. This provides stable and reliable compliance knowledge support for the accurate anomaly reasoning of large models.

[0010] In one optional implementation, the step of obtaining multimodal meeting organization car rental data includes: obtaining initial meeting organization car rental related data; and sequentially performing data cleaning and standardization processing on the initial meeting organization car rental related data to obtain multimodal meeting organization car rental data.

[0011] The method provided in this optional implementation performs data cleaning and standardization on the initial meeting car rental related data in sequence. This can effectively remove redundant and erroneous information from the original multi-source data, fill in missing fields, improve the accuracy and completeness of the basic data, and at the same time unify the format specifications of different types of data, eliminate the format differences between multimodal data, provide high-quality and standardized data input for subsequent feature fusion processing and large model anomaly identification, and ensure the stability and accuracy of anomaly judgment results.

[0012] In one optional implementation, the step of performing feature fusion processing on multimodal meeting organization and car rental data to obtain multimodal fusion features includes: acquiring multiple types of abnormal target query vectors; performing feature extraction and normalization processing on the meeting organization and car rental data of each modality in the multimodal meeting organization and car rental data to obtain the feature vector of the corresponding modality; determining the original attention score between the feature vector of the corresponding modality and the various types of abnormal target query vectors based on the feature vector of each modality and the various types of abnormal target query vectors; normalizing the original attention score between the feature vector of each modality and the various types of abnormal target query vectors to obtain the corresponding target attention weight; and determining the multimodal fusion features based on the feature vectors of multiple modalities and the target attention weights between the feature vectors of each modality and the various types of abnormal target query vectors.

[0013] The method provided in this optional implementation sequentially extracts and normalizes features from each modality of car rental data, and calculates attention weights by combining multiple types of abnormal target query vectors, ultimately generating multimodal fusion features. This method can dynamically adjust the weight ratio of each modality of data for different types of compliance anomalies, strengthen the judgment role of highly correlated data, effectively break down information barriers between different modalities of data, realize the linkage analysis of multi-source data, significantly improve the accuracy and adaptability of anomaly identification in complex scenarios, reduce false positives and false negatives, and also provide high-quality feature inputs for subsequent anomaly reasoning in large semantic understanding models, ensuring the stability and consistency of the judgment results.

[0014] Secondly, the present invention provides an abnormal meeting car rental monitoring device, the device comprising: an acquisition module for acquiring multimodal meeting organization car rental data; a first processing module for performing feature fusion processing on the multimodal meeting organization car rental data to obtain multimodal fusion features; and a first determination module for inputting the multimodal fusion features and a meeting car rental compliance knowledge graph into a pre-constructed semantic understanding model so that the semantic understanding model outputs a preliminary abnormal list.

[0015] In an optional implementation, the above-mentioned device further includes: a second processing module, used to perform standardized annotation processing on each abnormal information in the preliminary abnormality list to obtain a standardized evidence package classified by abnormality; and a second determining module, used to input the standardized evidence package and a pre-built meeting car rental compliance report template into a pre-built report generation big language model, so that the report generation big language model outputs monitoring result report text data.

[0016] Thirdly, the present invention provides an electronic device, comprising: a memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the computer instructions to perform the abnormal meeting car rental monitoring method of the first aspect or any corresponding embodiment described above.

[0017] Fourthly, the present invention provides a computer-readable storage medium storing computer instructions for causing a computer to execute the abnormal meeting car rental monitoring method of the first aspect or any corresponding embodiment described above.

[0018] Fifthly, the present invention provides a computer program product, including computer instructions for causing a computer to execute the abnormal meeting car rental monitoring method described in the first aspect or any corresponding embodiment thereof. Attached Figure Description

[0019] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0020] Figure 1 This is a schematic diagram of an application scenario according to an embodiment of the present invention; Figure 2 This is a schematic flowchart of the first method for monitoring abnormal meeting car rentals according to an embodiment of the present invention; Figure 3 This is a schematic diagram of the second process of the abnormal meeting car rental monitoring method according to an embodiment of the present invention; Figure 4 This is a schematic diagram of the third process of the abnormal meeting car rental monitoring method according to an embodiment of the present invention; Figure 5 This is a structural block diagram of an abnormal meeting car rental monitoring device according to an embodiment of the present invention; Figure 6 This is a schematic diagram of the hardware structure of an electronic device according to an embodiment of the present invention. Detailed Implementation

[0021] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, 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, 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.

[0022] It is understood that before using the technical solutions disclosed in the various embodiments of the present invention, users should be informed of the types, scope of use, and usage scenarios of the personal information involved in the present invention and their authorization should be obtained in accordance with relevant laws and regulations through appropriate means.

[0023] The terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0024] As an optional application scenario of this invention, the specific application environment architecture or specific hardware architecture on which the abnormal meeting car rental monitoring method depends is described here. For example... Figure 1 As shown, the architecture system may include at least one terminal device and at least one server. Figure 1 The system is illustrated in the example, which includes a computer 101, a mobile terminal 102, and a server 103, and the terminal devices such as the computer 101 and the mobile terminal 102 are connected to the server 103 through a network 110.

[0025] Specifically, the terminal device can be a smartphone, tablet, laptop, PDA, desktop computer, game console, smart TV, smart wearable device, in-vehicle terminal, VR (Virtual Reality) device, AR (Augmented Reality) device, etc. Server 103 can be a standalone physical server, a server cluster, a distributed system, or a cloud server providing cloud services. Network 110 can be a wired or wireless network, examples of which include, but are not limited to, the Internet, corporate intranet, local area network, wide area network, mobile communication network, and combinations thereof.

[0026] In related technologies, compliance checks for conference vehicle rentals often employ a combination of manual labor and simple tools. The core process involves, in sequence, collecting basic data via an interface, comparing data against preset fixed rules, manually screening for suspected anomalies, and having a specialist compile and write a report. However, this approach suffers from significant shortcomings in anomaly detection. Simple tools, relying on preset rules for data comparison, struggle to handle the complex and ever-changing demands of conference vehicle rental scenarios. Complex anomalies outside the scope of these rules require staff to meticulously review rental contracts, itinerary records, and expense receipts. This process is not only time-consuming and labor-intensive with low efficiency, but the results are also directly influenced by the staff's professional level and experience, leading to inconsistent anomaly detection accuracy and a high risk of misjudgments and missed detections.

[0027] In view of this, the present application provides an abnormal meeting car rental monitoring method that can be applied to a single server to monitor abnormal meeting car rentals. The method provided in this application provides a complete data foundation for full-dimensional compliance verification by acquiring multimodal meeting organization car rental data, avoiding blind spots caused by single data dimensions. By performing feature fusion processing on multimodal data, dynamic weight adaptation of different types of data is achieved, which can flexibly adapt to various complex business scenarios of meeting car rentals and has intelligent identification capabilities for implicit and complex anomalies that traditional rules cannot cover. The multimodal fusion features and the meeting car rental compliance knowledge graph are input into a pre-built semantic understanding model, which automatically completes compliance verification and anomaly judgment and outputs a standardized preliminary anomaly list. This eliminates the need for staff to manually review and verify massive amounts of various types of data, significantly reducing the time spent on anomaly investigation, lowering manpower costs, and eliminating the dependence of verification results on the professional level and experience of staff. It can stably output unified standard judgment results, effectively reducing the misjudgment and omission problems that are prone to occur in manual verification, and improving the professionalism, stability, and reliability of meeting car rental compliance checks overall.

[0028] According to an embodiment of the present invention, an embodiment of a method for monitoring abnormal meeting car rentals is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0029] This embodiment provides a method for monitoring abnormal meeting car rentals, which can be used on the aforementioned server. Figure 2 This is a flowchart of an abnormal meeting car rental monitoring method according to an embodiment of the present invention, such as... Figure 2 As shown, the process includes the following steps: Step S201: Obtain car rental data for multimodal conference organization.

[0030] For example, multimodal refers to various forms of information carriers. Meeting organization car rental data refers to all business data generated throughout the entire process of car rental services related to corporate meetings, from application and approval, contract signing, service execution to payment settlement, covering all core business elements such as vehicle qualifications, driver information, itinerary arrangements, fee standards, and approval processes. In this embodiment, multimodal meeting organization car rental data refers to the meeting organization car rental related data corresponding to the order to be analyzed, including text modality, image modality, spatiotemporal trajectory modality, and voice modality meeting organization car rental data. Text-based meeting organization and car rental data can include, but is not limited to, tabular numerical data such as rental fees, rental period, number of attendees, and vehicle type, as well as text-based data such as rental contracts, approval documents, and expense reimbursement forms; image-based meeting organization and car rental data can include, but is not limited to, image-based data such as scanned copies of contracts, actual photos of vehicles, driver's professional qualification certificates, and expense receipts; spatiotemporal trajectory-based meeting organization and car rental data can include, but is not limited to, dynamic data with time and space attributes such as driving routes, start and end times, and stop points generated by vehicle GPS positioning; and voice-based meeting organization and car rental data can include, but is not limited to, voice-based data such as communication regarding car use and service feedback.

[0031] Step S202: Perform feature fusion processing on the multimodal meeting organization car rental data to obtain multimodal fusion features.

[0032] For example, in this embodiment of the application, features of car rental data for each modality of the meeting organization are first extracted, and then the features of different modalities are fused to obtain multimodal fused features.

[0033] Step S203: Input the multimodal fusion features and the conference car rental compliance knowledge graph into the pre-built semantic understanding model so that the semantic understanding model outputs a preliminary list of anomalies.

[0034] For example, multimodal fusion features and a meeting car rental compliance knowledge graph are input into a pre-built semantic understanding model. The model performs compliance verification and anomaly identification on the fusion features based on the compliance rules in the knowledge graph, and finally outputs a preliminary anomaly list. In this embodiment, a general large language model is used as the foundation, and data such as meeting car rental compliance rules, historical anomaly cases, and business texts are collected to construct a special training set. The foundation model is then subjected to scenario-specific adaptation training through supervised fine-tuning to enhance the compliance semantic parsing and anomaly identification capabilities. After verification and optimization, a semantic understanding model is obtained.

[0035] The abnormal meeting car rental monitoring method provided in this embodiment acquires multimodal meeting organization car rental data, providing a complete data foundation for full-dimensional compliance verification and avoiding blind spots caused by single data dimensions. By performing feature fusion processing on multimodal data, dynamic weight adaptation of different types of data is achieved, which can flexibly adapt to various complex business scenarios of meeting car rental. It also has the ability to intelligently identify implicit and complex anomalies that traditional rules cannot cover. The multimodal fusion features and the meeting car rental compliance knowledge graph are jointly input into a pre-built semantic understanding model. The model automatically completes compliance verification and anomaly judgment and outputs a standardized preliminary anomaly list. This eliminates the need for staff to manually review and verify massive amounts of various types of data, which not only greatly reduces the time spent on anomaly investigation and lowers the cost of manpower, but also eliminates the dependence of the verification results on the professional level and experience of the staff. It can stably output unified standard judgment results, effectively reducing the misjudgment and omission problems that are prone to occur in manual verification, and improving the professionalism, stability and reliability of meeting car rental compliance inspection.

[0036] This embodiment provides a method for monitoring abnormal meeting car rentals, which can be used on the aforementioned server. Figure 3 This is a flowchart of an abnormal meeting car rental monitoring method according to an embodiment of the present invention, such as... Figure 3 As shown, the process includes the following steps: Step S301: Obtain car rental data for multimodal meeting organization. For details, please refer to [link / reference]. Figure 2 Step S201 of the illustrated embodiment will not be described again here.

[0037] Step S302: Perform feature fusion processing on the multimodal meeting organization car rental data to obtain multimodal fusion features.

[0038] In some alternative implementations, the aforementioned knowledge graph on meeting car rental compliance is constructed through the following steps: Step a1: Obtain the conference car rental rules dataset.

[0039] For example, in this embodiment of the application, various types of rule-based raw data are collected and aggregated to form a unified raw data set for constructing a compliance determination system for conference car rental. The conference car rental rules are all compliance requirements that govern car rental business in conference scenarios, including internal corporate car rental management regulations, relevant laws and regulations on official vehicles, compliance standards for the car rental industry, and control clauses such as vehicle qualifications, fee standards, and itinerary approval.

[0040] Step a2 involves structuring the conference car rental rules dataset to obtain triplet data.

[0041] For example, in this embodiment of the application, knowledge extraction and structured modeling are performed on the unstructured rule text in the conference car rental rule dataset to automatically identify core entities such as enterprise departments, attendees, drivers, vehicles, and car rental orders. The relationships between entities are extracted, and the rule content is transformed and mapped according to the triple structure of entity-relationship-entity. At the same time, the corresponding compliance judgment rules are extracted to finally obtain triple data.

[0042] Step a3: Construct a knowledge graph of meeting car rental compliance based on triple data.

[0043] For example, in this embodiment of the application, the graph ontology is first defined, and an ontology framework consisting of an entity layer, a relation layer, and a rule layer is built. The entity layer covers business entities, compliance standard entities, and regulatory entities; the relation layer covers business relationships and compliance constraint relationships; and the rule layer covers violation judgment rules and compliance basis clauses. Then, the triple data is stored in the database, and independent indexes are established for entities, relationships, and constraint rules to support fast retrieval. At the same time, conflicting rules are labeled and priorities are set to eliminate multi-source knowledge conflicts and achieve entity alignment, ultimately constructing a meeting car rental compliance knowledge graph.

[0044] Specifically, step S302 includes: Step S3021: Obtain query vectors for multiple types of abnormal targets.

[0045] For example, in this embodiment of the application, compliance judgment rules and semantic features corresponding to various anomalies are extracted from the conference vehicle rental compliance knowledge graph. The feature information of each type of anomaly is converted into a corresponding vector representation through an encoding model to obtain multi-type anomaly target query vectors. Specifically, for each type of conference vehicle rental anomaly, such as unqualified personnel, excessive fees, deviation from the itinerary, mismatch between vehicle type and number of passengers, and temporary vehicle dispatch without approval, a dedicated anomaly target query vector is generated in advance based on the compliance knowledge graph.

[0046] Step S3022: Perform feature extraction and normalization on the car rental data of each modality in the multimodal meeting organization data to obtain the feature vector of the corresponding modality.

[0047] For example, for each modality of meeting organization car rental data, its data features are extracted, and then the extracted features are normalized to obtain the feature vector of the corresponding modality, including: text features. Image features Trajectory temporal characteristics Speech features All features are uniformly scaled to the same numerical range. In this embodiment, the single-modal feature extraction stage performs feature extraction on each modality of data separately, outputting single-modal feature vectors to ensure that the feature information of each modality of data is fully mined. For text data, the BERT model is used to extract features, capturing semantic information and compliance clause features in the text; for image and video data, the ResNet-5 model is used to extract features, obtaining visual features such as vehicle model, vehicle condition, driver qualifications, and license plate; for voice data, the Whisper model is used to extract features, capturing semantic features such as trip communication and service feedback within the voice.

[0048] Step S3023: Based on the feature vectors of each modality and the query vectors of multiple types of abnormal targets, determine the original attention score between the feature vectors of the corresponding modality and the query vectors of various types of abnormal targets.

[0049] For example, in this embodiment of the application, the matching degree between the feature vector of each modality and the query vector of various abnormal targets is calculated one by one, and the original score of a single modality is calculated using the scaled dot product attention method. The calculation formula is as follows:

[0050] in, This represents the query vector for the abnormal target currently to be verified. This represents the eigenvector of the m-th mode. ; express transpose, This indicates the dimension of the feature vector, used for scaling to avoid excessively large values.

[0051] If the current anomaly is "vehicle qualifications do not match" , (Images of driver's license and vehicle operating permit) and The inner product is extremely large. The value is very high; (Contract qualification clause text) Secondly, medium; , The correlation with certificates and qualifications is extremely low, and the score is close to 0.

[0052] Step S3024: Normalize the original attention scores between the feature vectors of each modality and the query vectors of various abnormal targets to obtain the corresponding target attention weights.

[0053] For example, in this embodiment of the application, Softmax normalization is used to generate standardized attention scores. Since the original score values ​​are not uniform, the Softmax function normalizes all modality scores to obtain standardized attention scores in the range of 0 to 1 with a sum of 1. :

[0054] in, This represents the standardized attention score corresponding to the m-th modality, i.e., the target attention weight, with a value ranging from 0 to 1, and for all modalities... The sum is 1; Represents the natural exponential function; The summation of the original attention scores for all modalities after applying the natural exponent is used as the denominator in the normalized calculation.

[0055] Step S3025: Determine the multimodal fusion features based on the feature vectors of multiple modalities, the target attention weights between the feature vectors of each modality and the query vectors of various abnormal targets.

[0056] For example, in this embodiment of the application, the normalized attention score is directly used as the weighting coefficient of each modality feature, and the original single-modality features are weighted and fused, as shown in the following formula:

[0057] in, This represents the multimodal fusion feature; the meanings of the other variables will not be elaborated further.

[0058] Step S303 involves inputting the multimodal fusion features and the conference car rental compliance knowledge graph into a pre-built semantic understanding model, so that the semantic understanding model outputs a preliminary anomaly list. For details, please refer to [link to relevant documentation]. Figure 2 Step S203 of the illustrated embodiment will not be described again here.

[0059] This embodiment provides a method for monitoring abnormal meeting car rentals, which can be used on the aforementioned server. Figure 4 This is a flowchart of an abnormal meeting car rental monitoring method according to an embodiment of the present invention, such as... Figure 4 As shown, the process includes the following steps: Step S401: Obtain car rental data for multimodal conference organization.

[0060] Specifically, step S401 includes: Step S4011: Obtain the initial meeting organization car rental related data.

[0061] For example, in this embodiment of the application, raw data on meeting organization and car rental from multiple sources are collected, and various modal information such as text, images, voice, and location are gathered to complete the initial data collection and obtain the initial meeting organization and car rental related data.

[0062] Step S4012: Perform data cleaning and standardization on the initial meeting organization car rental data to obtain multimodal meeting organization car rental data.

[0063] For example, in this embodiment, the preprocessing process includes structured data cleaning, unstructured data conversion, and data standardization. Structured data cleaning relies on data verification algorithms to remove redundant data, correct abnormal values ​​such as fee amounts and location trajectories, and supplement missing content such as car rental service confirmation information, ensuring the accuracy and completeness of the structured data. Unstructured data conversion utilizes optical character recognition technology to extract text content from scanned contracts and photos to form editable text, automatic speech recognition technology to convert speech to text, and image segmentation technology to annotate photo regions. Data standardization uniformly converts multimodal data of various formats into vector formats recognizable by generative large models. Text-type data, including the text obtained from recognition and conversion, is uniformly converted into BERT vectors; image and video keyframes are converted into ResNet50 vectors; and trajectory-type data is converted into temporal feature vectors, thus providing data support for the semantic fusion layer to conduct correlation analysis and anomaly identification.

[0064] Step S402 involves performing feature fusion processing on the multimodal meeting organization car rental data to obtain multimodal fusion features. For details, please refer to [link to relevant documentation]. Figure 3 Step S302 of the illustrated embodiment will not be described again here.

[0065] Step S403 involves inputting the multimodal fusion features and the conference car rental compliance knowledge graph into the pre-built semantic understanding model, so that the semantic understanding model outputs a preliminary anomaly list. For details, please refer to [link to relevant documentation]. Figure 3 Step S303 of the illustrated embodiment will not be described again here.

[0066] Step S404: Standardize and annotate each abnormal information in the preliminary abnormality list to obtain a standardized evidence package classified by abnormality.

[0067] For example, in this embodiment of the application, all abnormal information in the preliminary abnormal list is traversed, and unified standardized labeling is completed according to the preset compliance abnormality classification specifications. The multimodal original data and feature basis corresponding to each abnormality are associated and integrated to form a standardized evidence package under the corresponding abnormality classification.

[0068] Step S405: Input the standardized evidence package and the pre-built meeting car rental compliance report template into the pre-built report generation large language model so that the report generation large language model outputs monitoring result report text data.

[0069] For example, a standardized evidence package and a pre-built meeting car rental compliance report template are fed into a large language model for report generation. The model analyzes the abnormal information and supporting evidence within the evidence package, and, combined with the fixed textual framework of the report template, automatically generates the output monitoring result report text data. In this embodiment, domain-specific fine-tuning is performed based on the basic large language model. A training dataset is constructed using historical verification reports in the meeting car rental compliance field, various standardized evidence packages, and compliance judgment clauses. The model's adaptability to abnormal evidence identification, compliance conclusion derivation, and report writing is continuously optimized. Corresponding output constraint instructions are set in conjunction with the report template to complete the construction of the large language model for report generation.

[0070] This embodiment also provides an abnormal meeting car rental monitoring device, which is used to implement the above embodiments and preferred embodiments, and will not be repeated as already described. As used below, the term "module" can be a combination of software and / or hardware that implements a predetermined function. Although the device described in the following embodiments is preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.

[0071] This embodiment provides an abnormal meeting car rental monitoring device, such as... Figure 5 As shown, it includes: Module 501 is used to acquire car rental data for multimodal conferences. The first processing module 502 is used to perform feature fusion processing on the multimodal meeting organization car rental data to obtain multimodal fusion features; The first determining module 503 is used to input multimodal fusion features and conference car rental compliance knowledge graph into a pre-built semantic understanding model so that the semantic understanding model outputs a preliminary list of anomalies.

[0072] In some alternative embodiments, the above-described apparatus further includes: The second processing module is used to standardize and annotate each abnormal information in the preliminary abnormality list to obtain a standardized evidence package classified by abnormality. The second determination module is used to input the standardized evidence package and the pre-built meeting car rental compliance report template into the pre-built report generation large language model, so that the report generation large language model outputs monitoring result report text data.

[0073] In some alternative implementations, the meeting car rental compliance knowledge graph is constructed through the following steps: Obtain the conference car rental rules dataset; The conference car rental rules dataset was structured to obtain triplet data; A knowledge graph of meeting car rental compliance constructed based on triplet data.

[0074] In some optional implementations, the acquisition module 501 includes: The first acquisition submodule is used to acquire initial meeting organization car rental-related data; The first processing submodule is used to perform data cleaning and standardization on the initial meeting organization car rental data in sequence to obtain multimodal meeting organization car rental data.

[0075] In some alternative implementations, the first processing module 502 includes: The second acquisition submodule is used to acquire query vectors for multiple types of abnormal targets; The second processing submodule is used to extract and normalize the features of each modality of the multimodal meeting organization and car rental data to obtain the feature vector of the corresponding modality. The first determination submodule is used to determine the original attention score between the feature vector of the corresponding modality and the query vectors of various abnormal targets based on the feature vectors of each modality and the query vectors of multiple types of abnormal targets. The third processing submodule is used to normalize the original attention scores between the feature vectors of each modality and the query vectors of various abnormal targets to obtain the corresponding target attention weights. The second determination submodule is used to determine multimodal fusion features based on feature vectors of multiple modalities and target attention weights between feature vectors of each modality and query vectors of various abnormal targets.

[0076] The abnormal meeting car rental monitoring device provided in this embodiment of the invention can execute the abnormal meeting car rental monitoring method provided in any embodiment of the invention, and has the corresponding functional modules and beneficial effects of the method. Further functional descriptions of the various modules and units are the same as in the corresponding embodiments described above, and will not be repeated here.

[0077] Figure 6 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention.

[0078] The following is a detailed reference. Figure 6 This diagram illustrates a suitable structural design for implementing an electronic device according to embodiments of the present invention. The electronic device may include a processor (e.g., a central processing unit, graphics processor, etc.) 601, which can perform various appropriate actions and processes based on a program stored in read-only memory (ROM) 602 or a program loaded from memory 608 into random access memory (RAM) 603. RAM 603 also stores various programs and data required for the operation of the electronic device. The processor 601, ROM 602, and RAM 603 are interconnected via a bus 604. An input / output (I / O) interface 605 is also connected to the bus 604.

[0079] Typically, the following devices can be connected to I / O interface 605: input devices 606 including, for example, touchscreens, touchpads, keyboards, mice, cameras, microphones, accelerometers, gyroscopes, etc.; output devices 607 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; memory devices 608 including, for example, magnetic tapes, hard disks, etc.; and communication devices 609. Communication device 609 allows electronic devices to communicate wirelessly or wiredly with other devices to exchange data. Although Figure 6 Electronic devices with various devices are shown, but it should be understood that it is not required to implement or have all of the devices shown, and more or fewer devices may be implemented or have instead.

[0080] In particular, according to embodiments of the present invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of the present invention include a computer program product comprising a computer program carried on a non-transitory computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device 609, or installed from a memory 608, or installed from a ROM 602. When the computer program is executed by the processor 601, it performs the functions defined in the abnormal meeting car rental monitoring method of the embodiments of the present invention.

[0081] Figure 6 The electronic device shown is merely an example and should not be construed as limiting the functionality and scope of use of the embodiments of the present invention.

[0082] This invention also provides a computer-readable storage medium. The methods described above according to embodiments of the invention can be implemented in hardware or firmware, or implemented as computer code that can be recorded on a storage medium, or implemented as computer code downloaded via a network and originally stored on a remote storage medium or a non-transitory machine-readable storage medium and then stored on a local storage medium. Thus, the methods described herein can be processed by software stored on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. The storage medium can be a magnetic disk, optical disk, read-only memory, random access memory, flash memory, hard disk, or solid-state drive, etc.; further, the storage medium can also include combinations of the above types of memory. It is understood that computers, processors, microprocessor controllers, or programmable hardware include storage components capable of storing or receiving software or computer code. When the software or computer code is accessed and executed by the computer, processor, or hardware, the abnormal meeting car rental monitoring method shown in the above embodiments is implemented.

[0083] A portion of this invention can be applied as a computer program product, such as computer program instructions, which, when executed by a computer, can invoke or provide the methods and / or technical solutions according to the invention through the operation of the computer. Those skilled in the art will understand that the forms in which computer program instructions exist in a computer-readable medium include, but are not limited to, source files, executable files, installation package files, etc. Correspondingly, the ways in which computer program instructions are executed by a computer include, but are not limited to: the computer directly executing the instructions, or the computer compiling the instructions and then executing the corresponding compiled program, or the computer reading and executing the instructions, or the computer reading and installing the instructions and then executing the corresponding installed program. Here, the computer-readable medium can be any available computer-readable storage medium or communication medium accessible to a computer.

[0084] Although embodiments of the invention have been described in conjunction with the accompanying drawings, those skilled in the art can make various modifications and variations without departing from the spirit and scope of the invention, and such modifications and variations all fall within the scope defined by the appended claims.

Claims

1. A method for monitoring abnormal meeting car rentals, characterized in that, The method includes: Acquire car rental data for multimodal meeting organization; The multimodal meeting organization car rental data is subjected to feature fusion processing to obtain multimodal fusion features; The multimodal fusion features and the conference car rental compliance knowledge graph are input into a pre-built semantic understanding model so that the semantic understanding model outputs a preliminary list of anomalies.

2. The method according to claim 1, characterized in that, The method includes: The abnormal information in the preliminary abnormality list is standardized and labeled to obtain a standardized evidence package classified by abnormality. The standardized evidence package and the pre-built meeting car rental compliance report template are input into the pre-built report generation large language model, so that the report generation large language model outputs monitoring result report text data.

3. The method according to claim 1 or 2, characterized in that, The knowledge graph for conference vehicle rental compliance was constructed through the following steps: Obtain the conference car rental rules dataset; The conference car rental rule dataset is structured to obtain triplet data; A knowledge graph of meeting car rental compliance was constructed based on the triplet data.

4. The method according to claim 1 or 2, characterized in that, The steps for acquiring car rental data for multimodal conferences include: Obtain initial meeting organization and car rental related data; The initial meeting organization car rental data was sequentially cleaned and standardized to obtain multimodal meeting organization car rental data.

5. The method according to claim 1 or 2, characterized in that, The step of performing feature fusion processing on the multimodal meeting organization car rental data to obtain multimodal fused features includes: Obtain query vectors for multiple types of anomalies; Feature extraction and normalization are performed on the car rental data of each modality in the multimodal meeting organization data to obtain the feature vector of the corresponding modality; Based on the feature vectors of each modality and the query vectors of the various types of abnormal targets, the original attention score between the feature vectors of the corresponding modality and the query vectors of the various types of abnormal targets is determined; The original attention scores between the feature vectors of each modality and the query vectors of various abnormal targets are normalized to obtain the corresponding target attention weights; Multimodal fusion features are determined based on the feature vectors of multiple modalities and the target attention weights between the feature vectors of each modality and various abnormal target query vectors.

6. An abnormal meeting car rental monitoring device, characterized in that, The device includes: The acquisition module is used to acquire car rental data for multimodal conferences. The first processing module is used to perform feature fusion processing on the multimodal meeting organization car rental data to obtain multimodal fusion features; The first determining module is used to input the multimodal fusion features and the conference car rental compliance knowledge graph into a pre-built semantic understanding model, so that the semantic understanding model outputs a preliminary anomaly list.

7. The apparatus according to claim 6, characterized in that, The device further includes: The second processing module is used to standardize and annotate each abnormal information in the preliminary abnormality list to obtain a standardized evidence package classified by abnormality. The second determining module is used to input the standardized evidence package and the pre-built meeting car rental compliance report template into the pre-built report generation large language model, so that the report generation large language model outputs monitoring result report text data.

8. An electronic device, characterized in that, include: A memory and a processor are communicatively connected, the memory stores computer instructions, and the processor executes the abnormal meeting car rental monitoring method according to any one of claims 1 to 5 by executing the computer instructions.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions for causing the computer to execute the abnormal meeting car rental monitoring method according to any one of claims 1 to 5.

10. A computer program product, characterized in that, Includes computer instructions for causing a computer to execute the abnormal meeting car rental monitoring method according to any one of claims 1 to 5.

Images