Method for checking authenticity of content, related devices and program product

Abstract

The present disclosure provides a method for verifying the authenticity of content, related devices and program products, relating to artificial intelligence technology fields such as content security, information retrieval, prompt word engineering, data governance, etc. A specific embodiment of the method includes: in response to the presence of at least three to-be-verified elements in the to-be-verified content, at least two to-be-verified element sets including at least two to-be-verified elements are constructed based on the to-be-verified elements, and the to-be-verified elements included between each to-be-verified element set are different; for each to-be-verified element set, based on the to-be-verified elements included in the to-be-verified element set, a first query information corresponding to the to-be-verified element set is constructed; using the first query information, evidence information is recalled; in response to the evidence information being able to cover all to-be-verified elements, the evidence information is used to verify each to-be-verified element, and a verification result for the to-be-verified content is generated. Thus, the quality of the authenticity verification of the content can be improved.

Description

Technical Field

[0001] This disclosure relates to the field of computer technology, specifically to artificial intelligence technologies such as content security, information retrieval, prompt word engineering, and data governance, and particularly to methods, devices, electronic devices, computer-readable storage media, and computer program products for verifying the authenticity of content. Background Technology

[0002] With the development of internet content production methods and content generation technologies, the sources of online content are becoming increasingly diverse. For example, online content can be uploaded manually by users, operators, or business personnel, or it can be generated and provided by large language models and other content generation systems.

[0003] Against this backdrop, to prevent erroneous, low-quality, or unreliable content from entering online platforms, platform operators and other content managers and maintainers typically need to verify the authenticity of relevant content to ensure content quality and service credibility. Therefore, how to efficiently and accurately verify the authenticity of content is a matter of concern and urgent need. Summary of the Invention

[0004] This disclosure provides a method, apparatus, electronic device, computer-readable storage medium, and computer program product for verifying the authenticity of content.

[0005] In a first aspect, embodiments of this disclosure propose a method for verifying the authenticity of content, comprising: responding to the existence of at least three elements to be verified in the content to be verified, constructing at least two sets of elements to be verified based on the elements to be verified, wherein each set of elements to be verified includes at least two elements to be verified, and the elements to be verified included in each set of elements to be verified are different; for each set of elements to be verified, constructing first query information corresponding to the set of elements to be verified based on the elements to be verified included in the set of elements to be verified; using the first query information to recall evidence information; responding to the evidence information being able to cover all elements to be verified, using the evidence information to verify each element to be verified, and generating a verification result for the content to be verified.

[0006] Secondly, embodiments of this disclosure propose an apparatus for verifying the authenticity of content, comprising: a first element set construction unit configured to construct at least two sets of elements to be verified based on the elements to be verified in response to the existence of at least three elements to be verified in the content to be verified, wherein each set of elements to be verified includes at least two elements to be verified, and the elements to be verified included in each set of elements to be verified are different; a first query information construction unit configured to construct first query information corresponding to each set of elements to be verified based on the elements to be verified included in the set of elements to be verified; an evidence information recall unit configured to recall evidence information using the first query information; and a first content verification unit configured to verify each element to be verified using the evidence information in response to the evidence information being able to cover all elements to be verified, thereby generating a verification result for the content to be verified.

[0007] Thirdly, embodiments of this disclosure provide an electronic device comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to implement a method for verifying the authenticity of content as described in any implementation of the first aspect.

[0008] Fourthly, embodiments of this disclosure provide a non-transitory computer-readable storage medium storing computer instructions that, when executed by a computer, enable the implementation of a method for verifying the authenticity of content as described in any implementation of the first aspect.

[0009] Fifthly, embodiments of this disclosure provide a computer program product including a computer program, which, when executed by a processor, can implement a method for verifying the authenticity of content as described in any implementation of the first aspect.

[0010] The method, apparatus, electronic device, computer-readable storage medium, and computer program product for verifying the authenticity of content provided in this disclosure firstly, in response to the existence of at least three elements to be verified in the content to be verified, at least two sets of elements to be verified are constructed based on the elements to be verified, wherein each set of elements to be verified includes at least two elements to be verified, and the elements to be verified included in each set of elements to be verified are different; then, for each set of elements to be verified, first query information corresponding to the set of elements to be verified is constructed based on the elements to be verified included in the set of elements to be verified; next, evidence information is recalled using the first query information; finally, in response to the evidence information being able to cover all elements to be verified, each element to be verified is verified using the evidence information, and a verification result for the content to be verified is generated.

[0011] This disclosure, by constructing an element set and recalling evidence based on that set, can take into account different factual dimensions and the relationships between elements, thereby improving the matching degree and coverage of evidence information with the content to be verified. Furthermore, by performing verification only after the evidence information covers all elements to be verified, this disclosure improves the completeness and traceability of the verification basis, reduces the risk of misjudgment, and thus enhances the accuracy and reliability of content authenticity verification.

[0012] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of this disclosure, nor is it intended to limit the scope of this disclosure. Other features of this disclosure will become readily apparent from the following description. Attached Figure Description

[0013] Other features, objects, and advantages of this disclosure will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings: Figure 1 This is an exemplary system architecture to which this disclosure can be applied; Figure 2 A flowchart illustrating a process for verifying the authenticity of content, provided as an embodiment of this disclosure; Figure 3 A flowchart illustrating a process for generating a verification result, provided in an embodiment of this disclosure; Figure 4 A flowchart illustrating another process for verifying the authenticity of content provided in this embodiment of the disclosure; Figure 5 A flowchart illustrating the process of verifying the authenticity of content in a specific application scenario, as provided in this embodiment of the disclosure; Figure 6 A structural block diagram of an apparatus for verifying the authenticity of content provided in this embodiment of the disclosure; Figure 7 This is a schematic diagram of the structure of an electronic device suitable for performing a method to verify the authenticity of content, as provided in an embodiment of this disclosure. Detailed Implementation

[0014] The exemplary embodiments of this disclosure are described below with reference to the accompanying drawings, including various details of the embodiments to aid understanding; these should be considered merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of this disclosure. Similarly, for clarity and brevity, descriptions of well-known functions and structures are omitted in the following description. It should be noted that, unless otherwise specified, the embodiments and features described in this disclosure can be combined with each other.

[0015] Furthermore, if the technical solutions disclosed herein involve information that falls under the category of user personal information (for example, user personal information is included in the content to be verified in the subsequent parts of this disclosure), the acquisition, storage, use, processing, transportation, provision, and disclosure of such information shall comply with the relevant laws and regulations and shall not violate public order and good morals.

[0016] Figure 1 An exemplary system architecture 100 is shown, illustrating embodiments of methods, apparatuses, electronic devices, and computer-readable storage media for verifying the authenticity of content, to which the present disclosure can be applied.

[0017] like Figure 1 As shown, system architecture 100 may include terminal devices 101, 102, and 103, a network 104, and a server 105. Network 104 serves as the medium for providing communication links between terminal devices 101, 102, and 103 and server 105. Network 104 may include various connection types, such as wired or wireless communication links, or fiber optic cables, etc.

[0018] Users can use terminal devices 101, 102, and 103 to interact with server 105 via network 104 to receive or send messages, etc. Various applications for enabling information communication between the terminal devices 101, 102, and 103 and server 105 can be installed. These applications include content authenticity verification applications, content generation applications, and instant messaging applications.

[0019] Terminal devices 101, 102, and 103 and server 105 can be either hardware or software. When terminal devices 101, 102, and 103 are hardware, they can be various electronic devices with displays, including but not limited to smartphones, tablets, laptops, and desktop computers. When terminal devices 101, 102, and 103 are software, they can be installed in the aforementioned electronic devices, and can be implemented as multiple software programs or software modules, or as a single software program or software module; no specific limitation is made here. When server 105 is hardware, it can be implemented as a distributed server cluster composed of multiple servers, or as a single server. When server 105 is software, it can be implemented as multiple software programs or software modules, or as a single software program or software module; no specific limitation is made here.

[0020] Server 105 can provide various services through its built-in applications. Taking a content authenticity verification application as an example, when running this application, server 105 can achieve the following: First, it receives content to be verified from terminal devices 101, 102, and 103 via network 104. Then, if there are at least three elements to be verified in the content, server 105 can respond by constructing at least two sets of elements to be verified, where each set includes at least two elements, and the elements included in each set differ. Then, for each set, server 105 constructs a first query information corresponding to the set based on the elements included in that set. Then, server 105 uses the first query information to recall evidence information. Finally, in response to the evidence information covering all elements to be verified, server 105 uses the evidence information to verify each element and generates a verification result for the content to be verified.

[0021] It should be noted that, in addition to being obtained from terminal devices 101, 102, and 103 via network 104, the content to be verified can also be pre-stored locally on server 105 through various means. Therefore, when server 105 detects that this data is already stored locally (for example, when starting to process a previously retained verification task that verifies the content to be verified that has already been obtained and maintained locally), it can choose to directly obtain this data from locally. In this case, the exemplary system architecture 100 may also exclude terminal devices 101, 102, and 103 and network 104.

[0022] Since determining the elements to be verified and recalling evidence information included in the content to be verified often requires significant computing resources and strong computing power, the methods for verifying the authenticity of content provided in the subsequent embodiments of this disclosure are generally executed by a server 105 with strong computing power and abundant computing resources. Correspondingly, the device for verifying the authenticity of content is also generally located in the server 105. However, it should also be noted that when terminal devices 101, 102, and 103 also have sufficient computing power and resources, they can also complete the aforementioned calculations performed by the server 105 through the content authenticity verification application installed on them, and thus output the same results as the server 105. Especially when multiple terminal devices with different computing capabilities exist simultaneously, but the content authenticity verification application determines that the terminal device has strong computing power and abundant remaining computing resources, the terminal device can be allowed to perform the aforementioned calculations, thereby appropriately reducing the computing pressure on the server 105. Accordingly, the device for verifying the authenticity of content can also be located in the terminal devices 101, 102, and 103. In this case, the exemplary system architecture 100 may also exclude the server 105 and the network 104.

[0023] It should be understood that Figure 1 The number of terminal devices, networks, and servers shown is merely illustrative. Depending on implementation needs, any number of terminal devices, networks, and servers can be included.

[0024] Please refer to Figure 2 , Figure 2 A flowchart of a process for verifying the authenticity of content provided in an embodiment of this disclosure includes process 200.

[0025] Process 200 specifically includes the following steps: Step 201: In response to the existence of at least three elements to be verified in the content to be verified, construct at least two sets of elements to be verified based on the elements to be verified; In the embodiments of this disclosure, the executing entity of the method for verifying the authenticity of content in this step (e.g., Figure 1 The server 105 shown can determine the number of elements to be verified after obtaining the content to be verified.

[0026] The element to be verified can be understood as a specific, smallest unit or point of content that needs to be verified. For example, the executing entity can break down the content entity to be verified into its constituent entities and use these entities to determine the elements to be verified.

[0027] For example, for content to be verified such as "Person A graduated from YY University with major ZZ in XX year", the executing entity can separate the entity into several entities such as "Person A", "graduated in XX year", "graduated from YY University", and "major ZZ" through entity decomposition, and use them to determine the elements to be verified.

[0028] It should be understood that in practice, based on different verification strategies, verification standards, and verification granularities, one can not only choose to determine the elements to be verified based on only one specific entity or by using a combination of at least two entities, but also choose to use a part of the extracted entity instead of all of it as the elements to be verified based on different entity extraction standards.

[0029] For example, in situations where certain entities lack authenticity verification value on their own, at least two entities can be combined to form a verification element. For instance, the entity "Person A" may lack authenticity verification value on its own, but it can be combined with "graduated from XX university" to generate a verification element. Furthermore, in different scenarios, the content corresponding to certain entities may be too commonplace and lack verification value; the executing entity may not include them as verification elements (e.g., widely accepted formulas included in the verification content, or a parameter in a formula with known significance).

[0030] In some embodiments, after obtaining the initial, original content to be verified, the executing entity may process the initial, original content to be verified according to predetermined standardized processing rules to remove invalid or interfering information, thereby obtaining the aforementioned content to be verified. This improves the quality of the subsequently segmented and determined elements to be verified, avoiding the identification of invalid or interfering information as elements to be verified. For example, the executing entity can process the initial, original content to be verified through character cleaning, format recognition, paragraph segmentation, etc.

[0031] Next, after determining the elements to be verified included in the content to be verified, the executing entity can detect the number of elements to be verified. If the number of elements to be verified is greater than or equal to three, that is, at least three elements to be verified have been split from the content to be verified, the executing entity can respond by constructing at least two sets of elements to be verified based on the elements to be verified.

[0032] During this process, the executing entity can select at least two elements from the aforementioned elements to be verified to form a set of elements to be verified. Furthermore, the elements to be verified included in each set of elements to be verified are different. That is, after determining the final sets of elements to be verified, there are no completely identical or duplicate sets of elements to be verified among these sets.

[0033] Furthermore, during this process, the executing entity should form at least two sets of elements to be verified, so that, based on the different elements to be verified contained in each set, first query information with different information search logic and search direction can be constructed, thereby enabling the executing entity to recall evidence information from different query directions through different first query information.

[0034] For example, if the content to be verified contains elements A, B, and C, the executing entity can determine the set of elements to be verified 1, which includes (elements A and B); the set of elements to be verified 2, which includes (elements B and C); and the set of elements to be verified 3, which includes (elements A, B, and C).

[0035] In practice, the range of values ​​with a lower limit of 2 and an upper limit of the total number of elements to be verified can be used as the value range for the execution subject. Accordingly, for any set of elements to be verified, the execution subject can select a value from this range as the number of elements to be verified included in the set, and select elements to be verified according to this number to form the set of elements to be verified.

[0036] In some embodiments, the executing entity may choose to form a set of elements to be verified by exhaustively considering all elements to be verified. For example, the executing entity may choose to form sets of elements to be verified by ensuring that each element to be verified exists in at least one set of elements to be verified. This ensures that, in the subsequent process of constructing the corresponding first query information for these sets of elements to be verified, the constructed first query information can cover and involve all elements to be verified, thereby improving the quality and coverage of the recalled evidence information.

[0037] Furthermore, since the set of elements to be verified will be used to construct the corresponding first query information, the executing entity can also classify the elements to be verified according to their "part of speech" and "component in the sentence" during the process of forming the set of elements to be verified. Elements that have the properties and functions of pointing to an object, such as subjects or objects, are maintained in each set of elements to be verified (e.g., person A). This is to avoid constructing first query information that lacks actual pointing and cannot be used due to a lack of information.

[0038] It should also be noted that, as discussed above, in different embodiments, the content to be verified can be obtained directly from the local storage device by the aforementioned execution entity, or it can be obtained from a non-local storage device (e.g., Figure 1The data can be obtained from the terminal devices 101, 102, and 103 shown. The local storage device can be a data storage module located within the execution entity, such as a server hard drive. In this case, the content to be verified can be quickly read locally. The non-local storage device can also be any other electronic device configured to store data, such as user terminals. In this case, the execution entity can obtain the required content to be verified by sending a retrieval command to the electronic device.

[0039] In some embodiments, if there are elements to be verified that point to the properties or functions of an object, these elements are typically highly versatile and may appear in multiple or even individual sets of elements to be verified. Therefore, when determining the number of elements to be verified, such as when detecting whether there are at least three elements to be verified in the content to be verified, the executing entity does not include these elements in the quantity statistics. Thus, when the content to be verified includes only one or two elements besides these types of elements, making it difficult to construct multiple sets of elements to be verified, the executing entity can avoid repeatedly performing combination, matching, or verification processing based on these general elements. This reduces invalid or low-value set construction processes, lowers computational complexity, saves computational resources, and improves overall verification efficiency.

[0040] In some embodiments, since the set of elements to be verified will be used to construct the first query information for querying and searching for evidence information, the specific elements to be verified included in the set of elements to be verified, as well as the number of elements to be verified, will directly affect the overall quantity and quality of the first query information. Therefore, if there are at least three elements to be verified in the content to be verified in this step, the executing entity can respond by choosing to read the actual number of elements to be verified first.

[0041] Then, a target quantity is determined based on the actual quantity to control and manage the quantity of the set of elements to be verified. This target quantity can be positively correlated with the actual quantity. For example, a mapping relationship between the target quantity and the actual quantity can be set in advance based on such a positive correlation. For instance, based on the possible numerical ranges of the actual quantity, a corresponding target quantity can be mapped to each numerical range, so that subsequent execution entities can map the corresponding target quantity based on the numerical range in which the actual quantity falls.

[0042] Then, the executing entity constructs a target number of sets of elements to be verified based on the elements to be verified.

[0043] Therefore, this approach allows the executing entity to adaptively determine the number of elements to be verified in the set based on the actual number of elements to be verified. This ensures that the subsequent evidence retrieval process using the constructed initial query information can cover different factual dimensions and element combinations, while avoiding the construction of excessive redundant sets when the number of elements to be verified is small. This guarantees the sufficiency of evidence retrieval while reducing invalid queries and computational resource consumption, thereby improving the processing efficiency and reliability of content authenticity verification.

[0044] Step 202: For each set of elements to be verified, construct the first query information corresponding to the set of elements to be verified based on the elements to be verified included in the set; In the embodiments of this disclosure, as discussed above, after the execution entity constructs the set of elements to be verified based on step 201, the execution entity can construct the first query information corresponding to the set of elements to be verified based on the elements to be verified included in the set of elements to be verified.

[0045] For example, the executing entity can generate first query information for querying the elements to be verified included in the set of elements to be verified by calling a pre-trained large model or generative model with content generation capabilities, such as a Large Language Model (LLM).

[0046] LLMs can be trained on large amounts of text data and perform a wide range of tasks, including text summarization, translation, sentiment analysis, and more. A key characteristic of LLMs is their large scale; they typically include a large number of parameters to help them learn complex patterns in language data. These models are often based on deep learning architectures, such as transformers, which contributes to their superior performance across various natural language processing tasks.

[0047] For example, the executing entity can use pre-configured prompts to instruct or invoke the LLM to generate first query information for querying all elements to be verified in the set of elements to be verified.

[0048] For example, a set of elements to be verified, including "Person A", "Graduated in XX year", and "Graduated from YY University", can be processed by LLM into, for example, the first query information "Did Person A graduate from YY University in XX year?"

[0049] Furthermore, for LLM, the "guide word" can be omitted through default configuration. For example, for the purpose of generating evidence information related to the provided elements to be verified for querying, searching, and recalling the elements to be verified, LLM can, based on the default configuration, naturally understand that it needs to construct the corresponding first query information based on these elements to be verified.

[0050] In this step, the executing entity can generate first query information corresponding to each set of elements to be verified by repeating the above process. Accordingly, each first query information can be used to recall evidence information related to the elements to be verified in its corresponding set of elements to be verified.

[0051] In some embodiments, the specific presentation form of the first query information can be diverse, so as to adapt it to different types of search engines, recall engines or information retrieval systems and obtain better retrieval and recall effects.

[0052] For example, the first query can be a natural language query sentence constructed based on the elements to be verified, such as: "Did Person A graduate from YY University in XX year?", "Did Person A graduate from YY University?", "Where did Person A graduate from in XX year?", etc. Thus, by adopting natural language query sentences, it is possible to better adapt to search engines that support semantic understanding or question-answering retrieval.

[0053] For example, the first query can also be a keyword search or Boolean search based on the elements to be verified, such as: "Person A AND XX year graduated AND YY university". Therefore, by using keyword search or Boolean search, it is possible to better adapt to traditional search engines, database retrieval systems, or structured information retrieval systems, thereby improving the retrieval efficiency and matching accuracy of relevant evidence information.

[0054] Step 203: Retrieve evidence information using the first query information; In the embodiments of this disclosure, after constructing the first query information based on the above step 202, the executing entity can use this first query information to recall evidence information that can match at least part of the content in the first query information (or, at least one of the elements to be verified targeted by the first query information) from pre-configured knowledge bases, online knowledge bases and other data and information sources.

[0055] In this step, after the executing entity completes the recall of evidence information, it can determine whether the currently recalled evidence information can cover all the elements to be verified (for example, the content of each element to be verified can be included and verified by at least one piece of evidence information).

[0056] If the evidence information recalled at this time can cover all the elements to be verified, the executing entity can respond to this and execute step 204.

[0057] Step 204: Use the evidence information to verify each element to be verified and generate verification results for the content to be verified.

[0058] In the embodiments of this disclosure, after recalling the evidence information based on step 203 above, the executing entity can detect whether this evidence information can hit and cover all the elements to be verified. For example, after recalling the evidence information, the executing entity can mark the elements to be verified that the evidence information points to and hits. Accordingly, the executing entity can subsequently determine the coverage of each element to be verified by the evidence information by checking whether each element to be verified is associated with at least one marking result.

[0059] If the current evidence information can cover all the elements to be verified, the executing entity can respond by using the current evidence information to verify each element and generate a verification result for the content to be verified.

[0060] Typically, the executing entity can choose to use evidence information to verify the elements to be verified that were hit in the first query information that recalled the evidence information, in order to improve verification efficiency.

[0061] The verification result of the element to be verified can be at least whether the content corresponding to the element to be verified is true or false, that is, true or false (or, it can also be said to be false).

[0062] In some embodiments, in order to make fuller use of the evidence information, the executing entity may also choose to use the evidence information to verify the elements to be verified that are not included in the first query information used to recall the evidence information (for example, the executing entity may use keyword retrieval or semantic relevance retrieval to determine these elements to be verified from the elements to be verified outside the first query information).

[0063] For ease of description, the verification result corresponding to each element to be verified can be called a "verification sub-result". Accordingly, after the executing entity completes the verification of each element to be verified and generates the corresponding verification sub-result, the executing entity can summarize and combine these verification sub-results to generate the final verification result for the content to be verified.

[0064] In some embodiments, the executing entity can represent and present the true or false (fake) results of the elements to be verified by adding indicator markers corresponding to each element to be verified in the content to be verified.

[0065] The method for verifying the authenticity of content provided in this disclosure firstly, in response to the existence of at least three elements to be verified in the content to be verified, at least two sets of elements to be verified are constructed based on these elements. Each set of elements to be verified includes at least two elements to be verified, and the elements to be verified differ between sets. Then, for each set of elements to be verified, first query information corresponding to the set is constructed based on the elements to be verified included in that set. Next, evidence information is retrieved using the first query information. Finally, in response to the evidence information covering all elements to be verified, each element to be verified is verified using the evidence information, generating a verification result for the content to be verified. Therefore, by constructing element sets and retrieving evidence based on these element sets, different factual dimensions and the relationships between elements can be considered, improving the matching degree and coverage of evidence information with the content to be verified. Furthermore, this disclosure improves the completeness and traceability of the verification basis by performing verification only after the evidence information covers all elements to be verified, reducing the risk of misjudgment and thus improving the accuracy and reliability of content authenticity verification.

[0066] In some embodiments, to achieve more accurate and higher-quality verification of the elements to be verified, a matching verification strategy can be assigned to each element based on its element type. This allows each element to be verified to undergo authenticity verification in a manner that aligns with its semantic attributes, factual attributes, and verification requirements. Accordingly, verification models for implementing and executing these verification strategies can be pre-trained and deployed to corresponding verification channels. This enables the executing entity to perform verification on elements of different element types using different verification channels and verification models during the verification process.

[0067] To facilitate discussion of this situation, the following can be referenced together. Figure 3 Please provide an explanation. Figure 3 A flowchart illustrating a process for generating a verification result, as provided in this embodiment of the disclosure, includes process 300. For example, process 300 can be an alternative or alternative implementation of step 204 described above.

[0068] Process 300 specifically includes the following steps: Step 301: In response to the fact that the evidence information can cover all the elements to be verified, the elements to be verified are assigned to the verification channels according to the element type of each element. Specifically, as discussed above, if the evidence information can cover all the elements to be verified, the executing entity can respond by first allocating the elements to be verified to the verification channels based on the element type of each element.

[0069] In practice, the classification strategy for element types can be based on factual types, communication number (address) types, etc., or it can be based on classification strategies such as text, images, audio, etc.

[0070] Accordingly, the verification channel can be configured as discussed above, with a verification model pre-trained for the element type and corresponding verification strategy. For example, for factual types, the corresponding verification model can be pre-trained based on the sample element to be verified and the corresponding sample evidence information.

[0071] Step 302: Provide the associated evidence information covering the element to be verified in the evidence information to the verification channel corresponding to the element to be verified; Specifically, after assigning the element to be verified to the corresponding verification channel, the executing entity can provide the evidence information associated with the element to be verified (e.g., the evidence information recalled by the first query information composed of the element to be verified) to the verification channel so that the subsequent verification model can use this evidence information to verify the element to be verified in the corresponding verification channel.

[0072] It should be understood that, if the aforementioned evidence information is to be utilized more fully, the executing entity may choose to provide, along with the evidence information that is not associated with the corresponding first query information but is associated, to the corresponding verification channel.

[0073] Accordingly, for ease of description, the evidence information related to the element to be verified and which can be used to verify the element to be verified can also be described here and thereafter as the “related evidence information” of the element to be verified.

[0074] Step 303: In each verification channel, the corresponding deployed verification model in the verification channel is used to verify the elements to be verified, and the verification sub-results for the elements to be verified in the verification channel are generated. Specifically, as discussed above, based on steps 301 and 302, the corresponding elements to be verified and the corresponding associated evidence information are allocated to each verification channel. The executing entity can use the verification model deployed in the verification channel to verify the elements to be verified and generate verification sub-results for the elements to be verified in the verification channel.

[0075] Then, as discussed above, the executing entity combines the results of each verification sub-result through subsequent step 304 to form a verification result for the content to be verified, which will not be repeated here.

[0076] Step 304: Combine the results of each verification sub-result to form a verification result for the content to be verified.

[0077] In some embodiments, considering that the verification models are often trained and deployed for a single feature type, a basic model with lower model size and parameter complexity can be used to train these verification models. This reduces the training difficulty of the verification models while also reducing the resource consumption for their deployment. Correspondingly, in such cases, to ensure the verification effect and quality, a fallback verification model with a larger model size and parameter complexity than the aforementioned verification models, capable of verifying multiple parameter types using corresponding verification strategies, can be used as a "backup."

[0078] Therefore, in some embodiments, the executing entity verifies the elements to be verified by utilizing the corresponding deployed verification model in each verification channel, and generates verification sub-results for the elements to be verified in each verification channel. In each verification channel, the corresponding deployed verification model can be used to verify the elements to be verified, and verification sub-results for the elements to be verified and a certainty index for the verification sub-results can be generated together.

[0079] For example, the executing entity can obtain the certainty index (e.g., confidence, degree of certainty) of the validation model for the processed validation sub-results by validating the classification head and attention head deployed in the validation model.

[0080] Then, if the certainty index output by the target verification channel is less than the threshold (usually set based on the standard that the verification model considers the output results reliable and trustworthy), the executing entity can respond by using a re-verification model to re-verify the elements to be verified in the target verification channel based on the associated evidence information in the target verification channel, generating a re-verification sub-result. This re-verification model can be a "backup" model with a larger model size than the corresponding verification model, as discussed above. In other words, this re-verification model is a "model" that can be trained according to the verification strategy corresponding to the verification model and has stronger model capabilities.

[0081] Subsequently, if the result of the re-inspection verification sub-result differs from the result output by the target verification channel, the executing entity can choose to replace the result output by the target verification channel with the result of the re-inspection verification sub-result. This ensures the generation and verification quality of the verification sub-results.

[0082] In some embodiments, the evidence information recalled above (e.g., the evidence information recalled in step 203) may not be sufficient or able to fully cover all the elements to be verified. In such cases, the executing entity can recall evidence information in multiple rounds by constructing query information in multiple rounds to ensure the quality of the evidence information.

[0083] For ease of discussion, please refer to the following: Figure 4 . Figure 4 A flowchart of another process for verifying the authenticity of content provided in an embodiment of this disclosure, including process 400.

[0084] Process 400 specifically includes the following steps: Step 401: In response to the existence of at least three elements to be verified in the content to be verified, construct at least two sets of elements to be verified based on the elements to be verified; Step 402: For each set of elements to be verified, construct the first query information corresponding to the set of elements to be verified based on the elements to be verified included in the set of elements to be verified; Step 403: Retrieve evidence information using the first query information; Steps 401-403 above are the same as those mentioned above. Figure 2 Steps 201-203 shown are the same. For the same parts, please refer to the corresponding parts of the previous embodiment. They will not be repeated here.

[0085] Next, as discussed in process 200, if the evidence information currently recalled can cover all the elements to be verified, the executing entity can respond to this and execute step 404, which will not be discussed again here.

[0086] Step 404: Use the evidence information to verify each element to be verified and generate verification results for the content to be verified.

[0087] If the existing evidence information that has been recalled does not cover all the elements to be verified, the executing entity can respond to this, that is, respond to the fact that the evidence information does not cover all the elements to be verified, and execute step 405.

[0088] Step 405: Identify any missing elements to be verified that were not covered by the evidence information; Specifically, if the existing evidence information that has been recalled cannot cover all the elements to be verified, the executing entity can read out those elements that were not hit and identify them as omitted elements to be verified.

[0089] Step 406: Construct a set of missing elements to be verified based on the missing elements to be verified; Specifically, the executing entity can construct a set of missing elements to be verified using these missing elements, as discussed in step 201 above. For example, the executing entity can form and construct a set of missing elements to be verified by combining at least two missing elements.

[0090] In some embodiments, after constructing the missing elements to be verified, the executing entity can first determine the semantic completeness and the completeness of the reference of the set of missing elements to be verified (for example, the executing entity can determine whether there is a lack of reference by detecting whether there are missing elements that can be used as subjects or objects in the currently formed set of missing elements to be verified). For example, whether the lack of a reference caused the missing elements such as "graduated in XX year" or "graduated from XX university".

[0091] If such a situation exists, the executing entity can fill in these elements to be verified that can point to specific objects and are used to determine the target, so as to form the final "set of missing elements to be verified".

[0092] In some embodiments, the executing entity can pre-construct such pointing relationships by building the node relationships between the various elements to be verified. For example, the executing entity can construct pointing (or combination, association) relationships between the various elements to be verified based on the position of the element to be verified in the original text, title, context parent node, citation relationship, and relevant supplementary explanation, so as to avoid the situation of "lack of pointing" caused by forming a set of elements to be verified or a set of missing elements of a "fragment" separately in the future.

[0093] Step 407: For each set of missing elements to be verified, construct the second query information corresponding to the set of missing elements to be verified based on the missing elements included in the set of missing elements to be verified; Specifically, similar to the discussion in step 202 above, after obtaining the set of missing elements to be verified based on step 406, the executing entity can construct query information corresponding to each set of missing elements to be verified based on the missing elements included in the set. For ease of description, the query information here can be described as second query information.

[0094] Step 408: Utilize the second query information to recall supplementary evidence information; Specifically, after constructing the second query information based on step 407 above, the executing entity can continue to use the second query information to retrieve supplementary evidence information from pre-configured knowledge bases, online knowledge bases and other data and information sources.

[0095] Then, in this step, the executing entity can determine whether the recalled supplementary evidence information and the previously recalled evidence information together can cover all the elements to be verified.

[0096] If so, the executing entity can respond to this and continue with step 409, similar to step 204 above, to verify each element to be verified using evidentiary information and supplementary evidentiary information, and generate a verification result for the content to be verified.

[0097] Step 409: Use the evidence information and supplementary evidence information to verify each element to be verified, and generate verification results for the content to be verified.

[0098] Therefore, the executing entity can adopt a multi-round approach to construct query information. In cases where some elements to be verified are not effectively covered due to improper combination of the set of elements to be verified, insufficient construction of the first query information, or deviation in the query direction, supplementary query information can be generated in a timely manner and corresponding evidence information can be recalled. This enhances the completeness and robustness of the evidence information recall process, reduces the risk of evidence omission caused by inaccurate expression of the first query information, unreasonable combination of elements, or a single search direction, and enables subsequent authenticity verification to be based on a more sufficient and comprehensive evidence foundation, thereby improving the accuracy, credibility, and stability of the content quality verification results.

[0099] Accordingly, if after step 408 there are still cases where not all elements to be verified can be covered, the executing entity can jump to step 405 to re-identify the missing elements to be verified and continue to recall the supplementary evidence information in the next round until full coverage is achieved. This will not be repeated here.

[0100] In some embodiments, if the element to be verified is simultaneously covered by both the first evidence information and the second evidence information—for example, after steps 203 and 403, if the executing entity detects that the element to be verified is simultaneously covered by both the first evidence information and the second evidence information—the executing entity can detect a conflict between the first evidence information and the second evidence information. For example, it can detect whether the content provided by the first evidence information and the second evidence information at the element to be verified is conflicting or contradictory.

[0101] If there is a conflict between the first piece of evidence and the second piece of evidence, the executing entity may respond by marking the element to be verified as an omitted element to be verified, so that the element to be verified may be included in the supplementary search scope in subsequent rounds.

[0102] This allows for the retrieval of more relevant evidence for elements under verification that are disputed or inconsistent, thereby expanding the evidentiary base and improving the sufficiency and diversity of evidence coverage. In the face of inconsistencies among multiple sources of evidence, this approach, through continuous supplementary retrieval and evidence expansion, yields more comprehensive and reliable judgment criteria, thus reducing the possibility of erroneous or biased verification and improving the quality of verification.

[0103] In some embodiments, when marking missing elements to be verified in the above manner, in order to avoid the evidence information provided by low-quality or low-credibility evidence sources interfering with the evidence information provided by high-quality, authoritative data sources, and thus causing elements to be verified that can actually be effectively verified by authoritative data sources to be mistakenly marked as missing elements to be verified, the executing entity may further introduce a credibility level judgment mechanism for evidence information during the conflict detection process.

[0104] Specifically, in the process of marking the first element to be verified as an omitted element to be verified when it is simultaneously covered by both first and second evidence information, and a conflict exists between the first and second evidence information, the executing entity can further determine whether the credibility levels of both the first and second evidence information are less than a first-level threshold. If the credibility levels of both the first and second evidence information are less than the first-level threshold, it indicates that the overall credibility of the currently conflicting evidence information is low, making it difficult to directly use as a reliable basis for verifying the first element to be verified. Therefore, the first element to be verified can be marked as an omitted element to be verified so that supplementary searches can be performed on it in subsequent rounds.

[0105] The first-level threshold can be pre-set based on evaluation criteria of high-quality data sources, authoritative data sources, or credible publishing entities. For example, the executing entity can determine the credibility level of evidence information based on indicators such as the publishing entity, data source type, publication time, historical credibility records, citation frequency, and official certification status. Accordingly, when the credibility level of a piece of evidence information is less than the first-level threshold, it can be considered as evidence information with low credibility or that does not meet the requirements of authority. Conversely, when the credibility level of a piece of evidence information is greater than or equal to the first-level threshold, it can be considered as evidence information with high credibility and can serve as a reliable basis for verifying the authenticity of the elements to be verified to a certain extent.

[0106] Accordingly, if the first element to be verified is covered by both the first evidence information and the second evidence information, and there is a conflict between the first evidence information and the second evidence information, and the credibility level of both the first evidence information and the second evidence information is less than the first level threshold, then the executing entity may, in response, mark the first element to be verified as an omitted element to be verified.

[0107] In some embodiments, in the evidence information recall rounds after the first round, if the target missing element to be verified is also covered by both the first supplementary evidence information and the second supplementary evidence information, and there is a conflict between the first supplementary evidence information and the second supplementary evidence information, the executing entity may choose to read the first credibility level of the first supplementary evidence information and the second credibility level of the second supplementary evidence information.

[0108] Then, since the missing element to be verified has already been supplemented in at least one subsequent supplementary round, the executing entity can select at least one of the first and second supplementary evidence information to retain based on the first and second confidence levels. In other words, the executing entity can choose to retain the first and / or second supplementary evidence information according to the confidence level of each supplementary evidence information, as the basis for subsequent verification of the missing element to be verified.

[0109] This approach allows the implementing entity to filter and retain supplementary evidence based on its credibility level, even when supplementary retrieval has been conducted to address missing elements in the target data but conflicts still exist between related supplementary evidence. This avoids continuously triggering new supplementary retrieval processes. Consequently, it prevents the system from endlessly performing supplementary retrievals due to conflicting or inconsistent evidence, reducing the risks of cyclical recalls, resource waste, or process blockages. Furthermore, retaining supplementary evidence with higher credibility levels provides a relatively reliable evidentiary basis for subsequent authenticity verification, thereby improving system stability and the usability of verification results.

[0110] That is, after reading the first credibility level of the first supplementary evidence information and the second credibility level of the second supplementary evidence information, if the first credibility level is greater than the second credibility level, the executing entity may, in response, retain the first supplementary evidence information; or

[0111] If the second level of credibility is greater than the first level of credibility, the executing entity may respond by retaining the second supplementary evidence information.

[0112] In some embodiments, during the process of constructing a set of missing verification elements based on the missing verification elements, the executing entity may also choose to generate supplementary verification elements based on the synonymous rewriting of the missing verification elements. For example, for "graduated from XX university", the executing entity can rewrite the missing verification elements into synonymous supplementary verification elements such as "resume" and "educational experience".

[0113] Then, in the process of constructing the set of missing elements to be verified based on the missing elements to be verified, the set of missing elements to be verified is constructed based on the missing elements to be verified and the supplementary elements to be verified.

[0114] Accordingly, in the process of constructing the second query information corresponding to each set of missing elements to be verified based on the missing elements included in the set of missing elements to be verified, the executing entity can further construct the second query information corresponding to each set of missing elements to be verified based on the missing elements and supplementary elements included in the set of missing elements to be verified.

[0115] Therefore, by rewriting the missing elements to be verified using synonyms and generating supplementary elements to be verified during the process of constructing a set of missing elements to be verified, the expression of the missing elements to be verified can be expanded. This allows the executing entity to no longer rely solely on the original expression when constructing the second query information, but to simultaneously use multiple expressions that are semantically similar to or equivalent to the missing element to be verified for retrieval. This enhances the semantic expansion capability and evidence recall capability of supplementary retrieval of missing elements to be verified, reduces evidence omissions caused by expression differences, and improves the coverage, completeness, and relevance of supplementary evidence information.

[0116] In some embodiments, in order to enable the second query information constructed based on the set of missing elements to be verified to have a wider search range and a stronger ability to recall evidence information, the upper limit of the number of elements in the set of missing elements to be verified can be limited to be less than the upper limit of the number of elements in the set of elements to be verified.

[0117] For example, if the set of elements to be verified can include a maximum of 5 elements, then the maximum number of elements in the set of missing elements to be verified can be limited to 3. By reducing the number of elements in the set of missing elements to be verified, the amount of information that the second query information constructed based on the set of missing elements to be verified needs to hit simultaneously can be reduced, avoiding narrowing the search scope due to too many query conditions or overly strong constraints.

[0118] This allows for the retrieval of supplementary evidence within a broader search space when using the second query information, thereby improving the coverage of potentially relevant evidence.

[0119] In some embodiments, if there are sub-elements in the omitted elements to be verified that can be covered by evidence information, the executing entity may, in response, remove the sub-elements from the omitted elements to be verified.

[0120] For example, if the missing element to be verified is "Person A graduated from University B and previously worked at Company C", and the currently recalled evidence information can cover the sub-element "Person A graduated from University B" but not the sub-element "previously worked at Company C", then the executing entity can remove "Person A graduated from University B" from the missing elements to be verified and retain "previously worked at Company C" as the updated missing element to be verified, so that subsequent rounds can construct query information and conduct supplementary searches for the uncovered part.

[0121] This allows for fine-grained updates to missing verification elements, enabling the status of these elements to be dynamically adjusted based on the recall results of evidence information. This avoids redundant searches for content that has already been effectively covered or verified in subsequent rounds. This not only reduces invalid queries and duplicate recalls but also lowers the consumption of computational, retrieval, and network access resources.

[0122] Based on any of the above embodiments, in the process of using evidence information to verify each element to be verified and generating a verification result for the content to be verified (for example, in the process of executing step 204), as an alternative or alternative, the executing entity may also first determine how many pieces of evidence information cover each element to be verified.

[0123] If the element to be verified is covered by only one related piece of evidence information, the executing entity can respond by using the evidence information to generate a first verification sub-result for the content to be verified.

[0124] If the element to be verified is covered by at least two related pieces of evidence, the executing entity can respond by reading the credibility level of each piece of related evidence, similar to what was discussed above.

[0125] If the associated evidence information includes only one target associated evidence information with a credibility level greater than or equal to the second-level threshold (its setting logic can be similar to the first-level threshold mentioned above, but the specific values ​​can be the same or different), then the executing entity can respond by using the target associated evidence information to generate a first verification sub-result for the content to be verified.

[0126] If the evidence information includes at least two target-related evidence information with a credibility level greater than or equal to the second-level threshold, that is, both "related evidence information" are credible and authoritative "target-related evidence information", then the executing entity can use the target-related evidence information to verify the element to be verified, and generate a second verification sub-result corresponding to each target-related evidence information for the element to be verified.

[0127] Then, by combining the first and second verification sub-results, a verification result is formed for the content to be verified. For example, if the first and second verification sub-results are the same, the executing entity can retain only one of them; if they are different, the executing entity can retain both and provide the two target-related evidence information used, as well as the verification sub-results corresponding to each target-related evidence information, in different display units in the final verification result.

[0128] Therefore, this approach enables the adoption of differentiated and hierarchical verification strategies for different elements to be verified based on the amount of evidence coverage and the level of evidence credibility. This ensures verification efficiency while improving the accuracy, reliability, and interpretability of the verification results for the authenticity of the content to be verified.

[0129] To enhance understanding, this disclosure also provides an implementation scheme for a specific application scenario. Please refer to this scheme for ease of discussion. Figure 5 . Figure 5 The flowchart illustrates a process for verifying the authenticity of content in a specific application scenario, as provided in this embodiment of the disclosure, including process 500. For example, process 500 may be executed and implemented by server 105 as illustrated in architecture 100.

[0130] In process 500, server 105 (not shown in the figure) can read the number of elements to be verified after obtaining the content to be verified 510. For example, it can read the elements to be verified 511, 512, 513, 514 and 515.

[0131] Accordingly, since the number of elements to be verified is "5", which is greater than "3", the server 105 can respond by executing S501 to construct at least two sets of elements to be verified based on the elements to be verified (i.e., elements 511, 512, 513, 514 and 515).

[0132] For example, server 105 can construct a set of elements to be verified, 521, 522, 523...52N (where N is a positive integer).

[0133] Then, server 105 can continue to execute S502 to construct the first query information corresponding to each set of elements to be verified (i.e., each of the sets of elements to be verified 521, 522, 523...52N) based on the elements to be verified included in the set of elements to be verified.

[0134] For example, based on different "query forms", server 105 can construct query information 531 for the set of elements to be verified 521, construct query information 532 for the set of elements to be verified 522, ... construct query information 53N for the set of elements to be verified 52N.

[0135] Then, server 105 can retrieve evidence information by executing S503 using this evidence query information. For example, server 105 can retrieve evidence information 541, 542...549.

[0136] For example, if the evidence information 541, 542...549 can completely cover the elements to be verified 521, 522, 523...52N, in such a case, the server 105 can respond to this and continue to execute S504 to use the evidence information 541, 542...549 to verify the elements to be verified 521, 522, 523...52N, and generate a verification result 550 for the content to be verified 510.

[0137] For example, in the verification result 550, it can be verified that Zhang San graduated from University A with major B in 2012 as true, and that he once worked for Company C as false. Accordingly, in the verification result 550, the true or false (fake) results of the elements to be verified can be represented and presented by adding indicator labels to each element to be verified.

[0138] It should be understood that in the example of process 500 above, the number of elements to be verified, the number of sets of elements to be verified constructed, the number of first query information, and the number of recalled evidence information are all exemplary choices for ease of understanding and are not intended to limit the number of these objects.

[0139] Further reference Figure 6 As an implementation of the methods shown in the above figures, this disclosure provides an embodiment of a device for verifying the authenticity of content. This device embodiment is similar to... Figure 2 Corresponding to the method embodiments shown, this device can be specifically applied to various electronic devices.

[0140] like Figure 6 As shown, the device 600 for verifying the authenticity of content in this embodiment may include: a first element set construction unit 601, a first query information construction unit 602, an evidence information recall unit 603, and a first content verification unit 604. The first element set construction unit 601 is configured to, in response to the existence of at least three elements to be verified in the content to be verified, construct at least two sets of elements to be verified based on the elements to be verified, wherein each set of elements to be verified includes at least two elements to be verified, and the elements to be verified included in each set of elements to be verified differ. The first query information construction unit 602 is configured to, for each set of elements to be verified, construct first query information corresponding to the set of elements to be verified based on the elements to be verified included in the set. The evidence information recall unit 603 is configured to, using the first query information, recall evidence information. The first content verification unit 604 is configured to, in response to the evidence information covering all elements to be verified, verify each element to be verified using the evidence information, and generate a verification result for the content to be verified.

[0141] In this embodiment, the specific processing and technical effects of the first element set construction unit 601, the first query information construction unit 602, the evidence information retrieval unit 603, and the first content verification unit 604 in the device 600 for verifying the authenticity of content can be referred to respectively. Figure 2 The relevant descriptions of steps 201-204 in the corresponding embodiments will not be repeated here.

[0142] In some optional implementations of this embodiment, the apparatus 600 further includes: an omission element determination unit, configured to determine omission elements to be verified that were not covered by the evidence information in response to the evidence information not covering all the elements to be verified; a second element set construction unit, configured to construct a set of omission elements to be verified based on the omission elements to be verified; a second query information construction unit, configured to construct second query information corresponding to each set of omission elements to be verified based on the omission elements to be verified included in the set of omission elements to be verified; an evidence information re-recall unit, configured to recall supplementary evidence information using the second query information; and a second content verification unit, configured to verify each element to be verified using the evidence information and the supplementary evidence information in response to the supplementary evidence information and the evidence information covering all the elements to be verified, and generate a verification result for the content to be verified.

[0143] In some optional implementations of this embodiment, the device 600 further includes: a missing element marking unit, configured to mark the element to be verified as a missing element in response to the element to be verified being simultaneously covered by the first evidence information and the second evidence information, and a conflict between the first evidence information and the second evidence information.

[0144] In some optional implementations of this embodiment, the missing element marking unit is further configured to mark the first element to be verified as a missing element to be verified in response to the fact that the first evidence information and the second evidence information are simultaneously covered by the first evidence information and the second evidence information, and there is a conflict between the first evidence information and the second evidence information, and the credibility level of the first evidence information and the second evidence information is less than the first level threshold.

[0145] In some optional implementations of this embodiment, the device 600 further includes: a credibility level reading unit, configured to read the first credibility level of the first supplementary evidence information and the second credibility level of the second supplementary evidence information in response to a situation where the target omitted verification element is simultaneously covered by the first supplementary evidence information and the second supplementary evidence information, and there is a conflict between the first supplementary evidence information and the second supplementary evidence information; and a supplementary evidence information retention unit, configured to retain the first supplementary evidence information in response to the first credibility level being greater than the second credibility level; or to retain the second supplementary evidence information in response to the second credibility level being greater than the first credibility level.

[0146] In some optional implementations of this embodiment, the apparatus 600 further includes: a supplementary element generation unit configured to generate supplementary elements to be verified based on a synonymous rewrite of the missing elements to be verified; a second element set construction unit further configured to construct a set of missing elements to be verified based on the missing elements to be verified and the supplementary elements to be verified; and a second query information construction unit further configured to, for each set of missing elements to be verified, construct second query information corresponding to the set of missing elements to be verified based on the missing elements to be verified and the supplementary elements to be verified included in the set of missing elements to be verified.

[0147] In some optional implementations of this embodiment, the upper limit of the number of elements in the missing elements set to be verified is less than the elements set to be verified.

[0148] In some optional implementations of this embodiment, the apparatus 600 further includes: a missing element adjustment unit, configured to remove sub-elements from the missing elements to be verified in response to the existence of sub-elements that can be covered by evidence information in the missing elements to be verified.

[0149] In some optional implementations of this embodiment, the verification of each element to be verified using evidence information to generate a verification result for the content to be verified includes: for each element to be verified: in response to the element to be verified being covered by only one related evidence information in the evidence information, generating a first verification sub-result for the content to be verified using the evidence information; or in response to the element to be verified being covered by at least two related evidence information, reading the credibility level of each related evidence information; in response to the related evidence information including only one target related evidence information with a credibility level greater than or equal to a second level threshold, generating a first verification sub-result for the content to be verified using the target related evidence information; or in response to the evidence information including at least two target related evidence information with a credibility level greater than or equal to a second level threshold; verifying the element to be verified using the target related evidence information respectively, generating a second verification sub-result for the element to be verified corresponding to each target related evidence information; and forming a verification result for the content to be verified by combining the first verification sub-result and the second verification sub-result.

[0150] In some optional implementations of this embodiment, the first content verification unit 604 includes: a channel allocation subunit, configured to allocate the elements to be verified to a verification channel based on the element type of each element to be verified in response to the evidence information covering all elements to be verified; an evidence information providing subunit, configured to provide the associated evidence information covering the elements to be verified to the verification channel corresponding to the elements to be verified; a content verification subunit, configured to verify the elements to be verified in each verification channel using the corresponding deployed verification model in the verification channel, and generate verification sub-results for the elements to be verified in the verification channel; and a verification sub-result combination unit, configured to combine the various verification sub-results to form a verification result for the content to be verified.

[0151] In some optional implementations of this embodiment, the content verification subunit includes: a content verification module configured to verify the elements to be verified in each verification channel using the corresponding deployed verification model in the verification channel, generating verification sub-results for the elements to be verified and a certainty index for the verification sub-results; a content re-examination module configured to, in response to the certainty index output by the target verification channel being less than the index threshold, use a re-verification model to re-examine the elements to be verified in the target verification channel based on the associated evidence information in the target verification channel, generating a re-examination verification sub-result, wherein the model size of the re-examination verification model is larger than that of the verification model; and a verification result replacement module configured to, in response to the re-examination verification sub-result being different from the verification sub-result output by the target verification channel, replace the verification sub-result output by the target verification channel with the re-examination verification sub-result.

[0152] In some optional implementations of this embodiment, the first element set construction unit 601 includes: an actual quantity reading subunit, configured to read the actual quantity of the elements to be verified in response to the existence of at least three elements to be verified in the content to be verified; a target quantity determination subunit, configured to determine a target quantity based on the actual quantity; and an element set construction subunit, configured to construct a set of elements to be verified with a target quantity based on the elements to be verified.

[0153] This embodiment exists as a device embodiment corresponding to the above method embodiment. The device for verifying the authenticity of content provided in this embodiment, by constructing an element set and recalling evidence based on the element set, can take into account different factual dimensions and the relationships between elements, thereby improving the matching degree and coverage of evidence information with the content to be verified. Furthermore, this disclosure improves the completeness and traceability of the verification basis by performing verification only after the evidence information covers all elements to be verified, reducing the risk of misjudgment, thereby improving the accuracy and reliability of content authenticity verification.

[0154] According to embodiments of this disclosure, this disclosure also provides an electronic device, a readable storage medium, and a computer program product.

[0155] Figure 7 A schematic block diagram of an example electronic device 700 that can be used to implement embodiments of the present disclosure is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the present disclosure described and / or claimed herein.

[0156] like Figure 7 As shown, device 700 includes a computing unit 701, which can perform various appropriate actions and processes based on a computer program stored in read-only memory (ROM) 702 or a computer program loaded from storage unit 708 into random access memory (RAM) 703. RAM 703 may also store various programs and data required for the operation of device 700. The computing unit 701, ROM 702, and RAM 703 are interconnected via bus 704. Input / output (I / O) interface 705 is also connected to bus 704.

[0157] Multiple components in device 700 are connected to I / O interface 705, including: input unit 706, such as keyboard, mouse, etc.; output unit 707, such as various types of monitors, speakers, etc.; storage unit 708, such as disk, optical disk, etc.; and communication unit 709, such as network card, modem, wireless transceiver, etc. Communication unit 709 allows device 700 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0158] The computing unit 701 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 701 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 701 performs the various methods and processes described above, such as methods for verifying the authenticity of content. For example, in some embodiments, the method for verifying the authenticity of content may be implemented as a computer software program tangibly contained in a machine-readable medium, such as storage unit 708. In some embodiments, part or all of the computer program may be loaded and / or installed on device 700 via ROM 702 and / or communication unit 709. When the computer program is loaded into RAM 703 and executed by the computing unit 701, one or more steps of the method for verifying the authenticity of content described above may be performed. Alternatively, in other embodiments, the computing unit 701 may be configured to perform the method for verifying the authenticity of content by any other suitable means (e.g., by means of firmware).

[0159] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0160] The program code used to implement the methods of this disclosure may be written in any combination of one or more programming languages. This program code may be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing apparatus, such that when executed by the processor or controller, the program code causes the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program code may be executed entirely on a machine, partially on a machine, as a standalone software package partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0161] In the context of this disclosure, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0162] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device for displaying information to the user (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor); and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the computer. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0163] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as a data server), or computing systems that include middleware components (e.g., an application server), or computing systems that include frontend components (e.g., a user computer with a graphical user interface or web browser through which a user can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., a communication network). Examples of communication networks include local area networks (LANs), wide area networks (WANs), and the Internet.

[0164] Computer systems can include clients and servers. Clients and servers are generally geographically separated and typically interact via communication networks. The client-server relationship is established by computer programs running on the respective computers and having a client-server relationship with each other. Servers can be cloud servers, also known as cloud computing servers or cloud hosts, which are hosting products within the cloud computing service ecosystem to address the management difficulties and weak business scalability inherent in traditional physical hosts and Virtual Private Servers (VPS) services. Servers can also be categorized as distributed system servers or servers incorporating blockchain technology.

[0165] According to the technical solution of this disclosure, by constructing an element set and recalling evidence based on the element set, it is possible to take into account different factual dimensions and the relationships between elements, thereby improving the matching degree and coverage of evidence information with the content to be verified. Furthermore, this disclosure improves the completeness and traceability of the verification basis by performing verification only after the evidence information covers all elements to be verified, reducing the risk of misjudgment and thus enhancing the accuracy and reliability of content authenticity verification.

[0166] It should be understood that the various forms of processes shown above can be used to rearrange, add, or delete steps. For example, the steps described in this disclosure can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution provided in this disclosure can be achieved, and this is not limited herein.

[0167] The specific embodiments described above do not constitute a limitation on the scope of protection of this disclosure. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this disclosure should be included within the scope of protection of this disclosure.

Claims

1. A method for verifying the authenticity of content, comprising: In response to the existence of at least three elements to be verified in the content to be verified, at least two sets of elements to be verified are constructed based on the elements to be verified, wherein each set of elements to be verified includes at least two of the elements to be verified, and the elements to be verified included in each set of elements to be verified are different; For each set of elements to be verified, based on the elements to be verified included in the set of elements to be verified, a first query information corresponding to the set of elements to be verified is constructed; Using the first query information, retrieve evidence information; In response to the fact that the evidence information can cover all the elements to be verified, the evidence information is used to verify each element to be verified, and a verification result is generated for the content to be verified.

2. The method according to claim 1, further comprising: In response to the fact that the evidence information does not cover all of the elements to be verified, the missing elements to be verified that were not hit by the evidence information are identified. Construct a set of missing elements to be verified based on the missing elements to be verified; For each set of missing elements to be verified, based on the missing elements to be verified included in the set of missing elements to be verified, a second query information corresponding to the set of missing elements to be verified is constructed; Using the second query information, supplementary evidence information is retrieved; In response to the fact that the supplementary evidence information and the evidence information can cover all the elements to be verified, the evidence information and the supplementary evidence information are used to verify each of the elements to be verified, and a verification result is generated for the content to be verified.

3. The method according to claim 2, further comprising: In response to the fact that the element to be verified is simultaneously covered by the first evidence information and the second evidence information, and there is a conflict between the first evidence information and the second evidence information, the element to be verified is marked as the omitted element to be verified.

4. The method according to claim 3, wherein, The response that the first element to be verified is simultaneously covered by the first evidence information and the second evidence information, and that there is a conflict between the first evidence information and the second evidence information, marking the first element to be verified as the omitted element to be verified, includes: If the first element to be verified is simultaneously covered by the first evidence information and the second evidence information, and there is a conflict between the first evidence information and the second evidence information, and the credibility level of both the first evidence information and the second evidence information is less than the first level threshold, the first element to be verified is marked as the omitted element to be verified.

5. The method according to claim 3, further comprising: In response to the fact that the missing elements to be verified of the target are simultaneously covered by the first supplementary evidence information and the second supplementary evidence information, and there is a conflict between the first supplementary evidence information and the second supplementary evidence information, the first confidence level of the first supplementary evidence information and the second confidence level of the second supplementary evidence information are read. In response to the first confidence level being greater than the second confidence level, the first supplementary evidence information is retained; or In response to the second confidence level being greater than the first confidence level, the second supplementary evidence information is retained.

6. The method according to claim 2, further comprising: Based on the synonymous rewriting of the omitted elements to be verified, supplementary elements to be verified are generated. as well as The construction of the set of missing elements to be verified based on the missing elements to be verified includes: A set of missing verification elements is constructed based on the missing verification elements and the supplementary verification elements; and For each set of missing elements to be verified, based on the missing elements included in the set, second query information corresponding to the set of missing elements to be verified is constructed, including: For each set of missing elements to be verified, based on the missing elements to be verified and the supplementary elements to be verified included in the set of missing elements to be verified, a second query information corresponding to the set of missing elements to be verified is constructed.

7. The method according to claim 2, wherein, The upper limit of the number of elements in the set of missing elements to be verified is less than the set of elements to be verified.

8. The method according to claim 2, further comprising: In response to the existence of a sub-element among the omitted elements to be verified that can be covered by the evidence information, the sub-element is removed from the omitted elements to be verified.

9. The method according to claim 1, wherein, The evidence information is used to verify each of the elements to be verified, generating a verification result for the content to be verified, including: For each of the elements to be verified: In response to the fact that the element to be verified is covered by only one related piece of evidence information, a first verification sub-result is generated for the content to be verified using the evidence information; or In response to the element to be verified being covered by at least two pieces of related evidence information, the credibility level of each piece of related evidence information is read. In response to the fact that the associated evidence information includes only one target associated evidence information whose credibility level is greater than or equal to the second level threshold, the first verification sub-result for the content to be verified is generated using the target associated evidence information; or In response to the fact that the evidence information includes at least two pieces of target-related evidence information whose credibility level is greater than or equal to the second level threshold; The elements to be verified are verified using the target-related evidence information respectively, and second verification sub-results corresponding to the target-related evidence information for the elements to be verified are generated. By combining the first verification result and the second verification result, a verification result is formed for the content to be verified.

10. The method according to claim 1, wherein, In response to the fact that the evidence information can cover all the elements to be verified, the evidence information is used to verify each of the elements to be verified, and a verification result is generated for the content to be verified, including: In response to the fact that the evidence information can cover all the elements to be verified, the elements to be verified are assigned to the verification channels accordingly based on the element type of each element to be verified. The associated evidence information covering the element to be verified is provided to the verification channel corresponding to the element to be verified. In each of the verification channels, the verification model deployed in the corresponding verification channel is used to verify the element to be verified, and a verification sub-result for the element to be verified in the verification channel is generated. The results of each of the verification sub-results are combined to form a verification result for the content to be verified.

11. The method according to claim 10, wherein, In each of the verification channels, the element to be verified is verified using the corresponding deployed verification model in the verification channel, and a verification sub-result for the element to be verified is generated in the verification channel, including: In each of the verification channels, the verification model deployed in the corresponding verification channel is used to verify the element to be verified, and a verification sub-result for the element to be verified and a certainty index for the verification sub-result are generated in the verification channel. In response to the certainty index output by the target verification channel being less than the index threshold, the re-verification model is used to re-verify the elements to be verified in the target verification channel based on the associated evidence information in the target verification channel, and a re-verification sub-result is generated, wherein the model size of the re-verification model is larger than that of the verification model. In response to the difference between the re-verification result and the verification result output by the target verification channel, the re-verification result replaces the verification result output by the target verification channel.

12. The method according to any one of claims 1-11, wherein, The response is that there are at least three elements to be verified in the content to be verified, and at least two sets of elements to be verified are constructed based on the elements to be verified, including: In response to the presence of at least three elements to be verified in the content to be verified, the actual number of the elements to be verified is read. Determine the target quantity based on the actual quantity; Based on the elements to be verified, construct the target number of sets of elements to be verified.

13. An apparatus for verifying the authenticity of content, comprising: The first element set construction unit is configured to construct at least two element sets based on the elements to be verified in response to the existence of at least three elements to be verified in the content to be verified. The element sets to be verified include at least two of the elements to be verified, and the elements to be verified included in each element set to be verified are different. The first query information construction unit is configured to construct, for each set of elements to be verified, first query information corresponding to the set of elements to be verified based on the elements to be verified included in the set of elements to be verified; The evidence information recall unit is configured to recall evidence information using the first query information; The first content verification unit is configured to, in response to the evidence information being able to cover all the elements to be verified, use the evidence information to verify each of the elements to be verified and generate a verification result for the content to be verified.

14. An electronic device comprising: At least one processor; as well as A memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, which, when executed by the at least one processor, enables the at least one processor to perform the method for verifying the authenticity of content as described in any one of claims 1-12.

15. A non-transitory computer-readable storage medium storing computer instructions, said computer instructions being configured to cause the computer to perform the method for verifying the authenticity of content as described in any one of claims 1-12.

16. A computer program product comprising a computer program that, when executed by a processor, implements the method for verifying the authenticity of content according to any one of claims 1-12.

Images