An information processing method that structurally demonstrates the range of candidates that cannot be established based on the preconditions of the establishment conditions.

Abstract

The present invention provides an information processing method that allows a third party to confirm that there are no other candidates that could be valid besides the one obtained as a result of the determination, without having to re-execute the search process or the feasibility determination. [Solution] The method acquires input information related to multiple elements, generates a set of elements to be processed, generates candidates that can be derived from the set of elements to be processed, determines whether the derived candidates are valid or not based on predetermined validity conditions, if a candidate that does not satisfy the validity conditions is obtained, other candidates that have a predetermined relationship with that candidate are also treated as invalid, identifies the range of invalid candidates, selects a representative candidate from among the candidates that satisfy the validity conditions based on predetermined ordering rules, generates a candidate absence trail expressed in a form that can be verified by a third party, and outputs or retains the representative candidate and the candidate absence trail externally.

Description

Technical Field

[0001] The present invention relates to an information processing method for determining whether a candidate is valid based on a combination of a plurality of elements or conditions, and particularly to an information processing method for enabling a third party to verify, without recalculation, that there are no candidates other than the determination result that can be established.

Background Art

[0002] Conventionally, in an information processing method for determining validity based on a combination of a plurality of elements or conditions, a method of enumerating or sequentially generating candidates in a search space and evaluating the establishment conditions for each has generally been used.

[0003] However, in these conventional techniques, in many cases, in order for a third party to confirm the validity of the determination result, it is necessary to reproduce the search process leading to the determination or re-execute equivalent calculations. That is, when the determination result depends on the calculation process in the execution entity and only the result is presented externally, it is difficult to lightly verify whether the result was obtained after appropriately excluding other candidates.

[0004] Conventionally, such verification has generally been performed by reproducing the input data and calculation procedure and re-executing the same calculation. However, as the number of candidates and the search space increase, there is a problem that it becomes inefficient from the viewpoints of memory resources and communication resources due to the calculation cost required for verification and the dependence on the execution environment. Moreover, many of the conventional techniques aim to directly determine what the correct fact or answer is, and the fact that there are no candidates other than the presented result that can be established has not been sufficiently studied in a framework that externalizes it as a structure indicating that the range of candidates that cannot be established is covered and enables verification without recalculation.

[0005] In particular, for applications such as auditing, evaluation, and certification, it is more important for a third party to be able to confirm that there are no other viable candidates (absence of candidates) than the determination process itself, yet sufficient technical means focusing on this point have not been provided.

[0006] Furthermore, this application does not specify the means for calculating the result itself, but rather focuses on the fact that it is possible to confirm that no result other than the given result can be obtained without relying on recalculation. This invention does not control the search order or the amount of the candidate search, but rather demonstrates that the range of results that cannot be obtained is determined as a logical structure. [Prior art documents] [Non-patent literature]

[0007] As an information processing method that determines whether something is true or false based on a combination of multiple elements or conditions, A known method involves enumerating or sequentially generating candidates within the search space and evaluating the conditions for each to be true. For example, methods such as branch-and-bound methods, search tree-based methods, and constraint satisfaction problems (CSPs) have been proposed.

[0008] Furthermore, techniques for reducing computational complexity by early elimination of unnecessary candidates during the search process are also widely known. These techniques aim to suppress the increase in the number of candidates by optimizing the search order and applying evaluation conditions early.

[0009] Furthermore, as a technique for confirming the validity of the judgment results, Common methods include saving a log of the search process, or reproducing the input information and calculation procedure to re-execute the same calculation. [Overview of the project] [Problems that the invention aims to solve]

[0010] Given the background described above, conventional information processing methods require a third party to re-execute the search process or feasibility determination used in the judgment in order to verify whether the presented judgment result is valid, making it difficult to avoid computational costs, dependence on the execution environment, or internal state. Furthermore, for candidates that could not be established, it was necessary to either explain the reasons for their failure individually or disclose the search process itself. This led to the problem of explanatory and evidence information becoming excessively large as the number of candidates increased.

[0011] The present invention does not aim to determine what constitutes a true fact, but rather to provide an information processing method that externalizes the fact that there are no other possible candidates besides the presented result as information indicating that the range of impossible candidates is structurally covered, and that this can be verified by a third party without recalculation. In this specification, "structurally covered" means that, based on the conditions for establishment and the premise relationships between candidates, it is logically determined that it cannot be established, and at the very least, based on the conditions for establishment and the ordering rules, it is verifiable that there are no candidates that can be logically prioritized over the representative candidate.

[0012] In other words, the object of the present invention is not merely to improve the efficiency of the search process (reduce computational load) when obtaining a result that satisfies the conditions from among multiple candidates, but rather to externalize information that structurally covers the range of candidates that cannot be satisfied, thereby generating or acquiring evidence of candidate absence that shows that candidates other than the result in question cannot be satisfied, without disclosing or reproducing the entire internal calculation process leading to the judgment result, and using this, the validity of the judgment result can be confirmed independently of the entity that performed the calculation. [Means for solving the problem]

[0013] The information processing method according to the present invention is configured to definitively select candidates that satisfy certain conditions based on input information consisting of multiple elements. Specifically, the information processing device acquires input information that may contain omissions or differences in order, performs processing to absorb differences in the notation, order, or structure of the elements, and generates a normalized set of elements to be processed.

[0014] The information processing device may determine whether a candidate that can be derived from the normalized set of elements is valid based on predetermined validity conditions. If a candidate is determined not to satisfy the validity conditions, it may treat any undetermined candidates that have a predetermined relationship with that candidate as logically invalid candidates and refrain from generating or determining the validity of such undetermined candidates. If a candidate is deemed invalid, candidates defined on the premise of that candidate's validity lose their logical basis for validity. In this specification, "predetermined relationship" refers to a relationship that can exist between candidates, and may include, for example, a premise relationship, a derivation relationship, a causal dependency relationship, or a relationship of inheritance of reasons for negation.

[0015] Furthermore, the information processing device stores or presents information regarding candidates that are determined not to satisfy the conditions for success, or candidates treated as undetermined candidates, in association with the conditions for success and the ordering rules used in selecting the representative candidate. This does not represent a sequential search order or execution history, but is configured to be obtainable as evidence of candidate absence, indicating that the range of candidates that cannot be successful is structurally covered. As a result, a third party can refer to the normalized set of elements, the conditions for success, and the ordering rules to confirm that a representative candidate has been selected and that no other candidates can be selected, without having to re-execute candidate generation or success / failure determination.

[0016] Even if multiple candidates satisfy the conditions for a condition to be met, the information processing device uniquely selects a representative candidate based on predetermined ordering rules (which may include, for example, lexicographical order, magnitude of hash values, or a combination thereof), and generates and outputs identification information that is definitively determined based on that representative candidate. The candidate absence trail can, if necessary, be generated as summary information that summarizes and expresses sets of unmet candidates according to common reasons for rejection, without individually listing them, and this summary information can also be structured by hierarchically linking them. Furthermore, depending on the verification purposes of a third party, a part of this hierarchical structure may be made selectively obtainable or discloseable.

[0017] [Definition of Terms] In this specification, 1. Evidence of candidate absence This refers to evidence that, when obtaining a result that satisfies the conditions from among multiple candidates, the range of candidates that cannot be satisfied is structurally covered by externalizing information, thereby demonstrating that candidates other than the result obtained cannot be satisfied without disclosing or reproducing the entire internal calculation process that led to the determination result. 2. Summary Information This refers to information that expresses the range of candidates that cannot be established, without individually listing them, based on the conditions for establishment and the relationships between candidates, and is used to show that the range is structurally covered by other summary information through hierarchical association. 3. Lower-level information This refers to information that is associated with the aforementioned summary information and shows the basis or breakdown of the scope of candidates that cannot be established as represented by the summary information, and which may be selectively presented in response to a verification request from a third party. 4. Summary Values This refers to a definitively determined value or identifier associated with a summary node, used to indicate the range of impossible candidates represented by the summary information. 5. Summary Node This refers to a node that constitutes the summary information and corresponds to a range of candidates that cannot be established or a unit that corresponds to a reason for rejection. 6. Prerequisites When a certain set of establishment conditions is not satisfied, it refers to a relationship where the determination of establishment or non - establishment based on another set of establishment conditions defined on the premise of the set of establishment conditions logically loses its foundation for establishment. Let it be so.

Advantages of the Invention

[0018] According to the present invention, regarding the result of search or determination of establishment or non - establishment performed to obtain candidates that satisfy the establishment conditions, the fact that there are no candidates other than the presented result that can be established is externalized as evidence of absence of candidates indicating that the range of candidates that cannot be established is structurally covered. As a result, a third party can confirm that there are no candidates other than the candidates obtained as the determination result that can be established (exclusivity) without re - executing the search process or the determination of establishment or non - establishment. The third party may be a subject not involved in the generation or evaluation of the candidates.

[0019] Also, according to the present invention, for candidates that cannot be established, without listing the reasons for non - establishment for each individual candidate, based on the relationship between the establishment conditions and the candidates, evidence of absence of candidates can be constructed in a form indicating that a plurality of candidates cannot be established collectively. Therefore, it is possible to suppress the enlargement of explanatory information and evidence information as the number of candidates increases, and it becomes possible to efficiently present the determination result from the viewpoints of memory resources and communication resources.

[0020] Furthermore, in the present invention, when a candidate that does not satisfy the establishment conditions is obtained, other candidates having a predetermined relationship with the candidate are treated as candidates that cannot be established, and the generation or determination of establishment or non - establishment of the other candidates is not performed. As a result, the processing load on the determination subject can be suppressed, and the omission not only improves efficiency but can also function as a component for obtaining evidence of absence of candidates covering the range of candidates that cannot be established. This is a configuration for determining the boundary of the non - establishment region.

[0021] Furthermore, according to the present invention, since the conditions for establishment are configured as multiple stages having prerequisite relationships, when it is found that a certain prerequisite is not met, the candidate range defined based on that prerequisite is integrally determined as a region where the condition cannot be established, and that range is structurally covered as a boundary.

[0022] In other words, according to the present invention, the search load and the verification load can be separated, so that the entity making the determination needs to search for candidates and determine whether they are valid or not, while the entity verifying the determination result only needs to refer to evidence of the absence of candidates. Therefore, there is no need to re-execute the search process or the determination of whether they are valid or not in order to confirm the validity of the determination result, and the validity of the result can be confirmed without relying on the determining entity. In addition, since the verification can reduce its dependence on the computational environment and internal state at the time of determination, it can suppress inconsistencies caused by differences in the execution environment of the verifying entity.

[0023] In addition, by saving or sharing the candidate absence trail according to the present invention together with the determination result, it becomes possible to retrospectively verify whether the result was obtained appropriately at any point after the determination. Therefore, in applications such as auditing, evaluation, and certification, it has the effect of allowing the validity of the result to be confirmed over a long period of time, regardless of the computing environment or internal state at the time of determination.

[0024] Furthermore, the present invention does not aim to determine what constitutes a true fact, but rather to provide a framework that allows for verification that there are no other possible candidates besides the presented result. Therefore, even when used in conjunction with technologies that determine primary facts or states, it becomes possible to verify, from an independent perspective, whether the judgments or choices made by those technologies based on the established facts or premises were appropriate. In other words, the present invention functions complementaryly by allowing a third party to verify the validity of the results derived under the premises obtained by the technology responsible for determining the facts, without altering those premises and without requiring recalculations.

[0025] Furthermore, the aforementioned summary information or evidence of candidate non-existence may include a configuration that allows a third party to verify only that the scope of the candidate being unfeasible is determined, without disclosing the specific details of the unfeasible candidate. Furthermore, the present invention does not limit the method of performing exploration or evaluation itself, and can be implemented independently or in combination with other information processing methods, as a configuration that allows the validity of the results obtained therefrom to be verified.

[0026] [Summary of technical effects] By externalizing the range of candidates that cannot be established as evidence of candidate absence based on establishment conditions, prerequisites, and order rules, a third party can definitively verify the exclusivity of representative candidates without relying on the judging body's search process, internal state, or recalculation.

[0027] [Supplementary explanation: Computational complexity] In general, if the number of candidates that can be derived from the input can be large, then enumerating all candidates will result in an exponential increase in the number of candidates with respect to the number of elements. In contrast, the information processing method according to the present invention can be configured such that, when a candidate that does not satisfy the conditions for establishment is obtained, other candidates having a predetermined relationship with that candidate are also treated as unable to be established, and no generation of such other candidates or determination of whether they can be established is performed. Therefore, there is no need to individually evaluate candidates that are not viable, and compared to evaluating all candidates exhaustively, it is possible to determine a representative candidate within a practical time and resource constraint while suppressing the increase in computational load required for candidate generation and feasibility determination. However, the key point of this invention is not the reduction of computational complexity itself, but rather the externalization of evidence of candidate absence that shows that the range where the formula cannot be applied is structurally covered, making it verifiable by a third party without recalculation. [Brief explanation of the drawing]

[0028] [Figure 1] This figure shows an example of the functional configuration of the information processing device according to the present invention.

[0029] [Figure 2]This figure shows an example of the processing procedure for the information processing method according to the present invention.

[0030] [Figure 3] This diagram illustrates an example of a concept in which a third party verifies the decision using the candidate for representative and evidence of the candidate's absence.

[0031] [Figure 4] This figure shows an example of a hierarchical structure in which summary information used to generate candidate absence trails is linked.

[0032] [Figure 5] This figure shows an example of the relationship between summary nodes, summary values, and subordinate information. [Modes for carrying out the invention]

[0033] (Overall structure) The following describes one embodiment for implementing the information processing method according to the present invention. The purpose of this embodiment is to determine a candidate that satisfies the conditions for success based on the input information, and to demonstrate in a way that can be verified by a third party that no other candidates can satisfy the conditions.

[0034] Hereafter, based on the device configuration shown in Figure 1 and the processing flow shown in Figure 2, the selection of representative candidates, the generation of evidence of candidate absence, and the presentation method for third-party verification in this embodiment will be explained.

[0035] (Device configuration) The information processing device according to this embodiment is configured to determine a representative candidate that satisfies the conditions for success based on input information, and to show in a way that can be verified by a third party that no other representative candidate can succeed.

[0036] 101: Input Information Acquisition Unit The input information acquisition unit acquires input information related to multiple elements. The input information may include a set of elements to be evaluated, relationships between elements, or information to which the conditions for fulfillment apply, and may include omissions, duplicates, or differences in order.

[0037] 102: Input Information Normalization Unit The input information normalization unit processes the input information acquired by the input information acquisition unit to absorb differences in the notation, order, or structure of the elements, and generates a set of elements to be processed. The set of elements subject to this processing is treated as the reference set used in the subsequent candidate derivation and feasibility determination.

[0038] 103: Candidate Derivation Unit The candidate derivation unit generates candidates that can be derived from the set of elements to be processed. Candidates can be constructed as combinations containing a predetermined number or more elements, but it is not necessary to enumerate all candidates in advance; they may be derived sequentially as the feasibility determination progresses. A predetermined number or more refers to combinations that satisfy the minimum number of elements.

[0039] 104: Establishment possibility judgment part The feasibility determination unit determines whether a candidate derived by the candidate derivation unit is feasible or not based on predetermined conditions for its validity. The conditions for a candidate to be valid can be defined based on the number of elements included, their composition, or the relationships between those elements.

[0040] 105: Candidate handling department that cannot be established The Candidate Handling Unit, based on the determination results of the Feasibility Determination Unit, will treat any other candidates that have a predetermined relationship with the candidate that does not meet the feasibility criteria as also being unfeasible. In this case, it is acceptable not to perform a determination on whether the other candidates can be generated or established.

[0041] 106:Candidate absence range identification part The candidate absence range identification unit identifies the range of candidates that cannot be established, based on the candidates that the establishment feasibility determination unit has treated as unestablishable. This range can be structurally represented based on the conditions for its establishment and the relationships between the candidates, without listing individual candidates.

[0042] 107: Existence candidate determination unit The establishment candidate existence determination unit determines, based on the determination result by the establishment feasibility determination unit, whether or not there are candidates that satisfy the establishment conditions. If no candidate meets the requirements for success, the process of selecting a representative candidate may be omitted.

[0043] 108: Representative Candidate Selection Department The representative candidate selection unit selects a representative candidate based on predetermined order rules when there is one or more candidates who meet the qualification conditions. The ordering rule can be defined by the order of the elements constituting the candidate, identification information, or a definitive value calculated based thereon, and is configured so that a representative candidate is uniquely determined even if there are multiple candidates that satisfy the same validity conditions.

[0044] 109: Candidate Absence Evidence Generation Unit The candidate absence evidence generation unit generates a candidate absence evidence that expresses the range of candidates that cannot be established, as identified by the candidate absence range identification unit, in a form that can be verified by a third party. The evidence of candidate absence can be configured in correspondence with the conditions for establishment and the aforementioned ordering rules, and can be made obtainable as information indicating that the range of candidates that cannot be established is adequately covered.

[0045] 110: Result output section The results output unit outputs or retains the representative candidate and the evidence of the candidate's absence to an external source. This allows a third party, independent of the entity that generated the judgment result, to refer to the evidence of candidate absence and confirm, without recalculation, that no candidate other than the representative candidate can be elected.

[0046] (Processing flow) Below, an example of the processing flow of the information processing method according to this embodiment will be described with reference to Figure 2.

[0047] S101: Get input information The information processing device acquires input information related to multiple elements (S101). The input information may include a set of elements to be evaluated, relationships between elements, or information to which the conditions for fulfillment apply.

[0048] S102: Normalization of input information The information processing device performs a normalization process on the acquired input information to absorb differences in the notation, order, or structure of the elements, and generates a set of elements to be processed (S102). The set of elements subject to this processing is treated as the reference set used in the subsequent candidate derivation and feasibility determination.

[0049] S103: Candidate Derivation The information processing device generates candidates that can be derived from the set of elements to be processed (S103). Candidates may be composed of combinations containing a predetermined number or more elements, but it is not necessary to enumerate all candidates in advance; they may be derived sequentially as the subsequent determination of whether or not they are valid progresses.

[0050] S104: Judgment whether or not it holds true The information processing device determines whether the derived candidates are valid or not based on predetermined conditions for their validity (S104).

[0051] S105: Handling of candidates who cannot be elected If, as a result of the determination of whether a candidate can be established, a candidate handling unit that does not meet the conditions for establishment is obtained, the unit that handles unestablishable candidates will also treat other candidates that have a predetermined relationship with the said candidate as unestablishable (S105). In this case, it is acceptable not to perform a determination on whether the other candidates can be generated or established.

[0052] S106: Identifying areas where candidates are absent The information processing device identifies the range of candidates that cannot be established based on the candidates that were treated as impossible (S106). This range can be expressed as a structure based on the conditions for its establishment and the relationships between the candidates, without listing individual candidates.

[0053] S107: Determination of whether or not there are candidates for establishment. The unit that determines whether or not a candidate exists that satisfies the conditions for establishment determines whether or not such a candidate exists (S107). If no candidates satisfy the conditions for success, the process may be terminated. If there is one or more candidates that satisfy the conditions for success, the process proceeds to S108.

[0054] S108: Selection of Representative Candidates The information processing device selects a representative candidate from among the candidates that satisfy the conditions for establishment, based on a predetermined ordering rule (S108). According to this ordering rule, even if there are multiple candidates who satisfy the same conditions for success, the representative candidate will be uniquely determined.

[0055] S109: Generating a candidate absence trail The information processing device associates the range of candidates that cannot be met as identified in S106 with the conditions for meeting the conditions and the order rules, and generates a candidate absence trail expressed in a form that can be verified by a third party (S109).

[0056] S110: Result output The information processing device outputs or retains the representative candidate and the evidence of the candidate's absence to an external source (S110). This allows a third party, independent of the entity that generated the judgment result, to refer to the evidence of candidate absence and confirm, without recalculation, that no candidate other than the representative candidate can be elected.

[0057] (Example of an ordering rule) The aforementioned ordering rule is structured as a definitive rule for uniquely selecting a representative candidate. Specific examples include, for instance, a ranking rule that quantifies and compares candidates, a comparison rule using a priority vector based on multiple evaluation axes, a comparison of normalized code sequences, or an externally agreed-upon comparison rule.

[0058] (Specific examples of third-party verification) The following describes one method of verifying the validity of a judgment result by a third party different from the entity that generated the result, with reference to Figure 3.

[0059] In this embodiment, the third party receives evidence of candidate absence, which structurally indicates the range of candidates who are not eligible to represent the group, along with information about the representative candidates. The evidence of candidate absence may include information representing the conditions for establishment, the relationships between candidates, and the range of candidates that cannot be established based on these. It does not represent a sequential search order or execution history, and does not need to include information that requires the calculation process or reproduction of the determination process in the determination process.

[0060] The third party will verify, based on the conditions for establishment and the relationships between candidates included in the evidence of the candidate's absence, whether the scope in which candidates other than the representative candidate cannot be established is adequately covered. This verification does not represent a sequential search order or execution history, but rather is performed by confirming the consistency and structural validity of the candidate absence evidence.

[0061] For example, a third party can determine whether the representative candidate has been appropriately selected by confirming that the information representing the range of candidates who cannot be elected is structured based on relationships corresponding to candidates who do not meet the election conditions, and that the range includes candidates other than the representative candidate.

[0062] Thus, according to this embodiment, a third party can confirm that there are no other candidates who could be elected besides the representative candidate, without having to rely on the entity that generated the determination result, and without having to refer to the details of the determination process or internal state performed by that entity. Therefore, the present invention provides a configuration that separates the execution entity of the determination process from the verification entity, and enables verification without recalculation between the two.

[0063] Furthermore, third-party verification is not limited to being completed in a single step upon receipt of the candidate absence evidence. It may also be conducted as an interactive process that includes a step in which the entity that generated the judgment result presents subordinate information to the summary information, or additional explanatory information corresponding to the summary information, in response to a request from the third party. A third party can, based on the additional information provided, progressively verify the validity of the candidate's absence trail, limited to the necessary scope.

[0064] Furthermore, the evidence of candidate absence may be structured as a hierarchical structure (e.g., a tree structure or a similar hierarchical structure) in which summary information corresponding to common reasons for rejection or common relationships is hierarchically linked as nodes, in order to indicate the scope of candidates that cannot be established without individually listing the scope of candidates that cannot be established. This allows higher-level summary information to indicate the scope of candidates that cannot be established by gradually integrating the scope of rejection based on lower-level summary information, and a third party can confirm that the scope of candidates that cannot be established is adequately covered by gradually switching the level of reference from higher-level summary information to lower-level summary information as needed, without involving excessive information disclosure or recalculation.

[0065] The “summary information that allows verification that it is not inconsistent with the representative candidate” included in the evidence of the absence of the candidate means that it includes at least information that allows confirmation that, in light of the conditions for establishment and the order rules, a candidate who could be selected in preference to the representative candidate is not included in the range of candidates who cannot be established, or (if such a candidate exists) that such a candidate is excluded for reasons of denial. For example, if the ordering rule is a rule that assigns an overall order to candidates that satisfy the conditions for success, then a third party can mechanically confirm, based on the summary information, that candidates with a higher priority than the representative candidate are not included in the range of candidates that cannot be selected due to reasons for rejection, or that such candidates do not exist. This means that the evidence of candidate absence does not merely show a history of unsuccessful candidates, but also allows for verification that the selection of the representative candidate is logically consistent in light of both the conditions for success and the order rules.

[0066] The evidence of candidate absence or the summary information is not necessarily limited to a specific format, but in one embodiment, it may be composed of aggregated information definitively calculated according to predetermined rules from lower-level summary information corresponding to multiple grounds for rejection. In this case, verification by a third party is performed, for example, by obtaining (a) a normalized set of elements, (b) a representative candidate, (c) denial reasons and scope information contained in the candidate absence trail, and (d) the conditions for establishment and ordering rules as input, and by confirming that (e) if there are candidates whose scope of unestablishable candidates corresponding to the denial reasons exceeds the representative candidate in light of the ordering rules, then those candidates are covered (i.e., excluded) as an unestablishable scope due to the denial reasons, and (f) confirming that the representative candidate satisfies the conditions for establishment. This allows a third party to verify, based on evidence of candidate absence, that there are no other candidates that could be selected besides the representative candidate, without having to re-generate candidates or re-determine their suitability. Scope information can be expressed based on the summary information.

[0067] In one embodiment, if the condition for fulfillment is defined as a monotonous condition relating to the number or composition of elements included in a candidate, the range information may be expressed as an inclusion condition indicating that a candidate can not be fulfilled if a portion of the element set included in a candidate that is determined not to satisfy the condition for fulfillment is excluded. In this case, the range information structurally indicates the range of candidates for which the reason for rejection is inherited, without individually listing candidates that cannot be valid.

[0068] In another embodiment, if the ordering rule is a rule that gives a lexicographical order or other overall order based on the elemental representation or sequence of elements of the candidates, the range information may be expressed as information that identifies an ordering range in which there may be candidates that can take precedence over the representative candidate in light of the ordering rule, and that indicates that candidates belonging to that ordering range cannot be established for reasons of denial. This allows third parties to verify, based on the relevant scope information, that candidates superior to the candidate for the representative position have been excluded based on the reasons for rejection.

[0069] (An embodiment of a hierarchical summary structure) Below, with reference to Figure 4, we will describe one form of hierarchically summarizing and storing information representing the range of candidates that cannot be established.

[0070] In this embodiment, the information processing device generates summary information corresponding to candidates or groups of candidates that are treated as not being valid during the determination process based on the conditions for validity. This summary information does not list individual candidates, but rather is structured as a unit representing the range of candidates that cannot be established, based on the conditions for establishment and the relationships between candidates.

[0071] The information processing device stores multiple summary information items in a hierarchical manner, relating them according to the stages of the conditions for acceptance, the presupposition relationships between candidates (relationships in which the non-acceptance of one candidate logically eliminates the basis for acceptance of other candidates), or the progress of the judgment. As a result, the range of candidates that cannot be established is gradually integrated from lower-level summary information to higher-level summary information, and the range of negations for the entire set of candidates is represented as a hierarchical structure.

[0072] By using this hierarchical summary structure, a third party can understand that no candidates other than the representative candidate can be elected by checking the top-level summary information, and, if necessary, can refer to the lower-level summary information to confirm which level of election criteria the scope of candidates who cannot be elected is based on. In other words, the level of detail of the information used for verification can be selected according to the verifier's purpose and authority.

[0073] Furthermore, since the summary information generated during the intermediate stages of the determination process can be reused in subsequent determination processes, if the range of candidates that cannot be determined has already been covered, there is no need to perform the determination process again for candidates belonging to that range. As a result, the range of negations generated as the judgment process progresses is accumulated, and the range of invalid conditions in the candidate set is determined step by step.

[0074] Thus, according to this embodiment, by maintaining the range of unacceptable candidates as a hierarchical summary structure, the range of unacceptable candidates can be expressed in an efficient and verifiable manner without having to enumerate the reasons for the unacceptability of each individual candidate. As a result, it has the effect of suppressing the amount of information required to verify the judgment results while enabling verification that includes the stages of the conditions for success and the status of achievement.

[0075] The verification body, without referring to the candidate generation process, search history, or intermediate stages of the determination, uses only the summary information contained in the normalized primary fact set, representative candidates, and candidate absence trails as input to verify whether the candidate absence trails cover a range of candidates that cannot be established based on the prescribed ordering rules and establishment conditions.

[0076] Furthermore, the hierarchical summary structure may be configured to allow selective acquisition or disclosure of a portion of the summary information (for example, a portion corresponding to a specific condition for establishment or a reason for denial) depending on the verification objectives of a third party. This allows third parties to conduct verification only to the necessary extent, without having to obtain all the grounds or scope of denial, and to complete the verification while suppressing the disclosure of unnecessary information.

[0077] As one embodiment, the candidate absence trail includes negative summary information corresponding to a candidate that has been determined not to meet the establishment conditions, and a summary structure indicating that the negative summary information is inherited by other candidates based on the premise relationship. The verification body verifies that there are no candidates other than the representative candidate by confirming that there is no inconsistency based on the ordering rule between the summary information corresponding to the representative candidate and the negative summary information. Furthermore, in this specification, "inheritance of grounds for rejection" or "premise relationship" includes a monotonous relationship in which, when the conditions for establishment of a certain candidate are rejected, the grounds for rejection automatically reject the establishment of other candidates that are defined on the premise of that candidate.

[0078] (Specific examples of hierarchical summarization) The following describes an example of the relationship between summary nodes, summary values, and subordinate information, with reference to Figure 5. The summary information included in the candidate absence trail can be configured as a hierarchical structure including, for example, multiple summary nodes (nodes constituting the summary information) that represent the range of candidates that cannot be met based on the conditions or ordering rules.

[0079] Each summary node includes identification information corresponding to the range of rejection reasons or impossible candidates represented by that node, and a summary value that can be definitively calculated based on the lower-level summary nodes. By hierarchically associating these, the entire set of impossible candidates can be represented without individually listing them.

[0080] A third party can verify, based on the summary information, eligibility conditions, and order rules, that the scope in which candidates other than the representative candidate cannot be elected is adequately covered.

[0081] Furthermore, the summary value is not limited to a hash value, but may be configured as any aggregate value that can be uniquely determined based on the lower-level summary nodes.

[0082] (Storage and reuse of verification information) The following describes one embodiment of the information processing method according to the present invention, in which the determination result and information used for verification are stored and reused for subsequent verification.

[0083] In this embodiment, the information processing device stores information about representative candidates that satisfy the conditions for establishment, along with candidate absence trails (hereinafter also referred to as verification information as needed) that represent the range of candidates that cannot be established, in a predetermined storage area. The verification information does not need to include the judgment order, internal state, or individual reasons for failure generated during the judgment process, but may include information representing the range of candidates that cannot be met, based on the conditions for success and the relationships between candidates.

[0084] The saved verification information can be retrieved at any point after the judgment process has been executed, and can be accessed by a third party other than the entity that generated the judgment result. By referring to the verification information, a third party can confirm that there are no other candidates that could be successful besides the representative candidate, regardless of the computational environment, judgment order, or internal state at the time the judgment process was executed.

[0085] Furthermore, the verification information can be reused as audit logs, evaluation records, or evidence. For example, even if the validity of the judgment result becomes an issue at a later date, by referring to the verification information, it is possible to confirm whether the conditions for success were properly applied at the time of judgment and whether the range of candidates that could not be successful was adequately covered.

[0086] Furthermore, in this embodiment, even if the conditions for success or the evaluation criteria are changed after the execution of the judgment process, the verification information saved before the change can function as evidence demonstrating the validity of the judgment result based on the conditions for success at that time. This helps to prevent the validity of past judgment results from becoming unclear due to changes in judgment criteria or retrospective evaluations.

[0087] Furthermore, in this embodiment, when the input information or the set of elements to be processed is partially changed, only the summary information corresponding to the affected area needs to be updated, and a third party may verify this by checking the difference information corresponding to the updated part.

[0088] Thus, according to this embodiment, by saving and reusing the judgment result and verification information, a configuration is realized that allows the validity of the judgment result to be verified independently over time. As a result, the present invention is not limited to one-off decision processing, but is also effective in applications requiring long-term audits, accountability, or proof.

[0089] (Variation: Rule change (diff update)) Furthermore, if the conditions for establishment or the order rules are changed afterward, it is also possible to update only a portion of the summary information affected by the change and partially reconstruct the evidence of candidate absence.

[0090] (Variation: Human-in-the-loop) Furthermore, the determination of whether or not a proposal is valid is not limited to automatic determination by an information processing device, but may also include a process of treating a specific candidate as ineligible based on approval or rejection operations by a user or administrator. Rejection by an external entity can also be a type of reason for non-validation.

[0091] (Example: Counterfactual explanation) Furthermore, the information processing device may be configured to analyze, based on the information contained in the candidate absence trail, the reasons why a particular candidate did not qualify, or the lack of elements or conditions necessary for its qualification, and present this information to the user.

[0092] (Examples of application: Evaluation of exams and assignments that include written responses, etc.) The information processing method according to the present invention allows for the decisive selection of a representative candidate based on the conditions for success, while tolerating fluctuations and diversity in the input information. It can also externalize the validity of the selection as evidence of candidate absence in a way that can be verified by a third party. For example, it can be applied to the evaluation of tests and assignments including written answers, staged certification or qualification determination, content verification of application documents and reports, and verification of the validity of audited information. In this case, by constructing the conditions for success in stages, it is possible to distinguish and determine whether the input satisfies the minimum conditions that must be met, or whether it satisfies stricter conditions including additional conditions. This allows a third party to verify the range of conditions under which a representative candidate succeeded and the range of conditions under which they did not succeed, without requiring recalculations.

[0093] This allows us to go beyond simply determining pass or fail and instead show in an explainable way how far the input has progressed, thereby improving the transparency and reproducibility of the evaluation results.

[0094] (Example of use: Proof of the "justification for failure" in automated evaluation / judgment results) The present invention can be applied, for example, to an automated evaluation of essays, reports, design proposals, or specification proposals, enabling verification of whether the presented evaluation results are valid or not. In such applications, the evaluation result is presented as "pass" or "fail." However, especially in the case of a failure, it is not necessarily required to individually explain why the evaluated item failed. Rather, it is important to demonstrate that no other interpretation or structure would be acceptable under the evaluation criteria in question.

[0095] According to the present invention, the range of candidates that cannot be established based on the evaluation criteria is externalized as a negative structure, and by presenting this negative structure as evidence of candidate absence, a third party can verify the validity of the rejection judgment without having to re-execute the evaluation process. This allows the evaluator to demonstrate that the judgment was not arbitrary without disclosing the internal evaluation process or decision-making procedures used by the evaluator.

[0096] Furthermore, such applications are not intended to uniquely determine the facts concerning the subject of evaluation, nor do they require the ability to reconstruct the correct or optimal solution for the subject of evaluation. Therefore, the basis for the evaluation results is verification that there are no other possible candidates besides the presented result, and in this respect, the present invention solves a technical problem that differs from that of technologies aimed at determining facts. [Industrial applicability]

[0097] The present invention can be applied to the following uses in order to verify the validity of the judgment result without recalculation. The following are examples, and the present invention is not limited to these uses.

[0098] This invention can be applied to applications such as qualification examinations, hiring selections, and grant application reviews, where it is possible to verify that the decision to reject or not approve a candidate is not arbitrary, without having to list individual reasons.

[0099] The present invention can be applied to applications that enable verification in automated determination based on codified rules such as contract clauses, regulations, operational rules, or evaluation criteria, that there are no other possible interpretations or candidates under said rules.

[0100] The present invention can be applied to applications that allow for verification of the validity of judgment results obtained by external computation (outsourced computation) or off-chain computation without relying on the entity that performed the computation, and without requiring recalculation.

[0101] This invention can be applied to applications that allow verification that no other possible candidates exist for evaluation and judgment results obtained by a learning model or generation algorithm, etc., without disclosing the internal processes of the model.

[0102] This invention can be applied to situations requiring transparency, such as public procurement, audits, and regulatory compliance, by presenting minimal evidence of the absence of candidates that forms the basis of the determination, thereby enabling third parties to verify the validity of the determination.

[0103] The present invention can be applied to processes involving approval / rejection decisions, such as approval workflows, authentication, and access control, by structurally indicating a range of candidates that cannot be met under predetermined conditions, thereby enabling verification of the validity of the decision result.

[0104] The present invention can be applied to applications that enable proof of existence and detection of tampering at a given time by registering the generated identification information in a timestamp server, a known time-certification means, a tamper-detectable recording medium, etc. This can function as a concrete means of ensuring the effectiveness of long-term audits.

Claims

1. An information processing method performed by an information processing device, A process of obtaining input information consisting of multiple elements, which may include the absence or incorrect order of some elements, The process involves performing a process on the aforementioned input information to absorb differences in the notation, order, or structure of the elements, and generating a normalized set of elements (a set of elements to be processed). A step of determining whether a candidate containing a predetermined number or more elements, which can be derived from the set of elements to be processed, is valid based on predetermined conditions for validity, If a candidate is determined not to satisfy the aforementioned conditions for establishment, the process of treating other candidates that have a predetermined relationship with that candidate as logically unsuccessful based on the conditions for establishment, and omitting the generation or determination of whether or not they are successful, A step of selecting a representative candidate from among the candidates that satisfy the aforementioned conditions, based on a predetermined ordering rule, in such a way that even if there are multiple candidates that satisfy the same conditions, the representative candidate is uniquely determined. A step of generating or obtaining from an external source evidence of candidate absence, indicating that the range of candidates determined not to satisfy the aforementioned conditions for establishment and the range of candidates for which the establishment feasibility determination was omitted is structurally covered as a range of candidates that cannot be established. A step of generating definitively determined identification information based on the aforementioned representative candidate, A step of maintaining or presenting the aforementioned representative candidate and evidence of the candidate's absence in a form that can be acquired or disclosed by a third party independent of the entity performing the information processing method, Information processing methods including

2. In the information processing method described in claim 1, The aforementioned conditions for establishment are, This includes a condition that does not hold true if the number or composition of elements included in the candidate does not meet the prescribed requirements, Regarding candidates who are determined not to meet the aforementioned conditions for establishment, Other candidates that have a predetermined relationship with the candidate in question, and from which some of the elements included in the candidate in question have been removed, The reason for rejection, based on the failure to meet the conditions for establishment, acts as a dominant negative factor for the establishment of the other candidates as well. As it can be logically determined that the other candidate is not eligible without needing to determine the conditions for eligibility, An information processing method characterized by treating the other candidate as logically invalid and omitting the generation or determination of validity.

3. In the information processing method described in claim 2, The aforementioned representative candidate is, Among the candidates that satisfy the above conditions, An information processing method characterized in that a candidate is selected as having a configuration that would no longer satisfy the aforementioned conditions if any one element were removed from the candidate.

4. In the information processing method described in claim 1, In selecting the aforementioned representative candidates, Information regarding candidates that are determined not to meet the aforementioned conditions, or candidates that are generated or whose determination is omitted based on such candidates, To be held or presented in a form that can be obtained in correspondence with the aforementioned conditions for establishment and the aforementioned order rules, The retained or presented information is configured to be referable as the candidate absence evidence, A third party independent of the entity performing the aforementioned information processing method, based on the normalized set of elements, the conditions for establishment, and the order rules, An information processing method characterized by being able to verify that the aforementioned representative candidate has been selected and that no other candidate can be selected.

5. In the information processing method according to claim 1 or 4, Verification by a third party independent of the entity performing the aforementioned information processing method is, Without regenerating the candidates from the input information, This is carried out based on the normalized set of elements, the conditions for establishment, the ordering rules, and the evidence of candidate absence that can be obtained in association with the representative candidate. The evidence of the candidate's absence is, Candidates that are determined not to satisfy the aforementioned conditions for establishment, or candidates that are generated or whose determination is omitted based on such candidates, This includes information indicating the reasons why a candidate cannot be selected as a representative candidate in light of the aforementioned conditions for selection or the aforementioned order rules, This allows the third party to determine the appropriateness of the selection of the representative candidate. An information processing method characterized by enabling verification without requiring re-execution of candidate generation or feasibility determination.

6. In the information processing method described in claim 5, The evidence of the candidate's absence is, A set of candidates that are determined not to satisfy the aforementioned conditions, or candidates that are generated or whose determination is omitted based on such candidates, Without listing each candidate individually, It is generated as summary information that is grouped and expressed for each common reason for rejection based on the aforementioned conditions for establishment or the aforementioned order rules. The aforementioned summary information is, Summary information corresponding to multiple reasons for rejection is stored in a hierarchical manner, An information processing method characterized by enabling the third party to verify, based on the hierarchical structure, the range in which candidates other than the representative candidate cannot be selected.

7. In the information processing method described in claim 6, The aforementioned conditions for establishment are, It is constructed as a set of multiple sets of conditions that have presupposition relations such that the next set of conditions is defined on the premise that a certain set of conditions is satisfied. The determination of whether the aforementioned candidate is eligible or not is as follows: This is performed for each set of conditions that are defined stepwise based on the premise relationships among the aforementioned sets of conditions that are valid, The candidate absence trail or the summary information is, This includes information that can identify the set of conditions under which the aforementioned representative candidate is found to be found, and the set of conditions under which the candidate is not found to be found. As a result, the aforementioned third party, Based on the range of conditions for establishment that the aforementioned representative candidate has met and the range of conditions for establishment that have not met, An information processing method characterized by enabling verification of the staged establishment status of the representative candidate.

8. In the information processing method described in claim 6, The aforementioned hierarchical summary structure may be used in accordance with the verification purposes or requirements of the third party. An information processing method characterized by being configured to selectively acquire or disclose a portion of summary information representing the range of candidates that cannot be established.

9. In the information processing method described in claim 7, The aforementioned third-party verification is The process is not limited to being completed in one go upon receipt of the aforementioned candidate absence evidence, An information processing method characterized in that it may be carried out as an interactive process that includes a step of presenting additional information related to the candidate absence trail in response to a request from the third party.

10. In the information processing method described in claim 1, An information processing method characterized in that the predetermined relationship is a monotonous relationship in which the establishment of other candidates defined on the premise of a given candidate is negated based on the deletion or absence of an element included in that candidate.

11. In the information processing method described in claim 1, The evidence of the candidate's absence is, (i) Information that can identify the reason for rejection corresponding to a candidate that is determined not to satisfy the conditions for establishment, or a candidate that is generated or whose determination has been omitted based on such candidate, (ii) Range information representing the scope of candidates that cannot be established based on the grounds for denial, (iii) Summary information that allows verification that the representative candidate can be selected in light of the grounds for rejection. An information processing method characterized by including

12. In the information processing method described in claim 1, The aforementioned ordering rule is, This is a rule that assigns a total order to multiple candidates that satisfy the aforementioned conditions. An information processing method characterized in that a representative candidate is uniquely selected based on the said ordering rule.

13. In the information processing method described in claim 11, The information processing method is characterized in that the range information is expressed based on summary information that is expressed collectively for each common reason for rejection based on the conditions for establishment or order rules described in claim 6.

Images