Method and device for determining inconsistency based on evidence reverse reference and order relation aggregation

Abstract

The invention provides an inconsistency determination method and device based on evidence reverse reference and order relation aggregation, and relates to the field of evidence matching, and the method comprises the steps: obtaining target evidence data collected by an unmanned aerial vehicle, and extracting a textocures set containing a file identifier, interval positioning and context information from the target evidence data; performing standardized conversion on the unstructured data in the textocycles set to generate structured evidence data containing evidence ID, file identification, interval coordinates, text content and context features, and constructing a target index based on the file identification and the interval coordinates according to the structured evidence data; and matching the to-be-processed conflict text with the set of the structured evidence data based on the target index to obtain an initial matching result, and calculating the text similarity based on the initial matching result by utilizing an editing distance and a local fingerprint algorithm.

Description

Technical Field

[0001] This application relates to the field of evidence matching technology, and in particular to a method and apparatus for determining inconsistencies based on evidence inverse reference and order relation aggregation. Background Technology

[0002] Currently, in the calculation of audit conclusions and evidence matching, with the expansion of data scale and the increase in text complexity, conflict detection and conclusion generation technologies based on algorithm models have become a key support for improving audit efficiency. These technologies are widely used in scenarios such as tax audits and compliance reviews. For example, existing technologies use Natural Language Processing (NLP) to extract key information from text and identify potential conflicting conclusions, then combine this with a multi-label classification model to label the conflicting content to finally obtain the audit conclusion. However, the efficiency of generating audit conclusions using this method in existing technologies is relatively low. Summary of the Invention

[0003] The purpose of this invention is to provide a method and apparatus for determining inconsistencies based on evidence inverse reference and order relation aggregation, so as to solve the technical problem of low efficiency in generating audit conclusions.

[0004] In a first aspect, this application provides a method for determining inconsistencies based on the aggregation of evidence reciprocal reference and order relation, applied to an electronic terminal, wherein the electronic terminal is communicatively connected to a drone, and the method includes: Acquire target evidence data collected by the drone, and extract a set of text_occurrences containing file identifiers, interval positioning and context information from the target evidence data; The unstructured data in the text_occurrences set is standardized and transformed to generate structured evidence data containing evidence ID, file identifier, interval coordinates, text content and context features. A target index based on file identifier and interval coordinates is constructed based on the structured evidence data. Based on the target index, the conflicting text to be processed is matched with the set of structured evidence data to obtain an initial matching result. Based on the initial matching result, the text similarity is calculated using edit distance and local fingerprint algorithms. In combination with a preset confidence threshold, valid matching results are filtered from the initial matching result and invalid texts that are not in the set of structured evidence data are removed. In response to the number of valid texts in the valid matching results being less than a specified number, it is determined that the conflicting text lacks supporting evidence and the conflict record corresponding to the conflicting text is deleted, resulting in the filtered target conflicting text. The tags of the target conflicting text are prioritized and aggregated according to a preset order relationship to obtain the aggregation result. The highest priority inconsistency result is selected from the aggregation result as the final result, and the structured evidence data associated with the final result is written back to the result field to output a single result that is locatable and has evidence.

[0005] In one possible implementation, the drone is equipped with a video acquisition device and a positioning device; the target evidence data further includes video data acquired by the video acquisition device, a timestamp corresponding to the video data, and positioning data obtained by the positioning device; wherein, the positioning data represents the acquisition location of the video data, and the timestamp represents the acquisition time of the video data; after acquiring the target evidence data acquired by the drone, the method further includes: Filter sub-video portions of multi-party communication text from the video data; wherein, the sub-video portions correspond to multiple text generation tasks; Speech recognition is performed on the audio content in the sub-video portion to obtain speech recognition results, and personnel distance detection is performed on the image content in the sub-video portion to obtain distance detection results; Based on the speech recognition results and the division of the sub-video portion into multiple text generation task objects; the multiple text generation task objects include contract clause generation tasks, review text generation tasks, and audit text generation tasks.

[0006] In one possible implementation, after the step of dividing the sub-video portion into multiple text generation task objects based on the speech recognition result and the step of dividing the sub-video portion into multiple text generation task objects, the method further includes: For each text generation task object, based on the sub-video portion, the identities of the communication parties involved in the target text generation task object and the text processing items proposed by each identity for the target text generation task are determined through voiceprint recognition, facial recognition, and lip-reading recognition; wherein, the text processing items include text proposal items, text modification items, and text deletion items for the target text; the communication parties involved in the target text generation task object include the text proposer, text modifier, and text deleter of the target text generation task object; Based on the target timestamp and target location data of the target speech content corresponding to the target text, the identity of the communicator, the text submission items, the text deletion items, the text modification items, and the number of text modifications and the content of the modification corresponding to the text modification items, the generation process of the target text is traced to obtain the text generation evidence data of the target text.

[0007] In one possible implementation, constructing a target index based on file identifiers and interval coordinates according to the structured evidence data includes: Based on the file identifier and the interval coordinates, a target index for querying is constructed, and a set of structured evidence units corresponding to the structured evidence data is output; wherein, the set of structured evidence units is used for subsequent module calls.

[0008] In one possible implementation, the process of matching the conflicting text to be processed with the set of structured evidence data based on the target index to obtain an initial matching result, calculating text similarity based on the initial matching result using edit distance and local fingerprint algorithms, and filtering valid matching results from the initial matching result and removing invalid text not within the set of structured evidence data using a preset confidence threshold, includes: The module starts by inputting conflicting text to be processed and the set of structured evidence units, and uses a specified algorithm and information threshold to filter valid matches to obtain text matching calculation results; wherein, the specified algorithm includes edit distance algorithm and / or local fingerprint algorithm; Based on the text matching calculation results, invalid texts that are not in the set or whose similarity does not meet the specified standard are deleted, and the remaining valid texts are obtained. The conflicting text is retained when the number of remaining valid texts is greater than or equal to a specified number, and the conflicting text is deleted when the number of remaining valid texts is less than the specified number, and a set of filtered valid conflicting texts is output.

[0009] In one possible implementation, the step of using a specified algorithm and an information threshold to filter effective matches and obtain text matching calculation results includes: The text matching result is obtained by performing an effective matching calculation using a specified algorithm and an information threshold according to the following formula:

[0010] in, Indicates conflicting text With evidence text Edit distance between; and These represent the character lengths of the conflicting text and the evidence text, respectively. This represents the similarity calculated based on the local fingerprint algorithm; This represents the coefficient used to balance edit distance and fingerprint similarity; This represents the final comprehensive text similarity result in the text matching calculation.

[0011] In one possible implementation, the inconsistency results of the tags of the target conflicting text are prioritized and aggregated according to a preset order relationship to obtain an aggregation result. The highest priority inconsistency result is selected from the aggregation result as the final result, and the structured evidence data associated with the final result is written back to the result field to output a single, locatable, and verifiable result, including: By starting the module, the filtered set of valid conflicting texts is input. During the conclusion priority determination process, the target conflicting texts are marked with target priorities according to the order that error cases take precedence over questionable cases and questionable cases take precedence over correct cases. Based on the marked target priorities, the conclusion with the highest priority is selected as the final conclusion through conclusion aggregation. The final conclusion is bound to the target file identifier and target interval coordinates of the corresponding target evidence unit to obtain key evidence information; Based on the key evidence information, the final conclusion with evidence location is written to the target storage location to write back the conclusion field and output a single conclusion that the business can directly consume; wherein, the single conclusion includes traceability information.

[0012] Secondly, this application provides an inconsistency determination device based on evidence inverse reference and order relation aggregation, applied to an electronic terminal, wherein the electronic terminal is communicatively connected to a drone, and the device includes: The acquisition module is used to acquire target evidence data collected by the UAV and extract a set of text_occurrences containing file identifiers, interval positioning and context information from the target evidence data. The generation module is used to standardize and transform the unstructured data in the text_occurrences set to generate structured evidence data containing evidence ID, file identifier, interval coordinates, text content and context features, and to construct a target index based on the file identifier and interval coordinates based on the structured evidence data. The matching module is used to match the conflicting text to be processed with the set of structured evidence data based on the target index to obtain an initial matching result. Based on the initial matching result, the module calculates the text similarity using the edit distance and local fingerprint algorithms, and filters the valid matching results from the initial matching result and removes invalid text that is not in the set of structured evidence data by combining a preset confidence threshold. The aggregation module is used to respond to the fact that the number of valid texts in the valid matching results is less than a specified number, determine that the conflicting text lacks supporting evidence and delete the conflict record corresponding to the conflicting text, obtain the filtered target conflicting text, perform priority judgment and aggregation of the inconsistency results on the tags of the target conflicting text according to a preset order relationship, obtain the aggregation result, select the inconsistency result with the highest priority from the aggregation result as the final result, and write the structured evidence data associated with the final result back to the result field to output a single result that is locatable and has evidence.

[0013] Thirdly, this application also provides an electronic device, including a memory and a processor, wherein the memory stores a computer program that can run on the processor, and the processor executes the computer program to implement the method described in the first aspect above.

[0014] Fourthly, this application also provides a computer-readable storage medium storing computer-executable instructions that, when invoked and executed by a processor, cause the processor to perform the method described in the first aspect above.

[0015] This application brings the following beneficial effects: This application provides a method and apparatus for determining inconsistencies based on evidence inverse indices and order relation aggregation. It can acquire target evidence data collected by a UAV and extract a set of text_occurrences containing file identifiers, interval locations, and contextual information from the target evidence data. The unstructured data in the text_occurrences set is standardized to generate structured evidence data containing evidence IDs, file identifiers, interval coordinates, text content, and contextual features. A target index based on file identifiers and interval coordinates is constructed based on the structured evidence data. The conflicting text to be processed is matched with the set of structured evidence data based on the target index to obtain an initial matching result. Based on the initial matching result, text similarity is calculated using edit distance and local fingerprint algorithms. A preset confidence threshold is used to filter valid matching results from the initial matching result and remove invalid text not within the set of structured evidence data. In response to the number of valid texts in the valid matching results being less than a specified number, the conflicting text is determined to lack supporting evidence, and the conflict record corresponding to the conflicting text is deleted, resulting in the filtered target conflicting text. The tags of the target conflicting text are prioritized and aggregated according to a preset order relationship to obtain the aggregation result. The highest priority inconsistency result is selected from the aggregation result as the final result, and the structured evidence data associated with the final result is written back to the result field to output a single, locatable, and verifiable result. In this solution, by filtering conflicting texts through the evidence occurrence set and aggregating the tags of the remaining conflicts according to a preset order relationship, the traceability of conflicting texts and the output of a single conclusion can be achieved through evidence back-pointing verification and ordered relationship aggregation. This ensures that each conflict can be accurately located in the original text, making each conclusion locatable and verifiable. At the same time, it automatically aggregates to form a single usable conclusion, improving the efficiency of audit conclusion generation and solving the technical problem of low efficiency in audit conclusion generation.

[0016] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the specific embodiments of this application or 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 this application. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0018] Figure 1 A flowchart illustrating the inconsistency determination method based on evidence back-reference and order relation aggregation provided in this application embodiment; Figure 2 Another flowchart illustrating the inconsistency determination method based on evidence back inference and order relation aggregation provided in this application embodiment; Figure 3 A schematic diagram of the structure of each module corresponding to the inconsistency determination method based on evidence back inference and order relation aggregation provided in the embodiments of this application; Figure 4 The flowchart of text evidence extraction and structuring in the inconsistency determination method based on evidence inverse reference and order relation aggregation provided in the embodiments of this application; Figure 5 The flowchart of the inconsistency determination method based on evidence inverse reference and order relation aggregation provided in the embodiments of this application is as follows: Figure 6 The flowchart of the inconsistency determination method based on evidence reversal and order relation aggregation provided in the embodiments of this application is as follows: Figure 7 A schematic diagram of the structure of an inconsistency determination device based on evidence back-reference and order relation aggregation provided in this application embodiment; Figure 8 A schematic diagram of the structure of an electronic device provided in an embodiment of this application is shown. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0020] The terms "comprising" and "having," and any variations thereof, used in the embodiments of this application, are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that includes a series of steps or units is not limited to the steps or units listed, but may optionally include other steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products, or devices.

[0021] Currently, the core structure of audit conclusion calculation and evidence matching typically includes a text information extraction module, a conflict identification module, and a conclusion output module. The information extraction module extracts key content from the original text and forms a preliminary dataset. The conflict identification module compares the extracted content based on preset rules or machine learning algorithms to identify inconsistencies. The conclusion output module directly feeds back the identified conflicts and their corresponding tags to the business side. From a signal transmission perspective, the original text data is processed by the extraction module and then transmitted to the conflict identification module. The identification results, carrying multi-dimensional tags, are then transmitted to the output module, forming a complete technical chain. However, existing technologies have significant shortcomings: On the one hand, the conflict identification module lacks a precise correlation mechanism with the original text extraction results, often resulting in the identification of conflicting texts without corresponding locations in the original text. For example, some deep learning-based models may generate virtual conflict content not mentioned in the original text due to training data bias or semantic understanding limitations, making it impossible for investigators to trace and verify the source. On the other hand, the multi-label processing mechanism has shortcomings. For multiple labels such as "error," "questionable," and "correct" corresponding to the same conflict, there is a lack of clear aggregation rules, and only a set of multiple labels can be output. Business personnel need to manually screen and integrate these labels, which greatly reduces the direct usability of the conclusions. The root cause of these problems is that existing technologies have not established an evidence counter-inference verification mechanism and an orderly label aggregation logic, making it difficult to meet the core requirements of investigation business for the authenticity and efficiency of conclusions.

[0022] First, existing conflict identification technologies lack precise correlation verification with the original text extraction results, often making it impossible to locate and trace the source of the generated conflict text in the original text. Investigators must spend a significant amount of extra time verifying the authenticity of the conflict, which is not only inefficient but also increases labor costs. Second, there is a lack of effective aggregation rules for multiple labels of the same conflict; only a set of multiple labels is output, requiring manual screening and integration to form a usable conclusion, further extending the business cycle. Therefore, the current approach to generating investigation conclusions is inefficient.

[0023] Based on this, the embodiments of this application provide a method and apparatus for determining inconsistencies based on evidence back inference and order relation aggregation, which can solve technical problems such as low efficiency in generating audit conclusions.

[0024] The embodiments of the present invention will be further described below with reference to the accompanying drawings.

[0025] Figure 1 This is a flowchart illustrating a method for determining inconsistencies based on evidence reciprocal reference and order relation aggregation, provided in an embodiment of this application. The method is applied to an electronic terminal, which is communicatively connected to a drone. Figure 1 As shown, the method includes: Step S110: Obtain target evidence data collected by the drone, and extract the text_occurrences set containing file identifiers, interval positioning and context information from the target evidence data.

[0026] In one possible implementation, after acquiring the target evidence data collected by the drone, such as Figure 2 and Figure 4 As shown, the text_occurrences set, which contains file identifiers, range locations, and context information, is read from the extraction stage.

[0027] Step S120: Standardize and transform the unstructured data in the text_occurrences set to generate structured evidence data containing evidence ID, file identifier, interval coordinates, text content and context features. Construct a target index based on file identifier and interval coordinates based on the structured evidence data.

[0028] As one possible implementation method, such as Figure 4 As shown, constructing a target index based on file identifiers and interval coordinates based on structured evidence data can specifically include the following steps: constructing a target index for querying based on file identifiers and interval coordinates, and outputting a set of structured evidence units corresponding to the structured evidence data; wherein, the set of structured evidence units is used for subsequent module calls.

[0029] For example, such as Figure 3 and Figure 4 As shown, the unstructured data in the set is standardized and transformed to generate structured evidence units containing evidence ID, file identifier, interval coordinates, text content and contextual features. An index based on file identifier and interval information is constructed to provide traceable basic data support for subsequent conflict text matching.

[0030] Step S130: Based on the target index, the conflict text to be processed is matched with the set of structured evidence data to obtain the initial matching result. Based on the initial matching result, the text similarity is calculated using the edit distance and local fingerprint algorithm. Combined with the preset confidence threshold, the valid matching results are filtered from the initial matching results and invalid texts that are not in the set of structured evidence data are removed.

[0031] In one possible implementation, the conflicting text to be processed is matched against a set of structured evidence data based on the target index to obtain initial matching results. Based on these initial matching results, text similarity is calculated using edit distance and local fingerprint algorithms. Then, a preset confidence threshold is used to filter valid matching results from the initial matching results and remove invalid text not included in the set of structured evidence data. Figure 5 As shown, the specific steps may include the following: The module starts by inputting conflicting text and a set of structured evidence units, and uses a specified algorithm and information threshold to filter valid matches to obtain text matching calculation results; the specified algorithm includes edit distance algorithm and / or local fingerprint algorithm. Based on the text matching calculation results, invalid texts that are not in the set or whose similarity does not meet the specified standard are deleted, and the remaining valid texts are obtained. When the number of remaining valid texts is greater than or equal to the specified number, conflicting texts are retained, and when the number of remaining valid texts is less than the specified number, conflicting texts are deleted, and the filtered set of valid conflicting texts is output.

[0032] like Figure 2 , Figure 3 and Figure 5 As shown, the conflict text to be processed is matched with the set of structured evidence units. The text similarity is calculated using the edit distance or local fingerprint algorithm and the effective matching results are filtered in combination with the preset confidence threshold. Invalid texts not in the set are removed. If the number of valid texts remaining after filtering is less than 2, it is determined that the conflict lacks sufficient evidence support and the conflict record is directly deleted.

[0033] In one optional implementation, a specified algorithm and an information threshold are used to filter valid matches to obtain the text matching calculation result, which may specifically include the following steps: The text matching result is obtained by performing an effective matching calculation using a specified algorithm and an information threshold according to the following formula:

[0034] in, Indicates conflicting text With evidence text Edit distance between; and These represent the character lengths of the conflict text and the evidence text, respectively. This represents the similarity calculated based on the local fingerprint algorithm; This represents the coefficient used to balance edit distance and fingerprint similarity; This represents the final comprehensive text similarity result in the text matching calculation.

[0035] In this embodiment of the application, by performing effective matching calculations using the above calculation formula, the accuracy of the final comprehensive text similarity result in the obtained text matching calculation results can be greatly improved.

[0036] In step S140, in response to the number of valid texts in the valid matching results being less than a specified number, it is determined that the conflicting text lacks supporting evidence and the conflicting record corresponding to the conflicting text is deleted, thus obtaining the filtered target conflicting text. The tags of the target conflicting text are prioritized and aggregated according to a preset order relationship to obtain the aggregation result. The highest priority inconsistency result is selected from the aggregation result as the final result, and the structured evidence data associated with the final result is written back to the result field to output a single result that is locatable and has evidence.

[0037] As an optional implementation, the tags of the target conflicting text are prioritized and aggregated according to a preset order to obtain an aggregated result. The highest priority inconsistency result is selected from the aggregated results as the final result, and the corresponding structured evidence data associated with the final result is written back to the result field to output a single, locatable, and verifiable result. Figure 6 As shown, the specific steps may include the following: The module starts by inputting a set of filtered and valid conflicting texts. During the conclusion priority determination process, the conflicting texts are marked with target priorities according to the order that erroneous cases take precedence over questionable cases and questionable cases take precedence over correct cases. Based on the marked target priorities, the conclusion with the highest priority is selected as the final conclusion through conclusion aggregation. The final conclusion is bound to the target file identifier and target interval coordinates of the corresponding target evidence unit to obtain key evidence information; based on the key evidence information, the final conclusion with evidence location is written to the target storage location to write back the conclusion field and output a single conclusion that can be directly consumed by the business; the single conclusion includes traceability information.

[0038] like Figure 2 , Figure 3 and Figure 6 As shown, by filtering and retaining conflicting texts, the conclusion priority is determined and aggregated according to the preset order relationship of "error > doubt > correct". The conclusion with the highest priority is selected as the final result, and the corresponding evidence unit information (such as file identifier and interval coordinates) associated with the conclusion is written back to the conclusion field, so as to realize the output of a single conclusion that can be located and has evidence, and meet the direct consumption needs of the business.

[0039] In this embodiment, conflicting texts are filtered by evidence occurrence set and the remaining conflicting tags are aggregated according to a preset order relationship. Through evidence back-point verification and ordered relationship aggregation, the source of conflicting texts can be traced and a single conclusion can be output, ensuring that each conflict can be accurately located in the original text and that each conclusion is locatable and evidenced. At the same time, it automatically aggregates to form a single usable conclusion, improving the efficiency and usability of audit conclusion generation and solving the technical problem of low efficiency in audit conclusion generation.

[0040] In some embodiments, the drone is equipped with a video acquisition device and a positioning device; the target evidence data further includes video data acquired by the video acquisition device, a timestamp corresponding to the video data, and positioning data obtained by the positioning device; wherein, the positioning data represents the acquisition location of the video data, and the timestamp represents the acquisition time of the video data; after acquiring the target evidence data acquired by the drone, the method may further include the following steps: The process involves filtering sub-video segments of multi-party communication text from video data; each sub-video segment corresponds to multiple text generation tasks; speech recognition is performed on the audio content in the sub-video segments to obtain speech recognition results; and personnel distance detection is performed on the image content in the sub-video segments to obtain distance detection results. Based on the speech recognition results, the sub-video portion is divided into multiple text generation task objects; these multiple text generation task objects include contract clause generation task, review text generation task, and audit text generation task.

[0041] In this embodiment, video, audio, spatial location (location data), and temporal information (timestamps) are organically combined to form an evidence data body with spatiotemporal context, enhancing the credibility and completeness of the evidence. By identifying "multi-party communication" scenarios, key sub-video segments are automatically extracted, and different types of text generation tasks (such as contract terms, review opinions, and audit reports) are divided accordingly, achieving end-to-end automated processing from raw audio and video to professional documents. Therefore, this embodiment realizes intelligent semantic analysis and task-oriented automatic generation of structured text from multimodal evidence data (video, audio, location, and timestamps) collected by UAVs, thereby efficiently and accurately supporting the construction of evidence chains and document generation in legal, compliance, or administrative audit scenarios. Based on UAV multimodal perception data, automated, structured, and traceable intelligent text generation for specific business scenarios (such as contracts, reviews, and audits) is achieved, significantly improving the efficiency of evidence utilization and decision support capabilities.

[0042] In some embodiments, after dividing the sub-video portion into multiple text generation task objects based on the speech recognition results, the method may further include the following steps: For each text generation task object, based on the sub-video portion, the identities of the communication parties involved in the target text generation task object and the text processing items proposed by each identity for the target text generation task are determined through voiceprint recognition, facial recognition, and lip-reading recognition. Among them, the text processing items include text proposal items, text modification items, and text deletion items of the target text; the communication parties involved in the target text generation task object include the text proposer, text modifier, and text deleter of the target text generation task object. Based on the target timestamp and target location data of the target speech content corresponding to the target text, the identity of the communicator, text submission items, text deletion items, text modification items, and the number of text modifications and the content of the modification corresponding to the text modification items, the generation process of the target text is traced to obtain text generation evidence data of the target text.

[0043] By fusing voiceprint recognition, facial recognition, and lip-reading recognition—three biometric features—the system accurately identifies the real communicator for each voice or text operation (submission, modification, deletion), effectively solving the attribution problem of "who said what" and "who modified which part" in multi-person dialogue scenarios. By structuring text processing behaviors into three categories—submission, modification, and deletion—and recording the number of modifications and specific content, the text evolution process becomes traceable, replayable, and comparable, avoiding version confusion or unclear responsibility. It provides high-precision, verifiable, and dynamic traceability of the entire process of collaborative text generation or modification by multiple parties (such as contracts, audit opinions, and audit documents), constructing a complete text generation evidence chain with identity authentication, behavior recording, and spatiotemporal anchoring. By combining timestamps and location data, each text operation is bound to a specific time point and physical location (such as the coordinates of a drone shot), forming irrefutable spatiotemporal contextual evidence, significantly improving the evidentiary value in legal or administrative scenarios. The final output of this application's embodiments is not only the target text, but also a structured evidence package containing five dimensions of information: identity, behavior, content, time, and location. This package can be used in demanding scenarios such as compliance review, dispute evidence presentation, and liability determination. Therefore, it achieves full-element, full-chain, and verifiable intelligent traceability of the multi-party collaborative text generation process, providing dynamic evidence support with judicial-grade credibility for key scenarios such as law, compliance, and investigation.

[0044] Figure 7 A schematic diagram of a device for determining inconsistencies based on evidence reciprocity and order relation aggregation is provided. This device can be applied to an electronic terminal that is communicatively connected to a drone. Figure 7 As shown, the inconsistency determination device 700 based on evidence reciprocity and order relation aggregation includes: The acquisition module 701 is used to acquire target evidence data collected by the UAV, and extract the te7t_occurrences set containing file identifier, interval positioning and context information from the target evidence data; The generation module 702 is used to standardize and transform the unstructured data in the te7t_occurrences set to generate structured evidence data containing evidence ID, file identifier, interval coordinates, text content and context features, and to construct a target index based on the file identifier and interval coordinates based on the structured evidence data. The matching module 703 is used to match the conflicting text to be processed with the set of structured evidence data based on the target index to obtain an initial matching result. Based on the initial matching result, the text similarity is calculated using the edit distance and local fingerprint algorithm. The module also filters valid matching results from the initial matching result and removes invalid text that is not in the set of structured evidence data, in conjunction with a preset confidence threshold. The aggregation module 704 is used to respond to the fact that the number of valid texts in the valid matching results is less than a specified number, determine that the conflicting text lacks supporting evidence and delete the conflict record corresponding to the conflicting text, obtain the filtered target conflicting text, perform priority judgment and aggregation of the inconsistency results on the tags of the target conflicting text according to a preset order relationship, obtain the aggregation result, select the inconsistency result with the highest priority from the aggregation result as the final result, and write the structured evidence data associated with the final result back to the result field to output a single result that is locatable and has evidence.

[0045] The inconsistency determination apparatus based on evidence back-reference and order relation aggregation provided in this application has the same technical features as the inconsistency determination method based on evidence back-reference and order relation aggregation provided in the above embodiments, so it can also solve the same technical problems and achieve the same technical effects.

[0046] An electronic device provided in this application embodiment, such as Figure 8 As shown, the electronic device 800 includes a processor 802 and a memory 801. The memory stores a computer program that can run on the processor. When the processor executes the computer program, it implements the steps of the method provided in the above embodiments.

[0047] See Figure 8 The electronic device also includes a bus 803 and a communication interface 804. The processor 802, the communication interface 804 and the memory 801 are connected through the bus 803. The processor 802 is used to execute executable modules, such as computer programs, stored in the memory 801.

[0048] The memory 801 may include high-speed random access memory (RAM) or non-volatile memory, such as at least one disk storage device. Communication between this system network element and at least one other network element is achieved through at least one communication interface 804 (which can be wired or wireless), such as the Internet, wide area network, local area network, or metropolitan area network.

[0049] Bus 803 can be an ISA bus, PCI bus, or EISA bus, etc. The bus can be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 8 The symbol is represented by a single double-headed arrow, but this does not mean that there is only one bus or one type of bus.

[0050] The memory 801 is used to store programs. After receiving an execution instruction, the processor 802 executes the program. The method executed by the apparatus defined by the process disclosed in any of the preceding embodiments of this application can be applied to the processor 802 or implemented by the processor 802.

[0051] The processor 802 may be an integrated circuit chip with signal processing capabilities. In implementation, each step of the above method can be completed by the integrated logic circuitry in the hardware of the processor 802 or by instructions in software form. The processor 802 may be a general-purpose processor, including a Central Processing Unit (CPU), a Network Processor (NP), etc.; it may also be a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field-Programmable Gate Array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this application. The general-purpose processor may be a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of this application can be directly manifested as execution by a hardware decoding processor, or execution by a combination of hardware and software modules in the decoding processor. The software module can reside in a mature storage medium in the field, such as random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, or registers. This storage medium is located in memory 801, and processor 802 reads the information from memory 801 and, in conjunction with its hardware, completes the steps of the above method.

[0052] Corresponding to the above-described method for determining inconsistencies based on evidence back inference and order relation aggregation, this application also provides a computer-readable storage medium storing computer-executable instructions. When the computer-executable instructions are invoked and executed by a processor, the computer-executable instructions cause the processor to perform the steps of the above-described method for determining inconsistencies based on evidence back inference and order relation aggregation.

[0053] The inconsistency determination device based on evidence back-pointing and order relation aggregation provided in this application embodiment can be specific hardware on the device or software or firmware installed on the device. The implementation principle and technical effects of the device provided in this application embodiment are the same as those in the foregoing method embodiments. For the sake of brevity, any parts not mentioned in the device embodiment can be referred to the corresponding content in the foregoing method embodiments. Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can all be referred to the corresponding processes in the above method embodiments, and will not be repeated here.

[0054] In the embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. The apparatus embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. Furthermore, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Additionally, the displayed or discussed mutual couplings, direct couplings, or communication connections may be through some communication interfaces; indirect couplings or communication connections between devices or units may be electrical, mechanical, or other forms.

[0055] For example, the flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of apparatus, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than those marked in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram and / or flowchart, and combinations of blocks in block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or action, or using a combination of dedicated hardware and computer instructions.

[0056] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0057] In addition, the functional units in the embodiments provided in this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0058] If the aforementioned function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the inconsistency determination method based on evidence back-pointing and order relation aggregation described in various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0059] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. In addition, the terms "first", "second", "third", etc. are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0060] Finally, it should be noted that the above-described embodiments are merely specific implementations of this application, used to illustrate the technical solutions of this application, and not to limit them. The protection scope of this application is not limited thereto. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that any person skilled in the art can still modify or easily conceive of changes to the technical solutions described in the foregoing embodiments, or make equivalent substitutions for some of the technical features, within the scope of the technology disclosed in this application; and these modifications, changes, or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application. All should be covered within the protection scope of this application. Therefore, the protection scope of this application should be determined by the protection scope of the claims.

Claims

1. A method for determining inconsistencies based on evidence reciprocity and order relation aggregation, characterized in that, The method is applied to an electronic terminal that is communicatively connected to a drone, and includes: Acquire target evidence data collected by the drone, and extract a set of text_occurrences containing file identifiers, interval positioning and context information from the target evidence data; The unstructured data in the text_occurrences set is standardized and transformed to generate structured evidence data containing evidence ID, file identifier, interval coordinates, text content and context features. A target index based on file identifier and interval coordinates is constructed based on the structured evidence data. Based on the target index, the conflicting text to be processed is matched with the set of structured evidence data to obtain an initial matching result. Based on the initial matching result, the text similarity is calculated using edit distance and local fingerprint algorithms. In combination with a preset confidence threshold, valid matching results are filtered from the initial matching result and invalid texts that are not in the set of structured evidence data are removed. In response to the number of valid texts in the valid matching results being less than a specified number, it is determined that the conflicting text lacks supporting evidence and the conflict record corresponding to the conflicting text is deleted, resulting in the filtered target conflicting text. The tags of the target conflicting text are prioritized and aggregated according to a preset order relationship to obtain the aggregation result. The highest priority inconsistency result is selected from the aggregation result as the final result, and the structured evidence data associated with the final result is written back to the result field to output a single result that is locatable and has evidence.

2. The method according to claim 1, characterized in that, The drone is equipped with a video acquisition device and a positioning device; the target evidence data also includes video data acquired by the video acquisition device, a timestamp corresponding to the video data, and positioning data obtained by the positioning device; wherein, the positioning data indicates the acquisition location of the video data, and the timestamp indicates the acquisition time of the video data; after acquiring the target evidence data acquired by the drone, the method further includes: Filter sub-video portions of multi-party communication text from the video data; wherein, the sub-video portions correspond to multiple text generation tasks; Speech recognition is performed on the audio content in the sub-video portion to obtain speech recognition results, and personnel distance detection is performed on the image content in the sub-video portion to obtain distance detection results; Based on the speech recognition results and the division of the sub-video portion into multiple text generation task objects; the multiple text generation task objects include contract clause generation tasks, review text generation tasks, and audit text generation tasks.

3. The method according to claim 2, characterized in that, After the step of dividing the sub-video portion into multiple text generation task objects based on the speech recognition result, the method further includes: For each text generation task object, based on the sub-video portion, the identities of the communication parties involved in the target text generation task object and the text processing items proposed by each identity for the target text generation task are determined through voiceprint recognition, facial recognition, and lip-reading recognition; wherein, the text processing items include text proposal items, text modification items, and text deletion items for the target text; the communication parties involved in the target text generation task object include the text proposer, text modifier, and text deleter of the target text generation task object; Based on the target timestamp and target location data of the target speech content corresponding to the target text, the identity of the communicator, the text submission items, the text deletion items, the text modification items, and the number of text modifications and the content of the modification corresponding to the text modification items, the generation process of the target text is traced to obtain the text generation evidence data of the target text.

4. The method according to claim 1, characterized in that, The step of constructing a target index based on file identifiers and interval coordinates according to the structured evidence data includes: Based on the file identifier and the interval coordinates, a target index for querying is constructed, and a set of structured evidence units corresponding to the structured evidence data is output; wherein, the set of structured evidence units is used for subsequent module calls.

5. The method according to claim 4, characterized in that, The process of matching the conflicting text to be processed with the set of structured evidence data based on the target index to obtain an initial matching result, calculating text similarity based on the initial matching result using edit distance and local fingerprint algorithms, and filtering valid matching results from the initial matching result and removing invalid text not in the set of structured evidence data based on a preset confidence threshold, includes: The module starts by inputting conflicting text to be processed and the set of structured evidence units, and uses a specified algorithm and information threshold to filter valid matches to obtain text matching calculation results; wherein, the specified algorithm includes edit distance algorithm and / or local fingerprint algorithm; Based on the text matching calculation results, invalid texts that are not in the set or whose similarity does not meet the specified standard are deleted, and the remaining valid texts are obtained. The conflicting text is retained when the number of remaining valid texts is greater than or equal to a specified number, and the conflicting text is deleted when the number of remaining valid texts is less than the specified number, and a set of filtered valid conflicting texts is output.

6. The method according to claim 5, characterized in that, The process of using a specified algorithm and information thresholds to filter effective matches and obtain text matching calculation results includes: The text matching result is obtained by performing an effective matching calculation using a specified algorithm and an information threshold according to the following formula: in, Indicates conflicting text With evidence text Edit distance between; and These represent the character lengths of the conflicting text and the evidence text, respectively. This represents the similarity calculated based on the local fingerprint algorithm; This represents the coefficient used to balance edit distance and fingerprint similarity; This represents the final comprehensive text similarity result in the text matching calculation.

7. The method according to claim 5, characterized in that, The process involves prioritizing and aggregating inconsistencies in the tags of the target conflicting text according to a preset order, obtaining an aggregation result, selecting the highest-priority inconsistency result from the aggregation result as the final result, and writing the corresponding structured evidence data associated with the final result back to the result field to output a single, locatable, and verifiable result, including: By starting the module, the filtered set of valid conflicting texts is input. During the conclusion priority determination process, the target conflicting texts are marked with target priorities according to the order that error cases take precedence over questionable cases and questionable cases take precedence over correct cases. Based on the marked target priorities, the conclusion with the highest priority is selected as the final conclusion through conclusion aggregation. The final conclusion is bound to the target file identifier and target interval coordinates of the corresponding target evidence unit to obtain key evidence information; Based on the key evidence information, the final conclusion with evidence location is written to the target storage location to write back the conclusion field and output a single conclusion that the business can directly consume; wherein, the single conclusion includes traceability information.

8. A device for determining inconsistencies based on evidence inverse reference and order relation aggregation, characterized in that, The device is applied to an electronic terminal that is communicatively connected to a drone, and includes: The acquisition module is used to acquire target evidence data collected by the UAV and extract a set of text_occurrences containing file identifiers, interval positioning and context information from the target evidence data. The generation module is used to standardize and transform the unstructured data in the text_occurrences set to generate structured evidence data containing evidence ID, file identifier, interval coordinates, text content and context features, and to construct a target index based on the file identifier and interval coordinates based on the structured evidence data. The matching module is used to match the conflicting text to be processed with the set of structured evidence data based on the target index to obtain an initial matching result. Based on the initial matching result, the module calculates the text similarity using the edit distance and local fingerprint algorithms, and filters the valid matching results from the initial matching result and removes invalid text that is not in the set of structured evidence data by combining a preset confidence threshold. The aggregation module is used to respond to the fact that the number of valid texts in the valid matching results is less than a specified number, determine that the conflicting text lacks supporting evidence and delete the conflict record corresponding to the conflicting text, obtain the filtered target conflicting text, perform priority judgment and aggregation of the inconsistency results on the tags of the target conflicting text according to a preset order relationship, obtain the aggregation result, select the inconsistency result with the highest priority from the aggregation result as the final result, and write the structured evidence data associated with the final result back to the result field to output a single result that is locatable and has evidence.

9. An electronic device comprising a memory and a processor, wherein the memory stores a computer program executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions that, when invoked and executed by a processor, cause the processor to perform the method according to any one of claims 1 to 7.

Images