Support device, support method, and program

Abstract

The present invention provides a support device for assisting in the determination of the truth value of input information, comprising: a target object acquisition unit for acquiring a target object to be determined; a query determination unit for determining a search query according to whether the target object is single-modal or multimodal; a result acquisition unit for acquiring the results of related information relating to the target object using the determined search query; and an output unit for outputting the results. The present invention also provides a support method for assisting in the determination of the truth value of input information, comprising: a step of a computer acquiring a target object to be determined; a step of a computer determining a search query according to whether the target object is single-modal or multimodal; a step of a computer acquiring the results of related information relating to the target object using the determined search query; and a step of a computer outputting the results.

Description

Technical Field

[0001] The present invention relates to a support device, a support method, and a program.

Background Art

[0002] Conventionally, it is known to "perform output based on the reliability of a web article" (see, for example, Patent Document 1). Patent Document 1 Japanese Patent Application Laid-Open No. 2025-041060

[0003] Provide a support device for assisting a user in determining authenticity.

Summary of the Invention

[0004] In a first aspect of the present invention, there is provided a support device for assisting in determining the authenticity of input information, the support device including: an object acquisition unit that acquires an object to be the subject of authenticity determination; a query determination unit that determines a search query according to whether the object is single-modal or multi-modal; an investigation result acquisition unit that acquires an investigation result of related information related to the object using the determined search query; and an output unit that outputs the investigation result.

[0005] In the above support device, when the object includes any one modality type of text format, image format, audio format, or video format, the query determination unit may determine that the object is single-modal. When the object includes at least two modality types of text format, image format, audio format, or video format, the query determination unit may determine that the object is multi-modal.

[0006] In any of the above support devices, the query determination unit may decompose the object determined to be single-modal into a plurality of fact elements, and generate a plurality of search queries according to the plurality of fact elements.

[0007] In any of the above-described support devices, the investigation result acquisition unit may acquire at least one investigation result for each of the multiple factual elements using the multiple search queries. The output unit may output at least one investigation result for each of the multiple factual elements.

[0008] In any of the above-mentioned support devices, if the query determination unit determines that the target object is multimodal, it may generate different search queries for each of the multiple modality types included in the target object.

[0009] In any of the above-described support devices, the investigation result acquisition unit may acquire at least one investigation result for each of the multiple modality types using the search query for each of the multiple modality types. The output unit may output at least one investigation result for each of the multiple modality types.

[0010] Any of the above-described support devices may include an estimation unit that integrates the investigation results for each of the multiple modality types to estimate the authenticity of the target object.

[0011] In any of the above-described support devices, if the query determination unit determines that the target object is multimodal, it may generate a hybrid query common to the multiple modality types of the target object based on the multiple modality types of the target object.

[0012] In any of the above-described support devices, the query determination unit may determine the target object as the search query. The investigation result acquisition unit may use the target object to acquire the investigation result.

[0013] Any of the above support devices may include a change instruction receiving unit that receives instructions to change the search query. The query determination unit may change the determined search query in response to the change instruction.

[0014] Any of the above support devices may include a sharing instruction receiving unit that receives a sharing instruction for sharing the survey results. The output unit may output the survey results to the sharing destination corresponding to the sharing instruction.

[0015] Any of the above-mentioned support devices may include an access verification unit that accesses the URL of the survey result and verifies whether or not the URL is accessible.

[0016] Any of the above-mentioned support devices may include a reliability determination unit that determines the reliability of the survey results based on at least one of the contents of the survey results and the contents of the source of the survey results.

[0017] In any of the above-described support devices, the reliability determination unit may determine whether the source of the investigation results is a reliable source that satisfies predetermined conditions. If the output unit determines that the source of the investigation results is reliable, it may output the investigation results with an identifier indicating the reliability level.

[0018] In any of the above-mentioned support devices, if the confidence level of the investigation results is below an arbitrary threshold value, the query determination unit may determine a new search query using the information from the investigation results.

[0019] In any of the above-described support devices, the investigation result acquisition unit may acquire multiple investigation results. Any of the above-described support devices may include a consistency evaluation unit that compares the contents of the multiple investigation results and evaluates the consistency of the multiple investigation results.

[0020] In any of the above-described support devices, the consistency evaluation unit may evaluate the consistency between the first investigation result related to the target object and the second investigation result related to the target object when the target object is single-modal.

[0021] In any of the above support devices, when the target object is multimodal, the consistency evaluation unit may evaluate the consistency between a first investigation result related to the first modality type of the target object and a second investigation result related to the second modality type of the target object.

[0022] In any of the above support devices, the output unit may determine whether to output the investigation result according to the evaluation result of the consistency.

[0023] In any of the above support devices, the output unit may output a URL or link indicating the location of the information source as the investigation result.

[0024] In any of the above support devices, the output unit may output attribute information including at least one of person information, location information, or time information as the investigation result.

[0025] In any of the above support devices, the investigation result acquisition unit may acquire the investigation result based on a specified whitelist or blacklist.

[0026] In any of the above support devices, the investigation result acquisition unit may acquire the investigation result based on specified period information.

[0027] Any of the above support devices may include an opinion analysis unit that analyzes at least one of positive opinions or negative opinions on the input information. The output unit may output the analysis result of the opinion analysis unit.

[0028] In any of the above support devices, the output unit may output at least one of the authenticity determination result of the target object, the reliability score of the investigation result, or the basis list of the investigation result.

[0029] In any of the above support devices, the output unit may output at least one of the title, source information, or date information of the investigation result.

[0030] The output unit may output the degree of coincidence between the content of the target object and the survey result.

[0031] The output unit may sort a plurality of the survey results based on at least one of the degree of coincidence or the date information of the survey result.

[0032] Any of the above support devices may include a storage unit for storing at least one of the input information, the target object, or the survey result.

[0033] In any of the above support devices, the output unit may output a list of at least a part or all of at least one of the past input information, the target object, or the survey result stored in the storage unit.

[0034] In any of the above support devices, the output unit may output at least one of statistical information regarding the creator of the input information, the regional distribution of the surveyed content, a word cloud, statistical information regarding the target object, statistical information regarding the creator of the target object, or the ratio of the topics of the content, from at least one of the input information, the target object, or the survey result stored in the storage unit.

[0035] In any of the above support devices, the target object acquisition unit may acquire a plurality of target objects. The survey result acquisition unit may acquire the survey results of the plurality of target objects. The output unit may output the survey results of two or more similar target objects among the plurality of target objects in an associated manner.

[0036] In any of the above support devices, the target object acquisition unit may extract and acquire the target object from the input information in response to receiving the input information.

[0037] In a second aspect of the present invention, a support method is provided for supporting the determination of the truth value of input information in any of the above-described support devices, the support method comprising: a step of a computer acquiring a target object to be determined to be true or false; a step of a computer determining a search query according to whether the target object is single-modal or multimodal; a step of a computer using the determined search query to acquire the results of the investigation of related information relating to the target object; and a step of a computer outputting the results of the investigation.

[0038] In a third aspect of the present invention, a program is provided that, when executed by a computer in any of the above-mentioned support devices, causes the computer to assist in determining the truth value of input information, the program causes the computer to acquire a target object to be determined as truth value, to determine a search query according to whether the target object is single-modal or multimodal, to acquire the results of the investigation of related information related to the target object using the determined search query, and to output the results of the investigation.

[0039] It should be noted that the above summary of the invention does not enumerate all of its features. Furthermore, subcombinations of these features may also constitute an invention. [Brief explanation of the drawing]

[0040] [Figure 1A] This is a block diagram showing an overview of the configuration of the support device 100. [Figure 1B] An example of the operation flowchart of the support device 100 is shown. [Figure 2A] An example of the input screen for the "In" information is shown. [Figure 2B] An example of the input screen for the "In" information is shown. [Figure 3A] Here is an example of a single-modal target object Ot. [Figure 3B] An example of how to obtain the survey results Rs is shown below. [Figure 4A]An example of a multimodal target object Ot is shown. [Figure 4B] An example of how to obtain the survey results Rs is shown below. [Figure 4C] An example of how to obtain the survey results Rs is shown below. [Figure 5A] This is an example of the analysis results of the text contained in the target object Ot. [Figure 5B] This is an example of the analysis results for videos contained in the target object Ot. [Figure 5C] This is an example of the analysis results of the audio contained in the target object Ot. [Figure 5D] This is an example of the analysis results of the text contained in the target object Ot. [Figure 6A] This is a block diagram showing a modified example of the support device 100. [Figure 6B] An example of the configuration of the analysis unit 60 is shown. [Figure 7] This is an example of the analysis results of the text contained in the target object Ot. [Figure 8A] This shows an example of the results of an evaluation of the consistency between the text survey results Rs and the video survey results Rs. [Figure 8B] An example of the consistency evaluation results for survey results Rs is shown. [Figure 9] This shows an example of how to output the results Rs for multiple target objects Ot. [Figure 10] This is a block diagram showing a modified example of the support device 100. [Figure 11] An example of a filtering method for survey results Rs is shown below. [Figure 12] This is a block diagram showing a modified example of the support device 100. [Figure 13] An example of the information output by the output unit 40 is shown below. [Figure 14] An example of the information output by the output unit 40 is shown below. [Figure 15] Examples of a computer 2200 in which multiple aspects of the present invention may be embodied in whole or in part are shown. [Modes for carrying out the invention]

[0041] The present invention will be described below through embodiments, but these embodiments are not intended to limit the scope of the claims. Furthermore, not all combinations of features described in the embodiments are necessarily essential to the solution of the invention.

[0042] Figure 1A is a block diagram showing an overview of the configuration of the support device 100. The support device 100 comprises a target object acquisition unit 10, a query determination unit 20, a results acquisition unit 30, and an output unit 40. The support device 100 assists in determining the truth value of input information In input to the support device 100. The support device 100 may output results Rs corresponding to the input information In to assist the user in determining the truth value of the input information In.

[0043] The target object acquisition unit 10 acquires the target object Ot that is the subject of truth value determination. The target object acquisition unit 10 may extract and acquire the target object Ot from the input information In. The target object acquisition unit 10 may acquire the target object Ot entered by the user. That is, the target object acquisition unit 10 may extract text, images, audio, or video as the target object Ot from the URL entered as input information In, or it may acquire the entered text, images, audio, or video as the target object Ot.

[0044] The input information In includes the target object Ot. The input information In may include comments on the target object Ot and may include information about the poster of the target object Ot. The input information In may be a web page specified by a URL, a post on a social networking service (SNS), or a video page on a video sharing site.

[0045] The target object Ot may include at least one modality type: text, image, audio, or video. The target object Ot may include a combination of two or more modality types: text, image, audio, or video. The target object Ot may be a news article posted on a website, the text, images, videos, or audio of a social media post, or a video from a video sharing site. The target object Ot may be both text and video posted on social media.

[0046] The target object acquisition unit 10 may extract the target object Ot from the input information In using any machine learning model. The target object acquisition unit 10 may automatically identify and acquire the target object Ot, which contains the core factual claims of the subject of investigation, from the input information In. For example, the machine learning model may include at least one of a Convolutional Neural Network (CNN), a Recurrent Neural Network (RNN), a Long Short-Term Memory (LSTM), or a Transformer. However, the machine learning model used by the target object acquisition unit 10 to acquire the target object Ot is not limited to these.

[0047] The query determination unit 20 determines a search query to obtain the survey results Rs. The query determination unit 20 may generate a search query or select one from multiple candidates. The query determination unit 20 may also determine the target object Ot as the search query, as will be described later. The query determination unit 20 may determine a search query depending on whether the target object Ot is single-modal or multimodal. That is, the query determination unit 20 may use different search queries depending on whether the target object Ot is single-modal or multimodal. The query determination unit 20 may provide the determined search query to the survey results acquisition unit 30.

[0048] The query determination unit 20 may determine whether the target object Ot is single-modal or multimodal. The query determination unit 20 may determine that the target object Ot is single-modal if it contains one modality type from among text format, image format, audio format, or video format. The query determination unit 20 may determine that the target object Ot is multimodal if it contains at least two modality types from among text format, image format, audio format, or video format. The query determination unit 20 may determine that the target object Ot contains both text format and video format modality types if the target object Ot contains audio in the video. The query determination unit 20 may determine that the target object Ot contains both audio format and video format modality types if the target object Ot contains audio in the video.

[0049] The query determination unit 20 may use any generative model to generate a search query corresponding to the target object Ot and obtain the search results Rs. The search query may include at least one of text, images, audio, or video. For example, the generative model may include at least one of a Large Language Model (LLM) or a Large Vision Language Model (VLM). The query determination unit 20 may use the VLM to recognize a specific building, logo, or action in an image and generate a search query. The query determination unit 20 may obtain text using OCR (Optical Character Recognition) or transcribe audio into text using speech recognition technology.

[0050] The results acquisition unit 30 acquires results Rs of related information related to the target object Ot using the search query determined by the query determination unit 20. The related information may be information related to the factual part of the target object Ot. The content of the results Rs may include related information. The results acquisition unit 30 may acquire results Rs using a generative model such as a Large Language Model (LLM). The results acquisition unit 30 may use the same generative model as the query determination unit 20. The results acquisition unit 30 may acquire multiple results Rs. The results acquisition unit 30 may acquire results Rs using any web search system with a search query determined by the query determination unit 20.

[0051] The research results Rs include search results extracted using the search query. The search results may be web pages matching the search query, or social media posts. The search results may also include press releases from public institutions or commentary articles by experts. The research results Rs may include URLs or links indicating the location of the information sources that constitute the search results. The search results may include related objects associated with the target object Ot. Related objects may be at least one of text, images, audio, or video. Related objects are examples of related information.

[0052] The research results Rs may include, in addition to the search results, a summary of the search results, an excerpt of the search results, or an explanation of the search results. In other words, the research results Rs may include information processed from the search results obtained using the search query. The research results Rs may include a list of web pages that match the search query. The processing of the search results may be achieved using any generative model. The research results Rs may include claims based on facts extracted from the search results.

[0053] The survey results Rs may include attribute information, which may include at least one of the following: person information, place information, or time information. The attribute information may be data such as the names of the people involved, the locations related to the events being investigated, or the time when the events being investigated occurred, extracted from the survey results Rs. The survey results acquisition unit 30 may acquire one or more related objects related to the target object Ot as a search result of the image search. The survey results acquisition unit 30 may acquire the publication date, title, or thumbnail of the related objects. The survey results acquisition unit 30 may extract attribute information from the related objects. Attribute information is an example of related information.

[0054] If the target object Ot contains context information, the investigation result acquisition unit 30 may or may not consider the context information of the target object Ot. The context information of the target object Ot may include text information such as text, summary, overview, or comment information associated with the target object Ot. The query determination unit 20 may determine the search query by considering the target object Ot and the context information of the target object Ot, or it may determine the search query without considering the context information of the target object Ot.

[0055] If the target object Ot contains both factual and opinional portions, the investigation result acquisition unit 30 may separate the factual and opinional portions of the target object Ot and acquire the investigation result Rs. The factual portion of the target object Ot is the part expressed as objective facts. The opinional portion of the target object Ot is the part expressed subjectively as the poster's opinion. The query determination unit 20 may generate a search query based only on the factual portion of the target object Ot, without reflecting the opinional portion of the target object Ot in the search query. By making only the factual portion of the target object Ot the subject of truth / false determination, the investigation result Rs necessary for truth / false determination can be acquired more efficiently.

[0056] The output unit 40 outputs the survey results Rs. The output unit 40 may output the survey results Rs to the display unit 45. The output unit 40 may output the survey results Rs to the storage unit 50, or to an external source of the support device 100. The output unit 40 may output the survey results Rs as audio.

[0057] The display unit 45 displays the survey results Rs output by the output unit 40. The specific display screen of the display unit 45 will be described later. The display unit 45 may be provided on the support device 100 or on the outside of the support device 100. The display unit 45 may be any display provided on the outside of the support device 100, or it may be the user's mobile terminal or the like.

[0058] The storage unit 50 stores at least one of the following: input information In, target object Ot, or investigation result Rs. The storage unit 50 may store investigation result Rs previously acquired by the support device 100, and target object Ot associated with the investigation result Rs. The storage unit 50 may be provided as a component of the support device 100, or it may be provided outside the support device 100. The storage unit 50 may store information such as the result of the truth value determination of the investigation result Rs, confidence level, or consistency, as described later.

[0059] The support device 100 may switch search queries according to the modality type of the target object Ot. In this example, the support device 100 can determine an appropriate search query according to the situation and obtain the search results Rs, even when combinations of text, images, audio, or video are mixed, such as in social networking services. The support device 100 may also dynamically change the verification depth according to the quality of the search results Rs.

[0060] The support device 100 may assist the user in determining the truth value of the target object Ot by providing the user with the investigation results Rs, which can serve as the basis for the truth value determination, rather than providing the user with the result of the truth value determination of the target object Ot. This enables a well-founded and objective truth value determination. Alternatively, the support device 100 may provide the user with the result of the truth value determination based on the investigation results Rs. In addition to the result of the truth value determination, the support device 100 may also provide the user with the investigation results Rs that serve as the basis for the determination.

[0061] The user may determine the truth value of the target object Ot by referring to the investigation results Rs of the support device 100. In addition to the result of the truth value determination, the user can easily perform evidence-based truth value determination by referring to the supporting investigation results Rs.

[0062] Figure 1B shows an example of the operation flowchart of the support device 100. In step S100, the target object Ot is acquired. In step S102, a search query is determined depending on whether the target object is single-modal or multimodal. Multiple search queries may be determined in step S102. In step S104, the search queries are used to acquire the investigation results Rs of related information associated with the target object Ot. After acquiring the investigation results Rs, the search query may be changed to acquire further investigation results Rs. The search query may also be changed according to the acquired investigation results Rs. In step S106, the investigation results Rs are output.

[0063] Figure 2A shows an example of the input screen for input information In. In this example, the URL of the link destination for input information In is entered into the search box of the truth / false determination support application. The URL of the link destination for input information In contains the target object Ot. In this example, the support device 100 is a truth / false determination support application that runs on a smartphone app, but the method of execution of the support device 100 is not limited to this. The support device 100 may be executed on any information processing device such as a personal computer. The user may copy and paste the source URL from the web page containing the target object Ot.

[0064] Figure 2B shows an example of the input screen for input information In. In this example, the support device 100 performs a truth value determination of the target object Ot directly input by the user. In this example, the support device 100 is described as being executed on any information processing device such as a personal computer, but the execution method of the support device 100 is not limited to this.

[0065] The query determination unit 20 may determine the target object Ot as the search query. That is, the query determination unit 20 may determine the target object Ot, which is text, image, audio, or video itself, as the search query. The results acquisition unit 30 may use the target object Ot to acquire the results Rs. For example, the query determination unit 20 may determine the image, which is the target object Ot, as the search query. The results acquisition unit 30 may use reverse image search technology to perform an image search on related objects associated with the target object Ot and acquire the results Rs. The query determination unit 20 may determine the video, which is the target object Ot, as the search query. The query determination unit 20 may perform an image search using frames extracted from the video.

[0066] The investigation result acquisition unit 30 may extract and acquire the target object Ot from the input information In in response to receiving the input information In. Receiving the input information In may include automatically extracting the target object Ot from the received input information In. That is, the investigation result acquisition unit 30 may start extracting the target object Ot without receiving a specific prompt from the user to extract the target object Ot from the input information In. As a result, the support device 100 in this example can standardize truth / false determination without requiring prompt input from the user. However, the support device 100 may start the process of extracting the target object Ot from the input information In after the execution button for truth / false determination processing is pressed.

[0067] Figure 3A shows an example of a single-modal target object Ot. In this example, the target object Ot contains text written in a social media post. The target object Ot in this example contains the text, "According to a speech given by Mr. A in Country B on December 20, 2023, he claimed, based on paper C, that the vaccine contains a microchip." The input information In may include the URL of the social media post, the URL of a media sharing site such as a video sharing site, or the account information of the poster.

[0068] The query determination unit 20 may decompose the target object Ot, which has been determined to be single modal, into multiple factual elements and generate multiple search queries corresponding to the multiple factual elements. The query determination unit 20 may decompose the target object Ot into factual elements such as when, where, who, what, why, and how. This allows search results to be obtained for each of the multiple factual elements. The search result acquisition unit 30 may use the multiple search queries to acquire at least one search result Rs for each of the multiple factual elements.

[0069] The output unit 40 may output multiple survey result Rs for a single factual element. The output unit 40 may output at least one survey result Rs for each of the multiple factual elements. The output unit 40 may output one survey result Rs for each of the multiple factual elements. The output unit 40 may output multiple survey result Rs for each of the multiple factual elements. The output unit 40 may output one or more survey result Rs in common for multiple factual elements.

[0070] Figure 3B shows an example of how to obtain the survey results Rs. This example describes how to obtain the survey results Rs when the target object Ot is single-modal. The support device 100 in this example extracts multiple factual elements from the target object Ot, determines a search query for each factual element, and obtains the survey results Rs. The extraction of factual elements from the target object Ot may be performed by the target object acquisition unit 10 or by the query determination unit 20. The support device 100 may similarly extract multiple factual elements from the target object Ot, determine a search query for each factual element, and obtain the survey results Rs even when the target object Ot is multimodal.

[0071] In this example, the support device 100 divides the target object Ot into three factual elements. The target object Ot is text-based information stating, "According to a speech given by Mr. A in Country B on December 20, 2023, he claimed, based on paper C, that the vaccine contains a microchip." The support device 100 extracts "Mr. A's speech in Country B on December 20, 2023" as factual element 1. The support device 100 extracts "the fact that paper C contains evidence that the vaccine contains a microchip" as factual element 2. The support device 100 extracts "the fact that Mr. A claims the vaccine contains a microchip" as factual element 3. The query determination unit 20 determines search queries 1 to 3 for each of factual elements 1 to 3. The search result acquisition unit 30 uses search queries 1 to 3 to acquire the search result Rs for each factual element.

[0072] Figure 4A shows an example of a multimodal target object Ot. In this example, the target object Ot contains both text and video information. The target object Ot in this example includes the text "A tsunami is coming." and a video of the tsunami, which are included in the SNS posting screen. The input information In may include the URL of the SNS post and may include the poster's account information.

[0073] If the query determination unit 20 determines that the target object Ot is multimodal, it may generate different search queries for each of the multiple modality types included in the target object Ot. In this example, the query determination unit 20 determines a search query for text-formatted information and a search query for video-formatted information. This allows search results to be obtained for each of the multiple modality types. The results acquisition unit 30 may use the multiple search queries to acquire the results Rs for each of the multiple modality types.

[0074] Figure 4B shows an example of how to obtain the survey results Rs. This example describes how to obtain the survey results Rs when the target object Ot is multimodal. The support device 100 in this example extracts multiple modality types from the target object Ot, determines a search query for each modality type, and obtains the survey results Rs. The extraction of modality types from the target object Ot may be performed by the target object acquisition unit 10 or by the query determination unit 20.

[0075] In this example, the support device 100 extracts two modality types from the input information In. The support device 100 extracts text-formatted information and videos from the SNS posts, which are the input information In. The query determination unit 20 may generate individual search queries for each extracted modality type. In this example, the query determination unit 20 determines search query 1 for modality type 1 and search query 2 for modality type 2.

[0076] The results acquisition unit 30 may acquire at least one results Rs for each of the multiple modality types using search queries for each of the multiple modality types. In this example, the results acquisition unit 30 acquires results Rs based on search query 1 and search query 2, respectively. The output unit 40 may output at least one results Rs for each of the multiple modality types.

[0077] Figure 4C shows an example of how to obtain the survey results Rs. If the query determination unit 20 determines that the target object Ot is multimodal, it may generate a hybrid query common to multiple modality types based on the multiple modality types of the target object Ot. For example, the query determination unit 20 generates a hybrid query by integrating keywords contained in text-format information with features extracted from image-format, audio-format, or video-format information. In one example, the query determination unit 20 extracts features such as character names, locations, or signs from image-format information and generates a search query by integrating them with keywords contained in text-format information.

[0078] Figure 5A shows an example of the analysis results of the text contained in the target object Ot. In this example, the output unit 40 displays "Accurate," which is the identifier for truth / false determination, as the analysis result, and outputs the reason for the determination. The output unit 40 may display the investigation results Rs in a list format, which may include at least one of the following: an overview of the information source, a URL indicating the location of the information source, a link corresponding to the URL, or the title of the news article. In this example, the title of the news article itself is the link corresponding to the URL.

[0079] Figure 5B shows an example of the analysis results for videos included in the target object Ot. In this example, output unit 40 outputs the investigation result Rs for the modality type of the video format. Output unit 40 in this example displays "Inaccurate," which is the identifier for truth / false determination, as the analysis result, and outputs the reason for the determination. Output unit 40 may output a summary of the video content. Output unit 40 in this example outputs the claims based on facts extracted from the search results as the investigation result Rs. Output unit 40 may display at least one of the following as the investigation result Rs: an overview of the information source, a URL indicating the location of the information source, and a link or news article title corresponding to the URL, in list format.

[0080] The output unit 40 may output metadata for the target object Ot or the investigation results Rs. The metadata may include media EXIF ​​information, file extension, video length, creator, C2PA information, watermark, etc. EXIF ​​information is metadata recorded in conjunction with image files. EXIF ​​information may include shooting date and time, camera model, aperture value, ISO sensitivity, GPS location information, etc. C2PA information is metadata for proving the history and authenticity of digital content. The watermark may include information regarding the copyright of the image or video.

[0081] The support device 100 may link the text analysis results and video analysis results with time information such as timestamps or contextual information and display them. If the target object Ot is multimodal, the support device 100 may analyze the correspondence between factual claims contained in text format information and visual information obtained from video format or image format information. The output unit 40 may output the text analysis results and video analysis results to the display unit 45 in synchronization along the time axis.

[0082] Figure 5C shows an example of the analysis results of the audio contained in the target object Ot. In this example, the output unit 40 outputs the investigation results Rs for the modality type of the audio format. In this example, the output unit 40 displays "Inaccurate," which is the identifier for truth / false determination, as the analysis result, and outputs the reason for the determination. The output unit 40 may display the result of evaluating the validity of the claims contained in the audio data as the reason for the determination of inaccuracy. The output unit 40 may also display a summary of the transcript, which is the result of analyzing the content of the audio. The output unit 40 may display at least one of the following as the investigation results Rs in list format: an overview of the information source, a URL indicating the location of the information source, and a link corresponding to the URL or the title of a news article.

[0083] The results acquisition unit 30 may acquire multiple claims based on facts extracted from audio information, and the results Rs for each claim. The output unit 40 may output, for each extracted fact-based claim, the name of the supporting information source and a URL indicating the location of the information in list format. The output unit 40 may display the title or link of an explanatory article by a reliable organization as the results Rs.

[0084] Figure 5D shows an example of the analysis results of text contained in the target object Ot. In this example, the output unit 40 outputs at least one research result Rs for each factual element of the target object Ot. In this example, the output unit 40 displays "Accurate," which is the identifier for truth / false determination, as the analysis result, and outputs the reasoning behind the determination for each factual element. The output unit 40 may output multiple research result Rs for each factual element. The output unit 40 may also add explanations, such as a summary of the research result Rs, for each factual element. In this example, the output unit 40 outputs research result Rs for the target object Ot in single-modal text format, but is not limited to this. The target object Ot may be single-modal or multimodal, as long as it contains multiple factual elements. The target object Ot may contain one or more arbitrary modality types from text format, image format, audio format, or video format.

[0085] Figure 6A is a block diagram showing a modified example of the support device 100. The support device 100 comprises a target object acquisition unit 10, a query determination unit 20, a results acquisition unit 30, an output unit 40, and an analysis unit 60. The support device 100 may also include a display unit 45 and a storage unit 50. In this example, the differences from the support device 100 in Figure 1A will be mainly explained.

[0086] The analysis unit 60 may perform information analysis in cooperation with at least one of the target object acquisition unit 10, the query determination unit 20, the investigation result acquisition unit 30, or the output unit 40. The analysis unit 60 may analyze the input information In, the target object Ot, the investigation result Rs, or the contents of the search results.

[0087] Figure 6B shows an example of the configuration of the analysis unit 60. In this example, the analysis unit 60 includes an estimation unit 62, an access verification unit 64, a reliability determination unit 66, and a consistency evaluation unit 68.

[0088] The estimation unit 62 integrates the survey results Rs for each of the multiple modality types to estimate the truth or falsity of the target object Ot. For example, the estimation unit 62 determines whether there are any inconsistencies between the text-format survey results Rs and the video-format survey results Rs, and determines the validity of the target object Ot as a whole. The determination result estimated by the estimation unit 62 may be displayed on the display unit 45 via the output unit 40. The estimation unit 62 may estimate the truth or falsity of the target object Ot as "True," "False," or "Unknown." The estimation unit 62 may also estimate the truth or falsity of the target object Ot as "Accurate," "Inaccurate," or "Unknown."

[0089] The access verification unit 64 accesses the URL of the survey results Rs and verifies whether the URL is accessible. The access verification unit 64 may maintain only valid information accessible to the user as the survey results Rs by checking in advance for broken links or access restrictions. The access verification unit 64 may notify the survey results acquisition unit 30 or output unit 40 of the verification results.

[0090] The reliability determination unit 66 determines whether the source of the survey results Rs is a reliable source that meets predetermined conditions. The reliability determination unit 66 may determine the reliability of the survey results Rs based on at least one of the content of the survey results Rs or the content of the source of the survey results Rs. The reliability determination unit 66 may determine the reliability based on the domain of the source, the author, the site type such as whether it is a public institution or not. For example, the reliability determination unit 66 may assign a higher reliability to a site type that is a public institution or a news site than to a social networking service or a personal blog. If the survey results Rs is a social networking service post, the reliability determination unit 66 may determine the reliability based on whether the poster is anonymous or not. If the survey results Rs is a social networking service post, the reliability determination unit 66 may determine the reliability based on the number of followers of the poster, or based on the content of the poster's comment section. The reliability determination unit 66 may determine the reliability of the poster of the survey results Rs based on the reliability of the poster who commented in the comment section of the poster of the survey results Rs. The confidence level determination unit 66 may calculate a confidence score, which is a numerical representation of the determined confidence level.

[0091] The output unit 40 may output the survey results Rs with an identifier indicating the level of confidence if it determines that the source of the survey results Rs is reliable. The identifier may be displayed on the display unit 45 in the form of an icon or check mark indicating that it is from a public source.

[0092] The output unit 40 outputs at least one of the truth / false determination result for the target object Ot, the confidence score of the investigation result Rs, or the list of evidence for the investigation result Rs. The output unit 40 may output the final truth / false determination result obtained by the analysis unit 60 or the estimation unit 62 using an identifier such as "correct" or "inaccurate". The output unit 40 may output in list format the confidence score assigned to each investigation result Rs and the details of the evidence information that serves as the source of the information, along with their correspondence.

[0093] The query determination unit 20 may determine a new search query using the information from the survey results Rs if the confidence level of the survey results Rs is below an arbitrary threshold value. The query determination unit 20 may perform a new search using the new search query if the confidence level determination unit 66 determines that the confidence level of the survey results Rs is low. The query determination unit 20 may regenerate a search query targeting the same target object Ot as the original search query, or it may regenerate a search query targeting different information than the original target object Ot.

[0094] The query determination unit 20 may regenerate advanced search queries for conducting a more detailed investigation, using the less reliable investigation result Rs as clues. For example, the query determination unit 20 may extract the information source or proper noun of the investigation result Rs that was determined to have low reliability and generate a search query to investigate the reliability of the investigation result Rs that was determined to have low reliability. The query determination unit 20 may generate a search query for re-searching and supplement the less reliable information, repeating the search until the reliability meets the criteria.

[0095] The query determination unit 20 may modify the search query or search conditions to obtain more reliable search results Rs if the reliability of the top-ranking search results Rs is low. To obtain more reliable search results Rs, the query determination unit 20 may limit the websites extracted as search results to domains of public institutions, etc.

[0096] The consistency evaluation unit 68 compares the contents of multiple survey results Rs to evaluate the consistency between them. The consistency evaluation unit 68 may also compare the contents of multiple search results to evaluate the consistency between them. By evaluating the consistency of the supporting information, the consistency evaluation unit 68 can assess the consensus on the subject matter, such as whether all the survey results Rs refer to the same facts or whether there are differing opinions. Furthermore, by evaluating the consistency of the supporting information, the consistency evaluation unit 68 can distinguish between a large number of reliable survey results Rs and contradictory survey results Rs, such as outliers. The consistency evaluation unit 68 may deduct points from the confidence score of survey results Rs that it has determined to be contradictory, assign a "caution" label to them, or omit output from the output unit 40.

[0097] The consistency evaluation unit 68 may evaluate the consistency of the input information In. For example, the consistency evaluation unit 68 may compare the content of the posted text in the input information In with the video analysis results to evaluate the consistency between the posted text and the video. The consistency evaluation unit 68 may also evaluate the consistency of the input information In by comparing the target object Ot with the metadata.

[0098] For example, if the text claims to describe a current event, but the metadata or video content indicates that the event is old, the consistency evaluation unit 68 may evaluate the input information In as inconsistent. The consistency evaluation unit 68 may also determine that the input information In is inconsistent if the text of the input information In is a partial excerpt of a statement from a video that presents facts that are not true. The consistency evaluation unit 68 may also determine that the input information In is inconsistent if, when the input information In includes both text and audio, the text and the transcript of the audio do not match. The consistency evaluation unit 68 may also determine that the input information In is inconsistent if, when the input information In includes both text and audio in a foreign language, the text and the translated content do not match.

[0099] The consistency evaluation unit 68 may evaluate the consistency of the input information In using the survey results Rs. For example, if the input information In includes a video with captions, the consistency evaluation unit 68 evaluates whether the content of the captions and the content of the video are consistent based on the video in the survey results Rs. If, upon referring to the survey results Rs, it is found that the video is intended to explain facts different from those of the captions, the consistency evaluation unit 68 may evaluate that the input information In is inconsistent.

[0100] The consistency evaluation unit 68 may evaluate the consistency between the first investigation result related to the target object Ot and the second investigation result related to the target object Ot when the target object Ot is single modal. The consistency evaluation unit 68 may detect inconsistencies in claims obtained from different investigation results Rs within a single modality format. For example, the consistency evaluation unit 68 may check for factual inconsistencies between multiple news sites in a text-only post.

[0101] The consistency evaluation unit 68 may evaluate the consistency between the first investigation result related to the first modality type of the target object Ot and the second investigation result related to the second modality type of the target object Ot when the target object Ot is multimodal. For example, the consistency evaluation unit 68 may evaluate the consistency between the investigation result Rs for text-format information and the investigation result Rs for video-format information. This allows the consistency evaluation unit 68 to evaluate whether content has been swapped between different formats of information.

[0102] The output unit 40 may decide whether or not to output the survey results Rs according to the consistency evaluation result. The output unit 40 may hide survey results Rs that the consistency evaluation unit 68 has determined to have a consistency lower than a predetermined standard. The output unit 40 may display survey results Rs that the consistency evaluation unit 68 has determined to have a consistency of or above a predetermined standard. The output unit 40 may prioritize outputting highly reliable information in which agreement is found among multiple survey results Rs.

[0103] The output unit 40 may output in a way that highlights the inconsistent parts. The output unit 40 may display the inconsistent parts in a different color from the others. The output unit 40 may highlight the playback time of the inconsistent parts in the video.

[0104] The analysis unit 60 may analyze information such as the survey results Rs using any machine learning model. For example, the machine learning model may include at least one of a Convolutional Neural Network (CNN), a Recurrent Neural Network (RNN), a Long Short-Term Memory (LSTM), or a Transformer. The machine learning model may also include at least one of a Large Language Model (LLM) or a Large Vision Language Model (VLM). However, the machine learning model used by the analysis unit 60 for determining confidence is not limited to these. The estimation unit 62, the confidence determination unit 66, and the consistency evaluation unit 68 may perform processing using the machine learning model.

[0105] Figure 7 shows an example of the analysis results of the text contained in the target object Ot. The estimation unit 62 may integrate multiple survey results Rs to estimate the truth value of the target object Ot. In this example, the output unit 40 displays "Accurate," which is the identifier for truth value determination, as the analysis result, and outputs the reason for the determination. In this example, the output unit 40 outputs the confidence level of the survey results Rs.

[0106] The estimation unit 62 estimates the truth value of the text contained in the target object Ot. In this example, the estimation unit 62 determines whether the claim that a special heavy rain warning was issued on the morning of September 21, 2024 is true or false, based on multiple acquired survey results Rs. The estimation unit 62 may update the truth value determination result of the target object Ot. For example, the estimation unit 62 ultimately compares the survey results Rs with the most reliable result with the target object Ot to estimate the truth value of the target object Ot. This can further improve the accuracy of the truth value determination of the target object Ot.

[0107] The output unit 40 may display a numerical value such as "98%" or "95%" as a confidence level identifier for each survey result Rs. The confidence level determination unit 66 may calculate the confidence level based on domain information, the date of publication of the information, or the accuracy of the content.

[0108] The output unit 40 may display an icon indicating that the URL is a public source. If the URL is not accessible, the output unit 40 may output a verification result such as "Inaccessible" associated with the survey result Rs. The access verification unit 64 may ensure the validity of the information that forms the basis for determining its truth or falsity by confirming whether the information is accessible or not.

[0109] Figure 8A shows an example of the consistency evaluation result between the text survey results Rs and the video survey results Rs. The consistency evaluation unit 68 may compare the contents of multiple survey results Rs to evaluate the consistency of multiple survey results Rs. The consistency evaluation unit 68 analyzes whether the factual information obtained from multiple sources is consistent or contradictory. For example, the consistency evaluation unit 68 evaluates the consistency between a location identified from visual information such as a sign or landscape in the video and a location claimed in the text.

[0110] The consistency evaluation unit 68 may notify the output unit 40 of a mismatch alert if it detects a consistency mismatch. The consistency evaluation unit 68 may provide the output unit 40 with a consistency score based on the evaluation results. The output unit 40 may output an analysis report that includes a detailed explanation of the mismatch and specific points of inconsistency.

[0111] Figure 8B shows an example of the consistency evaluation results of the survey results Rs. The output unit 40 may output at least one of the title, source information, or date information of the survey results Rs. The source information may be text such as a name indicating the source of the survey results Rs, or it may be the URL of the source. The date information may be the update date or publication date of the survey results Rs. The output unit 40 may output at least one of the title, source information, or date information for each of multiple survey results Rs in a list format.

[0112] The output unit 40 may output the degree of agreement between the content of the target object Ot and the survey results Rs. The output unit 40 may display an identifier that numerically indicates the degree of agreement between the content of the target object Ot and each survey result Rs on the display unit 45, associating it with each survey result Rs. The output unit 40 may sort multiple survey results Rs based on at least one of the degree of agreement or the date information of the survey results Rs. When the support device 100 receives a sorting instruction from the user, it may sort and output multiple survey results Rs in order of decreasing degree of agreement or in order of date. If the target object Ot is a video or image, the survey result acquisition unit 30 may acquire the survey results Rs using reverse image search technology.

[0113] Figure 9 shows an example of how to output the results Rs for multiple target objects Ot. In this example, we will use three target objects Ot1 to Ot3 for explanation. Target object Ot2 is similar to target object Ot1, but target object Ot3 is not similar to target object Ot1. The similarity of target object Ot2 to target object Ot1 is 90%, and the similarity of target object Ot3 to target object Ot1 is 20%.

[0114] The target object acquisition unit 10 acquires multiple target objects Ot. The target object acquisition unit 10 may acquire multiple target objects Ot that are subject to truth value determination from multiple URLs or multiple media files specified by the user. The target object acquisition unit 10 may also extract multiple target objects Ot from a single input information In.

[0115] The query determination unit 20 determines the search query for each of the multiple target objects Ot. In this example, the query determination unit 20 determines search queries 1 to 3 for target objects Ot1 to Ot3, respectively.

[0116] The results acquisition unit 30 may acquire results Rs for multiple target objects Ot. The results acquisition unit 30 may acquire results Rs by searching for relevant information on the web using a search query determined for each of the multiple target objects Ot. The results acquisition unit 30 may store the acquired multiple results Rs in a state that can be identified for each target object Ot.

[0117] The output unit 40 may output the survey results Rs of two or more similar target objects Ot from among multiple target objects Ot, linking them together. The output unit 40 may use a large-scale language model to calculate the similarity of the content of multiple target objects Ot, and output target objects Ot with high similarity linked together. For example, the output unit 40 may group and present posts with similar content based on the similarity between content. In this example, the output unit 40 outputs the survey results Rs of target object Ot1 linked to the survey results Rs of target object Ot2.

[0118] The output unit 40 may display links or thumbnails on the screen to the survey results Rs of similar content or similar genres. The output unit 40 may group and display the survey results Rs of similar target objects Ot in a list to show the user the extent of information dissemination or relevance.

[0119] Furthermore, the output unit 40 may selectively display the result Rs based on the similarity between the result Rs. The output unit 40 may select and output a representative result Rs when multiple result Rs are similar. The output unit 40 may perform emphasis processing on at least one of the multiple result Rs and output it. The output unit 40 may change the display color of the result Rs to be emphasized, increase the size of the displayed result Rs, or add a symbol to emphasize the result Rs. The output unit 40 may add an explanation of the result Rs to be emphasized to the search results. The output unit 40 may highlight and display some of the information in the result Rs. The output unit 40 may display the result Rs in a way that allows the content of the result Rs to be intuitively understood without clicking on them. The method of emphasis processing is not limited to these.

[0120] The output unit 40 may determine the similarity of multiple survey results Rs based on the context information of the survey results Rs. The output unit 40 may increase the similarity of the survey results Rs when the content of the context information of the survey results Rs is similar to that of the survey results Rs. The output unit 40 may decrease the similarity of the survey results Rs when the content of the context information of the survey results Rs is far apart from that of the survey results Rs.

[0121] Figure 10 is a block diagram showing a modified example of the support device 100. The support device 100 comprises a target object acquisition unit 10, a query determination unit 20, a results acquisition unit 30, an output unit 40, and a user instruction reception unit 70. The support device 100 may also include a display unit 45 and a storage unit 50. In this example, the differences from the support device 100 in Figure 1A will be mainly explained.

[0122] The user instruction receiving unit 70 receives instructions from the user. The user instruction receiving unit 70 may notify the query determination unit 20 or the output unit 40 of the content of the received instructions. The user instruction receiving unit 70 may provide an interface for the user to manually intervene in processes that the support device 100 performs automatically. In this example, the user instruction receiving unit 70 has a change instruction receiving unit 72 and a shared instruction receiving unit 74.

[0123] The change instruction receiving unit 72 receives instructions to change the search query. The change instruction receiving unit 72 may receive instructions to change the search query from the user. The query determination unit 20 may change the determined search query in accordance with the change instruction. The query determination unit 20 may supply the changed search query to the search result acquisition unit 30. The query determination unit 20 may optimize the search conditions by reflecting search words arbitrarily specified by the user in the automatically generated search query.

[0124] The sharing instruction receiving unit 74 receives a sharing instruction to share the survey results Rs. The sharing instruction receiving unit 74 may receive a sharing instruction from a user to share the survey results Rs. The sharing instruction receiving unit 74 may accept a sharing instruction when the user presses the share button to SNS. The output unit 40 may output the survey results Rs to the sharing destination corresponding to the sharing instruction. The sharing destination may be an SNS, messaging platform, or email address. The output unit 40 may transfer the survey results Rs to the external service specified by the sharing instruction. The output unit 40 may convert the data structure of the information and output it in a format that can be viewed at the sharing destination.

[0125] The query determination unit 20 may accept search parameter specifications from the user. The query determination unit 20 may accept specifications for the number of searches, the search period, or the search language. Specifying the number of searches can determine the search depth. Specifying the search period can determine whether the search results are reposts from the past.

[0126] Figure 11 shows an example of a filtering method for the survey results Rs. In this example, the output unit 40 displays the filtering conditions on the display unit 45. The filtering conditions may be determined by the user or by the support device 100. The support device 100 may determine the filtering conditions according to the survey results Rs.

[0127] The output unit 40 may filter and output at least one of the multiple Rs results based on the confidence level of the Rs results. In this example, the output unit 40 adjusts the number of hits by filtering. The output unit 40 may also exclude and output Rs results whose confidence level is lower than a set threshold.

[0128] The investigation result acquisition unit 30 may acquire investigation result Rs based on specified period information. The investigation result acquisition unit 30 may include a specific year, month, or day range in the search criteria to collect relevant information. If the target object Ot refers to a specific past incident, the investigation result acquisition unit 30 may retrospectively acquire the public records from that time as investigation result Rs. The output unit 40 may output the investigation result Rs acquired by the investigation result acquisition unit 30 without limiting the period information, according to the specified period information.

[0129] The output unit 40 may filter and output at least one of the multiple survey results Rs based on a keyword. The keyword may be entered by the user. The output unit 40 may acquire the keyword before aligning the multiple survey results Rs, or it may acquire the keyword after aligning the multiple survey results Rs and then rearrange the multiple survey results Rs. The support device 100 may generate appropriate keywords from the search results of the survey results Rs and suggest them to the user. The support device 100 may filter the survey results Rs based on keywords such as "politician X's press conference" in order to verify the credibility of a politician's statement.

[0130] The output unit 40 may change the filtering conditions according to the number of hits in the survey results Rs. If there are many hits, the output unit 40 may tighten the filtering conditions to reduce the number of hits. If there are few hits, the output unit 40 may loosen the filtering conditions to increase the number of hits. The number of hits may be set by the user or may be set to any number in advance. The output unit 40 may adjust the filtering conditions so that the number of hits matches the set number.

[0131] The results acquisition unit 30 may acquire results Rs based on a specified whitelist or blacklist. The whitelist and blacklist may be specified by the user. The results acquisition unit 30 may specify any website to be on the whitelist or blacklist. The results acquisition unit 30 may specify the whitelist or blacklist by domain. The results acquisition unit 30 may specify the whitelist or blacklist by specifying one or more websites. The results acquisition unit 30 may acquire highly reliable websites as a whitelist, such as government agency websites, newspaper company websites, and official corporate websites. The results acquisition unit 30 may acquire less reliable websites as a blacklist, such as social networking services (SNS) where anonymous accounts are registered.

[0132] The output unit 40 may filter and output multiple research result Rs based on at least two combinations of the update date, publication date, context information, meta information, title, or confidence level of the multiple research result Rs. The output unit 40 may exclude from the search results if at least two of the update date, publication date, context information, meta information, title, or confidence level are close to and / or far from the target object Ot. This makes it possible to exclude research result Rs that are too close to the target object Ot and / or research result Rs that are too far from the target object Ot.

[0133] Figure 12 is a block diagram showing a modified example of the support device 100. The support device 100 comprises a target object acquisition unit 10, a query determination unit 20, a survey result acquisition unit 30, an output unit 40, and an opinion analysis unit 80. The support device 100 may also include a display unit 45 and a storage unit 50. In this example, the differences from the support device 100 in Figure 1A will be mainly explained.

[0134] The opinion analysis unit 80 analyzes at least one positive or negative opinion regarding the input information In. The opinion analysis unit 80 may analyze responses such as replies, comments, or quoted reposts to posts on social media, etc. The opinion analysis unit 80 may use any natural language processing model to quantify the tendency of public responses to the input information In to be positive or negative. The opinion analysis unit 80 may analyze how the information is socially perceived, rather than the truthfulness of the information itself.

[0135] The output unit 40 outputs the analysis results from the opinion analysis unit 80. The output unit 40 may output the ratio of positive opinions to negative opinions as a reaction ratio to the display unit 45. The output unit 40 may also visualize the analysis results in the form of a graph or chart to help the user intuitively grasp the overall picture of the reactions.

[0136] Figure 13 shows an example of the information output by the output unit 40. This figure shows an example of displaying statistical analysis information based on the survey results Rs stored in the storage unit 50.

[0137] The output unit 40 may output at least one of the following from the input information In, target object Ot, or survey results Rs stored in the storage unit 50: the regional distribution of the surveyed content, a word cloud, statistical information about the target object Ot, or the proportion of topics in the content. Beyond determining the truthfulness of individual content, the output unit 40 may present the trend of misinformation dissemination or frequently occurring keywords from a macro perspective.

[0138] Statistical information regarding the target object Ot may be a reaction ratio showing the proportion of agreement, disagreement, or other responses to the input information In. Statistical information may include any rankings regarding the target object Ot. Statistical information may also include a ranking of posters, which is the ranking of the number of times they have previously posted target object Ot whose truthfulness has been determined. A word cloud visually highlights terms that frequently appear within a large number of pieces of content determined to be misleading, with the higher the frequency, the larger the term. The content topic proportion shows the composition ratio of the fields (e.g., politics, health, technology, etc.) to which the surveyed content belongs, shown by category using a pie chart or similar method. Regional distribution shows, in the form of a heatmap or similar method, which regions a particular topic is concentrated in or disseminated in.

[0139] Figure 14 shows an example of the information output by the output unit 40. This figure shows an example of displaying the history information of the survey results Rs stored in the storage unit 50.

[0140] The output unit 40 may output a list of at least one part or all of the past input information In, target object Ot, or survey result Rs stored in the storage unit 50. The output unit 40 may extract survey result Rs from the historical information that match a specific period, a specific topic, or a specific confidence range specified by the user. The output unit 40 may output data accumulated in the storage unit 50. The output unit 40 may output target object Ot for which survey result Rs has been obtained in the past. In this example, the output unit 40 displays whether or not the truth value determination of the target object Ot has been completed. In addition to the target object Ot, the output unit 40 may also display a summary of the survey result Rs, etc.

[0141] The output unit 40 may output the investigation results Rs for each target object Ot. For example, when the user clicks on a target object Ot, the output unit 40 expands and displays the investigation results Rs for the clicked target object Ot. In this example, the support device 100 can inform the user that similar false information dissemination has occurred in the past by displaying the history information of the investigation results Rs.

[0142] Figure 15 shows an example of a computer 2200 in which multiple aspects of the present invention may be embodied in whole or in part. A program installed on the computer 2200 can cause the computer 2200 to function as an operation or one or more sections of an apparatus according to an embodiment of the present invention, or to execute such operation or one or more sections, and / or to cause the computer 2200 to execute a process or a stage of such process according to an embodiment of the present invention. Such a program may be executed by the CPU 2212 to cause the computer 2200 to perform a particular operation associated with some or all of the blocks in the flowcharts and block diagrams described herein.

[0143] The computer 2200 according to this embodiment includes a CPU 2212, 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.

[0144] 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 from the frame buffer provided in RAM 2214 or from itself, and displays the image data on the display device 2218.

[0145] 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 them to the hard disk drive 2224 via the RAM 2214. The IC card drive reads programs and data from the IC card and / or writes programs and data to the IC card.

[0146] The ROM 2230 stores boot programs and / or hardware-dependent programs of the computer 2200, which are executed by the computer 2200 upon activation. The input / output chip 2240 may also connect various input / output units to the input / output controller 2220 via parallel ports, serial ports, keyboard ports, mouse ports, etc.

[0147] The program is provided on a computer-readable medium such as a DVD-ROM 2201 or an IC card. The program is read from the computer-readable medium and installed on a hard disk drive 2224, RAM 2214, or ROM 2230, which are examples of computer-readable mediums, and executed by the CPU 2212. The information processing described within these programs is read by the computer 2200, resulting in coordination between the program and the various types of hardware resources described above. The apparatus or method may be configured to realize the manipulation or processing of information in accordance with the use of the computer 2200.

[0148] For example, when communication is performed between a computer 2200 and an external device, the CPU 2212 may execute a communication program loaded into RAM 2214 and, based on the processing described in the communication program, instruct the communication interface 2222 to perform communication processing. Under the control of the CPU 2212, the communication interface 2222 reads transmission data stored in a transmission buffer processing area provided in a recording medium such as RAM 2214, a hard disk drive 2224, a DVD-ROM 2201, or an IC card, transmits the read transmission data to the network, or writes received data received from the network to a reception buffer processing area provided on the recording medium.

[0149] Furthermore, the CPU 2212 may read all or necessary parts of files or databases stored on external storage media such as the hard disk drive 2224, DVD-ROM drive 2226 (DVD-ROM 2201), or IC card into the RAM 2214, and perform various types of processing on the data in the RAM 2214. The CPU 2212 then writes the processed data back to the external storage media.

[0150] 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 the data read from RAM 2214, including various types of operations, information processing, conditional judgments, conditional branching, unconditional branching, information retrieval / replacement, etc., as described throughout this disclosure and specified by the program instruction sequence, and write the results back to RAM 2214. The CPU 2212 may also retrieve information in files, databases, etc., within the recording medium. For example, if multiple entries are stored in the recording medium, each having an attribute value of a first attribute associated with an attribute value of a second attribute, the CPU 2212 may search among the multiple entries for an entry that matches the condition for which the attribute value of the first attribute is specified, read the attribute value of the second attribute stored in that entry, and thereby obtain the attribute value of the second attribute associated with the first attribute that satisfies a predetermined condition.

[0151] The programs or software modules described above may be stored on or near computer 2200 on a computer-readable medium. Alternatively, recording media such as hard disks or RAM provided within a server system connected to a dedicated communication network or the Internet can be used as computer-readable media, thereby providing programs to computer 2200 via the network.

[0152] 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 or improvements can be made to the above embodiments. It will be clear from the claims that such modified or improved forms may also be included in the technical scope of the present invention.

[0153] It should be noted that the execution order of operations, procedures, steps, and stages in the apparatus, systems, programs, and methods shown in the claims, specifications, and drawings is not explicitly stated as "before," "prior to," etc., and that these can be implemented in any order unless the output of a previous process is used in a later process. Even if the operation flow in the claims, specifications, and drawings is described using phrases such as "first," "next," etc. for convenience, it does not mean that it is essential to perform the operations in that order. [Explanation of Symbols]

[0154] 10...Target object acquisition unit, 20...Query determination unit, 30...Results acquisition unit, 40...Output unit, 45...Display unit, 50...Storage unit, 60...Analysis unit, 62...Estimation unit, 64...Access verification unit, 66...Reliability determination unit, 68...Consistency evaluation unit, 70...User instruction reception unit, 72...Change instruction reception unit, 74...Sharing instruction reception unit, 80...Opinion analysis unit, 100...Support device, 2200...Computer, 2 201...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 support device for assisting in determining the truth or falsity of input information, A target object acquisition unit that acquires the target object to be determined as true or false from the input information, A query determination unit that determines a search query depending on whether the target object is single-modal or multimodal, A results acquisition unit that uses the determined search query to acquire the results of the investigation of related information related to the target object, An output unit that outputs the results of the investigation, Equipped with, The aforementioned query determination unit, If the target object includes one of the following modality types: text, image, audio, or video, the target object is determined to be single-modal, the target object determined to be single-modal is broken down into multiple factual elements, and multiple search queries corresponding to the multiple factual elements are generated. If the target object includes at least two modality types from text, image, audio, or video, the target object is determined to be multimodal, and different search queries are generated for each of the multiple modality types included in the target object. Support equipment.

2. The investigation result acquisition unit uses the plurality of search queries to acquire at least one investigation result for each of the plurality of factual elements, The output unit outputs at least one of the investigation results for each of the plurality of factual elements. The support device according to claim 1.

3. A support device for determining the truth or falsity of input information, A target object acquisition unit that acquires the target object to be determined as true or false from the input information, A query determination unit that determines a search query depending on whether the target object is single-modal or multimodal, A results acquisition unit that uses the determined search query to acquire the results of the investigation of related information related to the target object, An output unit that outputs the results of the investigation, Equipped with, The aforementioned query determination unit, If the target object includes one of the following modality types: text, image, audio, or video, the target object is determined to be single-modal, and if the target object is single-modal, the target object is determined to be the search query. If the target object includes at least two modality types from text, image, audio, or video, the target object is determined to be multimodal, and different search queries are generated for each of the multiple modality types included in the target object. Support equipment.

4. The investigation result acquisition unit uses the search queries for each of the multiple modality types to acquire at least one investigation result for each of the multiple modality types. The output unit outputs at least one of the survey results for each of the multiple modality types. The support device according to claim 3.

5. The system includes an estimation unit that integrates the investigation results for each of the multiple modality types to estimate the authenticity of the target object. The output unit outputs the result of the estimation unit's determination of the truth value of the target object. The support device according to claim 3.

6. A support device for assisting in determining the truth or falsity of input information, A target object acquisition unit that acquires the target object to be determined as true or false from the input information, A query determination unit that determines a search query depending on whether the target object is single-modal or multimodal, A results acquisition unit that uses the determined search query to acquire the results of the investigation of related information related to the target object, An output unit that outputs the results of the investigation, Equipped with, The aforementioned query determination unit, If the target object includes one of the following modality types: text, image, audio, or video, the target object is determined to be single-modal, the target object determined to be single-modal is broken down into multiple factual elements, and multiple search queries corresponding to the multiple factual elements are generated. If the target object includes at least two modality types from text, image, audio, or video formats, the target object is determined to be multimodal. If the target object is determined to be multimodal, features extracted from the multiple modality types of the target object are integrated to generate a hybrid query common to the multiple modality types. Support equipment.

7. It includes a change instruction receiving unit that accepts instructions to change the search query, The query determination unit modifies the determined search query in accordance with the change instruction. The support device according to any one of claims 1 to 6.

8. The system includes a sharing instruction receiving unit that receives a sharing instruction to share the aforementioned survey results, The output unit outputs the survey results to the sharing destination corresponding to the sharing instruction. The support device according to any one of claims 1 to 6.

9. The system includes an access verification unit that accesses the URL of the aforementioned survey result and verifies whether or not the URL is accessible. The output unit outputs the investigation results according to the access verification results from the access verification unit. The support device according to any one of claims 1 to 6.

10. A reliability determination unit determines the reliability of the survey results based on at least one of the contents of the survey results and the contents of the source of the survey results, Equipped with, The output unit outputs the confidence score of the investigation results determined by the confidence determination unit. The support device according to any one of claims 1 to 6.

11. The reliability determination unit determines whether the source of the investigation results is a reliable source that satisfies predetermined conditions, The output unit, when it determines that the source of the investigation results is reliable, outputs the investigation results with an identifier indicating the level of reliability. The support device according to claim 10.

12. The query determination unit, if the confidence level of the survey results is below an arbitrary threshold, uses the information from the survey results to determine a new search query. The support device according to claim 10.

13. The aforementioned survey result acquisition unit acquires multiple survey results, The support device includes a consistency evaluation unit that compares the contents of the multiple survey results and evaluates the consistency of the multiple survey results, The output unit outputs the survey results based on the consistency of the plurality of survey results. The support device according to any one of claims 1 to 6.

14. The consistency evaluation unit evaluates the consistency between the first investigation result related to the target object and the second investigation result related to the target object when the target object is single modal. The support device according to claim 13.

15. The consistency evaluation unit evaluates the consistency between the first investigation result related to the first modality type of the target object and the second investigation result related to the second modality type of the target object when the target object is multimodal. The support device according to claim 13.

16. The output unit determines whether or not to output the results of the consistency evaluation. The support device according to claim 13.

17. The output unit outputs a URL or link indicating the location of the information source as a result of the investigation. The support device according to any one of claims 1 to 6.

18. The output unit outputs attribute information, which includes at least one of person information, location information, or time information, as a result of the investigation. The support device according to any one of claims 1 to 6.

19. The aforementioned investigation result acquisition unit acquires the investigation results based on the specified whitelist or blacklist. The support device according to any one of claims 1 to 6.

20. The aforementioned investigation result acquisition unit acquires the investigation results based on the specified period information. The support device according to any one of claims 1 to 6.

21. The unit comprises an opinion analysis unit that analyzes the response to the input information obtained from the related information and analyzes at least one of positive or negative opinions, The output unit outputs the analysis results from the opinion analysis unit. The support device according to any one of claims 1 to 6.

22. The output unit outputs a list of the basis for the survey results. The support device according to any one of claims 1 to 6.

23. The output unit outputs at least one of the title, source information, or date information of the investigation results. The support device according to any one of claims 1 to 6.

24. The output unit outputs the degree of agreement between the contents of the target object and the results of the investigation. The support device according to any one of claims 1 to 6.

25. The output unit sorts the plurality of survey results based on at least one of the degree of agreement or the date information of the survey results. The support device according to claim 24.

26. The system includes a storage unit for storing at least one of the input information, the target object, or the survey results. The support device according to any one of claims 1 to 6.

27. The output unit outputs a list of at least one part or all of the past input information, the target object, or the investigation results stored in the storage unit. The support device according to claim 26.

28. The output unit outputs at least one of the following from the input information, the target object, or the survey results stored in the storage unit: statistical information about the creator of the input information, the regional distribution of the surveyed content, a word cloud, statistical information about the target object, statistical information about the creator of the target object, or the percentage of topics in the content. The support device according to claim 26.

29. The aforementioned target object acquisition unit acquires multiple target objects, The aforementioned investigation result acquisition unit acquires the investigation results of the multiple target objects, The output unit outputs the investigation results of two or more similar target objects from among the plurality of target objects, relating them together. The support device according to any one of claims 1 to 6.

30. The target object acquisition unit, upon receiving the input information, extracts and acquires the target object from the input information. The support device according to any one of claims 1 to 6.

31. A support method for determining the truth value of input information, The computer obtains the target object to be determined as true or false from the input information, The steps include determining a search query depending on whether the target object is single-modal or multimodal, The computer obtains the results of the investigation of related information relating to the target object using the determined search query, The computer outputs the results of the investigation, Equipped with, The step of determining the search query is: If the target object includes one modality type from among text format, image format, audio format, or video format, the target object is determined to be single modal, the target object determined to be single modal is broken down into multiple factual elements, and multiple search queries corresponding to the multiple factual elements are generated. The steps include: determining that the target object is multimodal if it includes at least two modality types from text format, image format, audio format, or video format, and generating different search queries for each of the multiple modality types included in the target object; Support methods, including those mentioned above.

32. A support method for assisting in determining the truth or falsity of input information, The computer obtains the target object to be determined as true or false from the input information, The steps include determining a search query depending on whether the target object is single-modal or multimodal, The computer obtains the results of the investigation of related information relating to the target object using the determined search query, The computer outputs the results of the investigation, Equipped with, The step of determining the search query is: If the target object includes one modality type from among text format, image format, audio format, or video format, the step of determining that the target object is single modal, and if the target object is single modal, determining that the target object is the search query, The steps include: determining that the target object is multimodal if it includes at least two modality types from text format, image format, audio format, or video format, and generating different search queries for each of the multiple modality types included in the target object; Support methods, including those mentioned above.

33. A program that, when executed by a computer, causes the computer to assist in determining the truth or falsity of input information, and which causes the computer to The target object to be determined as true or false is obtained from the input information. Determine the search query based on whether the target object is single-modal or multimodal. Using the determined search query, obtain the results of the investigation of related information related to the target object. Output the results of the aforementioned survey, To determine the aforementioned search query, the computer will If the target object includes one modality type from text, image, audio, or video, the target object is determined to be single-modal, the target object determined to be single-modal is decomposed into multiple factual elements, and multiple search queries corresponding to the multiple factual elements are generated. The system includes determining that the target object is multimodal if it includes at least two modality types from among text, image, audio, or video formats, and generating different search queries for each of the multiple modality types included in the target object. program.

34. A program that, when executed by a computer, causes the computer to assist in determining the truth or falsity of input information, wherein the computer: The target object to be determined as true or false is obtained from the input information. Determine the search query based on whether the target object is single-modal or multimodal. Using the determined search query, obtain the results of the investigation of related information related to the target object. Output the results of the aforementioned survey, To determine the aforementioned search query, the computer will If the target object includes one of the following modality types: text, image, audio, or video, the target object is determined to be single-modal, and if the target object is single-modal, the target object is used to determine the search query. The system includes determining that the target object is multimodal if it includes at least two modality types from among text, image, audio, or video formats, and generating different search queries for each of the multiple modality types included in the target object. program.

Images