Authenticity determination support device, authenticity determination support method, program, and synthesis degree determination support device

Abstract

To provide an authenticity determination support device, a method, a program, and a synthesis degree determination support device.SOLUTION: A authenticity determination support device 100 includes: a target information acquisition section that acquires multimodal target information which is a target of authenticity determination; a relevant information acquisition section that acquires multimodal relevant information related to the target information; a calculation section that calculates a degree of relevance between each modal of the target information and a modal of the corresponding relevant information; and an output section that outputs the calculated degree of relevance for each modal as output information. A synthesis degree determination support device includes: a target information acquisition section that acquires multimodal target information which is a determination target of a degree of synthesis; a relevant information acquisition section that acquires relevant information related to the target information; a calculation section that calculates a degree of synthesis of the target information; a determination basis generation section that generates a determination basis of the degree of synthesis in a natural language for at least one modal of the target information on the basis of the relevant information; and an output section that outputs the determination basis.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to an authenticity determination support device, an authenticity determination support method, a program, and a synthesis degree determination support device. [Background technology]

[0002] Patent Document 1 describes "an information determination program, an information processing device, and an information determination method that enable the suppression of the spread of false information and erroneous information." [Prior art document] [Patent documents] [Patent Document 1] JP 2024-007165 A Summary of the Invention

[0003] In a first aspect of the present invention, a truth determination support device is provided that includes a target information acquisition unit that acquires multimodal target information to be determined as truthful or false, a related information acquisition unit that acquires multimodal related information related to the target information, a calculation unit that calculates the degree of relevance between each modal of the target information and the corresponding modal of the related information, and an output unit that outputs the calculated degree of relevance for each modal as output information.

[0004] The authenticity determination support device may include a authenticity determination unit that determines authenticity of the target information based on the relevance for each modal.

[0005] In any of the above authenticity determination support devices, the target information may include image data and text data.

[0006] In any of the above authenticity determination support devices, the related information may include image data and text data.

[0007] Any of the above authenticity determination support devices may include a degree of synthesis determination unit that determines the degree of synthesis of data included in the target information.

[0008] Any of the above-described authenticity determination support devices may include a determination basis generation unit that generates a determination basis for the output information in natural language.

[0009] In any of the above-described authenticity determination support devices, the determination basis generation unit may generate the determination basis for each modal of the target information in natural language.

[0010] In any of the above-described authenticity determination support devices, the determination basis generation unit may generate a determination basis for determining authenticity of the target information in natural language.

[0011] In any of the above-described authenticity determination support devices, the determination basis generation unit may generate the determination basis for the degree of synthesis of the target information in natural language.

[0012] In any of the above authenticity determination support devices, the determination basis generation unit may generate a reliability of the determination basis.

[0013] In any of the above-described authenticity determination support devices, the output unit may highlight the basis for determination and display it on a predetermined display unit.

[0014] In any of the above-described authenticity determination support devices, the output unit may cause a predetermined display unit to display the basis for the determination as a score for each of predetermined viewpoints.

[0015] In any of the above authenticity determination support devices, the related information acquisition section may output information on the source of the related information that served as the basis for the determination.

[0016] Any of the above-described authenticity determination support devices may include an information source designation unit that designates a target information source for acquiring the target information. The target information acquisition unit may acquire the target information from the target information source designated by the information source designation unit.

[0017] In any of the above-described authenticity determination support devices, the output unit may display the target information on a display unit predetermined for each modality.

[0018] In any of the above-described authenticity determination support devices, the output section may output the related information that supports objective information included in the target information.

[0019] In any of the above-described authenticity determination support devices, the target information acquisition unit may calculate a diffusion index of data from a target information source of the target information, and acquire the target information selected based on the diffusion index.

[0020] In any of the above-described authenticity determination support devices, the target information acquisition unit may calculate a degree of synthesis of data from the target information sources of the target information, and acquire the target information selected based on the degree of synthesis.

[0021] In any of the above-described authenticity determination support devices, the target information acquisition unit may extract objective information from the acquired target information.

[0022] In any of the above authenticity determination support devices, the calculation section may calculate, as the relevance, at least one of a similarity of image information, a similarity of person information, a coincidence of location information, or a coincidence of time information.

[0023] In a second aspect of the present invention, there is provided a method for supporting authenticity determination, comprising the steps of acquiring multimodal target information to be subjected to authenticity determination, acquiring multimodal related information related to the target information, calculating the degree of relevance between each modal of the target information and the corresponding modal of the related information, and outputting the calculated degree of relevance for each modal as output information.

[0024] In a third aspect of the present invention, there is provided a program which, when executed by a computer, causes the computer to acquire multimodal target information to be subjected to truth determination, acquire multimodal related information related to the target information, calculate the degree of relevance between each modality of the target information and the corresponding modality of the related information, and output the calculated degree of relevance for each modality as output information.

[0025] In a fourth aspect of the present invention, there is provided a synthesis degree determination support device comprising: a target information acquisition unit that acquires multimodal target information to be the subject of synthesis degree determination; a related information acquisition unit that acquires related information related to the target information; a calculation unit that calculates the synthesis degree of the target information; a judgment basis generation unit that generates, in natural language, a judgment basis for the synthesis degree for at least one modality of the target information based on the related information; and an output unit that outputs the judgment basis.

[0026] In the synthesis degree determination support device, the determination basis generation unit may generate the determination basis in natural language for each modal of the target information.

[0027] Any of the above-described combination degree determination support devices may include a combination method identification unit that identifies the combination method used to generate the target information.

[0028] In any of the above-described combination degree determination support devices, the determination basis generator may generate the determination basis using a predetermined generative model, and the device may include a model selector that selects, as the generative model, a machine learning model according to the content of the target information.

[0029] Any of the above-described synthesis degree determination support devices may include a specialization processing unit that causes a generative model for generating the determination basis to learn data corresponding to the target information.

[0030] The above summary of the invention does not list all of the features of the present invention, and subcombinations of these features may also be inventions. [Brief explanation of the drawings]

[0031] [Figure 1] 1 shows an outline of the configuration of an authenticity determination support device 100. [Figure 2] 10 shows an example of target information, related information, and a degree of relevance. [Figure 3] 10 shows a modified example of the authenticity determination support device 100. [Figure 4] An example of target information, related information, relevance, and judgment basis is shown below. [Figure 5] An example of target information, related information, relevance, truth determination, and determination basis is shown below. [Figure 6] An example of target information, a truth determination, and a basis for the determination is shown below. [Figure 7] Examples of target information, objective information, related information, truth / falseness judgment, judgment basis, and information source information are shown below. [Figure 8] An example of target information, degree of synthesis, and basis for judgment is shown below. [Figure 9] 10 shows a modified example of the authenticity determination support device 100. [Figure 10A] An example of acquisition of target information by the target information acquisition unit 110 will be described. [Figure 10B] An example of acquisition of target information by the target information acquisition unit 110 will be described. [Figure 10C] An example of acquisition of target information by the target information acquisition unit 110 will be described. [Figure 11] 1 shows an outline of the configuration of a synthesis degree determination support device 200. [Figure 12] An example of target information, related information, degree of synthesis, and basis for judgment is shown below. [Figure 13] 10 shows a modified example of the synthesis degree determination support device 200. [Figure 14] Examples of target information, related information, degree of synthesis, basis for judgment, and synthesis method are shown below. [Figure 15] 22 illustrates an example computer 2200 in which aspects of the present invention may be embodied, in whole or in part. DETAILED DESCRIPTION OF THE INVENTION

[0032] The present invention will be described below through embodiments of the invention, but the following embodiments do not limit the scope of the invention according to the claims. Furthermore, not all of the combinations of features described in the embodiments are necessarily essential to the solution of the invention.

[0033] FIG. 1 shows an overview of the configuration of an authenticity determination support device 100. The authenticity determination support device 100 of this example includes a target information acquisition unit 110, a related information acquisition unit 120, a calculation unit 130, and an output unit 140. The authenticity determination support device 100 may also include a display unit 150. Note that the blocks shown are functionally separated functional blocks and may not necessarily match the actual device configuration. In other words, blocks shown as one block in this diagram may not necessarily be configured by one device. Furthermore, blocks shown as separate blocks in this diagram may not necessarily be configured by separate devices. The same applies to blocks in other diagrams.

[0034] The authenticity determination support device 100 supports the authenticity determination of target information. The target information may include multimodal information. Multimodal information is information that includes multiple types of data. The multimodal information may include at least two or more of text data, image data, video data, audio data, experimental data, and precipitation probability data. For example, the multimodal information may include text data and image data, video data and audio data, audio data and text data, experimental data and text data, or precipitation probability data and text data. However, the multimodal information is not limited to these combinations.

[0035] The target information acquisition unit 110 acquires multimodal target information to be subjected to truth / falsehood judgment. The target information acquisition unit 110 may acquire the target information by displaying a screen for inputting the target information on a display device and having the user input the target information. However, the method for acquiring the target information by the target information acquisition unit 110 is not limited to this. The target information acquisition unit 110 may supply the acquired target information and / or information related to the target information to the related information acquisition unit 120 and the calculation unit 130.

[0036] The related information acquisition unit 120 acquires multimodal related information related to the target information. The related information acquisition unit 120 may acquire the related information from a predetermined database. The type of data included in the multimodal related information acquired by the related information acquisition unit 120 may be the same as the type of data included in the multimodal target information. For example, if the multimodal target information acquired by the target information acquisition unit 110 includes text data and image data, the multimodal related information acquired by the related information acquisition unit 120 may include text data and image data. The related information acquisition unit 120 may supply the acquired related information to the calculation unit 130.

[0037] The calculation unit 130 calculates the degree of relevance between each modal of the target information and the corresponding modal of the related information. The calculation unit 130 may supply the calculated degrees of relevance to the output unit 140. The calculation unit 130 may calculate the degree of relevance between a modal of the target information and a modal of the corresponding related information using a predetermined machine learning model. For example, the machine learning model includes at least one of a convolutional neural network (CNN), a recurrent neural network (RNN), a long short-term memory network (LSTM), or a transformer. However, the machine learning model used by the calculation unit 130 to calculate the degree of relevance is not limited to these.

[0038] For example, when the target information includes text data and image data, and the related information includes text data and image data, the calculation unit 130 may calculate the degree of association between the text data included in the target information and the text data included in the related information, and the degree of association between the image data included in the target information and the image data included in the related information. The same applies when the target information and the related information include data in other formats.

[0039] The output unit 140 outputs the calculated relevance for each modal as output information. That is, the output information output by the output unit 140 includes the relevance. However, the information included in the output information is not limited to this. For example, the output information may include target information, related information, the relevance, a truth determination result for the target information, a determination result for the degree of synthesis of the data included in the target information, and the basis for the calculation and determination. Details of the truth determination, the determination of the degree of synthesis, and the basis for the determination will be described later. The output unit 140 may display the output information on the display unit 150.

[0040] As described above, the authenticity determination support device 100 of this example can assist in determining whether the target information is authentic by acquiring related information related to the target information and calculating and outputting the relevance between the target information and the related information. For example, a user of the authenticity determination support device 100 can determine whether the target information is authentic by referring to the relevance for each modal that is output.

[0041] For example, when a government agency or a local government monitors a social networking service (SNS) in the event of a natural disaster, the authenticity determination support device 100 can be used to determine the authenticity of information on the SNS. This allows measures to be taken against the spread of false information.

[0042] Furthermore, the authenticity determination support device 100 of this embodiment targets multimodal target information for authenticity determination, thereby enabling support for multifaceted authenticity determination.

[0043] FIG. 2 shows an example of target information, related information, and relevance. In this example, the target information includes text data and image data. In this example, the related information includes text data and image data. The output unit 140 may output the target information, related information, and relevance as output information. The output unit 140 may display the output information on the display unit 150. The output unit 140 may display the target information on a modal basis on the display unit 150. For example, the display unit 150 may display the output information in a format as shown in FIG. 2. However, the output information output by the output unit 140 and the display method by the display unit 150 are not limited to this example.

[0044] The target information acquisition unit 110 may summarize the text data of the acquired target information. For example, the target information acquisition unit 110 summarizes text data that describes in detail that a large tsunami occurred in City A as "large tsunami in City A." The related information acquisition unit 120 may summarize text data of the acquired related information. For example, the related information acquisition unit 120 summarizes text data that describes in detail that a large tsunami occurred in City B as "large tsunami in City B." The same applies to "fish landing volume decreased in City C" and "houses collapsed in City A." However, the target information acquisition unit 110 and / or the related information acquisition unit 120 do not have to summarize the text data.

[0045] The calculation unit 130 calculates the degree of relevance between each modality of the target information and the corresponding modality of the related information. In this example, the calculation unit 130 calculates the degree of relevance between text data included in the target information and text data included in the related information, and the degree of relevance between image data included in the target information and image data included in the related information. The calculation unit 130 may classify information included in the text data and calculate the degree of relevance for each classified information. For example, the calculation unit 130 classifies information included in the text data into location information and time information and calculates the degree of relevance for each.

[0046] The calculation unit 130 may calculate at least one of the similarity of image information, the similarity of person information, the coincidence of location information, or the coincidence of time information as the relevance. In this example, the calculation unit 130 calculates the similarity of image information, the coincidence of location information, and the coincidence of time information. The type of relevance calculated by the calculation unit 130 may be determined depending on the multimodal information included in the target information and the related information.

[0047] Fig. 3 shows a modified example of the authenticity determination support device 100. The authenticity determination support device 100 of this example differs from the embodiment of Fig. 1 in that it includes an authenticity determination unit 160, a synthesis degree determination unit 170, and a determination basis generation unit 180. In this example, differences from the embodiment of Fig. 1 will be particularly described, and the rest may be the same as the embodiment of Fig. 1.

[0048] The truth determination unit 160 may determine the truth of the target information based on the relevance for each modal. The truth determination unit 160 may determine the truth of the target information using a binary value of "true" or "false," or may determine the truth of the target information using a score indicating the degree to which the target information is true or false. For example, when the target information is likely to be false, the truth determination unit 160 determines that the false information score indicating the possibility that the target information is false is 80%, for example. The truth determination unit 160 may supply the determination result of the truth of the target information to the output unit 140. The output unit 140 may output the determination result of the truth as output information.

[0049] The synthesis degree determination unit 170 may determine the degree of synthesis of the data included in the target information. The synthesis degree determination unit 170 may determine the degree of synthesis of the data included in the target information using a score indicating the degree of synthesis of the data included in the target information. For example, when there is a high possibility that the data included in the target information is synthesized, the synthesis degree determination unit 170 may determine that the synthesis score indicating the possibility that the data is synthesized is 80%. The synthesis degree determination unit 170 may supply the determination result of the synthesis degree of the data included in the target information to the output unit 140. The output unit 140 may output the determination result of the synthesis degree as output information.

[0050] The synthesis degree determination unit 170 may supply the determination result of the synthesis degree of the data included in the target information to the authenticity determination unit 160. The authenticity determination unit 160 may determine the authenticity of the target information based on the synthesis degree determined by the synthesis degree determination unit 170. For example, when the synthesis degree of the data included in the target information is high, the authenticity determination unit 160 may determine that the target information is likely to be false information.

[0051] The determination basis generation unit 180 may generate the determination basis of the output information in natural language. For example, the determination basis may include the reliability or authenticity information of the sender. The authenticity information may include a digital watermark, a digital signature, or provenance information conforming to C2PA. The determination basis generation unit 180 may generate a plurality of determination basis. The determination basis generation unit 180 may supply the generated determination basis to the output unit 140. The output unit 140 may output the determination basis as output information.

[0052] The determination basis generator 180 may generate the determination basis using a predetermined generative model. For example, the generative model includes at least one of a large language model (LLM) or a large vision language model (VLM). However, the generative model used by the determination basis generator 180 to generate the determination basis is not limited to these.

[0053] The determination basis generating unit 180 may generate a determination basis for each modal of the target information in natural language. For example, if the target information includes text data and image data, the determination basis generating unit 180 may generate a determination basis for the text data and a determination basis for the image data, respectively.

[0054] The determination basis generation unit 180 may generate, in natural language, a determination basis for determining whether the target information is true or false. The determination basis generation unit 180 may generate, in natural language, a determination basis for the degree of synthesis of the target information. In this way, the determination basis generation unit 180 may generate a determination basis for each piece of information according to the type of information included in the output information.

[0055] The determination basis generation unit 180 may generate the reliability of the determination basis. When the determination basis generation unit 180 generates a plurality of determination basis, the determination basis generation unit 180 may generate the reliability of each determination basis. The output unit 140 may output the reliability of the determination basis as output information.

[0056] The output unit 140 may emphasize the basis for the judgment and display it on the display unit 150. For example, the output unit 140 may emphasize the basis for the judgment and display it on the display unit 150 by writing the text in bold. As another example, the output unit 140 may emphasize the basis for the judgment and display it on the display unit 150 by superimposing a heat map on image data included in the target information. As yet another example, the output unit 140 may emphasize the basis for the judgment and display it on the display unit 150 by increasing the volume of audio data included in the target information, or by displaying a score that changes according to the playback time and increasing the score for a specific time interval. The same applies when the target information includes data in other formats.

[0057] The output section 140 may determine whether to highlight the basis for judgment and display it on the display section 150, depending on the reliability of the basis for judgment generated by the basis for judgment generation section 180. For example, the output section 140 highlights the basis for judgment whose reliability is higher than a predetermined reference reliability and displays it on the display section 150.

[0058] The output unit 140 may display the basis for the judgment as a score for each predetermined viewpoint on the display unit 150. The viewpoint may include at least one of the reliability of the text, the reliability of the image, the reliability of the audio, the comprehensiveness of the evidence relative to the objective information, or the matching rate of the evidence relative to the objective information.

[0059] The authenticity determination support device 100 of this example includes a determination basis generation unit 180 that generates a determination basis in natural language. This can improve the explainability of the calculation result calculated by the calculation unit 130, the authenticity determination result of the target information determined by the authenticity determination unit 160, and / or the determination result of the degree of synthesis determined by the synthesis degree determination unit 170.

[0060] FIG. 4 shows an example of target information, related information, relevance, and judgment basis. In this example, the target information includes text data and image data. In this example, the related information includes text data and image data. The output unit 140 may output the target information, related information, relevance, and judgment basis as output information. The output unit 140 may display the output information on the display unit 150. For example, the display unit 150 may display the output information in a format as shown in FIG. 4. However, the output information output by the output unit 140 and the display method by the display unit 150 are not limited to this example.

[0061] The calculation unit 130 may calculate at least one of the similarity of image information, the similarity of person information, the degree of agreement of location information, or the degree of agreement of time information as the relevance, and the determination basis generation unit 180 may generate a determination basis for each modal of target information in natural language. The calculation unit 130 of this example calculates the similarity of image information, the degree of agreement of location information, and the degree of agreement of time information. The determination basis generation unit 180 of this example generates a determination basis for image information, a determination basis for location information, and a determination basis for time information, respectively.

[0062] For example, since the image data included in the target information and the image data included in the related information in the upper row match, the calculation unit 130 calculates the image similarity to be 1, i.e., a perfect match, and the determination basis generation unit 180 generates the determination basis in natural language, "The images match perfectly." Since the location information included in the target information and the location information included in the related information in the upper row are different, the calculation unit 130 calculates the match degree of the location information as a mismatch, and the determination basis generation unit 180 generates the determination basis in natural language, "City A and City B do not match." Since the time information included in the target information and the time information included in the related information in the upper row are different, the calculation unit 130 calculates the match degree of the time information as a mismatch, and the determination basis generation unit 180 generates the determination basis in natural language, "20XX and 20YY do not match." The same applies to the related information in the middle and lower rows.

[0063] FIG. 5 shows an example of target information, related information, relevance, true / false determination, and determination basis. The target information in this example includes text data and image data. The related information in this example includes text data and image data. The output unit 140 may output the target information, related information, relevance, true / false determination, and determination basis as output information. The output unit 140 may display the output information on the display unit 150. For example, the display unit 150 may display the output information in a format as shown in FIG. 5. However, the output information output by the output unit 140 and the display method by the display unit 150 are not limited to this example.

[0064] The truth determination unit 160 may determine the truth of the target information based on the relevance for each modal. The determination basis generation unit 180 may generate, in natural language, the basis for determining the truth of the target information. For example, although the image data included in the target information matches the image data included in the related information in the upper row, the location information and time information are different, so the truth determination unit 160 determines that the target information is likely to be fake information. The determination basis generation unit 180 generates the determination basis in natural language as follows: "The location and time of the article whose image matches the target information do not match the location and time of the target information, and the target information is newer, so the target information is likely to be fake information that uses past images."

[0065] The truth determination unit 160 may determine the degree to which the target information is true or false using a score that indicates the degree to which the target information is true or false. In this example, the truth determination unit 160 determines the degree to which the target information is false using a false information score of 90%.

[0066] 6 shows an example of target information, a true / false judgment, and the basis for the judgment. The output unit 140 may display the basis for the judgment on the display unit 150 as a score for each predetermined viewpoint. The viewpoints in this example include the reliability of the text, the reliability of the image, the comprehensiveness of the evidence relative to the objective information, and the match rate of the evidence relative to the objective information.

[0067] 7 shows an example of target information, objective information, related information, truth / falsehood judgment, judgment grounds, and information source information. The target information in this example includes text data as at least one data item of the multimodal target information.

[0068] The target information acquisition unit 110 may extract objective information from the acquired target information. In this example, the target information includes objective information indicating time, location, and event, and subjective information indicating the sender's opinion. In this example, the target information acquisition unit 110 extracts the objective information "now" and "earthquake in City A" from the target information.

[0069] The related information acquisition unit 120 may acquire related information that supports the objective information included in the target information. The output unit 140 may output related information that supports the objective information included in the target information. In this example, the related information acquisition unit 120 acquires information indicating that an earthquake has occurred in City A as related information. In this example, the output unit 140 outputs related information indicating that an earthquake has occurred in City A.

[0070] For example, since the objective information included in the target information is supported by related information, the truth determination unit 160 determines that the target information is not false information. The determination basis generation unit 180 generates the determination basis in natural language, such as "All of the objective information in the text information is supported by data."

[0071] The related information acquisition unit 120 may output information about the source of the related information that served as the basis for the judgment. The information about the source may be information indicating the information source from which the related information acquisition unit 120 acquired the related information, and may include, for example, a URL from which the related information was acquired. In this example, the related information acquisition unit 120 outputs a URL as information about the source from which the related information indicating that an earthquake occurred in City A was acquired. The output unit 140 may output information about the source of the related information as output information.

[0072] The output unit 140 may output the target information, objective information, related information, truth / false determination, judgment grounds, and information source information as output information. The output unit 140 may display the output information on the display unit 150. For example, the display unit 150 may display the output information in a format as shown in Fig. 7. However, the output information output by the output unit 140 and the display method by the display unit 150 are not limited to this example.

[0073] 8 shows an example of target information, the degree of synthesis, and the basis for determination. The target information in this example includes image data as at least one data item of the multimodal target information.

[0074] The synthesis degree determination unit 170 may determine the degree of synthesis of data included in the target information. For example, the synthesis degree determination unit 170 may determine that the image data included in the target information is unlikely to be synthesized. The output unit 140 may display a heat map superimposed on the image data included in the target information, thereby emphasizing the basis for the determination and displaying it on the display unit 150.

[0075] The determination basis generating unit 180 may generate multiple determination basis. In this example, the determination basis generating unit 180 generates the determination basis "features with a high possibility of being composite account for 20% or less of the image" and the determination basis "no eye features unique to composite images are observed."

[0076] The determination basis generating section 180 may generate a reliability of each determination basis. The output section 140 may determine whether or not to highlight the determination basis and display it on the display section 150, depending on the reliability of the determination basis generated by the determination basis generating section 180. In this example, the output section 140 highlights the determination basis with a higher reliability among the determination basis and displays it on the display section 150 in bold.

[0077] The output unit 140 may output the target information, the degree of synthesis, and the basis for the judgment as output information. The output unit 140 may cause the display unit 150 to display the output information. For example, the display unit 150 may display the output information in a format as shown in Fig. 8. However, the output information output by the output unit 140 and the display method by the display unit 150 are not limited to this example.

[0078] Fig. 9 shows a modified example of the authenticity determination support device 100. The authenticity determination support device 100 of this example differs from the embodiment of Fig. 1 in that it includes an information source designation unit 190. In this example, differences from the embodiment of Fig. 1 will be particularly described, and the rest may be the same as the embodiment of Fig. 1.

[0079] The information source designation unit 190 may designate a target information source for acquiring the target information. The target information source may include a URL for acquiring the target information, a search keyword for acquiring the target information, or an SNS or the like that serves as an information source for acquiring the target information. The information source designation unit 190 may designate the target information source by displaying a screen for inputting the target information source on the display device and having the user input the target information source. However, the method for acquiring the target information by the target information acquisition unit 110 is not limited to this. The information source designation unit 190 may designate multiple target information sources. The target information acquisition unit 110 may acquire the target information from the target information source designated by the information source designation unit 190.

[0080] In the authenticity determination support device 100 of this example, the information source designation unit 190 designates the target information source. This narrows the search range for the target information from among a vast number of information sources, making it possible to efficiently support authenticity determination.

[0081] 10A shows an example of acquisition of target information by the target information acquisition unit 110. The information source designation unit 190 designates, for example, a URL as a target information source for acquiring the target information. The target information acquisition unit 110 may acquire the target information from the URL designated by the information source designation unit 190.

[0082] 10B shows an example of acquisition of target information by the target information acquisition unit 110. The target information acquisition unit 110 may calculate a diffusion index of data from the target information source of the target information and acquire selected target information based on the diffusion index. The diffusion index may be an index indicating the number of impressions, citations, shares, reposts, bookmarks, likes, favorites, plays, or views of the target information source, or the number of people tracking the transmission of the sender of the target information, etc.

[0083] The target information acquisition unit 110 may calculate a diffusion index of data from multiple target information sources. In this example, the target information acquisition unit 110 calculates a diffusion index of data from three target information sources. The target information acquisition unit 110 may select target information based on the diffusion index. In this example, the target information acquisition unit 110 acquires target information from the target information source with the largest diffusion index among the three target information sources.

[0084] If the diffusion index of a target information source is high, there is a great merit in taking measures against the spread of false information if the target information source is false. By having the target information acquisition unit 110 acquire target information selected based on the diffusion index, it is possible to prioritize information for which there is a great merit in taking measures and to take measures against the spread of false information.

[0085] 10C shows an example of acquisition of target information by the target information acquisition unit 110. The target information acquisition unit 110 may calculate a degree of synthesis of data from the target information sources of the target information, and acquire target information selected based on the degree of synthesis.

[0086] The target information acquisition unit 110 may calculate the degree of synthesis of data from multiple target information sources. In this example, the target information acquisition unit 110 calculates the degree of synthesis of data from three target information sources. The target information acquisition unit 110 may select target information based on the degree of synthesis. In this example, the target information acquisition unit 110 acquires target information from the target information source with the highest degree of synthesis among the three target information sources.

[0087] When the degree of synthesis of a target information source is high, the target information source is likely to contain false information, and there is a great merit in taking measures against the spread of false information. By acquiring target information selected by the target information acquisition unit 110 based on the degree of synthesis, it is possible to prioritize information for which there is a great merit in taking measures and to take measures against the spread of false information.

[0088] 1, 3, and 9 described in relation to FIG. 1 to FIG. 10C may be provided together with the functional blocks described in other figures. That is, the authenticity determination support device 100 may include all or any combination of the functional blocks described in FIG. 1, 3, and 9.

[0089] 11 shows an outline of the configuration of the combination degree determination support device 200. The combination degree determination support device 200 of this example includes a target information acquisition unit 210, a related information acquisition unit 220, a calculation unit 230, a determination basis generation unit 240, and an output unit 250.

[0090] The combination degree determination support device 200 supports the determination of the combination degree of target information. For example, the combination degree determination support device 200 supports the determination of the combination degree indicating whether the target information has been generated by a generation AI.

[0091] The target information acquisition unit 210 acquires multimodal target information to be subjected to determination of the degree of synthesis. The target information acquisition unit 210 may acquire the target information by displaying a screen for inputting the target information on a display device and having the user input the target information. However, the method of acquiring the target information by the target information acquisition unit 210 is not limited thereto. The target information acquisition unit 210 may acquire the target information from a pre-specified target information source, or may calculate a diffusion index of data from the target information source of the target information and acquire selected target information based on the diffusion index. The target information acquisition unit 210 may supply the acquired target information and / or information related to the target information to the related information acquisition unit 220 and the calculation unit 230.

[0092] The related information acquisition unit 220 acquires related information related to the target information. The related information acquisition unit 220 may acquire related information from a predetermined database. The related information acquisition unit 220 may acquire multimodal related information including the same type of data as the type of data included in the multimodal target information, or may acquire related information including only data included in the target information that is the target of determining the degree of synthesis. For example, the target information includes text data and image data. When the target of determining the degree of synthesis is image data, the related information acquisition unit 220 may acquire information including image data as related information related to the target information. The related information acquisition unit 220 may supply the acquired related information to the calculation unit 230.

[0093] The calculation unit 230 calculates the degree of synthesis of the target information. The calculation unit 230 may calculate the degree of synthesis of the target information using a predetermined machine learning model. For example, the machine learning model includes at least one of a convolutional neural network (CNN), a recurrent neural network (RNN), a long short-term memory network (LSTM), or a transformer. However, the machine learning model used by the calculation unit 230 to calculate the degree of synthesis is not limited to these.

[0094] The determination basis generator 240 generates, in natural language, a determination basis for the degree of synthesis for at least one modal of the target information based on the related information. The determination basis generator 240 may generate, in natural language, a determination basis for each modal of the target information. The determination basis generator 240 may generate a reliability of the determination basis.

[0095] The output section 250 outputs the basis for the determination. The output section 250 may highlight the basis for the determination and display it on a predetermined display section. The output section 250 may output the degree of synthesis calculated by the calculation section 230.

[0096] As described above, the combination degree determination support device 200 of this example can assist in determining the combination degree of the target information by acquiring related information related to the target information, calculating the combination degree of the target information, and outputting the basis for the determination. For example, a user of the combination degree determination support device 200 can determine the combination degree of the target information by referring to the output basis for the determination. Furthermore, since the combination degree determination support device 200 of this example outputs the basis for the determination, the explainability of the calculation result of the combination degree can be improved.

[0097] 12 shows an example of target information, related information, a degree of synthesis, and a basis for determination. For example, the synthesis degree determination support device 200 supports the determination of the degree of synthesis of image data included in multimodal target information. However, the data for which the synthesis degree determination support device 200 supports the determination of the degree of synthesis is not limited to this.

[0098] The target information acquisition unit 210 acquires multimodal target information including image data. In this example, the image data of the target information includes a person image. The related information acquisition unit 220 may acquire image data including a person image as related information related to the target information. For example, the image data included in the related information may include image data that is entirely synthesized, image data that is partially synthesized, or image data that is not synthesized.

[0099] The calculation unit 230 calculates the degree of synthesis of the image data of the target information. In this example, the calculation unit 230 calculates that the image data is likely to be synthesized. The determination basis generation unit 240 generates a determination basis for the degree of synthesis for the image data of the target information in natural language. In this example, the determination basis generation unit 240 generates the determination basis in natural language, such as "80% or more of the features in the image are likely to be synthesized."

[0100] The output unit 250 may display a heat map superimposed on the image data included in the target information, thereby emphasizing the basis for the judgment and displaying it on a predetermined display unit. The output unit 250 may output the reliability of the basis for the judgment.

[0101] Fig. 13 shows a modified example of the combination degree determination support device 200. The combination degree determination support device 200 of this example differs from the embodiment of Fig. 11 in that it includes a combination method identification unit 260, a model storage unit 270, a model selection unit 280, and a specialization processing unit 290. In this example, differences from the embodiment of Fig. 11 will be particularly described, and the rest may be the same as the embodiment of Fig. 11.

[0102] The compositing method identification unit 260 may identify the compositing method used to generate the target information. The compositing method identification unit 260 may identify the compositing model used to generate the target information. The output unit 250 may output the identified compositing method.

[0103] The model storage unit 270 may store a predetermined generative model 272 for generating a decision basis. For example, the generative model 272 includes at least one of a large language model (LLM) or a large vision model (VLM). However, the generative model 272 is not limited to these. The model storage unit 270 may store a plurality of generative models 272. The decision basis generator 240 may generate the decision basis using the generative model 272.

[0104] The model selection unit 280 may select a machine learning model according to the content of the target information as the generative model 272. The determination basis generation unit 240 may generate a determination basis using the generative model 272 selected by the model selection unit 280.

[0105] The model selection unit 280 may select the generative model 272 by displaying a screen on the display device for selecting the generative model 272 and having the user select the generative model 272. However, the method for selecting the generative model 272 by the model selection unit 280 is not limited to this.

[0106] The specialization processing unit 290 may train the generative model 272 for generating the decision basis on data corresponding to the target information. For example, the model storage unit 270 stores a plurality of generative models 272 corresponding to the contents of the target information. The specialization processing unit 290 may perform the specialization process by training each of the plurality of generative models 272 on data corresponding to the target information. For example, a generative model 272 specialized for earthquakes is trained by training the generative model on data specialized for earthquakes. In other words, the specialization processing unit 290 may perform fine-tuning of the generative model 272 using data corresponding to the target information.

[0107] The specialization processing unit 290 may display on the display device a screen for selecting the generation model 272 for which the specialization processing is to be performed, and may have the user select the generation model 272, thereby causing the selected generation model 272 to learn data corresponding to the target information. However, the method by which the specialization processing unit 290 selects the generation model 272 for which the specialization processing is to be performed is not limited to this.

[0108] 14 shows an example of target information, related information, degree of synthesis, basis for determination, and synthesis method. The synthesis method identification unit 260 may identify the synthesis method used to generate the target information. In this example, the synthesis method identification unit 260 determines that it is highly likely that the image data of the target information was generated using image generation model A.

[0109] 11 and 13 described in relation to FIG. 11 to FIG. 14 may be provided together with the functional blocks described in other figures. That is, the composition degree determination support device 200 may include all or any combination of the functional blocks described in FIG. 11 and FIG. 13.

[0110] The target information acquisition unit 210, the related information acquisition unit 220, and the output unit 250 included in the synthesis degree determination support device 200 may have the same configuration as the target information acquisition unit 110, the related information acquisition unit 120, and the output unit 140 included in the authenticity determination support device 100. Therefore, it can be understood that the synthesis degree determination support device 200 and the authenticity determination support device 100 can be configured in a composite manner as the same device.

[0111] 15 illustrates an example of a computer 2200 in which aspects of the present invention may be embodied, in whole or in part. Programs installed on the computer 2200 may cause the computer 2200 to function as or perform operations associated with an apparatus or one or more sections of the apparatus according to embodiments of the present invention, and / or to perform a process or steps of a process according to embodiments of the present invention. Such programs may be executed by the CPU 2212 to cause the computer 2200 to perform specific operations associated with some or all of the blocks in the flowcharts and block diagrams described herein.

[0112] A computer 2200 according to this embodiment includes a CPU 2212, a RAM 2214, a graphics controller 2216, and a display device 2218, which are interconnected by a host controller 2210. The computer 2200 also includes input / output units such as a communication interface 2222, a hard disk drive 2224, a DVD-ROM drive 2226, and an IC card drive, which are connected to the host controller 2210 via an input / output controller 2220. The computer also includes legacy input / output units such as a ROM 2230 and a keyboard 2242, which are connected to the input / output controller 2220 via an input / output chip 2240.

[0113] The CPU 2212 operates according to programs stored in the ROM 2230 and RAM 2214, thereby controlling each unit. The graphics controller 2216 acquires image data generated by the CPU 2212 into a frame buffer or the like provided in the RAM 2214 or into the graphics controller 2216 itself, and causes the image data to be displayed on the display device 2218.

[0114] The communication interface 2222 communicates with other electronic devices via a network. The hard disk drive 2224 stores programs and data used by the CPU 2212 in the computer 2200. The DVD-ROM drive 2226 reads programs or data from the DVD-ROM 2201 and provides the programs or data to the hard disk drive 2224 via the RAM 2214. The IC card drive reads programs and data from an IC card and / or writes programs and data to an IC card.

[0115] The ROM 2230 stores therein a boot program or the like that is executed by the computer 2200 upon activation, and / or programs that depend on the hardware of the computer 2200. The input / output chip 2240 may also connect various input / output units to the input / output controller 2220 via a parallel port, a serial port, a keyboard port, a mouse port, etc.

[0116] The programs are provided by a computer-readable medium such as a DVD-ROM 2201 or an IC card. The programs are read from the computer-readable medium, installed in the hard disk drive 2224, RAM 2214, or ROM 2230, which are also examples of computer-readable media, and executed by the CPU 2212. Information processing described in these programs is read by the computer 2200, and brings about cooperation between the programs and the various types of hardware resources described above. An apparatus or method may be configured by realizing information manipulation or processing in accordance with the use of the computer 2200.

[0117] For example, when communication is performed between the computer 2200 and an external device, the CPU 2212 may execute a communication program loaded into the RAM 2214 and instruct the communication interface 2222 to perform communication processing based on the processing described in the communication program. Under the control of the CPU 2212, the communication interface 2222 reads transmission data stored in a transmission buffer processing area provided in the RAM 2214, the hard disk drive 2224, the DVD-ROM 2201, or a recording medium such as an IC card, and transmits the read transmission data to the network, or writes reception data received from the network to a reception buffer processing area or the like provided on the recording medium.

[0118] Furthermore, the CPU 2212 may cause all or a necessary portion of a file or database stored on an external recording medium such as the hard disk drive 2224, the DVD-ROM drive 2226 (DVD-ROM 2201), an IC card, etc. to be read into the RAM 2214, and perform various types of processing on the data on the RAM 2214. The CPU 2212 then writes back the processed data to the external recording medium.

[0119] Various types of information, such as various types of programs, data, tables, and databases, may be stored on the recording medium and subjected to information processing. The CPU 2212 may perform various types of processing on data read from the RAM 2214, including various types of operations, information processing, conditional judgment, conditional branching, unconditional branching, information search / replacement, etc., as described throughout this disclosure and specified by the instruction sequences of the programs, and write the results back to the RAM 2214. The CPU 2212 may also search for information in a file, database, etc. on the recording medium. For example, if multiple entries each having an attribute value of a first attribute associated with an attribute value of a second attribute are stored on the recording medium, the CPU 2212 may search for an entry that matches a condition specified by the attribute value of the first attribute from among the multiple entries, read the attribute value of the second attribute stored in the entry, and thereby obtain the attribute value of the second attribute associated with the first attribute that satisfies a predetermined condition.

[0120] The above-described programs or software modules may be stored in a computer-readable medium on or near the computer 2200. A recording medium such as a hard disk or RAM provided in a server system connected to a dedicated communication network or the Internet can also be used as a computer-readable medium, thereby providing the programs to the computer 2200 via the network.

[0121] Although the present invention has been described above using embodiments, the technical scope of the present invention is not limited to the scope described in the above embodiments. It will be apparent to those skilled in the art that various modifications and improvements can be made to the above embodiments. It is clear from the claims that such modifications and improvements can also be included within the technical scope of the present invention.

[0122] It should be noted that the execution order of each process, such as operations, procedures, steps, and stages, in the devices, systems, programs, and methods shown in the claims, specifications, and drawings is not specifically stated as "before," "prior to," etc., and that the processes can be performed in any order unless the output of a previous process is used in a subsequent process. Even if the operational flow in the claims, specifications, and drawings is described using "first," "next," etc. for convenience, this does not mean that the processes must be performed in this order. [Explanation of symbols]

[0123] 100...Authenticity judgment support device, 110...Target information acquisition unit, 120...Related information acquisition unit, 130...Calculation unit, 140...Output unit, 150...Display unit, 160...Authenticity judgment unit, 170...Synthesis degree judgment unit, 180...Judgment basis generation unit, 190...Information source designation unit, 200...Authenticity degree judgment support device, 210...Target information acquisition unit, 220...Related information acquisition unit, 230...Calculation unit, 240...Judgment basis generation unit, 250...Output unit, 260...Synthesis method identification unit, 270...Model storage unit, 272...Generative model , 280···Model selection unit, 290···Specialized processing unit, 2200···Computer, 2201···DVD-ROM, 2210···Host controller, 2212···CPU, 2214···RAM, 2216···Graphics controller, 2218···Display device, 2220···Input / output controller, 2222···Communication interface, 2224···Hard disk drive, 2226···DVD-ROM drive, 2230···ROM, 2240···Input / output chip, 2242···Keyboard

Claims

1. a target information acquisition unit that acquires multimodal target information to be subjected to truth / falsehood judgment; a related information acquisition unit that acquires multimodal related information related to the target information; a calculation unit that calculates a degree of association between each modal of the target information and a corresponding modal of the related information; an output unit that outputs the calculated relevance for each modality as output information; An authenticity determination support device comprising:

2. a truth determination unit that determines the truth of the target information based on the relevance for each modal The authenticity determination support device according to claim 1.

3. The target information includes image data and text data. The authenticity determination support device according to claim 1.

4. The related information includes image data and text data. The authenticity determination support device according to claim 1.

5. A synthesis degree determination unit that determines the degree of synthesis of data included in the target information The authenticity determination support device according to claim 1.

6. A decision basis generation unit that generates the decision basis of the output information in natural language The authenticity determination support device according to claim 1.

7. The determination basis generation unit generates a determination basis for each modal of the target information in natural language. The authenticity determination support device according to claim 6.

8. The determination basis generation unit generates a determination basis for determining the truth or falsity of the target information in natural language. The authenticity determination support device according to claim 6.

9. The determination basis generation unit generates a determination basis for the degree of synthesis of the target information in natural language. The authenticity determination support device according to claim 6.

10. The determination basis generation unit generates a reliability of the determination basis. The authenticity determination support device according to claim 6.

11. The output unit displays the basis for the determination on a predetermined display unit in an emphasized manner. The authenticity determination support device according to claim 6.

12. The output unit displays the basis for the judgment on a predetermined display unit as a score for each predetermined viewpoint. The authenticity determination support device according to claim 6.

13. The related information acquisition unit outputs information on the source of the related information that served as the basis for the determination. The authenticity determination support device according to claim 6.

14. an information source specifying unit that specifies a target information source for acquiring the target information; The target information acquisition unit acquires the target information from the target information source designated by the information source designation unit. The authenticity determination support device according to any one of claims 1 to 13.

15. The output unit displays the target information on a display unit predetermined for each modality. The authenticity determination support device according to any one of claims 1 to 13.

16. The output unit outputs the related information that supports objective information included in the target information. The authenticity determination support device according to any one of claims 1 to 13.

17. The target information acquisition unit calculates a diffusion index of data from a target information source of the target information, and acquires the target information selected based on the diffusion index. The authenticity determination support device according to any one of claims 1 to 13.

18. The target information acquisition unit calculates a degree of synthesis of data from target information sources of the target information, and acquires the target information selected based on the degree of synthesis. The authenticity determination support device according to any one of claims 1 to 13.

19. The target information acquisition unit extracts objective information from the acquired target information. The authenticity determination support device according to any one of claims 1 to 13.

20. The calculation unit calculates, as the relevance level, at least one of a similarity level of image information, a similarity level of person information, a coincidence level of location information, or a coincidence level of time information. The authenticity determination support device according to any one of claims 1 to 13.

21. acquiring multimodal target information to be judged as true or false; acquiring multimodal related information related to the target information; calculating a degree of association between each modal element of the target information and a corresponding modal element of the related information; outputting the calculated relevance for each modality as output information; A method for supporting authenticity determination.

22. When executed by a computer, the computer: Acquire multimodal target information to be judged as true or false, obtaining multimodal related information related to the target information; calculating a degree of relevance between each modal of the target information and a corresponding modal of the related information; outputting the calculated relevance for each modality as output information; program.

23. an object information acquisition unit that acquires multimodal object information that is a target for determining a degree of synthesis; a related information acquisition unit that acquires related information related to the target information; a calculation unit that calculates a degree of synthesis of the target information; a determination basis generation unit that generates, in natural language, a determination basis for the degree of synthesis for at least one modal of the target information based on the related information; an output unit that outputs the basis for the determination; A synthesis degree determination support device comprising:

24. The determination basis generation unit generates the determination basis in natural language for each modal of the target information. The synthesis degree determination support device according to claim 23.

25. a synthesis method identification unit that identifies the synthesis method used to generate the target information; 25. The apparatus for supporting determination of a degree of synthesis according to claim 23 or 24.

26. the determination basis generation unit generates the determination basis using a predetermined generative model; A model selection unit is provided that selects a machine learning model according to the content of the target information as the generation model.

25. The apparatus for supporting determination of a degree of synthesis according to claim 23 or 24.

27. A specialized processing unit is provided for causing a generation model for generating the judgment basis to learn data corresponding to the target information.

25. The apparatus for supporting determination of a degree of synthesis according to claim 23 or 24.

Images