Multi-modal evidence chain analysis method and system

Abstract

The invention relates to a multi-modal evidence chain analysis method and system, and the method comprises the steps: determining a case description word set based on a case basic report, querying an associated evidence event set in a multi-modal evidence event set based on the case description word set, carrying out the pairwise combination correlation analysis of the associated evidence events in the associated evidence event set, and obtaining a multi-modal evidence chain analysis result; identifying a high association value set in the content association value set, identifying a center association value set corresponding to the center evidence event in the high association value set, sequentially identifying a high association event of each center association value in the center association value set to obtain a target evidence chain event set, and summarizing the target evidence chain event set. And obtaining a plurality of duplicate removal evidence chain event sets. According to the invention, the efficiency and accuracy of multi-modal evidence chain analysis can be improved.

Description

Technical Field

[0001] The present invention relates to the technical field of data analysis, and in particular to a multimodal evidence chain analysis method and system. Background Art

[0002] With the rapid development of information technology, especially the widespread popularization of the Internet and mobile communication technology, the types and sources of data have become increasingly diverse when conducting evidence data analysis in key fields such as law, security, and medicine. Therefore, in today's data analysis field, it is particularly important to analyze multimodal evidence chains and construct a comprehensive multimodal evidence chain.

[0003] At present, the analysis of multimodal evidence chains mainly relies on manual processing by relevant case personnel. This method not only consumes a lot of human resources, but also leads to low analysis efficiency due to the limitations of manual operation. In addition, human subjective judgment will also affect the accuracy of the analysis results. Summary of the invention

[0004] The present invention provides a multimodal evidence chain analysis method and system, the main purpose of which is to improve the efficiency and accuracy of multimodal evidence chain analysis.

[0005] To achieve the above purpose, the present invention provides a multimodal evidence chain analysis method, comprising:

[0006] Receiving an evidence chain analysis instruction, and obtaining a multimodal evidence event set based on the evidence chain analysis instruction, wherein the multimodal evidence event set includes text modal information, image modal information, and video modal information of the case involved, and the multimodal evidence event set is provided by relevant case personnel;

[0007] Based on a preset basic case report, determine a case description word set, and based on the case description word set, query a related evidence event set in the multimodal evidence event set;

[0008] Performing pairwise combination correlation analysis on the associated evidence events in the associated evidence event set to obtain a content correlation value set, wherein each content correlation value in the content correlation value set corresponds to two associated evidence events in the associated evidence event set;

[0009] According to a preset standard association value, identifying a highly associated value set in a content association value set, wherein the highly associated value in the highly associated value set is greater than or equal to the standard association value;

[0010] Extracting central evidence events in sequence from the associated evidence event set, identifying a central associated value set corresponding to the central evidence event in the highly associated value set, wherein the central associated value set contains multiple central associated values ​​or the central associated value set is an empty set;

[0011] If the central correlation value set is not an empty set, then sequentially identifying highly correlated events of each central correlation value in the central correlation value set to obtain a highly correlated event set, wherein the highly correlated event is another correlated evidence event corresponding to the central correlation value except the central evidence event;

[0012] Supplement the central evidence event to the highly correlated event set to obtain a target evidence chain event set, and summarize the target evidence chain event sets corresponding to each central evidence event to obtain multiple target evidence chain event sets;

[0013] Duplicate event sets are eliminated from the multiple target evidence chain event sets to obtain multiple deduplicated evidence chain event sets, and multimodal evidence chain analysis is completed based on the multiple deduplicated evidence chain event sets.

[0014] Optionally, the determining of the case description word set based on a preset case basic report includes:

[0015] Performing Chinese word segmentation on the basic case report to obtain an original descriptive word set, wherein the tool for performing Chinese word segmentation is an online word segmentation database;

[0016] Counting the original description frequency of each original description word in the original description word set to obtain an original description frequency set, and counting the common description frequency of each original description word in the original description word set in a preset common description word library to obtain a common description frequency set;

[0017] Extract target descriptive words in the original descriptive word set one by one;

[0018] According to the original description frequency set and the common description frequency set, the target description word is analyzed for its semantic keyness to obtain the description key value;

[0019] Determine whether the description key value is greater than a preset standard key value;

[0020] If the description key value is greater than the standard key value, the target description word is recorded as the case description word;

[0021] The case description words are summarized to obtain a case description word set.

[0022] Optionally, the performing a semantic key analysis on the target description word to obtain a description key value includes:

[0023] Identifying the target description frequency corresponding to the target descriptor in the original description frequency set, and identifying the target common frequency corresponding to the target descriptor in the common description frequency set;

[0024] Based on the target description frequency, target common frequency, original description frequency set and common description frequency set, the description key value is calculated using the following formula:

[0025]

[0026] Among them, K represents the description key value, N 1 Indicates the target description frequency, n 1 represents the number of original description frequencies in the original description frequency set, represents the i-th original description frequency in the original description frequency set, ln(*) represents the logarithmic function, N 2 Indicates the common frequency of the target, n 2 Indicates the number of common description frequencies in the common description frequency set, Represents the jth common description frequency in the common description frequency set.

[0027] Optionally, the step of searching for a related evidence event set in the multimodal evidence event set based on the case description word set includes:

[0028] Extracting multimodal evidence events in sequence from the multimodal evidence event set;

[0029] Identifying event description text in the multimodal evidence event, wherein the event description text comes from descriptive text information, image description text information, and video description text information in the multimodal evidence event;

[0030] Extracting case description words from the case description word set in turn, and determining whether the case description words appear in the event description text;

[0031] If the case description word appears in the event description text, the case description word is recorded as a related description word, and the related description frequency of the related description word appearing in the event description text is counted;

[0032] Summarizing the associated description words and the corresponding associated description frequencies respectively to obtain an associated description word set and an associated description frequency set, and identifying the number of associated words in the associated description word set;

[0033] According to the association description frequency set and the number of associated words, the event association value is calculated using the following formula:

[0034]

[0035] Among them, G represents the event association value, S represents the number of associated words, and m represents the number of associated description frequencies in the associated description frequency set. represents the kth associated description frequency in the associated description frequency set;

[0036] If the event association value is greater than a preset minimum association value, the multimodal evidence event is recorded as an associated evidence event;

[0037] The associated evidence events are summarized to obtain an associated evidence event set.

[0038] Optionally, performing pairwise combination correlation analysis on the associated evidence events in the associated evidence event set to obtain a content correlation value set includes:

[0039] Randomly extracting a first associated evidence event from the associated evidence event set, and removing the first associated evidence event from the associated evidence event set to obtain a removed evidence event set;

[0040] Sequentially extracting second related evidence events from the eliminated evidence event set;

[0041] Performing event correlation analysis on the first correlation evidence event and the extracted second correlation evidence event to obtain a content correlation value;

[0042] Summarize the content association values ​​to obtain a unit content association value set, and use the eliminated evidence event set as the associated evidence event set, return to the step of randomly extracting the first associated evidence event in the associated evidence event set, until the eliminated evidence event set is an empty set, and summarize the unit content association value set to obtain a content association value set.

[0043] Optionally, performing event correlation analysis on the first correlation evidence event and the second correlation evidence event to obtain a content correlation value includes:

[0044] Respectively identifying a first occurrence time and a first end time of the first associated evidence event, and a second occurrence time and a second end time of the second associated evidence event;

[0045] Based on the first occurrence time and the first end time, a first duration is calculated, and based on the second occurrence time and the second end time, a second duration is calculated;

[0046] Calculating a time association value according to the first occurrence time, the first end time, the second occurrence time, the second end time, the first duration, and the second duration;

[0047] Performing spatial correlation analysis on the first correlation evidence event and the second correlation evidence event to obtain a spatial correlation value;

[0048] According to the time association value and the space association value, the content association value is calculated using the following formula:

[0049]

[0050] Among them, G n Represents the content association value, Gtime represents the time-related value, G space Represents a spatially associated value.

[0051] Optionally, the calculating time-related value includes:

[0052] The minimum disturbance duration is set, and based on the minimum disturbance duration, the occurrence correlation factor, the end correlation factor and the duration factor are calculated using the first occurrence time, the second occurrence time, the first end time, the second end time, the first duration and the second duration using the following formula:

[0053]

[0054]

[0055] Among them, P start represents the correlation factor, |*| represents the absolute value symbol, T 1k represents the first occurrence time, T 2k represents the second occurrence time, Δt b represents the minimum disturbance duration, P end represents the end correlation factor, T 1s represents the first end time, T 2s represents the second end time, P during represents the duration factor, t 1 Indicates the first duration, t 2 Indicates the second duration;

[0056] The duration fluctuation factor of the minimum disturbance duration is set, and the time correlation value is calculated according to the occurrence correlation factor, the end correlation factor, the duration factor and the duration fluctuation factor, wherein the duration fluctuation factor and the time correlation value are respectively expressed as:

[0057]

[0058] G time =ΔP×(P start +P end +P during )

[0059] Among them, ΔP represents the duration fluctuation factor.

[0060] Optionally, performing spatial correlation analysis on the first correlation evidence event and the second correlation evidence event to obtain a spatial correlation value includes:

[0061] Obtaining a first associated text of the first associated evidence event and a second associated text of the second associated evidence event, and performing word segmentation on the first associated text and the second associated text respectively to obtain a first associated word set and a second associated word set;

[0062] Comparing the first associated word set with the second associated word set to obtain a common associated word set, and obtaining the number of common associated words in the common associated word set, wherein the common associated words in the common associated word set are associated words that appear in both the first associated word set and the second associated word set;

[0063] Determining whether the number of common associated words is less than a preset minimum number of associated words;

[0064] If the number of common related words is less than the minimum number of related words, the spatial relatedness value is recorded as zero;

[0065] If the number of common associated words is not less than the minimum number of associated words, the spatial association value is calculated based on the common associated word set.

[0066] Optionally, the calculating of the spatial association value based on the common association word set includes:

[0067] Extracting common related words in the common related word set in sequence;

[0068] respectively counting the first common frequency of the common associated words in the first associated word set and the second common frequency in the second associated word set;

[0069] The first public frequencies and the second public frequencies are respectively aggregated to obtain a first public frequency set and a second public frequency set;

[0070] Based on the first common frequency set and the second common frequency set, a spatial correlation value is calculated, wherein the spatial correlation value is expressed as:

[0071]

[0072] Among them, G space represents the spatial association value, U represents the number of the first common frequency in the first common frequency set or the number of the second common frequency in the second common frequency set, represents the e-th first common frequency in the first common frequency set, Represents the e-th common frequency in the second common frequency set.

[0073] To achieve the above object, the present invention also provides a multimodal evidence chain analysis system, comprising:

[0074] An associated evidence query module is used to receive an evidence chain analysis instruction, and obtain a multimodal evidence event set based on the evidence chain analysis instruction, wherein the multimodal evidence event set includes text modal information, image modal information, and video modal information of the case involved, and the multimodal evidence event set is provided by relevant case personnel, and based on a preset case basic report, a case description word set is determined, and based on the case description word set, an associated evidence event set is queried in the multimodal evidence event set;

[0075] An event association analysis module is used to perform a pairwise combination association analysis on the associated evidence events in the associated evidence event set to obtain a content association value set, wherein each content association value in the content association value set corresponds to two associated evidence events in the associated evidence event set, and to identify a highly associated value set in the content association value set according to a preset standard association value, wherein the highly associated value in the highly associated value set is greater than or equal to the standard association value;

[0076] A highly correlated identification module is used to sequentially extract central evidence events from a set of correlated evidence events, and identify a central correlated value set corresponding to the central evidence event in the highly correlated value set, wherein the central correlated value set contains multiple central correlated values ​​or the central correlated value set is an empty set. If the central correlated value set is not an empty set, then sequentially identify highly correlated events of each central correlated value in the central correlated value set to obtain a highly correlated event set, wherein the highly correlated event is another correlated evidence event corresponding to the central correlated value other than the central evidence event;

[0077] The target evidence integration module is used to supplement the central evidence event to a highly correlated event set to obtain a target evidence chain event set, summarize the target evidence chain event set corresponding to each central evidence event to obtain multiple target evidence chain event sets, and eliminate duplicate event sets from the multiple target evidence chain event sets to obtain multiple deduplicated evidence chain event sets.

[0078] In order to solve the above problem, the present invention further provides an electronic device, the electronic device comprising:

[0079] A memory storing at least one instruction;

[0080] The processor executes the instructions stored in the memory to implement the multimodal evidence chain analysis method described above.

[0081] In order to solve the above problems, the present invention also provides a computer-readable storage medium, in which at least one instruction is stored. The at least one instruction is executed by a processor in an electronic device to implement the multimodal evidence chain analysis method described above.

[0082] In order to solve the problems described in the background technology, the present invention first determines a case description word set based on the basic case report, and based on the case description word set, searches for a related evidence event set in a multimodal evidence event set. In this step, by determining the case description word set, evidence directly related to the case can be more accurately screened out from the multimodal evidence event set, thereby improving the efficiency and accuracy of evidence analysis. Next, a pairwise combination correlation analysis is performed on the related evidence events in the related evidence event set to obtain a content correlation value set. In this step, by quantifying the correlation between evidence events, it is possible to identify which evidence events have a strong connection, thereby providing an important reference for constructing an evidence chain. Among them, the correlation analysis includes temporal correlation analysis and spatial correlation analysis, thereby further improving the correlation. The accuracy of the multi-modal evidence chain analysis is improved by filtering out evidence with weak correlation by setting a threshold, and focusing on analyzing evidence with high correlation, thereby improving the efficiency of the analysis and the reliability of the chain of evidence. Finally, a target evidence chain event set is created for each central evidence event to obtain multiple target evidence chain event sets, and duplicate event sets in multiple target evidence chain event sets are eliminated to obtain multiple deduplicated evidence chain event sets, wherein the construction of the target evidence chain event set can more clearly display the logical relationship between the evidence, which enhances the reliability and persuasiveness of the evidence. At the same time, for complex cases involving a large amount of evidence and information, the target evidence chain event set can help analysts quickly locate key evidence, thereby reducing the complexity of the case and improving the efficiency of the evidence chain analysis. Therefore, the present invention can improve the efficiency and accuracy of multi-modal evidence chain analysis. BRIEF DESCRIPTION OF THE DRAWINGS

[0083] Figure 1 A schematic diagram of a process flow of a multimodal evidence chain analysis method provided by an embodiment of the present invention;

[0084] Figure 2 A functional module diagram of a multimodal evidence chain analysis system provided by an embodiment of the present invention;

[0085] Figure 3 A schematic diagram of the structure of an electronic device for implementing the multimodal evidence chain analysis method provided by one embodiment of the present invention.

[0086] Description of reference numerals:

[0087] 1. Electronic device; 10. Processor; 11. Memory; 12. Bus.

[0088] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0089] It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention.

[0090] Embodiment 1:

[0091] This embodiment provides a multimodal evidence chain analysis method. The execution subject of the multimodal evidence chain analysis method includes but is not limited to at least one of the electronic devices such as a server and a terminal that can be configured to execute the method provided in the embodiment of the present application. In other words, the multimodal evidence chain analysis method can be executed by software or hardware installed on a terminal device or a server device, and the software can be a blockchain platform. The server includes but is not limited to: a single server, a server cluster, a cloud server or a cloud server cluster, etc.

[0092] Reference Figure 1 FIG. 1 is a flow chart of a multimodal evidence chain analysis method provided in this embodiment, wherein the multimodal evidence chain analysis method includes:

[0093] S1. Receive an evidence chain analysis instruction, and obtain a multimodal evidence event set based on the evidence chain analysis instruction, wherein the multimodal evidence event set includes text modal information, image modal information, and video modal information of the case involved, and the multimodal evidence event set is provided by relevant case personnel.

[0094] It can be understood that the evidence chain analysis instruction refers to a manually initiated instruction to analyze the existing evidence information, and the multimodal evidence event set refers to a set of events related to this case, wherein each multimodal evidence event involves time, place, people and the course of events, and the time, place, people and events involved in the multimodal evidence event are recorded in text, images, videos, etc. The multimodal evidence event set is retrieved and sorted by the relevant case personnel of the case, wherein the relevant case personnel may be: investigators, lawyers, witnesses, etc.

[0095] For example, Xiao Zhang is an investigator of a certain online fraud case. One day, Xiao Zhang received an evidence chain analysis instruction and needed to conduct an evidence chain analysis on a certain online fraud case. Therefore, Xiao Zhang obtained the basic situation of the case through communication with relevant case personnel, and integrated the case data to obtain a multimodal evidence event set. Among them, a multimodal evidence event in the multimodal evidence event set is recorded as follows: "Internet user Xiao Liu transferred money to another netizen Xiao Yu on a certain social platform, where Xiao Liu initiated the transfer at time A and Xiao Yu received the transfer at time B", and a group of pictures were also recorded in the multimodal evidence event, where the caption of the picture is: " Figure 1 Transfer pictures for Xiao Liu and Xiao Yu", " Figure 2 This is the chat record between Xiao Liu and Xiao Yu."

[0096] S2. Based on a preset basic case report, determine a case description word set, and based on the case description word set, query the associated evidence event set in the multimodal evidence event set.

[0097] It can be understood that the basic case report refers to the descriptive text information about the case, and the basic case report is drawn up by the relevant case personnel through a simple summary of the case.

[0098] For example, Xiao Wang is an investigator of a certain case. One day, Xiao Wang received a report of an online fraud. After communicating with the reporter, Xiao Wang summarized the case and obtained a basic report on the case. The basic report on the case covered the causes and process of the case.

[0099] Specifically, the case description word set is determined based on the preset case basic report, including:

[0100] Performing Chinese word segmentation on the basic case report to obtain an original descriptive word set, wherein the tool for performing Chinese word segmentation is an online word segmentation database;

[0101] Counting the original description frequency of each original description word in the original description word set to obtain an original description frequency set, and counting the common description frequency of each original description word in the original description word set in a preset common description word library to obtain a common description frequency set;

[0102] Extract target descriptive words in the original descriptive word set one by one;

[0103] According to the original description frequency set and the common description frequency set, the target description word is analyzed for its semantic keyness to obtain the description key value;

[0104] Determine whether the description key value is greater than a preset standard key value;

[0105] If the description key value is greater than the standard key value, the target description word is recorded as the case description word;

[0106] The case description words are summarized to obtain a case description word set.

[0107] It can be understood that the original description word set refers to the collection of all words in the basic case report, and the online word segmentation database refers to a program or algorithm that can segment Chinese text segments, such as: the jieba word segmentation library, HanLP language library, SnowNLP language library, etc. in the Python language. The Chinese word segmentation of the basic case report can be performed by directly calling the relevant word segmentation library in the Python language environment.

[0108] It can be understood that the original description frequency refers to the number of times the original description word appears in the original description word set, and the common description frequency refers to the number of times the original description word appears in the common description word library, wherein the common description word library refers to an online database containing a large number of common words in the Chinese environment, such as: Baidu Index, Google Trends and other websites, which provide word frequency statistics based on network search data, which can reflect the frequency of use of vocabulary in a certain period of time, or the Modern Chinese Corpus of the National Language Resource Monitoring and Research Center, which contains a large amount of written and spoken corpus.

[0109] It should be explained that when performing a criticality analysis on the original descriptive words, the introduction of a common descriptive word library can filter out words that are commonly found in the Chinese context, such as "maybe", "but", "and", etc., thereby reducing the noise in the descriptive words and improving the efficiency and accuracy of the analysis. At the same time, by comparing the word frequencies in the original descriptive word set and the common descriptive word library, words that appear frequently in the basic case report but are uncommon in daily language can be identified, and these words have a very high generalization of the content of the basic case report.

[0110] For example, a basic report of a case is: "The victim went home alone at night, but the suspect followed her and used violence." In this basic report of the case, a common description vocabulary can be introduced. Since "but" and "and" appear frequently in the common description vocabulary, when calculating the description key value, these words will be filtered out due to their higher target common frequency, resulting in a lower calculated description key value. These words will be filtered out because they are very common in any text and are not very helpful for understanding the specific content of the case. The filtered keywords may include "victim", "alone", "go home", "suspect", "tracking", and "violence". These words can better summarize the main content of the case report, that is, a victim was followed and suffered violence by a suspect. In this way, the key information in the basic case report can be more accurately identified and analyzed.

[0111] Furthermore, the target descriptor refers to the original descriptor extracted from the original descriptor set, the description key value refers to a numerical value used to represent the criticality of the target descriptor to the basic case report, and the standard key value refers to an artificially set key value constant.

[0112] Specifically, the word meaning key degree analysis of the target description word to obtain the description key value includes:

[0113] Identifying the target description frequency corresponding to the target descriptor in the original description frequency set, and identifying the target common frequency corresponding to the target descriptor in the common description frequency set;

[0114] Based on the target description frequency, target common frequency, original description frequency set and common description frequency set, the description key value is calculated using the following formula:

[0115]

[0116] Among them, K represents the description key value, N 1 Indicates the target description frequency, n 1 represents the number of original description frequencies in the original description frequency set, represents the i-th original description frequency in the original description frequency set, ln(*) represents the logarithmic function, N 2 Indicates the common frequency of the target, n 2 Indicates the number of common description frequencies in the common description frequency set, Represents the jth common description frequency in the common description frequency set.

[0117] It can be understood that the target description frequency refers to the original description frequency corresponding to the target description word in the original description frequency set, and the target common frequency refers to the common description frequency corresponding to the target description word in the common description frequency set.

[0118] In detail, the step of searching for a related evidence event set in the multimodal evidence event set based on the case description word set includes:

[0119] Extracting multimodal evidence events in sequence from the multimodal evidence event set;

[0120] Identifying event description text in the multimodal evidence event, wherein the event description text comes from descriptive text information, image description text information, and video description text information in the multimodal evidence event;

[0121] Extracting case description words from the case description word set in turn, and determining whether the case description words appear in the event description text;

[0122] If the case description word appears in the event description text, the case description word is recorded as a related description word, and the related description frequency of the related description word appearing in the event description text is counted;

[0123] Summarizing the associated description words and the corresponding associated description frequencies respectively to obtain an associated description word set and an associated description frequency set, and identifying the number of associated words in the associated description word set;

[0124] According to the association description frequency set and the number of associated words, the event association value is calculated using the following formula:

[0125]

[0126] Among them, G represents the event association value, S represents the number of associated words, and m represents the number of associated description frequencies in the associated description frequency set. represents the kth associated description frequency in the associated description frequency set;

[0127] If the event association value is greater than a preset minimum association value, the multimodal evidence event is recorded as an associated evidence event;

[0128] The associated evidence events are summarized to obtain an associated evidence event set.

[0129] It can be understood that the event description text refers to the collection of all words and texts appearing in a multimodal evidence event. For example, an event description text includes descriptive text information, image description text information and video description text information, and the descriptive text information is a text description of the multimodal evidence event, the image description text information is a text description that explains the pictures involved in the multimodal evidence event, and the video description text information is a text description that explains the videos involved in the multimodal evidence event.

[0130] For example, a multimodal evidence event includes: 1. A descriptive text message that explains the process of this event: "Internet user Xiao Liu transferred money to another Internet user Xiao Yu on a social platform, where Xiao Liu initiated the transfer at time A and Xiao Yu received the transfer at time B", 2. The evidence chain event also includes multiple recorded images of Xiao Liu and Xiao Yu chatting and images of this transfer, and these images are all provided by the reporter. The image description text information for these images is: " Figure 1 Transfer pictures for Xiao Liu and Xiao Yu", " Figure 2 The chat record between Xiao Liu and Xiao Yu", the above descriptive text information and the image description text information constitute the event description text.

[0131] It can be understood that the detailed steps for determining whether the case description words appear in the event description text are: first, the event description text is segmented to obtain a set of event description words, wherein the step of segmenting the event description text is the same as the step of segmenting the Chinese words of the basic case report mentioned above, and will not be repeated here. Then, it is determined whether the case description words are in the set of event description words. If the case description words are in the event description words, it is considered that the case description words appear in the event description text. Otherwise, it is considered that the case description words do not appear in the event description text.

[0132] It should be explained that the association description frequency refers to the number of times the association description word appears in the event description text, the association word number refers to the number of association description words in the association description word set, the event association value refers to a numerical value used to represent the degree of association between a multimodal evidence event and a case description word set, and the minimum association value refers to an artificially set association value constant.

[0133] S3. Perform a pairwise combination correlation analysis on the associated evidence events in the associated evidence event set to obtain a content correlation value set, wherein each content correlation value in the content correlation value set corresponds to two associated evidence events in the associated evidence event set.

[0134] It can be understood that the pairwise combination correlation analysis refers to first combining the related evidence events in the related evidence event set in pairs and then performing correlation analysis on the two combined related evidence events, and the content correlation value refers to a numerical value used to represent the degree of content correlation between two different related evidence events.

[0135] Specifically, the related evidence events in the related evidence event set are subjected to a pairwise combination correlation analysis to obtain a content correlation value set, including:

[0136] Randomly extracting a first associated evidence event from the associated evidence event set, and removing the first associated evidence event from the associated evidence event set to obtain a removed evidence event set;

[0137] Sequentially extracting second related evidence events from the eliminated evidence event set;

[0138] Performing event correlation analysis on the first correlation evidence event and the extracted second correlation evidence event to obtain a content correlation value;

[0139] Summarize the content association values ​​to obtain a unit content association value set, and use the eliminated evidence event set as the associated evidence event set, return to the step of randomly extracting the first associated evidence event in the associated evidence event set, until the eliminated evidence event set is an empty set, and summarize the unit content association value set to obtain a content association value set.

[0140] It should be explained that the first related evidence event refers to the related evidence event extracted from the related evidence event set, the eliminated evidence event set refers to the related evidence event set after eliminating the first related evidence event, and the second related evidence event refers to the related evidence event extracted from the eliminated evidence event set.

[0141] In detail, the event correlation analysis of the first correlation evidence event and the second correlation evidence event to obtain the content correlation value includes:

[0142] Respectively identifying a first occurrence time and a first end time of the first associated evidence event, and a second occurrence time and a second end time of the second associated evidence event;

[0143] Based on the first occurrence time and the first end time, a first duration is calculated, and based on the second occurrence time and the second end time, a second duration is calculated;

[0144] Calculating a time association value according to the first occurrence time, the first end time, the second occurrence time, the second end time, the first duration, and the second duration;

[0145] Performing spatial correlation analysis on the first correlation evidence event and the second correlation evidence event to obtain a spatial correlation value;

[0146] According to the time association value and the space association value, the content association value is calculated using the following formula:

[0147]

[0148] Among them, G n Represents the content association value, G time represents the time-related value, G space Represents a spatially associated value.

[0149] It can be understood that the first occurrence time and the first end time refer to the start time and the end time of the event in the first related evidence event respectively, and the second occurrence time and the second end time refer to the start time and the end time of the event in the second related evidence event respectively.

[0150] For example, the record of a related evidence event is as follows: "Netizen Xiao Liu transferred money to another netizen Xiao Yu on a social platform, where the time when Xiao Liu initiated the transfer was time A, and the time when Xiao Yu received the transfer was time B". In this related evidence event, time A can be considered as the occurrence time, and time B can be considered as the end time.

[0151] It can be understood that the first duration refers to the time difference between the first occurrence moment and the first end moment, the second duration refers to the time difference between the second occurrence moment and the second end moment, the time association value refers to a numerical value used to represent the degree of temporal association between the first associated evidence event and the second associated evidence event, and the spatial association value refers to a numerical value of the degree of formal association between the first associated evidence event and the second associated evidence event.

[0152] In detail, the calculating time-related value includes:

[0153] The minimum disturbance duration is set, and based on the minimum disturbance duration, the occurrence correlation factor, the end correlation factor and the duration factor are calculated using the first occurrence time, the second occurrence time, the first end time, the second end time, the first duration and the second duration using the following formula:

[0154]

[0155] Among them, P start represents the correlation factor, |*| represents the absolute value symbol, T 1k represents the first occurrence time, T 2k represents the second occurrence time, Δt b represents the minimum disturbance duration, P end represents the end correlation factor, T 1s represents the first end time, T 2s represents the second end time, P during represents the duration factor, t 1 Indicates the first duration, t 2 Indicates the second duration;

[0156] The duration fluctuation factor of the minimum disturbance duration is set, and the time correlation value is calculated according to the occurrence correlation factor, the end correlation factor, the duration factor and the duration fluctuation factor, wherein the duration fluctuation factor and the time correlation value are respectively expressed as:

[0157]

[0158] G time =ΔP×(P start +P end +P during )

[0159] Among them, ΔP represents the duration fluctuation factor.

[0160] It should be explained that the minimum disturbance duration refers to an artificially set duration constant, which is used to represent the maximum time difference that can exist between the occurrence time or the end time of two related evidence events. When the duration corresponding to the absolute value of the difference between the occurrence time, end time and duration of the first related evidence event and the second related evidence event exceeds the minimum disturbance duration, the corresponding occurrence correlation factor, end correlation factor and duration factor can be considered to be zero, and when the occurrence correlation factor, end correlation factor and duration factor are all zero, it can be considered that the two events are not temporally correlated, wherein the occurrence correlation factor refers to a numerical value used to represent the degree of correlation between two related evidence events at the time of occurrence, the end correlation factor refers to a numerical value used to represent the degree of correlation between two related evidence events at the end time, and the duration factor refers to a numerical value used to represent the degree of correlation between two related evidence events at the duration of the event.

[0161] Furthermore, since the occurrence, end and duration of two different events are not exactly the same, there will be certain errors. Therefore, the minimum disturbance duration is introduced to quantify this error. However, the minimum disturbance duration only determines whether the above-mentioned occurrence correlation factor, end correlation factor, and duration factor are zero values, and does not refine the numerical relationship between these factors and the minimum disturbance duration. This will make it impossible to represent the specific duration deviation in the subsequently calculated time correlation value. Therefore, a duration fluctuation factor is introduced. The duration fluctuation factor is used to quantify the duration deviation between two related evidence events, wherein the duration deviation includes: the deviation in the occurrence time between the first related evidence event and the second related evidence event, the deviation in the end time between the first related evidence event and the second related evidence event, and the deviation in the duration between the first related evidence event and the second related evidence event.

[0162] In detail, the spatial correlation analysis of the first correlation evidence event and the second correlation evidence event to obtain the spatial correlation value includes:

[0163] Obtaining a first associated text of the first associated evidence event and a second associated text of the second associated evidence event, and performing word segmentation on the first associated text and the second associated text respectively to obtain a first associated word set and a second associated word set;

[0164] Comparing the first associated word set with the second associated word set to obtain a common associated word set, and obtaining the number of common associated words in the common associated word set, wherein the common associated words in the common associated word set are associated words that appear in both the first associated word set and the second associated word set;

[0165] Determining whether the number of common associated words is less than a preset minimum number of associated words;

[0166] If the number of common related words is less than the minimum number of related words, the spatial relatedness value is recorded as zero;

[0167] If the number of common associated words is not less than the minimum number of associated words, the spatial association value is calculated based on the common associated word set.

[0168] It can be understood that the first associated text refers to the description text in the first associated evidence event. The detailed explanation of the first associated text is the same as the explanation of the event description text in the evidence chain event, and will not be repeated here. The second associated text refers to the description text in the second associated evidence event. The first associated word set refers to the vocabulary set in the first associated text, and the second associated word set refers to the vocabulary set in the second associated text. The step of word segmentation has been given in detail in the above steps.

[0169] Furthermore, the number of common associated words refers to the number of common associated words in the common associated word set, and the minimum number of associated words refers to the minimum number of common associated words used to indicate that the first associated evidence event and the second associated evidence event are considered to have spatial correlation.

[0170] In detail, the calculation of the spatial association value based on the common association word set includes:

[0171] Extracting common related words in the common related word set in sequence;

[0172] respectively counting the first common frequency of the common associated words in the first associated word set and the second common frequency in the second associated word set;

[0173] The first public frequencies and the second public frequencies are respectively aggregated to obtain a first public frequency set and a second public frequency set;

[0174] Based on the first common frequency set and the second common frequency set, a spatial correlation value is calculated, wherein the spatial correlation value is expressed as:

[0175]

[0176] Among them, G space represents the spatial association value, U represents the number of the first common frequency in the first common frequency set or the number of the second common frequency in the second common frequency set, represents the e-th first common frequency in the first common frequency set, Represents the e-th common frequency in the second common frequency set.

[0177] It should be explained that the first common frequency refers to the number of times the common associated word appears in the first associated word set, and the second common frequency refers to the number of times the common associated word appears in the second associated word set.

[0178] S4. Identify a highly correlated value set in a content correlated value set according to a preset standard correlated value, wherein the highly correlated value in the highly correlated value set is greater than or equal to the standard correlated value.

[0179] It can be understood that the standard association value is an artificially set association value constant. When the content association value is greater than or equal to the standard association value, the content association value at this time can be considered to be a high association value, which means that the two associated evidence events corresponding to the content association value have a high correlation in content.

[0180] S5. Extract central evidence events in the associated evidence event set in sequence, and identify a central associated value set corresponding to the central evidence event in the highly associated value set, wherein the central associated value set contains multiple central associated values ​​or the central associated value set is an empty set.

[0181] It should be explained that the central evidence event refers to an associated evidence event in the associated evidence event set, the central association value refers to the high association value corresponding to the central evidence event, and the central association value is the content association value between the central evidence event and another associated evidence event.

[0182] S6. If the central correlation value set is not an empty set, then identify the highly correlated events of each central correlation value in the central correlation value set in turn to obtain a highly correlated event set, wherein the highly correlated event is another associated evidence event corresponding to the central correlation value except the central evidence event.

[0183] It can be understood that if the highly correlated value set is not an empty set, it means that the central evidence event corresponding to the highly correlated value is not an independent event, and there are one or more associated evidence events that are highly correlated with the central evidence event. Then the highly correlated event set is the collection of the one or more associated evidence events.

[0184] S7. Supplement the central evidence event to the highly correlated event set to obtain a target evidence chain event set, and summarize the target evidence chain event sets corresponding to each central evidence event to obtain multiple target evidence chain event sets.

[0185] It should be explained that the target evidence chain event set refers to a highly correlated event set after being supplemented by the central evidence event. It can be considered that all target evidence chain events in the target evidence chain event set are highly correlated in content.

[0186] S8. Remove duplicate event sets from the multiple target evidence chain event sets to obtain multiple deduplicated evidence chain event sets, and complete multimodal evidence chain analysis based on the multiple deduplicated evidence chain event sets.

[0187] It is clear that there will be repeated event sets in multiple target evidence chain event sets, where a repeated event set refers to a set that contains exactly the same target evidence chain events, and the repeated event set elimination refers to eliminating the repeated event sets in multiple target evidence chain event sets and retaining only one event set in the multiple repeated event sets.

[0188] Furthermore, by analyzing and organizing the multimodal evidence event set to obtain a deduplicated evidence chain event set, we can focus on analyzing those evidences that are directly and highly relevant to the case, avoiding wasting time on a large amount of irrelevant or low-relevance evidence, which improves the efficiency of the analysis. At the same time, the deduplicated evidence chain event set contains the key evidence and highly relevant evidence in the case. These evidences form a logical connection with each other, which helps to build a complete chain of evidence. Therefore, the deduplicated evidence chain event set can provide a clear evidence framework for the subsequent handling of the case.

[0189] Therefore, this embodiment first determines the case description word set based on the basic case report, and based on the case description word set, searches for the associated evidence event set in the multimodal evidence event set, so that through the determination of the case description word set, the evidence directly related to the case can be more accurately screened out from the multimodal evidence event set, thereby improving the efficiency and accuracy of evidence analysis;

[0190] Further, by performing a pairwise correlation analysis on the associated evidence events in the associated evidence event set, a content correlation value set can be obtained to quantify the correlation between the evidence events and identify which evidence events have a strong connection, which provides an important reference for building the chain of evidence. In this step, the accuracy of the correlation analysis is further improved based on the temporal correlation analysis and spatial correlation analysis. Then, by setting a threshold, those evidences with weak correlation are filtered out, and those highly related evidences are analyzed intensively, which improves the efficiency of the analysis and the reliability of the chain of evidence. Finally, a target evidence chain event set is created for each central evidence event to obtain multiple target evidence chain event sets, and the repeated event sets in the multiple target evidence chain event sets are eliminated to obtain multiple deduplicated evidence chain event sets. Among them, the construction of the target evidence chain event set can more clearly show the logical relationship between the evidence to enhance the reliability and persuasiveness of the evidence. At the same time, for complex cases involving a large amount of evidence and information, the target evidence chain event set can help analysts quickly locate key evidence, thereby reducing the complexity of the case and improving the efficiency of the evidence chain analysis.

[0191] Embodiment 2:

[0192] like Figure 2As shown, this embodiment provides a functional module diagram of a multimodal evidence chain analysis system, and the multimodal evidence chain analysis system 100 can be installed in an electronic device. According to the functions implemented, the multimodal evidence chain analysis system 100 can include an associated evidence query module 101, an event association analysis module 102, a highly associated identification module 103 and a target evidence integration module 104. The module described in the present invention can also be referred to as a unit, which refers to a series of computer program segments that can be executed by an electronic device processor and can complete fixed functions, which are stored in the memory of the electronic device.

[0193] The associated evidence query module 101 is used to receive an evidence chain analysis instruction, and obtain a multimodal evidence event set based on the evidence chain analysis instruction, wherein the multimodal evidence event set includes text modal information, image modal information, and video modal information of the case involved, and the multimodal evidence event set is provided by relevant case personnel, and based on a preset case basic report, a case description word set is determined, and based on the case description word set, an associated evidence event set is queried in the multimodal evidence event set;

[0194] The event association analysis module 102 is used to perform a pairwise combination association analysis on the associated evidence events in the associated evidence event set to obtain a content association value set, wherein each content association value in the content association value set corresponds to two associated evidence events in the associated evidence event set, and to identify a highly associated value set in the content association value set according to a preset standard association value, wherein the highly associated value in the highly associated value set is greater than or equal to the standard association value;

[0195] The highly correlated identification module 103 is used to sequentially extract central evidence events from the associated evidence event set, and identify a central correlation value set corresponding to the central evidence event in the highly correlated value set, wherein the central correlation value set contains multiple central correlation values ​​or the central correlation value set is an empty set. If the central correlation value set is not an empty set, the highly correlated events of each central correlation value in the central correlation value set are sequentially identified to obtain a highly correlated event set, wherein the highly correlated event is another associated evidence event corresponding to the central correlation value except the central evidence event;

[0196] The target evidence integration module 104 is used to supplement the central evidence event to a highly correlated event set to obtain a target evidence chain event set, summarize the target evidence chain event sets corresponding to each central evidence event to obtain multiple target evidence chain event sets, and eliminate duplicate event sets from the multiple target evidence chain event sets to obtain multiple deduplicated evidence chain event sets.

[0197] In detail, each module in the multimodal evidence chain analysis system 100 in the embodiment of the present invention is used in the same manner as described above. Figure 1 The multimodal evidence chain analysis method described in the previous section is a similar technical method and can produce the same technical effects, so I will not go into details here.

[0198] like Figure 3 , which is a schematic diagram of the structure of an electronic device for implementing a multimodal evidence chain analysis method provided by an embodiment of the present invention.

[0199] The electronic device 1 may include a processor 10, a memory 11 and a bus 12, and may also include a computer program stored in the memory 11 and executable on the processor 10, such as a multimodal evidence chain analysis method program.

[0200] Among them, the memory 11 includes at least one type of readable storage medium, and the readable storage medium includes flash memory, mobile hard disk, multimedia card, card-type memory (for example: SD or DX memory, etc.), magnetic memory, disk, optical disk, etc. In some embodiments, the memory 11 can be an internal storage unit of the electronic device 1, such as a mobile hard disk of the electronic device 1. In other embodiments, the memory 11 can also be an external storage device of the electronic device 1, such as a plug-in mobile hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (SecureDigital, SD) card, a flash card (Flash Card), etc. equipped on the electronic device 1. Further, the memory 11 also includes an internal storage unit of the electronic device 1 and an external storage device. The memory 11 can not only be used to store application software and various types of data installed in the electronic device 1, such as the code of the multimodal evidence chain analysis method program, etc., but also can be used to temporarily store data that has been output or is to be output.

[0201] The processor 10 may be composed of an integrated circuit in some embodiments, for example, a single packaged integrated circuit, or a plurality of packaged integrated circuits with the same or different functions, including one or more central processing units (CPUs), microprocessors, digital processing chips, graphics processors, and combinations of various control chips, etc. The processor 10 is the control core (Control Unit) of the electronic device, and uses various interfaces and lines to connect various components of the entire electronic device, and executes or executes programs or modules (such as multimodal evidence chain analysis method programs, etc.) stored in the memory 11, and calls data stored in the memory 11 to execute various functions of the electronic device 1 and process data.

[0202] The bus 12 may be a peripheral component interconnect (PCI) bus or an extended industry standard architecture (EISA) bus, etc. The bus 12 may be divided into an address bus, a data bus, a control bus, etc. The bus 12 is configured to realize connection and communication between the memory 11 and at least one processor 10, etc.

[0203] Figure 3 Only an electronic device with components is shown, and those skilled in the art will understand that Figure 3 The structure shown does not constitute a limitation on the electronic device 1, and may include fewer or more components than shown in the figure, or combine certain components, or arrange the components differently.

[0204] The multimodal evidence chain analysis method program stored in the memory 11 in the electronic device 1 is a combination of multiple instructions. When running in the processor 10, the multimodal evidence chain analysis method described in Example 1 can be implemented. Specifically, the specific implementation method of the processor 10 for the above instructions can refer to Figures 1 to 3 The description of the relevant steps in the corresponding embodiments will not be repeated here.

[0205] Furthermore, if the module / unit integrated in the electronic device 1 is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. The computer-readable storage medium can be volatile or non-volatile. For example, the computer-readable medium can include: any entity or system capable of carrying the computer program code, a recording medium, a USB flash drive, a mobile hard disk, a magnetic disk, an optical disk, a computer memory, and a read-only memory (ROM).

[0206] The present invention also provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor of an electronic device, it can implement the multimodal evidence chain analysis method described in Example 1.

[0207] It is obvious to those skilled in the art that the present invention is not limited to the details of the above exemplary embodiments, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0208] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the spirit and scope of the technical solution of the present invention.

Claims

1. A multimodal evidence chain analysis method, characterized in that: The method comprises: Receiving an evidence chain analysis instruction, and obtaining a multimodal evidence event set based on the evidence chain analysis instruction, wherein the multimodal evidence event set includes text modal information, image modal information, and video modal information of the case involved, and the multimodal evidence event set is provided by relevant case personnel; Based on a preset basic case report, determine a case description word set, and based on the case description word set, query a related evidence event set in the multimodal evidence event set; Performing pairwise combination correlation analysis on the associated evidence events in the associated evidence event set to obtain a content correlation value set, wherein each content correlation value in the content correlation value set corresponds to two associated evidence events in the associated evidence event set; According to a preset standard association value, identifying a highly associated value set in a content association value set, wherein the highly associated value in the highly associated value set is greater than or equal to the standard association value; Extracting central evidence events in sequence from the associated evidence event set, identifying a central associated value set corresponding to the central evidence event in the highly associated value set, wherein the central associated value set contains multiple central associated values ​​or the central associated value set is an empty set; If the central correlation value set is not an empty set, then sequentially identifying highly correlated events of each central correlation value in the central correlation value set to obtain a highly correlated event set, wherein the highly correlated event is another correlated evidence event corresponding to the central correlation value except the central evidence event; Supplement the central evidence event to the highly correlated event set to obtain a target evidence chain event set, and summarize the target evidence chain event sets corresponding to each central evidence event to obtain multiple target evidence chain event sets; Duplicate event sets are eliminated from the multiple target evidence chain event sets to obtain multiple deduplicated evidence chain event sets, and multimodal evidence chain analysis is completed based on the multiple deduplicated evidence chain event sets.

2. The multimodal evidence chain analysis method according to claim 1, characterized in that: The case description word set is determined based on the preset case basic report, including: Performing Chinese word segmentation on the basic case report to obtain an original descriptive word set, wherein the tool for performing Chinese word segmentation is an online word segmentation database; Counting the original description frequency of each original description word in the original description word set to obtain an original description frequency set, and counting the common description frequency of each original description word in the original description word set in a preset common description word library to obtain a common description frequency set; Extract target descriptive words in the original descriptive word set one by one; According to the original description frequency set and the common description frequency set, the target description word is analyzed for its semantic keyness to obtain the description key value; Determine whether the description key value is greater than a preset standard key value; If the description key value is greater than the standard key value, the target description word is recorded as the case description word; The case description words are summarized to obtain a case description word set.

3. The multimodal evidence chain analysis method according to claim 2, characterized in that: The step of performing a semantic key analysis on the target description word to obtain a description key value includes: Identifying the target description frequency corresponding to the target descriptor in the original description frequency set, and identifying the target common frequency corresponding to the target descriptor in the common description frequency set; Based on the target description frequency, target common frequency, original description frequency set and common description frequency set, the description key value is calculated using the following formula: Among them, K represents the description key value, N1 represents the target description frequency, and n1 represents the number of original description frequencies in the original description frequency set. represents the i-th original description frequency in the original description frequency set, ln(*) represents the logarithmic function, N2 represents the target common frequency, n2 represents the number of common description frequencies in the common description frequency set, Represents the jth common description frequency in the common description frequency set.

4. The multimodal evidence chain analysis method according to claim 3, characterized in that: The step of searching for a related evidence event set in the multimodal evidence event set based on the case description word set includes: Extracting multimodal evidence events in sequence from the multimodal evidence event set; Identifying event description text in the multimodal evidence event, wherein the event description text comes from descriptive text information, image description text information, and video description text information in the multimodal evidence event; Extracting case description words from the case description word set in turn, and determining whether the case description words appear in the event description text; If the case description word appears in the event description text, the case description word is recorded as a related description word, and the related description frequency of the related description word appearing in the event description text is counted; Summarizing the associated description words and the corresponding associated description frequencies respectively to obtain an associated description word set and an associated description frequency set, and identifying the number of associated words in the associated description word set; According to the association description frequency set and the number of associated words, the event association value is calculated using the following formula: Among them, G represents the event association value, S represents the number of associated words, and m represents the number of associated description frequencies in the associated description frequency set. represents the kth associated description frequency in the associated description frequency set; If the event association value is greater than a preset minimum association value, the multimodal evidence event is recorded as an associated evidence event; The associated evidence events are summarized to obtain an associated evidence event set.

5. The multimodal evidence chain analysis method according to claim 4, characterized in that: The above-mentioned performing pairwise combination correlation analysis on the related evidence events in the related evidence event set to obtain a content correlation value set includes: Randomly extracting a first associated evidence event from the associated evidence event set, and removing the first associated evidence event from the associated evidence event set to obtain a removed evidence event set; Sequentially extracting second related evidence events from the eliminated evidence event set; Performing event correlation analysis on the first correlation evidence event and the extracted second correlation evidence event to obtain a content correlation value; Summarize the content association values ​​to obtain a unit content association value set, and use the eliminated evidence event set as the associated evidence event set, return to the step of randomly extracting the first associated evidence event in the associated evidence event set, until the eliminated evidence event set is an empty set, and summarize the unit content association value set to obtain a content association value set.

6. The multimodal evidence chain analysis method according to claim 5, characterized in that: The performing event correlation analysis on the first correlation evidence event and the second correlation evidence event to obtain a content correlation value includes: Respectively identifying a first occurrence time and a first end time of the first associated evidence event, and a second occurrence time and a second end time of the second associated evidence event; Based on the first occurrence time and the first end time, a first duration is calculated, and based on the second occurrence time and the second end time, a second duration is calculated; Calculating a time association value according to the first occurrence time, the first end time, the second occurrence time, the second end time, the first duration, and the second duration; Performing spatial correlation analysis on the first correlation evidence event and the second correlation evidence event to obtain a spatial correlation value; According to the time association value and the space association value, the content association value is calculated using the following formula: Among them, G n Represents the content association value, G time represents the time-related value, G space Represents a spatially associated value.

7. The multimodal evidence chain analysis method according to claim 6, characterized in that: The calculation time associated value includes: The minimum disturbance duration is set, and based on the minimum disturbance duration, the occurrence correlation factor, the end correlation factor and the duration factor are calculated using the first occurrence time, the second occurrence time, the first end time, the second end time, the first duration and the second duration using the following formula: Among them, P start represents the correlation factor, |*| represents the absolute value symbol, T 1k represents the first occurrence time, T 2k represents the second occurrence time, Δt b represents the minimum disturbance duration, P end represents the end correlation factor, T 1s represents the first end time, T 2s represents the second end time, P during represents the duration factor, t1 represents the first duration, and t2 represents the second duration; The duration fluctuation factor of the minimum disturbance duration is set, and the time correlation value is calculated according to the occurrence correlation factor, the end correlation factor, the duration factor and the duration fluctuation factor, wherein the duration fluctuation factor and the time correlation value are respectively expressed as: G time =ΔP×(P start +P end +P during ) Among them, ΔP represents the duration fluctuation factor.

8. The multimodal evidence chain analysis method according to claim 7, characterized in that: The performing of spatial correlation analysis on the first correlation evidence event and the second correlation evidence event to obtain a spatial correlation value includes: Obtaining a first associated text of the first associated evidence event and a second associated text of the second associated evidence event, and performing word segmentation on the first associated text and the second associated text respectively to obtain a first associated word set and a second associated word set; Comparing the first associated word set with the second associated word set to obtain a common associated word set, and obtaining the number of common associated words in the common associated word set, wherein the common associated words in the common associated word set are associated words that appear in both the first associated word set and the second associated word set; Determining whether the number of common associated words is less than a preset minimum number of associated words; If the number of common related words is less than the minimum number of related words, the spatial relatedness value is recorded as zero; If the number of common associated words is not less than the minimum number of associated words, the spatial association value is calculated based on the common associated word set.

9. The multimodal evidence chain analysis method according to claim 8, characterized in that: The step of calculating the spatial association value based on the common association word set includes: Extracting common related words in the common related word set in sequence; respectively counting the first common frequency of the common associated words in the first associated word set and the second common frequency in the second associated word set; The first public frequencies and the second public frequencies are respectively aggregated to obtain a first public frequency set and a second public frequency set; Based on the first common frequency set and the second common frequency set, a spatial correlation value is calculated, wherein the spatial correlation value is expressed as: Among them, G space represents the spatial association value, U represents the number of the first common frequency in the first common frequency set or the number of the second common frequency in the second common frequency set, represents the e-th first common frequency in the first common frequency set, Represents the e-th common frequency in the second common frequency set.

10. A multimodal evidence chain analysis system, characterized in that: The system comprises: An associated evidence query module is used to receive an evidence chain analysis instruction, and obtain a multimodal evidence event set based on the evidence chain analysis instruction, wherein the multimodal evidence event set includes text modal information, image modal information, and video modal information of the case involved, and the multimodal evidence event set is provided by relevant case personnel, and based on a preset case basic report, a case description word set is determined, and based on the case description word set, an associated evidence event set is queried in the multimodal evidence event set; An event association analysis module is used to perform a pairwise combination association analysis on the associated evidence events in the associated evidence event set to obtain a content association value set, wherein each content association value in the content association value set corresponds to two associated evidence events in the associated evidence event set, and to identify a highly associated value set in the content association value set according to a preset standard association value, wherein the highly associated value in the highly associated value set is greater than or equal to the standard association value; A highly correlated identification module is used to sequentially extract central evidence events from a set of correlated evidence events, and identify a central correlated value set corresponding to the central evidence event in the highly correlated value set, wherein the central correlated value set contains multiple central correlated values ​​or the central correlated value set is an empty set. If the central correlated value set is not an empty set, then sequentially identify highly correlated events of each central correlated value in the central correlated value set to obtain a highly correlated event set, wherein the highly correlated event is another correlated evidence event corresponding to the central correlated value other than the central evidence event; The target evidence integration module is used to supplement the central evidence event to a highly correlated event set to obtain a target evidence chain event set, summarize the target evidence chain event set corresponding to each central evidence event to obtain multiple target evidence chain event sets, and eliminate duplicate event sets from the multiple target evidence chain event sets to obtain multiple deduplicated evidence chain event sets.

Images