Information processing device, information processing method, and program

Abstract

To provide an information processing device / method and a program which make a proofreading proposal of an expression from a point of view that sentences follow a rule for target sentences.SOLUTION: An information processing device 1 to detect a focused expression comprises a detector 142 to detect a first expression as a focused expression, which does not conform to a sentence expression rule, a determination part 143 to determine appropriateness of correction expression which corrects the first expression so as to conform to a sentence expression rule using a language model, and a proposal part 144 to output the information proposing the first expression correction based on the determined result by the determination part 142.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] The present invention relates to an information processing apparatus, an information processing method, and a program.

Background Art

[0002] Conventionally, a text analysis system for analyzing text has been known. For example, Patent Document 1 describes a text analysis system that converts acquired text data into a time-series signal, extracts features from the converted time-series signal, and detects a specific text based on the extracted features.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] On the other hand, depending on the industry, there are cases where it is necessary to create texts that comply with various rules, and such work usually includes a manual proofreading process. In such work, a system that can propose expression correction based on the above rules is desired, but even if the above text analysis system is applied, such expected processing cannot be performed.

[0005] One aspect of the present invention is to provide a technology capable of proposing expression correction for a target text from the viewpoint of rules that the text should follow.

Means for Solving the Problems

[0006] To solve the above problems, an information processing device according to one aspect of the present invention is an information processing device that detects an expression of interest that a user should focus on from a target document, comprising: a detection unit that refers to a text expression rule and detects a first expression that does not conform to the text expression rule as the expression of interest; a determination unit that uses a language model to determine the appropriateness of a modified expression obtained by modifying the first expression to conform to the text expression rule; and a proposal unit that outputs information that proposes correction of the first expression based on the determination result by the determination unit. [Effects of the Invention]

[0007] According to one aspect of the present invention, it is possible to propose revisions to the expression of a target document from the perspective of the rules that the document should follow. [Brief explanation of the drawing]

[0008] [Figure 1] This is a block diagram showing the configuration of an information processing device according to Embodiment 1 of the present invention. [Figure 2] This is a flowchart showing the processing flow executed by the information processing device according to Embodiment 1 of the present invention. [Figure 3] This is an example of the expression of interest in Embodiment 1 of the present invention. [Figure 4] This table shows an example of a glossary of terms and usage in Embodiment 1 of the present invention. [Figure 5] This figure shows an example of the output of the language model in Embodiment 1 of the present invention. [Figure 6] This figure shows an example of an image displayed on a display device by the proposed unit in Embodiment 1 of the present invention. [Figure 7] This is a block diagram showing the configuration of an information processing device according to Embodiment 2 of the present invention. [Figure 8] This is a block diagram illustrating the physical configuration of a computer used as an information processing device. [Modes for carrying out the invention]

[0009] [Embodiment 1] One embodiment of the present invention will be described in detail below.

[0010] (Overview of Information Processing Device 1) The information processing device 1 according to this embodiment detects expressions of interest that the user should focus on from the target document. The information processing device 1 according to this embodiment then determines whether or not modifications are necessary regarding the detected expressions of interest and makes suggestions based on the determination result.

[0011] As an example, the information processing device 1 detects expressions of interest among the expressions contained in the target document to determine whether or not they conform to the rules. Next, the information processing device 1 determines whether or not the detected expressions of interest conform to the rules and makes a suggestion based on the determination result. Examples of the modes proposed by the information processing device 1 include, but are not limited to, modes that display an image and modes that output sound.

[0012] Examples of target documents include, but are not limited to, documents related to the pharmaceutical field, documents related to the chemical field, and documents related to AI (Artificial Intelligence).

[0013] Furthermore, the supported document format is not limited to a specific format, and various formats are supported, including text documents and images containing text. Examples of users include the creator of the target document, the person who proofreads the target document, and the reviewer of the target document. Examples of rules include laws, industry rules, and company rules. However, these examples are not limiting to this embodiment.

[0014] (Configuration of Information Processing Device 1) Figure 1 is a block diagram showing the configuration of the information processing device 1 according to this embodiment. As shown in Figure 1, the information processing device 1 includes a storage unit 10, a communication unit 11, an input unit 12, an output unit 13, and a control unit 14.

[0015] The storage unit 10 stores data that is referenced by the control unit 14 described later. Examples of the data stored in the storage unit 10 include, but are not limited to, the target document TD and the output information OI. The output information OI will be described later.

[0016] The communication unit 11 is a communication module for communicating with other devices connected via a network not shown. The communication unit 11 supplies data acquired from other devices via the network to the control unit 14, or outputs data acquired from the control unit 14 to other devices via the network.

[0017] The specific configuration of the network for the communication unit 11 to communicate with other devices does not limit this embodiment. As an example, a wireless LAN (Local Area Network), a wired LAN, a WAN (Wide Area Network), a public switched telephone network, a mobile data communication network, or a combination of these networks can be used.

[0018] The input unit 12 is an interface for acquiring data from other connected devices. As an example, the input unit 12 acquires the target document TD from other connected devices. The input unit 12 supplies the data acquired from other devices to the control unit 14 described later.

[0019] The output unit 13 is an interface for outputting data to other connected devices. The output unit 13 outputs the data supplied from the control unit 14 described later to other devices. As an example, the output unit 13 outputs a proposed document including the target document TD and the output information OI to a connected display device. The proposed document will be described later.

[0020] (Control Unit 14) The control unit 14 controls each component of the information processing device 1. As an example, the control unit 14 stores the data supplied from the input unit 12 in the storage unit 10, or supplies the data stored in the storage unit 10 to the output unit 13.

[0021] Furthermore, as shown in Figure 1, the control unit 14 also functions as an acquisition unit 141, a detection unit 142, a determination unit 143, and a proposal unit 144.

[0022] The acquisition unit 141 acquires data via the communication unit 11 or the input unit 12. The acquisition unit 141 stores the acquired data in the storage unit 10.

[0023] The detection unit 142 acquires the target document TD from the storage unit 10 and detects the expression of interest that is included in the target document TD and is an expression of interest. The detection unit 142 supplies the detected expression of interest to the determination unit 143 and the proposal unit 144.

[0024] The following are examples of how the detection unit 142 detects the expression of interest. • Refer to the target document TD and the rules regarding the expressions used, and detect expressions included in the target document TD that do not conform to those rules. The target document (TD) is input to a trained model that takes a document as input and outputs a focus expression contained in that document. The expression output by the trained model is then detected as the focus expression. The trained model will be described later.

[0025] The determination unit 143 refers to the focus expression detected by the detection unit 142 and performs a determination process corresponding to the detected focus expression. The determination unit 143 supplies a determination result indicating the result of the determination process to the proposal unit 144.

[0026] Examples of determination processes performed by the determination unit 143 include the following: • Determine the appropriateness of the modified expression, which has been corrected to conform to the rules, based on the detected expression of interest. • Determine whether a predetermined expression is included in the vicinity of the detected expression of interest.

[0027] The proposal unit 144 outputs output information OI, which is information for making a proposal based on at least one of the focus expression detected by the detection unit 142 and the determination result by the determination unit 143. As an example, the proposal unit 144 generates a proposal document including the target document TD and the output information OI, and outputs it via the output unit 13.

[0028] Examples of output information OI output by the proposal unit 144 include the following: Information suggesting revisions to the wording. • Information suggesting that users check whether or not there is any information that should be included. • Information that suggests to the user to check expressions in the vicinity of the detected expression of interest.

[0029] Specific examples of the processing performed by the detection unit 142, the determination unit 143, and the proposal unit 144 will be described later.

[0030] (Processing flow by information processing device 1) The following describes the processing flow by the information processing device 1 according to this embodiment.

[0031] Figure 2 is a flowchart showing the processing flow executed by the information processing device 1 according to this embodiment.

[0032] (Step S11) In step S11, the acquisition unit 141 acquires the target document TD via the communication unit 11 or the input unit 12. The acquisition unit 141 stores the acquired target document TD in the storage unit 10.

[0033] (Step S12) In step S12, the detection unit 142 acquires the target document TD stored in the storage unit 10. Then, the detection unit 142 refers to the target document TD and detects the expression of interest. The detection unit 142 supplies the detected expression of interest to the determination unit 143 and the proposal unit 144.

[0034] The focus expressions detected by the detection unit 142 will be explained using Figure 3. Figure 3 shows an example of a focus expression in this embodiment. As shown in Figure 3, examples of focus expressions detected by the detection unit 142 include expressions EX1 to EX6.

[0035] (Detection of expression EX1) Expression EX1 is an expression related to the sentence expression rules, which are rules concerning the expressions to be used. For example, the detection unit 142 refers to the sentence expression rules and detects Expression EX1, which is an expression that does not conform to the sentence expression rules, as a focus expression. Hereafter, Expression EX1 will also be referred to as the "first expression".

[0036] An example of a rule for written expression is a glossary of words and terms that are not used. Examples of glossaries of words and terms include industry dictionaries, company dictionaries, and designated glossaries. However, this embodiment is not limited to these. An example of a glossary of words and terms is shown in Figure 4. Figure 4 is Table T1, which shows an example of a glossary of words and terms in this embodiment.

[0037] As shown in Table T1, the glossary of terms includes "terms that are not used." If a term that falls under the category of "terms that are not used" is included in the target document TD, the detection unit 142 detects that "term that is not used" as the focus expression for expression EX1. For example, if the target document TD contains the term "clearly," the detection unit 142 detects the term "clearly" as the focus expression for expression EX1. The "terms that are used" associated with the "terms that are not used" in Table T1 will be described later.

[0038] Furthermore, as described above, the detection unit 142 may detect the first representation using a trained model. For example, the detection unit 142 may be configured to input a target document TD to a trained model that takes a document as input and outputs the first representation contained in the document, and detect the first representation output from the trained model as the representation of interest. Examples of training data for training the trained model include a pair of a document and the first representation contained in the document. However, this is not limited to this exemplary embodiment. Hereinafter, the trained model will also be referred to as model M1.

[0039] Here, the specific configuration of Model M1 is not limited to this exemplary embodiment, but as an example, a CNN (Convolutional Neural Network), an RNN (Recurrent Neural Network), or a combination thereof can be used. Alternatively, non-neural network models such as Random Forests or Support Vector Machines may also be used.

[0040] (Detection of expression EX2) Expression EX2 is an expression related to the Japanese language. As an example, the detection unit 142 inputs the target document TD to a language model that estimates errors, omissions, and typos (incorrect input and incorrect conversion) of particles, and detects the expression of interest in Expression EX2 by referring to the output from the language model. Here, as the language model, an existing language model such as BERT (Bidirectional Encoder Representations from Transformers) can be used.

[0041] For example, if the target document TD contains the expression "to use," the detection unit 142 obtains output from the language model indicating that "to use" is presumed to be a misspelling of "to use." In this case, the detection unit 142 detects "to use" as the focus expression for expression EX2.

[0042] (Detection of expression EX3) Expression EX3 is an expression relating to a comparison with another company's product. If the target document TD is a document created by a company, the detection unit 142 detects expression EX3, which is an expression included in the target document TD and relates to a product of a company other than the company in question, as a focus expression. For example, if the target document TD is a document related to the pharmaceutical field and contains the names of drugs and ingredients of products from other companies, the detection unit 142 detects the drug names and ingredient names as focus expressions of expression EX3. Hereafter, expression EX3 will also be referred to as the "third expression".

[0043] As an example of how the detection unit 142 detects drug names and ingredient names as representation EX3, the detection unit 142 refers to a list of drug names and ingredient names, and if any of the drug names and ingredient names included in the list are included in the target document TD, it extracts those drug names and ingredient names.

[0044] Another example of how the detection unit 142 detects drug names and ingredient names as representation EX3 is as follows: First, the detection unit 142 inputs the target document TD into a rule-based AI that extracts katakana characters, refers to the output of the rule-based AI, and extracts the katakana characters contained in the target document TD.

[0045] Next, the detection unit 142 inputs the extracted katakana into a rule-based AI that outputs whether or not the input katakana matches the name of a drug or ingredient of any company. If the output of the rule-based AI indicates that the input katakana is a product of another company, the detection unit 142 detects the extracted katakana as the drug name or ingredient name of expression EX3.

[0046] Furthermore, the detection unit 142 may use rule-based AI that takes into account variations in the notation of extracted drug names and ingredient names (typos, missing characters, etc.). For example, the detection unit 142 may use rule-based AI that outputs whether or not the input katakana partially matches the drug name or ingredient name of any company.

[0047] Furthermore, the detection unit 142 may be configured to take a document as input and use a trained model to extract named entities contained in the document, taking into account variations in the notation of the extracted drug names and ingredient names. Examples of training data for training the trained model include a document and a pair of drug names and ingredient names with misspellings or missing characters.

[0048] (Detection of expression EX4) Expression EX4 is an expression related to statistics. For example, the detection unit 142 detects expressions related to statistics, such as "significant," "significant difference," and "significant," as focus expressions of expression EX4. As an example of how the detection unit 142 detects expressions related to statistics EX4, the detection unit 142 inputs the target document TD into a rule-based AI that detects expressions related to statistics, refers to the output of the rule-based AI, and detects expressions EX4 contained in the target document TD. Hereinafter, expression EX4 will also be referred to as the "fourth expression."

[0049] (Detection of expression EX5) Expression EX5 is an expression related to a conflict of interest (COI). As an example, the detection unit 142 detects an expression in which a paper is cited as a focus expression for expression EX5. Hereafter, expression EX5 will also be referred to as the "second expression".

[0050] The detection unit 142 may detect a second representation from the target document TD using a trained model (detection model) that takes a document as input and outputs a second representation contained in the document. Examples of training data for training the above trained model include a drug package insert and a set of references listed in the reference list contained in the package insert. However, this does not limit the present exemplary embodiment. Hereinafter, the trained model will also be referred to as model M2.

[0051] Here, the specific configuration of Model M2 is not limited to this exemplary embodiment, but as an example, a CNN, RNN, or a combination thereof can be used. Alternatively, non-neural network models such as random forests or support vector machines may also be used.

[0052] Alternatively, the detection unit 142 may detect a second expression from the target document TD using a rule-based AI that outputs a second expression contained in the document.

[0053] (Detection of expression EX6) Expression EX6 is an expression related to the Gregorian calendar or the Japanese calendar. As an example, the detection unit 142 detects the Gregorian or Japanese calendar expression EX6 as the expression of interest.

[0054] As an example of how the detection unit 142 detects expressions EX6 related to the Western calendar and the Japanese calendar, it inputs the target document TD to a rule-based AI that extracts the Western or Japanese calendar, refers to the output of the rule-based AI, and detects the expressions EX6 contained in the target document.

[0055] (Step S13) In step S13, the determination unit 143 executes a determination process corresponding to the expression of interest detected by the detection unit 142. The determination unit 143 generates a determination result indicating the result of the determination process. As an example, the processes of the determination unit 143 when the above-mentioned expressions EX1 to EX6 are detected by the detection unit 142 will be described.

[0056] (If expression EX1 is detected) In step S12, if expression EX1 is detected as a focus expression by the detection unit 142, the determination unit 143 uses a language model to determine the appropriateness of the modified expression obtained by modifying the detected expression EX1 to conform to the sentence expression rules. The determination unit 143 can use an existing language model such as BERT as the language model.

[0057] As an example, if Table T1 shown in Figure 4 is used as a rule for sentence expression, and the target document TD contains the sentence "I always make sure to put it away," then in step S12, the detection unit 142 detects "oku," which is included as a language not used in Table T1, as the focus expression for expression EX1.

[0058] Next, in step S13, the determination unit 143 refers to table T1 and generates the sentence "I always put it away and leave it there," which is corrected to the unused term "oku" and replaced with the used term "oku." The determination unit 143 then inputs both "I always put it away and leave it there," which includes expression EX1, and "I always put it away and leave it there," which includes the corrected expression, into the language model. An example of the language model's output in this case is shown in Figure 5.

[0059] Figure 5 shows an example of the output of the language model in this embodiment. As shown in Figure 5, the output of the language model includes a candidate sentence representing the input sentence and a "naturalness score," which is an index indicating whether the candidate sentence is a natural expression. In Figure 5, a higher "naturalness score" indicates that the expression is estimated to be more natural. The determination unit 143 determines that the appropriateness of the modified expression, which is obtained by modifying expression EX1 to conform to the sentence expression rules, is a "naturalness score" of "-25.82" for "I always make sure to put it away."

[0060] (If expression EX2 is detected) In step S12, if expression EX2 is detected as a focus expression by the detection unit 142, the determination unit 143 determines the appropriateness of the detected expression EX2 using a language model. Here, the determination unit 143 can use an existing language model such as BERT as the language model.

[0061] For example, in step S12, if the target document TD contains the expression "to use", the detection unit 142 obtains output from the language model indicating that "to use" is presumed to be a misspelling of "to use". The detection unit 142 detects the expression "to use" as the focus expression of expression EX2.

[0062] Next, in step S13, the determination unit 143 inputs the sentence containing the modified expression, in which "use" has been changed to "use," into the language model used for expression EX1 described above. The determination unit 143 then determines that the appropriateness of the modified expression is the value of the "naturalness score" included in the output from the language model.

[0063] (If expression EX3 is detected) In step S12, if expression EX3 is detected as a focus expression by the detection unit 142, the determination unit 143 determines whether a comparative expression is included in the vicinity of the detected expression EX3. Examples of comparative expressions include, but are not limited to, expressions that disparage expression EX3 related to products of other companies, or expressions that slander or defame.

[0064] For example, in step S12, if the target document TD contains a description of another company's product, the detection unit 142 detects the description of that other company's product as the focus expression of expression EX3.

[0065] Next, in step S13, the determination unit 143 inputs the target document TD to a rule-based AI that outputs whether or not a comparative expression is included in the vicinity of expression EX3. The determination unit 143 refers to the output of the rule-based AI and determines whether or not a comparative expression is included in the vicinity of expression EX3.

[0066] (If expression EX4 is detected) In step S12, if expression EX4 is detected as the expression of interest by the detection unit 142, the determination unit 143 determines whether or not statistically significant supporting information is included in the vicinity of the detected expression EX4. Examples of supporting information include, but are not limited to, expressions containing p-values, which are measures of the likelihood that differences between groups occur by chance.

[0067] For example, in step S12, if the target document TD contains the expression "significant," the detection unit 142 detects the expression "significant" as the focus expression of expression EX4.

[0068] Next, in step S13, the determination unit 143 inputs the target document TD to a rule-based AI that outputs whether or not a supporting expression is included in the vicinity of expression EX4. The determination unit 143 refers to the output of the rule-based AI and determines whether or not supporting information is included in the vicinity of expression EX4.

[0069] (If expression EX5 is detected) In step S12, if expression EX5 is detected as a focus expression by the detection unit 142, the determination unit 143 determines whether or not the target document TD contains disclosure information related to a conflict of interest. Examples of disclosure information include, but are not limited to, the names of companies involved in the paper and companies that funded the research in the paper, if expression EX5 is detected as an expression citing a paper.

[0070] For example, in step S12, if the target document TD contains an expression that cites a paper, the detection unit 142 detects the expression that cites the paper as the focus expression EX5.

[0071] Next, in step S13, the determination unit 143 inputs the target document TD to the rule-based AI, which outputs whether or not disclosure information is included. The determination unit 143 refers to the output of the rule-based AI and determines whether or not disclosure information is included.

[0072] (When Expression EX6 is detected) In step S12, when Expression EX6 is detected by the detection unit 142 as the target expression, the determination unit 143 determines whether there are a plurality of detected Expressions EX6. When it is determined that there are a plurality of detected Expressions EX6, the determination unit 143 determines whether each of the plurality of detected Expressions EX6 is unified into the Gregorian calendar or the Japanese calendar.

[0073] (Step S14) In step S14, the proposal unit 144 generates output information OI, which is information for making a proposal. Also, the proposal unit 144 stores the generated output information OI in the storage unit 10. Further, the proposal unit 144 generates a proposal document including the target document TD and the output information OI stored in the storage unit 10. As an example, each process of the proposal unit 144 when the above-described Expressions EX1 to EX6 are detected by the detection unit 142 will be described.

[0074] (When Expression EX1 is detected) In step S12, when Expression EX1 is detected by the detection unit 142 as the target expression, the proposal unit 144 generates output information OI that proposes the correction of Expression EX1 based on the determination result in step S13.

[0075] As an example, when the appropriateness of the corrected expression determined in step S13 is higher than a predetermined value, the proposal unit 144 generates output information OI indicating a message (e.g., "Is 'oku' the kanji character '置く' instead?") that proposes the correction of Expression EX1. On the other hand, when the appropriateness of the corrected expression determined in step S13 is lower than a predetermined value, the proposal unit 144 may not generate the output information OI, or may generate output information OI indicating a message (e.g., "Is it okay to leave it in hiragana?") that proposes to prompt the user to confirm Expression EX1.

[0076] As another example, if the appropriateness of the revised expression determined in step S13 is higher than the appropriateness of the original expression EX1, the suggestion unit 144 generates output information OI that suggests proofreading of expression EX1. On the other hand, if the appropriateness of the revised expression determined in step S13 is lower than the appropriateness of the original expression EX1, the suggestion unit 144 does not have to generate output information OI, or it may generate output information OI that suggests prompting the user to check expression EX1.

[0077] (If expression EX2 is detected) In step S12, if expression EX2 is detected as a focus expression by the detection unit 142, the proposal unit 144 generates output information OI that proposes a correction of expression EX2, based on the determination result determined in step S13, similar to the case where expression EX1 is detected.

[0078] In other words, if expression EX2 is detected as a focus expression by the detection unit 142, just as when expression EX1 is detected, if the appropriateness of the corrected expression determined in step S13 is higher than a predetermined value, or if the appropriateness of the corrected expression is higher than the appropriateness of expression EX2 before correction, the proposal unit 144 generates output information OI that proposes correction of expression EX2.

[0079] On the other hand, if the appropriateness of the modified expression determined in step S13 is lower than a predetermined value, or if the appropriateness of the modified expression is lower than the appropriateness of the expression EX2 before modification, the proposal unit 144 may not generate output information OI, or it may generate output information OI to suggest to the user that they check the expression EX2.

[0080] (If expression EX3 is detected) In step S12, if expression EX3 is detected as a focus expression by the detection unit 142, the suggestion unit 144 generates output information OI that suggests to the user to check expressions in the vicinity of expression EX3. As an example, the suggestion unit 144 generates output information OI that shows a message suggesting to the user to check expressions in the vicinity of expression EX3 (for example, "This mentions products from other companies, so please check to make sure there are no defamatory statements").

[0081] As another example, if the determination result in step S13 indicates that a comparison expression is included in the vicinity of expression EX3, the suggestion unit 144 generates output information OI that suggests the user to check the expressions in the vicinity of expression EX3 (for example, "There is an expression that compares it with a competitor's product, but is it defamatory?").

[0082] On the other hand, if the determination result in step S13 indicates that no comparison expression is included in the vicinity of expression EX3, the proposal unit 144 does not need to generate output information OI, or it may generate output information OI that suggests to the user to check the expressions in the vicinity of expression EX3.

[0083] (If expression EX4 is detected) In step S12, if expression EX4 is detected as a focus expression by the detection unit 142, the suggestion unit 144 generates output information OI that suggests to the user to check expressions in the vicinity of expression EX4. As an example, the suggestion unit 144 generates output information OI that indicates a message suggesting to the user to check expressions in the vicinity of expression EX4 (for example, "Please check if the p-value is listed").

[0084] As another example, if the determination result in step S13 indicates that the neighborhood of expression EX4 does not contain any supporting information that would serve as the basis for the statistics, the suggestion unit 144 generates output information OI that suggests to the user to check the expressions in the neighborhood of expression EX4 (for example, "It appears that the p-value is not listed, please list it").

[0085] On the other hand, if the determination result in step S13 indicates that the vicinity of expression EX4 contains supporting information that serves as the basis for the statistics, the proposal unit 144 may not generate output information OI, or it may generate output information OI that suggests to the user to check the expressions in the vicinity of expression EX4.

[0086] (If expression EX5 is detected) In step S12, if expression EX5 is detected as a focus expression by the detection unit 142, the suggestion unit 144 generates output information OI that suggests to the user to check for the presence or absence of disclosure information related to conflicts of interest in the target document TD. As an example, the suggestion unit 144 generates output information OI that indicates the presence or absence of disclosure information related to conflicts of interest in the target document TD (for example, "Is there any description indicating that you were involved in this research?").

[0087] As another example, if the determination result in step S13 indicates that the target document TD does not contain any disclosure information relating to a conflict of interest, the suggestion unit 144 generates output information OI that suggests to the user whether or not there is any disclosure information relating to a conflict of interest in the target document TD (for example, "It appears that there is no description indicating that you were involved in the research, so please check").

[0088] On the other hand, if the determination result in step S13 indicates that the target document TD contains disclosure information relating to a conflict of interest, the proposal unit 144 may not generate output information OI, or it may generate output information OI that suggests to the user that they check whether or not the target document TD contains disclosure information relating to a conflict of interest.

[0089] (If expression EX6 is detected) In step S12, if expression EX6 is detected as a focus expression by the detection unit 142, the suggestion unit 144 generates output information OI that suggests to the user to check whether the calendar is standardized to the Gregorian or Japanese calendar. As an example, the suggestion unit 144 generates output information OI that suggests to the user to check whether the target document TD is standardized to the Gregorian or Japanese calendar (for example, "Is the calendar standardized to the Gregorian or Japanese calendar?").

[0090] As another example, if the determination result in step S13 is that each of the multiple expressions EX6 is not unified in either the Gregorian or Japanese calendar, the suggestion unit 144 generates output information OI that suggests to the user to check whether the target document TD is unified in either the Gregorian or Japanese calendar.

[0091] On the other hand, if the determination result in step S13 is determined to be that each of the multiple expressions EX6 is unified to either the Gregorian calendar or the Japanese calendar, or that no multiple expressions EX6 are detected, the suggestion unit 144 does not need to generate output information OI, or it generates output information OI that suggests to the user to check whether the target document TD is unified to either the Gregorian calendar or the Japanese calendar.

[0092] (Step S15) In step S15, the proposal unit 144 outputs the generated proposal document (target document TD and output information OI) via the communication unit 11 or the output unit 13. An example of an image that the proposal unit 144 displays on a display device when the information processing device 1 is connected to a display device via the communication unit 11 or the output unit 13 will be explained with reference to Figure 6.

[0093] Figure 6 shows an example of an image displayed on a display device by the proposal unit 144 in this embodiment. The image shown in Figure 6 includes a group of checkboxes CB that accepts an operation to select which of the focus expressions to display, a target document TD, and a button BT that accepts an operation to input the target document TD to the information processing device 1.

[0094] The checkbox group CB contains checkboxes corresponding to each expression of interest. In the diagram shown in Figure 6, when the information processing device 1 receives an operation to check "Japanese expression", the suggestion unit 144 displays the target document TD in a manner that emphasizes "specifications," which was detected as the expression EX2 corresponding to "Japanese expression" among the detected expressions of interest. In other words, the suggestion unit 144 displays the target document TD on the display device in a manner that emphasizes the expression of interest.

[0095] Similarly, in the diagram shown in Figure 6, when the information processing device 1 receives an operation to check "comparative expression of competitor's product" and "statistical expression," the suggestion unit 144 displays the target document TD in a manner that emphasizes the expressions EX3 and EX4, respectively, which are detected as expressions corresponding to "comparative expression of competitor's product" and "statistical expression," from among the detected expressions of interest, namely "△△ (competitor's product) is inferior to 〇〇 (company's product)" and "significant difference."

[0096] Furthermore, the proposal unit 144 displays the proposed information, output information OI, in an area linked to the display area of ​​the expression of interest. In the diagram shown in Figure 6, the message MG1 "Is this a misuse of the term?" indicated by the output information OI is displayed in the area linked to the display area AP1 of the expression of interest "specifications".

[0097] Similarly, in the diagram shown in Figure 6, the message MG2 "The text refers to a competitor's product, but does this constitute defamation?" and the message MG3 "The p-value is not mentioned, but is it not necessary?" are displayed in the areas linked to the display areas AP2 and AP3 of the focus expression "△△ (competitor's product) is inferior to 〇〇 (our company's product)" and the focus expression "statistical difference," respectively, as indicated by the output information OI.

[0098] As described above, the information processing device 1 according to this embodiment is an information processing device 1 that detects expressions of interest that the user should focus on from a target document TD, and includes a detection unit 142 that refers to the text expression rules and detects expressions EX1 that do not conform to the text expression rules as expressions of interest, a determination unit 143 that uses a language model to determine the appropriateness of a modified expression obtained by modifying expression EX1 to conform to the text expression rules, and a proposal unit 144 that outputs output information OI that proposes correction of expression EX1 based on the determination result by the determination unit 143.Therefore, if an expression contained in a sentence contained in the target document TD does not conform to the text expression rules, the information processing device 1 according to this embodiment proposes to construct the expression, so that it can make correction suggestions for the expression of the target sentence from the perspective of the rules that the sentence should follow.

[0099] [Embodiment 2] Other embodiments of the present invention will be described below. For the sake of clarity, components having the same function as those described in the above embodiments will be denoted by the same reference numerals, and their descriptions will not be repeated.

[0100] (Overview of Information Processing Device 1A) In addition to the configuration of the above-described information processing device 1A, the information processing device 1A according to this embodiment includes a configuration that acquires a revision history CO indicating the modifications made by the user to the target document TD, and trains a trained model using training data that includes pairs of the target document TD and the revision history CO.

[0101] Examples of information included in the revision history CO include, but are not limited to, information indicating the parts (scope) of revisions made by the user, the category of the revision (for example, expression EX1, expression EX2, etc. shown in Figure 3), and the result of the revision (revised text).

[0102] (Configuration of Information Processing Device 1A) Figure 7 is a block diagram showing the configuration of the information processing device 1A according to this embodiment. As shown in Figure 7, the information processing device 1A includes a storage unit 10A, a communication unit 11, an input unit 12, an output unit 13, and a control unit 14A. The communication unit 11, the input unit 12, and the output unit 13 are as described above.

[0103] The storage unit 10A stores data that the control unit 14A, described later, will refer to. Examples of data stored in the storage unit 10A include, but are not limited to, the target document TD, output information OI, and revision history CO. The target document TD and output information OI are as described above.

[0104] (Control Unit 14A) The control unit 14A controls each component of the information processing device 1A. For example, the control unit 14A stores data acquired by the input unit 12 in the storage unit 10A, or supplies data stored in the storage unit 10A to the output unit 13.

[0105] Furthermore, as shown in Figure 7, the control unit 14A also functions as an acquisition unit 141, a detection unit 142, a determination unit 143, a suggestion unit 144, and a learning unit 145. The acquisition unit 141, detection unit 142, determination unit 143, and suggestion unit 144 are as described above.

[0106] The learning unit 145 acquires the target document TD and revision history CO stored in the memory unit 10A, and uses the training data, which includes the pairs of target document TD and revision history CO, to train the trained models (model M1 and model M2).

[0107] As an example, the learning unit 145 takes a document as input and outputs a first expression contained in that document, and inputs the target document TD to model M1. If the first expression output by model M1 is not included in the revision history CO, the learning unit 145 updates the parameters of model M1 so that it does not output that first expression.

[0108] As another example, the learning unit 145 takes a document as input and inputs the target document TD to model M2, which outputs a second expression contained in that document. If the second expression contained in the revision history CO is not output by model M2, the learning unit 145 updates the parameters of model M2 to output that second expression.

[0109] Thus, the information processing device 1A according to this embodiment includes a learning unit 145 that trains pre-trained models (model M1 and model M2) using training data that includes pairs of target documents TD and revision history CO. Therefore, the information processing device 1A according to this embodiment can further train the pre-trained models.

[0110] [Examples of implementation using software] The functions of the information processing device 1 (hereinafter referred to as "the device") are programs that cause the device to function as a computer, and these programs can be realized by programs that cause each control block of the device (particularly each part included in the control unit 14) to function as a computer.

[0111] In this case, the device includes a computer having at least one control device (e.g., a processor) and at least one storage device (e.g., memory) as hardware for executing the program. By executing the program using this control device and storage device, the functions described in each of the embodiments are realized.

[0112] The above program may be recorded on one or more computer-readable recording media, not temporary ones. These recording media may or may not be provided by the above device. In the latter case, the program may be supplied to the above device via any wired or wireless transmission medium.

[0113] Furthermore, some or all of the functions of each of the above control blocks can also be realized by logic circuits. For example, an integrated circuit in which logic circuits functioning as each of the above control blocks are formed is also included in the scope of the present invention. In addition, it is also possible to realize the functions of each of the above control blocks by, for example, a quantum computer.

[0114] Furthermore, each process described in the above embodiments may be performed by AI (Artificial Intelligence). In this case, the AI ​​may operate on the control device described above, or it may operate on other devices (for example, an edge computer or a cloud server).

[0115] (Physical configuration of information processing device 1) Figure 8 is a block diagram illustrating the physical configuration of a computer used as an information processing device 1. As shown in Figure 8, the information processing device 1 can be configured with a computer comprising a bus 110, a processor 101, a main memory 102, an auxiliary memory 103, a communication interface 104, and an input / output interface 105. The processor 101, main memory 102, auxiliary memory 103, communication interface 104, and input / output interface 105 are connected to each other via the bus 110. An input device 106 and an output device 107 are connected to the input / output interface 105.

[0116] Processor 101 can be, for example, a CPU (Central Processing Unit), a microprocessor, a digital signal processor, a microcontroller, or a combination thereof.

[0117] For example, semiconductor RAM (random access memory) can be used as the main memory 102.

[0118] For example, the auxiliary memory 103 may be flash memory, an HDD (Hard Disk Drive), an SSD (Solid State Drive), or a combination thereof. The auxiliary memory 103 stores a program that causes the processor 101 to execute the operations of the information processing device 1 described above. The processor 101 loads the program stored in the auxiliary memory 103 onto the main memory 102 and executes each instruction contained in the loaded program.

[0119] The communication interface 104 is an interface that connects to network N.

[0120] The input / output interface 105 can be, for example, a USB interface, a short-range communication interface such as infrared or Bluetooth®, or a combination thereof.

[0121] The input device 106 may be, for example, a keyboard, mouse, touchpad, microphone, or a combination thereof. The output device 107 may be, for example, a display, printer, speaker, or a combination thereof.

[0122] [summary] An information processing device according to one aspect of each embodiment is an information processing device that detects a focus expression that a user should focus on from a target document, and comprises: a detection unit that refers to a text expression rule and detects a first expression that does not conform to the text expression rule as the focus expression; a determination unit that uses a language model to determine the appropriateness of a modified expression obtained by modifying the first expression to conform to the text expression rule; and a proposal unit that outputs information that proposes correction of the first expression based on the determination result by the determination unit.

[0123] According to the above structure, it is possible to propose revisions to the expression of the target text from the perspective of the rules that the text should adhere to.

[0124] In the information processing device according to one aspect of each embodiment, the target document is preferably a document related to the pharmaceutical field.

[0125] According to the above structure, it is possible to propose revisions to the wording from the perspective of the rules that should be followed in documents in the pharmaceutical field.

[0126] In an information processing device according to one aspect of each embodiment, it is preferable that the detection unit further detects a second expression related to a conflict of interest as the focus expression, and the suggestion unit further outputs information suggesting to the user to confirm whether or not there is any disclosure information related to the conflict of interest in the target document.

[0127] According to the above structure, in documents where there are rules for disclosing information related to conflicts of interest, it is possible to propose revisions to the wording of the document from the perspective of the rules that the document must follow.

[0128] In an information processing device according to one aspect of each embodiment, it is preferable that the detection unit further detects the second expression from the target document using a detection model that takes a document as input and outputs the second expression.

[0129] According to the above configuration, expressions that do not conform to the rules can be effectively detected.

[0130] In an information processing device according to one aspect of each embodiment, the target document is a document created by a company, the detection unit further detects a third expression related to a product of a company other than the company in question as the focus expression, and the suggestion unit further outputs information suggesting to the user to check expressions in the vicinity of the third expression.

[0131] According to the above structure, in documents where there are rules regarding products of a company other than the company that created the target document, it is possible to propose revisions to the expression of the target document from the perspective of the rules that the document should adhere to.

[0132] In an information processing device according to one aspect of each embodiment, the determination unit preferably determines whether or not a comparison expression is included in the vicinity of the third expression, and the suggestion unit preferably outputs information suggesting to the user to check the nearby expression when it is determined that the comparison expression is included.

[0133] According to the above structure, in cases where a document contains expressions that do not conform to the rules when comparing products of companies other than the company that created the document, it is possible to propose revisions to the expression of the document from the perspective of the rules that the document should adhere to.

[0134] In an information processing device according to one aspect of each embodiment, it is preferable that the detection unit further detects a fourth expression related to statistics as the focus expression, and the suggestion unit further outputs information suggesting to the user to check expressions in the vicinity of the fourth expression.

[0135] According to the above structure, in texts that have rules regarding statistics, it is possible to propose revisions to the expression of the target text from the perspective of the rules that the text should adhere to.

[0136] In an information processing device according to one aspect of each embodiment, the determination unit preferably determines whether or not the basis information for the statistics is included in the vicinity of the fourth expression, and the suggestion preferably outputs information suggesting to the user to check the nearby expression when it is determined that the basis information is not included.

[0137] According to the above structure, in cases where a text does not conform to the rules because it does not contain information that serves as the basis for statistics, it is possible to propose revisions to the text from the perspective of the rules that the text should follow.

[0138] In an information processing device according to one aspect of each embodiment, it is preferable that the proposal unit displays the target document on the display device in a manner that emphasizes the expression of interest, and displays the proposed information in an area linked to the display area of ​​the expression of interest.

[0139] According to the above configuration, suggestions for proofreading the expression can be suitably presented to the user.

[0140] Information processing information relating to one aspect of each embodiment is an information processing method for detecting a focus expression that a user should focus on from a target document, comprising: a detection step of detecting a first expression that does not conform to the sentence expression rules as the focus expression by referring to sentence expression rules; a determination step of determining the appropriateness of a modified expression obtained by modifying the first expression to conform to the sentence expression rules using a language model; and a proposal step of outputting information that proposes correction of the first expression based on the determination result in the determination step.

[0141] According to the above structure, it is possible to propose revisions to the expression of the target text from the perspective of the rules that the text should adhere to.

[0142] A program according to one aspect of each embodiment is a program for causing a computer to function as the information processing device, and causes the computer to function as the detection unit, the determination unit, and the proposal unit.

[0143] According to the above structure, it is possible to propose revisions to the expression of the target text from the perspective of the rules that the text should adhere to.

[0144] The present invention is not limited to the embodiments described above, and various modifications are possible within the scope of the claims. Embodiments obtained by appropriately combining the technical means disclosed in different embodiments are also included in the technical scope of the present invention. [Explanation of Symbols]

[0145] 1. 1A Information Processing Device 10, 10A storage section 11 Communications Department 12 Input section 13 Output section 14, 14A Control Unit 141 Acquisition Department 142 Detection unit 143 Judgment section 144 proposals 145 Learning Department

Claims

1. An information processing device for detecting key expressions that users should focus on from a target document, A detection unit that, by referring to the rules of written expression, detects a first expression that does not conform to the rules of written expression as the expression of interest, A determination unit that uses a language model to determine the appropriateness of a modified expression obtained by modifying the first expression to conform to the sentence expression rules, Based on the determination result by the determination unit, a proposal unit outputs information suggesting to the user to check whether or not there is information that should be entered. Equipped with, The detection unit further detects a second expression related to a conflict of interest as the expression of interest, The aforementioned suggestion unit further outputs information suggesting to the user to check for the presence or absence of disclosure information relating to the conflict of interest in the subject document. Information processing device.

2. The aforementioned documents are documents related to the pharmaceutical field. The information processing apparatus according to claim 1.

3. The detection unit further detects the second expression from the target document using a detection model that takes a document as input and outputs the second expression. The information processing apparatus according to claim 1 or 2.

4. The aforementioned documents are documents created by companies, The detection unit further detects a third expression related to a product of a company other than the aforementioned company as the focus expression, The suggestion unit further outputs information suggesting to the user to check expressions in the vicinity of the third expression. An information processing apparatus according to any one of claims 1 to 3.

5. The determination unit determines whether or not a comparative expression is included in the vicinity of the third expression. The suggestion unit, when it is determined that the comparison expression is included, outputs information suggesting to the user to check the neighboring expression. The information processing apparatus according to claim 4.

6. The detection unit further detects a fourth expression related to statistics as the expression of interest, The suggestion unit further outputs information suggesting to the user to check expressions in the vicinity of the fourth expression. The information processing apparatus according to any one of claims 1 to 5.

7. The determination unit determines whether or not the background information that forms the basis of the statistics is included in the vicinity of the fourth expression. The proposal outputs information suggesting to the user to check the nearby expression if it is determined that the supporting information is not included. The information processing apparatus according to claim 6.

8. The aforementioned proposal section is, The target document is displayed on the display device in a manner that emphasizes the aforementioned expression of interest, The proposed information is displayed in the area linked to the display area of ​​the aforementioned focus expression. The information processing apparatus according to any one of claims 1 to 7.

9. An information processing device for detecting key expressions that users should focus on from a target document, A detection unit that, by referring to the rules of written expression, detects a first expression that does not conform to the rules of written expression as the expression of interest, A determination unit that uses a language model to determine the appropriateness of a modified expression obtained by modifying the first expression to conform to the sentence expression rules, A proposal unit outputs information that proposes corrections to the first expression based on the determination result by the determination unit, Equipped with, The detection unit further detects a second expression related to a conflict of interest as the expression of interest, The aforementioned suggestion unit further outputs information suggesting to the user to check for the presence or absence of disclosure information relating to the conflict of interest in the subject document. Information processing device.

10. An information processing device for detecting key expressions that users should focus on from a target document, A detection unit that, by referring to the rules of written expression, detects a first expression that does not conform to the rules of written expression as the expression of interest, A determination unit that uses a language model to determine the appropriateness of a modified expression obtained by modifying the first expression to conform to the sentence expression rules, A proposal unit outputs information that proposes corrections to the first expression based on the determination result by the determination unit, Equipped with, The aforementioned documents are documents created by companies, The detection unit further detects a third expression related to a product of a company other than the aforementioned company as the focus expression, The suggestion unit further outputs information suggesting to the user to check expressions in the vicinity of the third expression. Information processing device.

11. A program for causing a computer to function as an information processing device according to claim 1, 9, or 10, wherein the program causes the computer to function as the detection unit, the determination unit, and the proposal unit.

12. An information processing method in which an information processing device detects a particular expression that a user should focus on from a target document, A detection step in which a first expression that does not conform to the sentence expression rules is detected as the expression of interest by referring to the sentence expression rules, A determination step in which the appropriateness of the modified expression obtained by modifying the first expression to conform to the sentence expression rules is determined using a language model, A proposal step that outputs information suggesting corrections to the first expression based on the determination result in the determination step, Includes, In the detection step described above, a second expression related to a conflict of interest is further detected as the expression of interest, In the aforementioned proposal step, information is further output suggesting that the user check for the presence or absence of disclosure information relating to the conflict of interest in the target document. Information processing methods.

13. An information processing method in which an information processing device detects a particular expression that a user should focus on from a target document, A detection step in which a first expression that does not conform to the sentence expression rules is detected as the expression of interest by referring to the sentence expression rules, A determination step in which the appropriateness of the modified expression obtained by modifying the first expression to conform to the sentence expression rules is determined using a language model, A proposal step that outputs information suggesting corrections to the first expression based on the determination result in the determination step, Includes, The aforementioned documents are documents created by companies, In the detection step described above, a third expression related to a product of a company other than the aforementioned company is further detected as the expression of interest. In the aforementioned proposal step, information is further output suggesting that the user check the expressions in the vicinity of the third expression. Information processing methods.

Images