Information processing system, information processing device, information processing method, and computer program

Abstract

We provide an information processing system that can deliver highly useful information to users. [Solution] An information processing system S according to one embodiment includes: a document storage unit 201 that stores existing documents in advance; a processing unit 311 that derives a predetermined expression (a modified expression in this embodiment) or a predetermined document (a contradictory document in this embodiment) from a document to be processed that includes one or more documents; a secondary information acquisition unit 312 that acquires a similar expression similar to the predetermined expression derived by the processing unit 311, or a similar document similar to the predetermined document derived by the processing unit 311, from part or all of the existing document; and a provision information generation unit 313 that generates provision information based on the predetermined expression derived by the processing unit 311 and the similar expression acquired by the secondary information acquisition unit 312, or based on the predetermined document derived by the processing unit 311 and the similar document acquired by the secondary information acquisition unit 312.

Description

Technical Field

[0001] The present invention relates to an information processing system, an information processing apparatus, an information processing method, and a computer program, and more particularly to a technique for providing a user with highly useful information.

Background Art

[0002] Generative Artificial Intelligence is expected to be increasingly utilized in various fields in the future. Recently, many generative AIs that are being rapidly utilized are natural language processing models. A natural language processing model can understand, for example, a query (which can also be called a question or prompt) consisting of sentences created by humans and generate an answer to the query.

[0003] Various techniques related to natural language processing have been proposed in the past. For example, Patent Document 1 discloses a technique for improving the computational efficiency of natural language processing.

Prior Art Documents

Patent Documents

[0004]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0005] Currently, many generative AIs available to general users are managed in data centers owned by companies. When a user uses such a generative AI, the user sends a query from, for example, a terminal device that the user operates to the data center and obtains an answer to the query from the generative AI in the data center.

[0006] Generative AI managed by companies, as described above, learns a vast amount of general information and can output general answers to queries. However, it is usually difficult for them to provide non-general information specific to a user. For example, when a user queries a generative AI about something called A, the generative AI can easily provide general information about A, but it usually cannot provide information such as the specific relationship between A and the user.

[0007] To address the problems described above, a technology called RAG (Retrieval Augmented Generation) is useful. RAG can supplement the answers provided to users. In systems that utilize RAG, unique information held by the user is generally stored in a database, and when the user creates a query, the information in the database is added to the query as reference information before it is sent to the generating AI.

[0008] The inventors of this case believed that the generating AI could easily provide users with highly useful information by secondarily or supplementarily using the database to extract information based on the information from the answers to queries derived without using the database that holds user-specific information used in RAG.

[0009] This invention was conceived against the above background, and aims to provide an information processing system, information processing device, information processing method, and computer program that can provide users with highly useful information. [Means for solving the problem]

[0010] The present invention relates to the following aspects.

[0011] A text memory unit that stores existing text in advance, A processing unit that derives a predetermined expression or predetermined sentence from a document containing one or more sentences, A secondary information acquisition unit that acquires a similar expression to the predetermined expression derived by the processing unit, or a similar sentence to the predetermined sentence derived by the processing unit, from part or all of the existing sentence, An information processing system comprising: an information generation unit that generates information based on a predetermined expression derived by the processing unit and a similar expression acquired by the secondary information acquisition unit, or based on a predetermined sentence derived by the processing unit and a similar sentence acquired by the secondary information acquisition unit.

[0012] A processing unit that derives a predetermined sentence from a document containing multiple sentences, A secondary information acquisition unit obtains similar sentences from the sentence to be processed that are similar to the predetermined sentence derived by the processing unit, An information processing system comprising: a provisioning information generation unit that generates provisioning information based on the predetermined text derived by the processing unit and the similar text acquired by the secondary information acquisition unit.

[0013] A text memory unit that stores existing text in advance, A processing unit that derives a predetermined expression or predetermined sentence from a document containing one or more sentences, A secondary information acquisition unit that acquires a similar expression to the predetermined expression derived by the processing unit, or a similar sentence to the predetermined sentence derived by the processing unit, from part or all of the existing sentence, An information processing apparatus comprising: a provisioning information generation unit that generates provisioning information based on a predetermined expression derived by the processing unit and a similar expression acquired by the secondary information acquisition unit, or based on a predetermined sentence derived by the processing unit and a similar sentence acquired by the secondary information acquisition unit.

[0014] A processing unit that derives a predetermined sentence from a document containing multiple sentences, A secondary information acquisition unit obtains similar sentences from the sentence to be processed that are similar to the predetermined sentence derived by the processing unit, An information processing apparatus comprising: a provisioning information generation unit that generates provisioning information based on the predetermined text derived by the processing unit and the similar text acquired by the secondary information acquisition unit.

[0015] A processing step of deriving a predetermined expression or predetermined sentence from a document to be processed that contains one or more sentences, A secondary information acquisition step which acquires a similar expression similar to the predetermined expression derived in the processing step, or a similar sentence similar to the predetermined sentence derived in the processing step, from a part or all of an existing sentence different from the sentence to be processed. An information processing method comprising: a provisioning information generation step that generates provisioning information based on the predetermined expression derived in the processing step and the similar expression obtained in the secondary information acquisition step, or based on the predetermined sentence derived in the processing step and the similar sentence obtained in the secondary information acquisition step.

[0016] A processing step of deriving a predetermined sentence from a document containing multiple sentences, A secondary information acquisition step in which similar sentences to the predetermined sentence derived in the processing step are obtained from the sentence to be processed, An information processing method comprising: a provision information generation step, which generates provision information based on the predetermined text derived in the processing step and the similar text obtained in the secondary information acquisition step.

[0017] A processing step of deriving a predetermined expression or predetermined sentence from a document to be processed that contains one or more sentences, A secondary information acquisition step which acquires a similar expression similar to the predetermined expression derived in the processing step, or a similar sentence similar to the predetermined sentence derived in the processing step, from a part or all of an existing sentence different from the sentence to be processed. A computer program that causes a computer to execute a provided information generation step of generating provided information based on the predetermined expression derived in the processing step and the similar expression obtained in the secondary information acquisition step, or based on the predetermined sentence derived in the processing step and the similar sentence obtained in the secondary information acquisition step.

[0018] A processing step of deriving a predetermined sentence from a processing target sentence including a plurality of sentences A secondary information acquisition step of acquiring a similar sentence similar to the predetermined sentence derived in the processing step from the processing target sentence A computer program that causes a computer to execute a provided information generation step of generating provided information based on the predetermined sentence derived in the processing step and the similar sentence obtained in the secondary information acquisition step

Effect of the Invention

[0019] According to the present invention, highly useful information can be provided to the user.

Brief Description of the Drawings

[0020] [Figure 1] It is a diagram schematically showing the configuration of an information processing system according to the first embodiment. [Figure 2] It is a block diagram showing the functional configuration of each part constituting the information processing system shown in FIG. 1. [Figure 3] It is a flowchart for explaining the flow of the ambiguity check process performed in the information processing system shown in FIG. 1. [Figure 4] It is a flowchart for explaining the flow of the logical check process performed in the information processing system shown in FIG. 1. [Figure 5] It is a diagram specifically explaining the ambiguity check process performed in the information processing system shown in FIG. 1. [Figure 6] It is a diagram showing the processing result of the ambiguity check process described in FIG. 5. [Figure 7]This figure specifically illustrates the logic check process performed in the information processing system shown in Figure 1. [Figure 8] This figure shows the processing results of the logic check process explained in Figure 7. [Figure 9] This is a flowchart illustrating the processing of the information processing system according to the second embodiment. [Figure 10] This is a flowchart illustrating the processing of the information processing system according to the third embodiment. [Figure 11] This is a block diagram showing the functional configuration of the information processing system according to the fourth embodiment. [Figure 12] This is a block diagram showing the functional configuration of the information processing system according to the fifth embodiment. [Figure 13] This figure shows the hardware configuration of terminal devices and information processing devices that can constitute a server, which make up the information processing system according to each embodiment. [Modes for carrying out the invention]

[0021] The following describes each embodiment.

[0022] <<First Embodiment>> <Outline configuration of the information processing system> Figure 1 schematically shows the configuration of the information processing system S according to the first embodiment. The information processing system S includes a terminal device 10, a database device 20, and a server 30.

[0023] The terminal device 10, the database device 20, and the server 30 are connected to each other via a network 40, enabling them to send and receive information. The network 40 is a communication network including, for example, the Internet, a WAN (Wide Area Network), or a LAN (Local Area Network). The terminal device 10, the database device 20, and the server 30 can connect to the network 40 wirelessly or via wired connections and communicate with external devices. The functional configuration of the information processing system S is described in detail below.

[0024] <Functional Configuration of Information Processing System> Figure 2 is a block diagram showing the functional configuration of each component (terminal device 10, database device 20, and server 30) that constitute the information processing system S. The information processing system S performs predetermined processing on a document to be processed, which includes one or more documents generated or held by the terminal device 10. Specifically, the predetermined processing performed by the information processing system S includes a process to check for ambiguity in the document (ambiguity check process) and a process to detect whether there are two or more contradictory documents among multiple documents (logical consistency check process).

[0025] Furthermore, the texts handled in the information processing system S refer to information for communication expressed in natural language that can be handled (displayed, stored, etc.) on a computer.

[0026] In the information processing system S according to this embodiment, when a user requests (instructs) the execution of the above-mentioned ambiguity check process for a document using the terminal device 10, the terminal device 10 sends the request to the server 30. In this embodiment, the request from the terminal device 10 is first sent to the database device 20, and some or all of the information stored in the database device 20 is added to the execution request. Subsequently, the ambiguity check execution request with the added information is sent to the server 30. The server 30 then derives a response (execution result) to the request for execution of the ambiguity check process and obtains information associated with the response from the information provided by the database device 20. Subsequently, the server 30 sends information (provided information) based on the information of the response initially derived and the information associated with the response obtained afterward to the terminal device 10. The above-mentioned logical consistency check process is also performed using the same procedure as the ambiguity check described above.

[0027] As described above, the information processing system S aims to provide users with highly useful information by deriving responses (execution results) to requests for execution of ambiguity checks or logical checks performed by the server 30 as basic processing, as well as by obtaining further specific information from the database device 20. The following describes in detail each part of the information processing system S, in particular the functional configuration of each part.

[0028] (Terminal device) The terminal device 10 is a device operated by a user, such as a personal computer, smartphone, or tablet. The terminal device 10 includes an ambiguous expression correction and presentation unit 101 and a logic check unit 102.

[0029] The ambiguous expression correction suggestion unit 101 detects at least one of the ambiguous terms and phrases in the target document, which includes one or more sentences that have been subjected to ambiguity checking, and performs processing to suggest corrected expressions related to the ambiguity. When performing ambiguity checking, the ambiguous expression correction suggestion unit 101 generates an execution request for ambiguity checking processing (hereinafter sometimes referred to as a query) for the target document in response to user operations. Here, in this specification, "term" means a word used in a sentence, and in particular a word, including concepts such as common nouns, proper nouns, technical terms, industry terms, etc. On the other hand, "phrase" means a group that is a component of a sentence, consisting of two or more words (including particles, auxiliary verbs, etc.).

[0030] The ambiguity correction and presentation unit 101 sends an execution request for ambiguity checking to the server 30 via the database device 20. When the execution request for ambiguity checking goes through the database device 20, some or all of the information stored in the database device 20 is added to the execution request for ambiguity checking. Subsequently, the execution request for ambiguity checking with the added information is sent from the database device 20 to the server 30.

[0031] More specifically, the database device 20 is configured to store information, receive requests to perform ambiguity checking, and send the request with some or all of the information stored in the database device 20 added to it. The information stored in the database device 20 is, for example, text (existing text) that the user has arbitrarily stored in the database device 20.

[0032] The logic check unit 102 compares multiple sentences contained in the document targeted for logic check processing, checks (verifies) whether there are two or more sentences that are contradictory (logically opposed) in relation to each other, and performs processing to present the verification results and related information. In the logic check processing, in order to detect at least two or more contradictory (or potentially contradictory) sentences, the document targeted for logic check processing will contain at least two or more sentences. When performing logic check processing, the logic check unit 102 generates an execution request for logic check processing to verify the existence of contradictory sentences in response to instructions from the user. In the following, the explanation of "contradictory" sentences or relationships shall be synonymous with "potentially contradictory" sentences or relationships unless otherwise specified.

[0033] The logic check unit 102 sends a request to execute the logic check process to the server 30 via the database device 20. When the request to execute the logic check process passes through the database device 20, some or all of the information stored in the database device 20 is added to the request. Subsequently, the request to execute the logic check process with the added information is sent from the database device 20 to the server 30. In other words, the database device 20 is configured to add some or all of the information stored in the database device 20 to the request when it receives a request to execute the logic check process and then send it.

[0034] In this embodiment, as an example, the user interface is configured to allow the user to specify part or all of the text in pre-created document data and then instruct the system to request ambiguity checking. Therefore, in this embodiment, the text to be processed for ambiguity checking is one or more texts specified by the user from the document data before requesting the execution of ambiguity checking. Note that document data means electronic data containing at least one text.

[0035] Furthermore, in this embodiment, as an example, the user interface is configured to allow the user to input (select) multiple pre-created document data and then instruct the user to request a logical consistency check. In this embodiment, the target of the logical consistency check is the logic between a sentence contained in one of the multiple document data and a sentence contained in another document data. Therefore, in this embodiment, the documents to be processed for the logical consistency check are all of the sentences contained in the multiple document data. Note that the documents to be processed for the logical consistency check do not have to be defined as in the above example; for example, they may be two or more arbitrarily selected sentences.

[0036] As will be described in detail later, in this embodiment, ambiguity checking is performed by the learning model 301 of the server 30. The response to the request to execute the above ambiguity checking process derived by the server 30 may include information on terms and / or phrases that have ambiguity in the text to be processed, and information on at least one of one or more correcting terms, one or more correcting phrases, and one or more modifiers (hereinafter sometimes referred to as "correcting expressions") that correct the ambiguity of the terms and / or phrases that have ambiguity. On the other hand, the response to the request to execute the above ambiguity checking process may also be information indicating that there are no terms and / or phrases that have ambiguity in the text to be processed. Note that modifiers are a concept that includes nouns, noun-particle combinations (phrases), adjectives, and adverbs.

[0037] Furthermore, if the response to the request to execute the ambiguity check process includes information about ambiguous terms and / or phrases, and related information such as corrected expressions, in this embodiment, the server 30 obtains the information accompanying the response to the request to execute the ambiguity check process from the information stored in the database device 20. In this case, the server 30 transmits the provided information generated based on the response to the request to execute the ambiguity check process and the accompanying information to the terminal device 10 (ambiguous expression correction presentation unit 101).

[0038] When the ambiguous expression correction presentation unit 101 receives the above-mentioned information from the server 30, it presents information about ambiguous terms and / or phrases in the document to be processed, and related information such as corrected expressions, on, for example, the display of the terminal device 10. The information such as corrected expressions at this time is generated based on information about at least one of the corrected terms, corrected phrases, and corrected modifiers derived by the server 30 as the initial answer, and accompanying information obtained from the information stored in the database device 20.

[0039] The presentation method by the ambiguous expression correction presentation unit 101 is not particularly limited, but for example, it may be a method in which ambiguous terms and / or phrases are displayed in a space different from the display space of the text in the document data on the display, and text notifying that they are ambiguous, corrected expressions, etc. are displayed. In addition, along with the display of the above text, ambiguous terms and / or phrases in the text of the document data on the display may be highlighted.

[0040] The response from server 30 to the request to execute the logic check process means information about multiple sentences that may be contradictory to each other, and information indicating the existence of multiple sentences that may be contradictory to each other. A contradictory relationship means a state in which mutually exclusive facts are recognized between two sentences. Simple examples of contradictions include the relationship between "A is B" and "A is not B," or the relationship between "raise A" and "lower A." As will be described in detail later, in this embodiment, the logic check process is performed by the learning model 301 of server 30. The learning model 301 can detect contradictions, for example, by generating and classifying sentence vectors. On the other hand, the response to the request to execute the logic check process may also be information indicating that there are no sentences that may be contradictory in the sentence being processed.

[0041] Furthermore, if the response to the request to execute the logic check process includes information about the existence of two or more contradictory sentences, in this embodiment, the server 30 obtains information associated with the two or more contradictory sentences from the information stored in the database device 20. In this case, the server 30 transmits the provided information generated based on the response to the request to execute the logic check process and the associated information to the terminal device 10 (ambiguous expression correction presentation unit 101).

[0042] When the logic check unit 102 receives the above-mentioned information from the server 30, it displays information about the existence of two or more contradictory sentences and a warning about them, for example, on the display of the terminal device 10. Furthermore, in this embodiment, information obtained from the information stored in the database device 20, which is associated with the response to the request to execute the logic check process, is displayed. The manner in which the logic check unit 102 presents the information is not particularly limited, but for example, it may display text indicating the existence of multiple potentially contradictory sentences in a space different from the display space for the sentences in each of the multiple document data on the display, and also display that the above-mentioned associated information exists.

[0043] (Data server device) The database device 20 has a document storage unit 201, which pre-stores existing documents, for example, arbitrarily saved by the user. The database device 20 may also consist of an information processing device that has storage constituting the document storage unit 201 and processes the information stored in the storage. Alternatively, the database device 20 may be configured as a cloud database virtually configured on the cloud. Furthermore, the information stored in the document storage unit 201 may be stored via a route different from that of the user.

[0044] The existing documents stored in the document storage unit 201 are, as described above, information arbitrarily stored by the user, but it is desirable that they be documents highly related to the documents to be processed, which are subject to ambiguity checking and / or logical consistency checking. The existing documents may be documents in various formats, such as files created with Microsoft Word, Excel, PowerPoint, etc., or documents based on emails, etc. Alternatively, the existing documents may be documents arbitrarily stored by the operator of the information processing system S for each of several categories (IT industry, retail industry, manufacturer, etc.). In this case, depending on the user's industry, one of the documents may be selected from among the multiple existing documents and added to the request to execute ambiguity checking or logical consistency checking.

[0045] As mentioned above, for example, when the ambiguity correction suggestion unit 101 sends a request to the server 30 to execute an ambiguity check process, the request is sent to the server 30 via the database device 20. When the request to execute the ambiguity check process passes through the database device 20, information stored in the database device 20 is added to the request. Here, the information added to the request to execute the ambiguity check process is part or all of the existing document stored in the document storage unit 201.

[0046] In other words, as described above, the database device 20 is configured to store existing documents as information, receive requests to execute ambiguity check processing, and add part or all of the existing documents to the request before sending it. Furthermore, when the database device 20 receives a request to execute logical consistency check processing, it is configured to add part or all of the existing documents stored in the database device 20 to the request before sending it.

[0047] In this embodiment, the database device 20 adds all of the stored existing documents to the received ambiguity check execution request or logic check execution request and sends it to the server 30. However, in this case, if the amount of information in the existing documents is too large, the response speed of information transmission may decrease. Therefore, the database device 20 may add only a portion of the stored existing documents to the received ambiguity check execution request or logic check execution request. In this embodiment, a portion of existing documents that is highly related to the document to be processed, which is the target of the ambiguity check execution request or logic check execution request, is obtained. In this case, a portion of existing documents may be selected based on tag information indicating the classification of the document to be processed, or a portion of existing documents with a high degree of similarity to the document to be processed may be selected. Known methods such as cosine similarity or pattern matching may be used to determine the degree of similarity. Specifically, for example, if the similarity of the documents being compared is determined to be above a predetermined value, it may be determined that the documents being compared are similar.

[0048] In this embodiment, the learning model 301 of the server 30 performs the desired processing using natural language processing. With this in mind, the existing text described above is stored in the text storage unit 201 in a state converted into a text vector. This allows the learning model 301 to smoothly obtain the desired information from the existing text in the state converted into a text vector.

[0049] (server) Server 30 is a device that receives requests from terminal devices 10 to execute ambiguity check processing or logic check processing, and provides responses to those requests. Server 30 includes a learning model 301. The responses to the above requests are derived from the learning model 301.

[0050] The learning model 301 is a natural language processing model that understands sentences contained within the target text in the ambiguity check execution request or logic check execution request provided from the terminal device 10 via the database device 20, and functions to verify ambiguity and logical inconsistencies. Preferably, the learning model 301 is a BERT (Bidirectional Encoder Representations from Transformers) natural learning model.

[0051] The learning model 301 may be a general-purpose generative AI managed in a data center owned by the company. Alternatively, the learning model 301 may be a learning model created by the user and stored in the server 30, or a learning model created by an operator of the information processing system S, which is different from the user and the company that owns the data center.

[0052] The learning model 301 is pre-trained using training data, for example, training texts as input data, with at least one of the ambiguous terms and phrases in the training texts serving as the ground truth data. As a result, when the learning model 301 receives a text as input data, it can derive at least one of the ambiguous terms and phrases as the answer. The ambiguous terms and phrases that the learning model 301 derives are basically predetermined through pre-training.

[0053] Furthermore, the learning model 301 is pre-trained with training data in which, for example, training text is used as input data and corrected text (a modified version of the training text) is used as correct answer data. As a result, when the learning model 301 receives text as input data, it can derive a modified text as the answer. Specifically, in this embodiment, the learning model 301 can derive the differences and changes between the input text and the modified text as information such as corrected terms and phrases for ambiguous terms and / or phrases, and the addition of modifiers.

[0054] Furthermore, the learning model 301 is pre-trained with training data that, for example, takes a first training sentence and a second training sentence different from the first training sentence as input data, and uses tags that indicate the logical agreement (consistency) or contradiction (deviation) between the first and second training sentences as ground truth data. As a result, the learning model 301 can derive contradictions between sentences when it receives multiple sentences as input data.

[0055] Furthermore, when the learning model 301 derives information such as corrected terms, corrected phrases, or additions of modifiers for ambiguous terms and / or phrases detected in a text, it can obtain similar expressions similar to the derived corrected terms, corrected phrases, and modifiers from existing texts provided by the database device 20.

[0056] Furthermore, the learning model 301 derives information such as corrected terms, corrected phrases, and the addition of modifiers for ambiguous terms and / or phrases detected in the text as described above, and when it obtains similar expressions similar to the derived corrected terms, corrected phrases, and modifiers, it generates and transmits information to be provided to the terminal device 10 based on these.

[0057] Furthermore, the learning model 301 accepts multiple sentences as input data, and when it derives contradictions between sentences, it can obtain similar sentences from existing sentences provided by the database device 20 that are similar to each of the contradictory sentences it derives.

[0058] Then, as described above, the learning model 301 derives contradictions between sentences in the text, and when it obtains similar sentences to each of the contradictory sentences it has derived, it generates and transmits information to be provided to the terminal device 10 based on these.

[0059] As shown in Figure 2, the learning model 301 includes a processing unit 311, a secondary information acquisition unit 312, and a provided information generation unit 313, which are functionally separated.

[0060] The processing unit 311 derives a predetermined expression or a predetermined sentence from a processing target document that includes one or more sentences subject to ambiguity checking or logical consistency checking. In this embodiment, the predetermined expression means at least one of one or more correcting terms, one or more correcting phrases, and one or more modifiers (correcting expression) that correct the ambiguity relating to ambiguous terms and / or phrases in the processing target document subject to ambiguity checking. The predetermined sentence means two or more sentences that have a contradictory relationship, derived from the processing target document subject to logical consistency checking.

[0061] The processing unit 311 is implemented by some functions of the learning model 301, and more specifically by a function to derive at least one of the ambiguous terms and phrases in a text, a function to derive a modified expression that modifies at least one of the terms and phrases deemed ambiguous, and a function to derive two or more sentences that have a contradictory relationship from among multiple sentences.

[0062] The secondary information acquisition unit 312 acquires similar expressions from existing documents provided by the database device 20 if the processing unit 311 performs ambiguity checking and derives at least one of one or more correcting terms, one or more correcting phrases, and one or more modifiers that correct ambiguity in the document to be processed. Furthermore, if the processing unit 311 performs logical consistency checking and derives two or more contradictory sentences from the document to be processed, the secondary information acquisition unit 312 acquires similar expressions from some or all of the existing documents provided by the database device 20 that are similar to each of the derived contradictory sentences.

[0063] The function of the secondary information acquisition unit 312 to acquire similar expressions and similar sentences is realized by some functions of the learning model 301. The function of the secondary information acquisition unit 312 to acquire similar expressions can be realized, for example, by providing a query to the learning model 301. The query provided here may include, for example, a request to the learning model 301 to "derive information such as modified terms, modified phrases, and the addition of modifiers from the sentence to be processed, and then search for and acquire information similar to the derived information from existing sentences." In this embodiment, such a query is included in the request to execute the ambiguity check process.

[0064] Furthermore, the function of obtaining similar sentences by the secondary information acquisition unit 312 can also be realized by providing queries to the learning model 301. The queries provided here may include, for example, a request to the learning model 301 to "derive contradictory sentences, and then search for and obtain information similar to the derived sentences from existing sentences." In this embodiment, such queries are included in the request to execute the logic check process.

[0065] When obtaining the similar expressions and sentences described above, the similarity determination will be based on the generation AI set by the company if the learning model 301 is a general-purpose generation AI managed in the company's data center, but it is preferable to tune it as appropriate. Also, if the learning model 301 is, for example, a learning model created by a user and stored in the server 30, then known methods such as cosine similarity or pattern matching may be used to determine the similarity when obtaining the similar expressions and sentences described above.

[0066] When the processing unit 311 derives a modified expression as a predetermined expression during the ambiguity check process, the information generation unit 313 generates information to be provided to the terminal device 10 based on at least one of the one or more modified terms, one or more modified phrases, and one or more modifiers (hereinafter referred to as the modified expression) derived by the processing unit 311, and similar expressions acquired by the secondary information acquisition unit 312. Furthermore, when the processing unit 311 derives two or more sentences that have a contradictory relationship as predetermined sentences during the logic check process, the information generation unit 313 generates information based on the two or more sentences derived by the processing unit 311 and similar sentences acquired by the secondary information acquisition unit 312. The information generation unit 313 then transmits the generated information to the terminal device 10.

[0067] When the provided information generation unit 313 generates provided information regarding ambiguity after the ambiguity check process, it generates the provided information by adding some or all of the similar expressions acquired by the secondary information acquisition unit 312 to the modified expression derived by the processing unit 311, or by substituting some or all of the modified expression. In this embodiment, the provided information regarding ambiguity is generated by adding all of the similar expressions acquired by the secondary information acquisition unit 312 to the modified expression derived by the processing unit 311. However, provided information may also be generated by, for example, deleting some of the acquired similar expressions according to the display limit on the terminal device 10, and adding the remaining part to the modified expression. Alternatively, provided information may be generated by replacing all of the modified expression with all of the similar expressions according to the user's settings.

[0068] Furthermore, when the information generation unit 313 generates information related to logic after the logic check process, it generates information as information that includes two or more contradictory sentences derived by the processing unit 311 and similar sentences acquired by the secondary information acquisition unit 312.

[0069] The functions of the information generation unit 313 are realized by some functions of the learning model. The information generation unit 313 can be realized, for example, by providing the learning model 301 with a query that instructs the learning model 301 to generate information according to predetermined rules after obtaining similar expressions or similar sentences and to send it to the terminal device 10. In this embodiment, such queries are included in the request to execute the ambiguity check process and the request to execute the logic check process, respectively.

[0070] <Processing by an information processing system> The following describes an example of processing performed by the information processing system S. Figure 3 is a flowchart illustrating the flow of ambiguity checking performed by the information processing system S. Figure 4 is a flowchart illustrating the flow of logic checking performed by the information processing system S. First, the ambiguity checking process will be explained with reference to Figure 3.

[0071] (Ambiguous checking process) In Figure 3, a flowchart showing the processing of terminal device 10 is shown on the left, and a flowchart showing the processing of server 30 is shown on the right. Processing of terminal device 10 begins with the launch of a dedicated browser that accepts instructions for ambiguity check processing and logic check processing. Server 30 is always running and processes in conjunction with terminal device 10. Although a flowchart for the processing of database device 20 is not shown, database device 20 is also always running and processes in conjunction with terminal device 10.

[0072] In terminal device 10, the input of document data is first monitored in step S101. If document data is entered (YES in step S101), in step S102, it is monitored whether or not text has been specified by the user in the document data. If no document data is entered in step S101 (NO in step S101), the process moves to step S107, as shown in the transition "A" in the diagram, and the occurrence of a termination instruction or other event is monitored. If no termination instruction or other event occurs in step S107, the input of document data is monitored again in step S101.

[0073] In step S102, more specifically, it is monitored whether the user has specified part or all of the text contained in the document data. The manner in which the text is specified that is monitored in step S102 may be, for example, a manner in which the user specifies part or all of the text contained in the document data displayed on the screen using a mouse or touch panel.

[0074] If a document is specified in step S102 (YES in step S102), step S103 monitors whether the user has issued an instruction for ambiguity checking. The document specified in step S102 corresponds to the document to be processed. If no document is specified in step S102 (NO in step S102), the process moves to step S107, where a termination instruction or the occurrence of another event is monitored. If no termination instruction or other event occurs in step S107, the process proceeds to step S101 and then back to step S102 to monitor for document specification.

[0075] If the instruction for ambiguity checking is confirmed in step S103 (YES in step S103), then in step S104, the ambiguity expression correction suggestion unit 101 sends a request to execute ambiguity checking to the server 30 via the database device 20. When the request to execute ambiguity checking is sent via the database device 20, some or all of the information stored in the database device 20 is added to the request. Subsequently, the request to execute ambiguity checking with the added information is sent from the database device 20 to the server 30.

[0076] When the request to execute the ambiguity check process, sent in step S104, is sent to the server 30 via the database device 20, the server 30, which was waiting to receive the execution request from the terminal device 10 in step S121, confirms receipt and proceeds to step S122. Upon receiving the request to execute the ambiguity check process in step S121, the server 30 also receives part or all of the existing document, which is information stored in the document storage unit 201 of the database device 20. Then, in step S122, the learning model 301 derives the result of the ambiguity check process using the processing unit 311.

[0077] The result of the ambiguity check process derived by the learning model 301 in step S121 may include information about ambiguous terms and / or phrases in the text, and at least one of the following (modified expressions): information about modified terms or additions of modifiers to the ambiguous terms and / or phrases. Alternatively, the result of the ambiguity check process may indicate that no ambiguous terms and / or phrases exist.

[0078] In step S122, once the result (answer) of the ambiguity check process is derived, in step S123, it is determined whether or not there is a corrected expression in the result of the ambiguity check process. If it is confirmed in step S123 that there is a corrected expression (YES in step S123), in step S124, the learning model 301 obtains expressions similar to the corrected expression from existing documents provided by the database device 20. Specifically, when the learning model 301 derives information such as corrected terms, corrected phrases, or additions of modifiers for ambiguous terms and / or phrases detected in the document to be processed, the secondary information acquisition unit 312 obtains similar expressions similar to the derived corrected terms, corrected phrases, and modifiers from existing documents provided by the database device 20.

[0079] Subsequently, in step S125, the learning model 301 generates information to be provided to the terminal device 10 based on the modified expression derived by the processing unit 311 and the similar expression acquired by the secondary information acquisition unit 312. The learning model 301 then transmits the generated information to the terminal device 10. In this embodiment, the information regarding ambiguity is generated by adding all of the similar expressions acquired by the secondary information acquisition unit 312 to the modified expression derived by the processing unit 311. For example, if the addition of modifiers such as "of the business," "of the organization," and "of personnel" is derived as the modified expression, and the expressions "of the new business," "of the existing business," and "of the organizational structure" are acquired as similar expressions, the information provided will show six suggestions.

[0080] Furthermore, if it is not confirmed in step S123 that there is a corrected expression (NO in step S123), then in step S125, information is generated indicating that there are no ambiguous terms and / or phrases. After the transmission process is performed in step S125, the server 30 waits in step S121 for the receipt of a request to perform the process.

[0081] Returning to the processing on the terminal device 10 side, when the information provided as a result of the ambiguity check process is transmitted to the terminal device 10 in step S125, the ambiguity expression correction presentation unit 101 receives the information in step S105.

[0082] Then, in step S106, the ambiguity correction presentation unit 101, as a response to the request to execute the ambiguity check process, presents, for example, on the display, at least one of the following as correcting expressions that correct the ambiguity of the ambiguous term and / or phrase, one or more correcting terms, one or more correcting phrases, and one or more modifiers. The correcting expressions displayed here reflect similar expressions acquired by the secondary information acquisition unit 312. If there are no ambiguous terms and phrases, the ambiguity correction presentation unit 101 displays a message to that effect.

[0083] Subsequently, in step S107, the system monitors for a termination instruction or the occurrence of another event. If no termination instruction or other event occurs, the system monitors the text specification again in step S102 after going through step S101, and if a termination instruction or other event is confirmed, the ambiguity check process terminates.

[0084] (Logical logic check process) Next, the logic check process will be explained with reference to Figure 4. In Figure 4, the flowchart showing the processing of the terminal device 10 is shown on the left, and the flowchart showing the processing of the server 30 is shown on the right.

[0085] In terminal device 10, first, in step S201, the input of multiple related document data is monitored. If document data is input (YES in step S201), in step S202, it is monitored whether the user has issued an instruction for logical checking. All of the multiple sentences in the multiple document data input in step S201 correspond to the sentence to be processed. If no document data is input in step S201 (NO in step S201), the process moves to step S206, as shown in the transition "B" in the figure, and the occurrence of a termination instruction or other event is monitored. If no termination instruction or other event occurs in step S206, the input of document data is monitored again in step S201.

[0086] If the instruction for logical checking is confirmed in step S202 (YES in step S202), then in step S203, the logical checking unit 102 sends a request to execute the logical checking process to the server 30 via the database device 20. When the request to execute the logical checking process passes through the database device 20, some or all of the information stored in the database device 20 is added to the request. Subsequently, the request to execute the logical checking process with the added information is sent from the database device 20 to the server 30.

[0087] When the request to execute the logic check process, sent in step S203, is sent to the server 30 via the database device 20, the server 30, which had been waiting to receive the execution request from the terminal device 10, confirms receipt in step S221 (YES in step S221) and proceeds to step S222. The server 30, which receives the request to execute the logic check process in step S221, also receives some or all of the existing documents, which are information stored in the document storage unit 201 of the database device 20. Then, in step S222, the learning model 301, with the processing unit 311, derives the result of the logic check process (logic verification result) which verifies the existence of contradictory sentences among the multiple related document data received (those input in step S201).

[0088] In this embodiment, the logic check unit 102 uses the learning model 301 to compare a sentence contained in one of a plurality of related document data held in the terminal device 10 with a sentence contained in another document data, and detects whether or not a contradictory relationship exists between a sentence contained in one of the plurality of document data and a sentence contained in another document data. However, the processing by the logic check unit 102 is not limited to comparing sentences between multiple document data. For example, the processing by the logic check unit 102 may also detect documents with contradictory relationships by comparing multiple documents within a single document data, or by comparing multiple documents designated independently of document data.

[0089] The results of the logic check process derived by the learning model 301 may include information about multiple potentially contradictory sentences, and information indicating the existence of multiple potentially contradictory sentences. Alternatively, the results of the logic check process may indicate that no potentially contradictory sentences exist.

[0090] In step S222, once the result of the logic check process is derived, in step S223, it is determined whether the result of the logic check process contains information indicating the existence of multiple potentially contradictory sentences (contradictory sentences). If it is confirmed in step S223 that there are contradictory sentences (YES in step S223), in step S224, the learning model 301 retrieves similar sentences similar to the contradictory sentences from existing sentences provided by the database device 20 via the secondary information acquisition unit 312.

[0091] Subsequently, in step S225, the learning model 301 generates information to be provided to the terminal device 10 based on the contradictory sentences derived by the processing unit 311 and the similar sentences acquired by the secondary information acquisition unit 312. The learning model 301 then transmits the generated information to the terminal device 10. The information provided regarding the logic when contradictory sentences exist includes information on multiple sentences that may contradict each other, information indicating the existence of multiple sentences that may contradict each other, and similar sentences that are similar to the contradictory sentences. Similar sentences are, for example, sentences in emails that the user has exchanged with customers in the past that have similar content to the contradictory sentences. By recognizing such similar sentences together with the contradictory sentences, the user can more easily understand the background of the contradictory sentences and verify their correctness.

[0092] Furthermore, if no contradictory sentences are found in step S223 (NO in step S223), information is generated in step S225 indicating that there are no multiple potentially contradictory sentences. After the transmission process is performed in step S225, the server 30 waits in step S221 for the receipt of a processing request.

[0093] Returning to the processing on the terminal device 10 side, when the information provided as a result of the logic check process is transmitted to the terminal device 10 in step S225, the logic check unit 102 receives the information in step S204.

[0094] Then, in step S205, the logic check unit 102 presents information corresponding to the response to the request to execute the logic check process, for example, on the display. Here, if a contradictory sentence is derived from the document to be processed, in addition to information about the contradictory sentence, information about similar sentences obtained from existing documents that are similar to the contradictory sentence is presented. In this case, for example, if a potentially contradictory sentence is detected, a notification to that effect may be displayed on the display, and the relevant sentence may be highlighted. If there are no contradictory sentences, the logic check unit 102 presents a notification to that effect.

[0095] Subsequently, in step S206, the occurrence of a termination instruction or other event is monitored. If no termination instruction or other event occurs, the input of document data is monitored again in step S201, and if the occurrence of a termination instruction or other event is confirmed, the logic check process terminates.

[0096] (Specific image of the process) The specific images of the ambiguity check and logical consistency check processes in this embodiment will be explained below with reference to Figures 5 to 8.

[0097] First, Figure 5 is a diagram that specifically illustrates the ambiguity check process. Figure 5(A) shows an image of the interface displayed on the terminal device 10's display when the ambiguity check process is executed. In Figure 5(A), reference numeral 51 indicates the title of the input document data (in the illustrated example, "Meeting Minutes"). Below the title 51, the text contained in the document data is displayed. The box indicated by reference numeral 52 indicates that part of the text has been specified by the user. When such a specification is made, a pop-up 53 prompting the execution of the ambiguity check process is displayed. Then, when the pop-up 53 is operated, for example, by the user's hand or cursor 54, a request to execute the ambiguity check process is generated.

[0098] Figure 5(B) shows an example in which the learning model 301 derives ambiguous expressions and corrected expressions from the text (processing text) indicated by the symbol 52 in Figure 5(A). In this example, the phrase "leading to a growth strategy" and the term "thing" are detected as ambiguous terms or phrases. As corrected expressions for the ambiguous terms or phrases, additional information such as modifiers "business," "organization," and "personnel" is derived for the phrase "leading to a growth strategy." In addition, suggested corrected terms such as "goal," "task," and "staffing" are derived for "thing."

[0099] Figure 5(C) shows an example of similar expressions to the revised expression derived in Figure 5(B) being obtained from existing text. In Figure 5(C), examples of existing text are shown as text from the first email 55, the second email 56, and the internal document 57, all of which were previously received by the user. In the example in Figure 5(C), the expressions "of existing businesses" 55A in the text of the first email 55, "of projects" 56A in the second email 56, and "of existing businesses" 57A and "of new businesses" 57B in the internal document 57 are obtained as similar expressions to the revised expression.

[0100] In the examples shown in Figures 5(A) to (C), the information provided by the server 30 to the terminal device 10 is generated based on, for example, additional information such as the modifiers “of the business,” “of the organization,” and “of personnel,” as modified expressions, and similar expressions such as the expression “of existing businesses” 55A, the expression “of projects” 56A, the expression “of existing businesses” 57A, and the expression “of new businesses” 57B.

[0101] Figure 6 shows how the response to a request to execute an ambiguity check process is displayed on the screen. The response shown in Figure 6 is based on the provided information. In Figure 6, below the title "Feedback" 61, the response to the ambiguity check for the text is shown as one or more of the following: an ambiguous term and / or phrase, one or more correcting terms, one or more correcting phrases, and one or more modifiers that correct the ambiguity related to the term and / or phrase. In Figure 6, the information indicated by symbols 62 and 64 corresponds to information about the ambiguous term and / or phrase. The information indicated by symbol 63 corresponds to one or more correcting terms, one or more correcting phrases, and one or more modifiers that correct the ambiguity. Here, in information 63, in addition to the correcting expression initially derived by the learning model 301, similar expressions obtained from existing texts from the database device 20 based on this are added. A user who receives such information can recognize and correct the ambiguity of the text more efficiently than when using only the correcting expression initially derived by the learning model 301.

[0102] Figure 7 is a diagram illustrating the logic check process in detail. Figure 7(A) shows an image of the interface displayed on the terminal device 10's display when the logic check process is executed. In Figure 7(A), the symbol 71 indicates the title of the multiple input document data (in the illustrated example, "B Corporation New Business Development"). The multiple document data are linked by title 71, for example. Below title 71, the text contained in the document data, the creation date, etc., are displayed.

[0103] For details, column 72 displays texts 72A and 72B concerning decisions in the first document data created on October 30, 2023. Column 73 displays text 73A concerning decisions in the second document data created on March 30, 2024. The interface also shows a logical check execution button 74. Operating the logical check execution button 74 generates a request to execute the logical check process.

[0104] Figure 7(B) shows an example in which the learning model 301 derives two potentially contradictory sentences from the multiple sentences (sentences to be processed) shown in Figure 7(A). In this example, the sentence "...lower the priority of product requirements" in the sentence on the left and the sentence "...raise the priority of product requirements" in the sentence on the right are derived as potentially contradictory sentences.

[0105] Figure 7(C) shows an example of similar sentences to potentially contradictory sentences derived in Figure 7(B) being obtained from existing sentences. In Figure 7(C), as examples of existing sentences, sentences from the third email 75 and the fourth email 76 that the user previously received are shown, and each of these sentences has been obtained as a similar sentence. The sentences from the third email 75 and the fourth email 76 were judged to have a high degree of similarity to the sentences shown in Figure 7(B) based on words such as "Company B," "product requirements," "Company S," and "Hashimoto."

[0106] In the examples shown in Figures 7(A) to (C), the information provided by the server 30 to the terminal device 10 is generated based on information from two potentially contradictory sentences, and the sentences in the third email 75 and the fourth email 76, which are similar sentences.

[0107] Figure 8 shows how the response to a request to execute a logic check process is displayed on the screen. The response shown in Figure 8 is based on the provided information. In Figure 8, a notification 80 indicating the presence of contradictory sentences is displayed at the top of the interface. Furthermore, potentially contradictory sentences are highlighted and surrounded by a frame indicated by a dashed line. Below the potentially contradictory sentences, a notification 81 containing information to be noted regarding these sentences is displayed, and below that, the sentences in the third email 75 and the fourth email 76, which are similar sentences as shown in Figure 7(C), are displayed. When a user receives such information, for example, when considering potentially contradictory sentences derived by the learning model 301, they can refer to similar sentences to efficiently examine contradictions and understand the intent of the sentences.

[0108] The information processing system S according to the first embodiment described above comprises: a document storage unit 201 that stores existing documents in advance; a processing unit 311 that derives a predetermined expression (a modified expression in this embodiment) or a predetermined document (a contradictory document in this embodiment) from a document to be processed that includes one or more documents; a secondary information acquisition unit 312 that acquires a similar expression similar to the predetermined expression derived by the processing unit 311, or a similar document similar to the predetermined document derived by the processing unit 311, from part or all of the existing document; and a provision information generation unit 313 that generates provision information based on the predetermined expression derived by the processing unit 311 and the similar expression acquired by the secondary information acquisition unit 312, or based on the predetermined document derived by the processing unit 311 and the similar document acquired by the secondary information acquisition unit 312.

[0109] In this configuration, the processing unit 311 generates information to be provided in response to the user's request based on a predetermined expression or sentence derived from the document to be processed in accordance with the user's request, and similar expressions or sentences that are similar to the predetermined expression or sentence obtained based on this. This allows us to provide users with highly useful information.

[0110] Specifically, for example, when the processing unit 311 in the learning model 301 derives an ambiguous term or phrase from the text to be processed and derives a corrected expression for it, the secondary information acquisition unit 312 additionally acquires similar expressions from existing texts in the database device 20 based on the corrected expression. Then, the user is provided with information based on the initial corrected expression and the similar expressions. A user who receives such information can efficiently recognize and correct the ambiguity of the text compared to when only the corrected expression initially derived by the learning model 301 is used. Also, when the processing unit 311 in the learning model 301 derives a potentially contradictory sentence from the text to be processed, the secondary information acquisition unit 312 additionally acquires similar expressions from existing texts in the database device 20 based on the derived contradictory sentence. Then, the user is provided with information based on the potentially contradictory sentence and the similar expressions. A user who receives such information can, for example, efficiently recognize contradictions and the intent of a sentence by referring to similar expressions when considering a potentially contradictory sentence.

[0111] <<Second Embodiment>> Figure 9 is a flowchart illustrating the processing of the information processing system according to the second embodiment. Common parts of the components in the embodiments described below with those of the first embodiment are denoted by the same reference numerals, and redundant explanations are omitted.

[0112] The difference between the second embodiment and the first embodiment is that when the request to execute the logic check process is sent to the server 30, the request is sent directly to the server 30 without going through the database device 20 (step S203'). Another difference between the second embodiment and the first embodiment is that when the learning model 301 in the server 30 derives potentially contradictory sentences from the document to be processed, in step S224', it obtains similar sentences to the contradictory sentences from the document to be processed. In other words, the processing by the secondary information acquisition unit 312 differs from that of the first embodiment. The information on similar sentences is then presented together with the information on the existence of potentially contradictory sentences. Even with this configuration, highly useful information can be provided to the user.

[0113] <<Third Embodiment>> Figure 10 is a flowchart illustrating the processing of the information processing system according to the third embodiment. Common parts of the components in the embodiments described below with those of the first embodiment are denoted by the same reference numerals, and redundant explanations are omitted.

[0114] In this embodiment, when a request to execute a logical check process is sent to the server 30, the request is sent to the server 30 via the database device 20. Then, after some or all of the information of existing documents stored in the database device 20 is added to the request to execute the logical check process, the request is sent to the server 30.

[0115] The difference between the third embodiment and the first embodiment is that, when the learning model 301 in the server 30 derives potentially contradictory sentences from the text to be processed, in step S224'', it obtains similar sentences to the contradictory sentences from existing texts and the text to be processed. Then, information on similar sentences is presented along with information on the existence of potentially contradictory sentences. Even with this configuration, highly useful information can be provided to the user.

[0116] <<Fourth Embodiment>> Figure 11 is a block diagram showing the functional configuration of the information processing system according to the fourth embodiment. Common parts of the components in the embodiments described below with those of the first embodiment are denoted by the same reference numerals, and redundant explanations are omitted.

[0117] In this embodiment, the terminal device 10 includes, in addition to the ambiguous expression correction and presentation unit 101 and the logic check unit 102, a document storage unit 201, a secondary information acquisition unit 312, and a provided information generation unit 313. On the other hand, there is no database device 20, and the server 30 includes a learning model 301 having the processing unit 311 described above.

[0118] In this configuration, after the learning model 301 sends the answer to the ambiguity check or logic check process to the terminal device 10, the terminal device 10 acquires similar expressions to the corrected expression for ambiguity or similar sentences to potentially contradictory sentences, and generates provided information that reflects these. According to this embodiment, there is an advantage in reducing the risk of information leakage because there is no need to release the information stored in the document storage unit 201 to the outside.

[0119] <<Fifth Embodiment>> Figure 12(A) is a block diagram showing the functional configuration of the information processing system according to the fifth embodiment, and Figure 12(B) is a diagram illustrating the information stored in the database device 20. In the embodiments described below, common parts with the components of the first embodiment are denoted by the same reference numerals, and redundant explanations are omitted.

[0120] In this embodiment, the terminal device 10 includes, in addition to the ambiguous expression correction and presentation unit 101 and the logic check unit 102, a voice input unit 103, a vector conversion unit 104, and a vectorized text transmission unit 105.

[0121] The audio input unit 103 inputs and stores audio data in the terminal device 10. The audio input unit 103 can, for example, capture audio via a microphone provided in the terminal device 10 and store it as audio data, or it can capture and store audio data recorded in an audio recorder in the terminal device 10. Specifically, by operating the audio input unit 103 on the terminal device 10 during a meeting, the content of what is said in the meeting can be stored in the terminal device 10 as audio data.

[0122] The vector conversion unit 104 converts the audio data provided by the audio input unit 103 into text and converts the text data into text vectors (vectorizes it). The vector conversion unit 104 performs the text conversion of audio data using, for example, an existing application, but the method is not particularly limited. The vector conversion unit 104 may also convert the audio data into text using an external learning model. Furthermore, in this example, the method used by the vector conversion unit 104 to vectorize the text data is an existing method. The vectorization of text by the vector conversion unit 104 can be performed using, for example, word2vec, doc2vec, BERT, etc., but the method is not particularly limited.

[0123] The vectorized document transmission unit 105 transmits the document vector converted by the vector conversion unit 104 to the database device 20. The vectorized document transmission unit 105 receives identification information (metadata) such as the classification and title of the document vector to be transmitted to the database device 20 from the user, and adds it to the document vector before transmitting it.

[0124] When the vectorized text transmission unit 105 transmits a text vector to the database device 20, the database device 20 stores the text vector as an existing text in the text storage unit 201.

[0125] Figure 12(B) shows an example of text vectors D1 to D4 stored as existing text in the text storage unit 201. In Figure 12(B), text vectors D1 to D4 are assigned titles as identification information entered by the user.

[0126] In this embodiment, similar to the first embodiment, when a request to execute a logic check process is sent from the terminal device 10 to the database device 20, one or more existing documents are selected from a plurality of existing documents based on the title of the document to be processed included in the logic check process, and added to the request to execute the logic check process. For example, in this embodiment, document vectors D1 and D4 are selected as existing documents that include the title "New Information Development for Company B" in the document data to be processed, and have identification information determined to be similar to this title, and are added to the request to execute the logic check process. Known methods such as cosine similarity or pattern matching may be used to determine the similarity between such a title and the identification information. Specifically, for example, if the similarity of the documents being compared is determined to be greater than or equal to a predetermined value, it may be determined that the documents being compared are similar.

[0127] As described above, when an existing document is added and a request to execute a logic check process is sent to the server 30, documents similar to two or more documents deemed contradictory within the document being processed may be retrieved from the document corresponding to the meeting content. In this case, the user can verify the correctness of the content of the contradictory documents from the document corresponding to the meeting content.

[0128] <<Example Hardware Configuration of Information Processing Equipment>> Figure 13 shows the hardware configuration of an information processing device 400 that can constitute a terminal device 10, a database device 20, and a server 30. The information processing device 400 comprises a processor 401, a ROM 402, a RAM 403, a communication interface 404, an operating device 405, and a display device 406, which are interconnected by a bus.

[0129] The processor 401 may be a CPU, a GPU, or a combination of a CPU and a GPU. The processor 401 executes a computer program stored in, for example, ROM 402 on RAM 403. The computer program is a program that can implement the above-mentioned functional configurations of, for example, the terminal device 10, the database device 20, and the server 30. The computer program may not be a single program, but may be implemented by a combination of multiple programs or scripts.

[0130] ROM402 is a non-temporary computer-readable storage medium that stores computer programs, data necessary for the execution of computer programs, and data generated by the execution of computer programs. These computer programs and data are read into RAM403 when the computer program is executed. ROM402 is, but is not limited to, a hard disk, optical disk, flash memory, and magnetic tape.

[0131] The computer program may be pre-installed on the information processing device 400, or it may be stored on a non-temporary computer-readable storage medium such as a CD-ROM. The computer program may also be uploaded to the internet. RAM 403 is, for example, DRAM or SRAM, but is not limited to these. Furthermore, the information processing device 400 may further include a storage device capable of reading and writing various types of information.

[0132] The communication interface 404 is a circuit for the terminal device 10, the database device 20, and the server 30 to communicate with external devices wirelessly or via wired connection. The operating device 405 is a circuit for inputting operation signals from input devices such as keyboards, mice, and touch panels to the terminal device 10, the database device 20, and the server 30. The display device 406 displays data stored in the terminal device 10, the database device 20, and the server 30, for example. The display device 406 may be an LCD (liquid crystal display), an organic electroluminescent display, or the like.

[0133] Furthermore, the information processing device 400 may consist of a single device or it may be configured as a system consisting of multiple interconnected computer devices.

[0134] The embodiments described above are merely examples of how the present invention can be implemented, and it is possible to carry out the present invention in various other forms. For example, various modifications, substitutions, omissions, or combinations thereof are possible without departing from the spirit of the present invention. Such modified, substituted, or omission forms are also included within the scope of the present invention, as well as within the scope of the invention and its equivalents as described in the claims. [Explanation of Symbols]

[0135] S... Information Processing System 10…Terminal device 101... Ambiguous expression correction presentation section 102...Logical Check Section 103...Voice input section 104...Vector transformation section 105... Vectorized text transmission unit 20…Database device 201...Text memory section 30… Server 301...Learning Model 311... Processing Unit 312...Secondary Information Acquisition Unit 313…Provided information generation unit 40…Network

Claims

1. A text memory unit that stores existing text in advance, A processing unit that derives a predetermined expression or predetermined sentence from a document containing one or more sentences, A secondary information acquisition unit that acquires a similar expression to the predetermined expression derived by the processing unit, or a similar sentence to the predetermined sentence derived by the processing unit, from part or all of the existing sentence, An information processing system comprising: an information generation unit that generates information based on a predetermined expression derived by the processing unit and a similar expression acquired by the secondary information acquisition unit, or based on a predetermined sentence derived by the processing unit and a similar sentence acquired by the secondary information acquisition unit.

2. The information processing system according to claim 1, wherein the processing unit derives at least one of one or more correcting terms, one or more correcting phrases, and one or more modifiers as a predetermined expression for correcting ambiguity relating to ambiguous terms and / or phrases in one or more sentences included in the document to be processed.

3. The information processing system according to claim 1, wherein the information provision generation unit generates the information provision by adding a part or all of the similar expressions acquired by the secondary information acquisition unit to the predetermined expression derived by the processing unit, or by substituting a part or all of the predetermined expression derived by the processing unit.

4. The information processing system according to claim 1, wherein the processing unit derives two or more sentences that have a contradictory relationship in a plurality of sentences in the document to be processed as the predetermined sentence.

5. The information processing system according to claim 4, wherein the information provision generation unit generates information as the information provision, which includes two or more sentences having a contradictory relationship as predetermined sentences and similar sentences acquired by the secondary information acquisition unit.

6. The information processing system according to claim 1, wherein the processing unit derives the predetermined expression or predetermined sentence from the text to be processed by natural language processing using a learning model.

7. The aforementioned existing text is stored in the text storage unit in a state converted into a text vector. The information processing system according to claim 6, wherein the secondary information acquisition unit acquires the similar expressions or similar sentences from the document storage unit by natural language processing using the learning model used in the processing unit.

8. The information processing system according to claim 1, wherein the processing unit and the secondary information acquisition unit are configured in the same information processing device.

9. The information processing system according to claim 1, comprising an information processing device having the processing unit and another information processing device having the secondary information acquisition unit.

10. A processing unit that derives a predetermined sentence from a document containing multiple sentences, A secondary information acquisition unit obtains similar sentences from the sentence to be processed that are similar to the predetermined sentence derived by the processing unit, An information processing system comprising: a provisioning information generation unit that generates provisioning information based on the predetermined text derived by the processing unit and the similar text acquired by the secondary information acquisition unit.

11. A text memory unit that stores existing text in advance, A processing unit that derives a predetermined expression or predetermined sentence from a document containing one or more sentences, A secondary information acquisition unit that acquires a similar expression to the predetermined expression derived by the processing unit, or a similar sentence to the predetermined sentence derived by the processing unit, from part or all of the existing sentence, An information processing apparatus comprising: a provisioning information generation unit that generates provisioning information based on a predetermined expression derived by the processing unit and a similar expression acquired by the secondary information acquisition unit, or based on a predetermined sentence derived by the processing unit and a similar sentence acquired by the secondary information acquisition unit.

12. A processing unit that derives a predetermined sentence from a document containing multiple sentences, A secondary information acquisition unit obtains similar sentences from the sentence to be processed that are similar to the predetermined sentence derived by the processing unit, An information processing apparatus comprising: a provisioning information generation unit that generates provisioning information based on the predetermined text derived by the processing unit and the similar text acquired by the secondary information acquisition unit.

13. A processing step of deriving a predetermined expression or predetermined sentence from a document to be processed that contains one or more sentences, A secondary information acquisition step which acquires a similar expression similar to the predetermined expression derived in the processing step, or a similar sentence similar to the predetermined sentence derived in the processing step, from a part or all of an existing sentence different from the sentence to be processed. An information processing method comprising: a provisioning information generation step that generates provisioning information based on the predetermined expression derived in the processing step and the similar expression obtained in the secondary information acquisition step, or based on the predetermined sentence derived in the processing step and the similar sentence obtained in the secondary information acquisition step.

14. A processing step of deriving a predetermined sentence from a document containing multiple sentences, A secondary information acquisition step in which similar sentences to the predetermined sentence derived in the processing step are obtained from the sentence to be processed, An information processing method comprising: a provision information generation step, which generates provision information based on the predetermined text derived in the processing step and the similar text obtained in the secondary information acquisition step.

15. A processing step of deriving a predetermined expression or predetermined sentence from a document to be processed that contains one or more sentences, A secondary information acquisition step which acquires a similar expression similar to the predetermined expression derived in the processing step, or a similar sentence similar to the predetermined sentence derived in the processing step, from a part or all of an existing sentence different from the sentence to be processed. A computer program that causes a computer to execute a provision information generation step, which generates provision information based on the predetermined expression derived in the processing step and the similar expression obtained in the secondary information acquisition step, or based on the predetermined sentence derived in the processing step and the similar sentence obtained in the secondary information acquisition step.

16. A processing step of deriving a predetermined sentence from a document containing multiple sentences, A secondary information acquisition step in which similar sentences to the predetermined sentence derived in the processing step are obtained from the sentence to be processed, A computer program that causes a computer to execute a provision information generation step, which generates provision information based on the predetermined sentence derived in the processing step and the similar sentence obtained in the secondary information acquisition step.

Images