Information processing device, information processing method, and computer program

Abstract

To provide an information processing apparatus for improving accuracy of company valuation regarding ESG, an information processing method, and a computer program.SOLUTION: In a company valuation system configured by connecting an information processing apparatus, a company disclosure document database, and an ESG-related document database to a communication network, an information processing apparatus 1 for calculating an ESG score which is scores of company valuation regarding ESG (Environment, Social, and Governance) includes an ESG-related word dictionary storage unit 11 which stores ESG-related word dictionary data formed by registering ESG-related words which are words related to ESG; a document acquisition unit 13 which acquires a company document including company information; and an ESG score calculation unit 14 which calculates a ratio of the ESG-related words included in the company document based on the ESG-related word dictionary data, and calculates an ESG score based on the ratio of the ESG-related words.SELECTED DRAWING: Figure 2

Description

Technical Field

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

Background Art

[0002] Conventionally, as a technique for evaluating whether a company is taking ESG (Environment, Social, Governance) - consideration measures, for example, the technique described in Patent Document 1 is known. The technique described in Patent Document 1 collects a plurality of news articles published on the Internet, classifies the collected plurality of news articles by date or company, performs morphological analysis of the news articles, vectorizes each news article using TF-IDF (Term Frequency-Inverse Document Frequency), analyzes the similarity between the news articles, and thus performs clustering on the news articles. Next, using machine learning, it calculates the probability of which of the three ESG items, "environment", "social", and "corporate governance", the news articles belonging to each cluster are classified into, extracts nouns from each news article, categorizes the extracted nouns, and calculates the ESG corporate evaluation score of the company based on the frequency of words included in each category item.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] However, the technique described in Patent Document 1 mentioned above has the following problems.

[0005] The technology described in Patent Document 1 calculates an ESG corporate evaluation score for a company from multiple news articles published on the internet. However, even if a company is taking ESG-conscious measures, these measures are not necessarily covered in news articles. For example, a well-known company may be featured in news articles, while a less well-known company may not. Therefore, the technology described in Patent Document 1 cannot evaluate all companies equally, and there is a risk that some companies may be overvalued or undervalued, potentially leading to discrepancies between ESG corporate evaluation scores and the actual situation.

[0006] Furthermore, the technology described in Patent Document 1 uses the frequency of occurrence of nouns belonging to a category when calculating the ESG company rating score. However, since the frequency of word occurrences tends to increase with the number of sentences in a news article, there was a possibility that the ESG company rating score would deviate from the actual ESG efforts of the company and overestimate its performance. In addition, since the amount of information contained in each word decreases as it is used in more sentences, using the frequency of word occurrences in the ESG company rating score may lead to an overestimation of the company's ESG efforts.

[0007] Furthermore, the technology described in Patent Document 1 uses pre-classified training data from the user to perform machine learning on ESG-related topics for a machine learning model that determines which of the three ESG categories—"environment," "social," and "corporate governance"—a news article belongs to. However, new terms related to ESG are created daily, and the technology described in Patent Document 1 has difficulty incorporating these newly created terms into its machine learning model. As a result, there was a possibility that the accuracy of ESG company evaluation scores would decrease.

[0008] This invention was made in consideration of these circumstances, and its purpose is to improve the accuracy of corporate evaluations related to ESG. [Means for solving the problem]

[0009] (1) One aspect of the present invention is an information processing device for calculating an ESG score, which is a corporate evaluation score related to ESG (Environment, Social, Governance), comprising: an ESG-related word dictionary storage unit that stores ESG-related word dictionary data in which ESG-related words are registered; a document acquisition unit that acquires corporate documents which are documents containing corporate information; and an ESG score calculation unit that calculates the proportion of ESG-related words contained in the corporate documents based on the ESG-related word dictionary data and calculates an ESG score based on the proportion of ESG-related words. The system includes an ESG-related word dictionary creation unit that retrieves ESG-related words from a document database, which is publicly available via a communication network and contains documents that include at least ESG-related text, to be added to the ESG-related word dictionary data stored in the ESG-related word dictionary storage unit, and the ESG-related word dictionary creation unit searches the document database for ESG-related words, and if it does not find any registered ESG-related words already included in the ESG-related word dictionary data stored in the ESG-related word dictionary storage unit, it deletes any undiscovered registered ESG-related words from the ESG-related word dictionary data stored in the ESG-related word dictionary storage unit. It is an information processing device. (2) One aspect of the present invention is the information processing device described in (1) above, wherein the timing for updating the ESG-related word dictionary data stored in the ESG-related word dictionary storage unit is the timing when an existing document stored in the document database is deleted from the document database. ( 3 ) One aspect of the present invention is that the corporate document is a document made public by the company, as described in (1) above. (2) It is an information processing device. ( 4 ) One aspect of the present invention is that the ESG-related word dictionary data includes representative words related to ESG, similar words similar to the representative words, and the similarity between the representative words and the similar words, and the ESG score calculation unit uses the similarity as a weighting coefficient when calculating the proportion of ESG-related words, as described in (1) above. (3) It is one of the following information processing devices. ( 5 ) One aspect of the present invention is that the ESG score calculation unit calculates the IDF (inverse document frequency) value of ESG-related words and uses the IDF value of ESG-related words as a weighting coefficient when calculating the proportion of ESG-related words, as described in (1) above. (4) It is one of the following information processing devices. ( 6 ) One aspect of the present invention is that the ESG-related word dictionary data includes a flag indicating the item to which the ESG-related word belongs among the three ESG items (environment, social, and corporate governance), and the ESG score calculation unit calculates the proportion of each ESG item related to the ESG-related word contained in the corporate document, and calculates the breakdown of each ESG item in the ESG score based on the proportion of each ESG item, from (1) to ( 5 It is one of the following information processing devices. ( 7 ) One aspect of the present invention is that the document acquisition unit acquires corporate documents belonging to the same category for each company, and the ESG score calculation unit calculates an ESG score for each company from the corporate documents belonging to the same category, from (1) above ( 6 It is one of the following information processing devices. ( 8 ) One aspect of the present invention is that the ESG-related word dictionary creation unit obtains similar words that are similar to representative words related to ESG from the document database, from (1) to ( 7 It is one of the following information processing devices. ( 9 ) In one aspect of the present invention, the ESG-related word dictionary creation unit assigns the same flag to similar words that are similar to a representative word related to ESG as a flag indicating the item to which an ESG-related word belongs among the three ESG items (environment, social, and corporate governance), the above ( 8 It is an information processing device.

[0010] (10) One aspect of the present invention is an information processing method for calculating an ESG score, which is a score for evaluating a company regarding ESG (Environment, Social, Governance), comprising: an ESG related word dictionary storage step in which an information processing device stores ESG related word dictionary data in which ESG related words are registered; a document acquisition step in which the information processing device acquires a corporate document which is a document containing corporate information; and an ESG score calculation step in which the information processing device calculates the proportion of ESG related words contained in the corporate document based on the ESG related word dictionary data and calculates an ESG score based on the proportion of ESG related words. The process includes: creating an ESG-related word dictionary by obtaining ESG-related words to add to the ESG-related word dictionary data from a document database that is publicly available via a communication network and contains documents that include at least ESG-related text; the ESG-related word dictionary creation step involves searching the document database for ESG-related words, and if no registered ESG-related words already included in the ESG-related word dictionary data are found, deleting the undiscovered registered ESG-related words from the ESG-related word dictionary data. It is an information processing method.

[0011] (11) One aspect of the present invention is an ESG-related word dictionary storage step of storing ESG-related word dictionary data in which ESG-related words, which are words related to ESG (Environment, Social, Governance), are registered in a computer that calculates an ESG score, which is an enterprise evaluation score related to ESG; a document acquisition step of acquiring an enterprise document, which is a document including enterprise information; an ESG score calculation step of calculating the ratio of ESG-related words included in the enterprise document based on the ESG-related word dictionary data and calculating an ESG score based on the ratio of ESG-related words. The system performs an ESG-related word dictionary creation step, which involves obtaining ESG-related words to be added to the ESG-related word dictionary data from a document database that is publicly available via a communication network and contains documents that include at least ESG-related text; and the ESG-related word dictionary creation step involves searching the document database for ESG-related words, and if no registered ESG-related words already included in the ESG-related word dictionary data are found, the system deletes the undiscovered registered ESG-related words from the ESG-related word dictionary data. It is a computer program.

Effect of the Invention

[0012] According to the present invention, an effect that the accuracy of enterprise evaluation related to ESG can be improved is obtained.

Brief Description of the Drawings

[0013] [Figure 1] It is a block diagram showing a configuration example of an ESG enterprise evaluation system according to an embodiment. [Figure 2] It is a block diagram showing a configuration example of an information processing apparatus according to an embodiment. [Figure 3] It is a diagram showing a configuration example of ESG-related word dictionary data according to an embodiment. [Figure 4] It is a flowchart showing an example of the procedure of an ESG-related word dictionary creation method according to an embodiment. [Figure 5] It is a flowchart showing an example of the procedure of an ESG score calculation method according to an embodiment. [Figure 6] It is an example of the morphological analysis result of an annual securities report according to an embodiment. [Figure 7] It is an example of ESG-related words extracted from an annual securities report according to an embodiment.

Modes for Carrying Out the Invention

[0014] Embodiments of the present invention will be described below with reference to the drawings. Figure 1 is a block diagram showing an example configuration of an ESG corporate evaluation system according to one embodiment. In Figure 1, the information processing device 1, the corporate public document database 2, and the ESG-related document database 3 are connected to a communication network NW. The communication network NW is a communication network that includes, for example, the internet, a wired LAN (Local Area Network), or a wireless LAN. The information processing device 1 accesses the corporate public document database 2 and the ESG-related document database 3 via the communication network NW.

[0015] Information processing device 1 performs information processing to calculate an ESG score, which is a company's evaluation score related to ESG (Environment, Social, and Governance). The ESG score can be used as one evaluation indicator when assessing whether a company is taking ESG-conscious measures into consideration.

[0016] Corporate Public Document Database 2 is a database that stores documents released by companies (corporate public documents). Corporate Public Document Database 2 stores corporate public documents belonging to the same category for each company. Corporate Public Document Database 2 stores, for example, each company's securities reports.

[0017] ESG-related document database 3 is a document database made public via a communication network NW, and is a database that stores at least ESG-related documents. ESG-related documents are documents that contain text related to ESG. ESG-related document database 3 may be a database that stores only ESG-related documents, or it may be a database that stores ESG-related documents and other documents other than ESG-related documents.

[0018] Figure 2 is a block diagram showing an example configuration of an information processing device according to this embodiment. In Figure 2, the information processing device 1 includes an ESG-related word dictionary storage unit 11, an ESG-related word dictionary creation unit 12, a document acquisition unit 13, and an ESG score calculation unit 14.

[0019] Each function of the information processing device 1 is realized by the information processing device 1, which is equipped with computer hardware such as a CPU (Central Processing Unit) and memory, and by the CPU executing computer programs stored in memory. The information processing device 1 may be configured using a general-purpose computer device, or it may be configured as a dedicated hardware device. For example, the information processing device 1 may be configured using a server computer connected to a communication network such as the Internet. Furthermore, each function of the information processing device 1 may be realized through cloud computing.

[0020] Furthermore, the information processing device 1 may be implemented by a single computer, or its functions may be distributed among multiple computers. For example, the ESG-related word dictionary creation unit 12 may be implemented as a separate device.

[0021] Furthermore, the information processing device 1 may be configured to, for example, establish a website using a WWW system.

[0022] The ESG-related word dictionary storage unit 11 stores ESG-related word dictionary data. The ESG-related word dictionary data is dictionary data in which ESG-related words, which are words related to ESG, are registered.

[0023] Figure 3 shows an example of the configuration of ESG-related word dictionary data according to this embodiment. In Figure 3, the ESG-related word dictionary data is data that includes ESG-related words. There are two types of ESG-related words: representative words and similar words. Representative words are ESG-related words that have been registered in advance by the user. Similar words are ESG-related words that are similar to representative words and have been added after the representative words have been registered. As illustrated in Figure 3, one similar word may be registered for one representative word, or multiple similar words may be registered. Also, there may be representative words for which no similar words have been registered. In this embodiment, the ESG-related word dictionary creation unit 12, which will be described later, adds ESG-related words similar to representative words as similar words to the ESG-related word dictionary data.

[0024] The ESG-related word dictionary data includes the similarity score to the representative word for each similar word. In this embodiment, when the ESG-related word dictionary creation unit 12, described later, adds ESG-related words similar to the representative word as similar words to the ESG-related word dictionary data, it includes the similarity score of those similar words in the ESG-related word dictionary data.

[0025] The ESG-related word dictionary data includes a flag for each ESG-related word indicating which of the three ESG categories (environment, social, and corporate governance) the word belongs to. ESG-related words with an E flag of "1" belong to "E (environment)". ESG-related words with an S flag of "1" belong to "S (social)". ESG-related words with a G flag of "1" belong to "G (corporate governance)". Furthermore, ESG-related words with multiple flags (E, S, and G) set to "1" belong to multiple categories corresponding to those flags. For example, the ESG-related word "best-in-class" has E, S, and G flags all set to "1", so it belongs to all three categories: "E (environment)", "S (social)", and "G (corporate governance)".

[0026] Similar words to the main word will be flagged with the same item from the three ESG categories (Environment, Social, and Governance) as the main word.

[0027] The ESG-related word dictionary creation unit 12 obtains ESG-related words from the ESG-related document database 3 to be added to the ESG-related word dictionary data stored in the ESG-related word dictionary storage unit 11. In this embodiment, the ESG-related word dictionary creation unit 12 obtains ESG-related words (similar words) that are similar to the representative words registered in the ESG-related word dictionary data from the ESG-related document database 3. The ESG-related word dictionary creation unit 12 associates the similar words obtained from the ESG-related document database 3 with the corresponding representative words and adds them to the ESG-related word dictionary data stored in the ESG-related word dictionary storage unit 11.

[0028] The document acquisition unit 13 acquires publicly available corporate documents. Publicly available corporate documents are documents made public by a company. In this embodiment, the document acquisition unit 13 acquires publicly available corporate documents from the publicly available corporate document database 2. The document acquisition unit 13 acquires, for example, securities reports from the publicly available corporate document database 2.

[0029] The ESG score calculation unit 14 calculates the proportion of ESG-related words contained in publicly available corporate documents based on the ESG-related word dictionary data stored in the ESG-related word dictionary storage unit 11, and calculates an ESG score based on the proportion of ESG-related words. The ESG score calculation unit 14 uses the publicly available corporate documents of the companies to be calculated for ESG score calculation, which are acquired by the document acquisition unit 13, to calculate the ESG score of the companies to be calculated for ESG score calculation. For example, the ESG score calculation unit 14 uses the securities reports of the companies to be calculated for ESG score calculation, which are acquired by the document acquisition unit 13, to calculate the ESG score of the companies to be calculated for ESG score calculation.

[0030] [How to create an ESG-related vocabulary dictionary] The method for creating an ESG-related word dictionary according to this embodiment will be explained with reference to Figure 4. Figure 4 is a flowchart showing an example of the procedure for creating an ESG-related word dictionary according to this embodiment.

[0031] (Step S1) The user inputs a representative word from among the ESG-related words into the information processing device 1. At this time, the user specifies the item to which the representative word belongs from among the three ESG items (environment, social, and corporate governance). For example, the information processing device 1 displays a representative word input screen on the terminal used by the user (user terminal). The representative word input screen includes an input field for the representative word and a field for specifying the item to which the representative word belongs. The user uses the representative word input screen displayed on the user terminal to input the representative word and specify the item to which the representative word belongs. The item to which the representative word belongs can be one or more from among the three ESG items (environment, social, and corporate governance).

[0032] The ESG-related word dictionary creation unit 12 includes representative words entered by the user in the ESG-related word dictionary data. The ESG-related word dictionary storage unit 11 also includes a flag from among the E flag, S flag, and G flag that indicates the item to which the representative word specified by the user belongs, associating it with the representative word and including it in the ESG-related word dictionary data. As a result, in the ESG-related word dictionary data exemplified in Figure 3, representative words among ESG-related words and flags (E flag, S flag, G flag) indicating the item to which the representative word belongs are registered. For example, it is expected that several dozen words will be registered as representative words. The ESG-related word dictionary storage unit 11 stores the ESG-related word dictionary data in which the representative words and flags indicating the item to which the representative words belong are registered.

[0033] (Step S2) The ESG-related word dictionary creation unit 12 determines whether it is time to update the ESG-related word dictionary data (update timing). The update timing is, for example, a fixed period.

[0034] Furthermore, the update timing may also be when the ESG-related document database 3 is updated. For example, the update timing may be when a new ESG-related document is added to the ESG-related document database 3, or when an existing ESG-related document stored in the ESG-related document database 3 is deleted from the ESG-related document database 3. By using the timing of the ESG-related document database 3 being updated as the update timing, newly created ESG-related words can be added to the ESG-related word dictionary data as quickly as possible, and outdated ESG-related words that are no longer used can be deleted from the ESG-related word dictionary data as quickly as possible.

[0035] If the result of step S2 indicates that it is time to update, proceed to step S3. If it is not time to update, repeat step S2 until it becomes time to update.

[0036] (Step S3) The ESG-related word dictionary creation unit 12 searches for similar words from the ESG-related document database 3.

[0037] Here, we will explain how to search for similar words. The ESG-related word dictionary creation unit 12 searches for similar words to representative words already registered in the ESG-related word dictionary data stored in the ESG-related word dictionary storage unit 11 from the ESG-related documents in the ESG-related document database 3.

[0038] In this embodiment, a pre-trained machine learning model is used to search for similar words. For example, the "word2vec" model is used as the machine learning model. The "word2vec" model is known as a natural language processing model that converts words in a text into numerical vectors to understand their meaning. The "word2vec" model uses a neural network to obtain vector representations of words from existing text. Furthermore, by vectorizing the words, the semantic similarity between words can be calculated.

[0039] The ESG-related word dictionary creation unit 12 uses the "word2vec" model to search for similar words to representative words among the words contained in ESG-related documents in the ESG-related document database 3. The ESG-related documents to be searched are, for example, all ESG-related documents in the ESG-related document database 3. Alternatively, the ESG-related documents to be searched may be a predetermined number of ESG-related documents randomly selected from the ESG-related document database 3.

[0040] The word2vec model searches for similar words, yielding similar words to the representative word and the similarity score between the representative word and the similar words. Furthermore, the word2vec model undergoes automatic learning through this search process.

[0041] (Step S4) The ESG-related word dictionary creation unit 12 determines whether or not a new similar word has been found as a result of the similar word search in Step S3. The ESG-related word dictionary creation unit 12 determines that a similar word is a new similar word if the similar word found by the similar word search in Step S3 is not included in the ESG-related word dictionary data stored in the ESG-related word dictionary storage unit 11. On the other hand, if the similar word found by the similar word search in Step S3 is included in the ESG-related word dictionary data stored in the ESG-related word dictionary storage unit 11, the unit determines that the similar word is not a new similar word. If a new similar word has been found, the unit proceeds to Step S5; otherwise, the unit proceeds to Step S6.

[0042] (Step S5) The ESG-related word dictionary creation unit 12 includes the newly discovered similar words found through the search for similar words in step S3, and the similarity between the newly discovered similar words and the representative word, in the ESG-related word dictionary data stored in the ESG-related word dictionary storage unit 11, associating them with the representative word. As a result, similar words and their similarity values ​​are added to the ESG-related word dictionary data shown in the example in Figure 3.

[0043] The ESG-related word dictionary creation unit 12 assigns a flag to similar words that have been added to the ESG-related word dictionary data stored in the ESG-related word dictionary storage unit 11, corresponding to the same item as the representative word among the three ESG items (environment, social, and corporate governance).

[0044] Furthermore, the ESG-related word dictionary creation unit 12 may include in the ESG-related word dictionary data only those similar words found for a given representative word whose similarity to that representative word is above a predetermined threshold. Alternatively, the ESG-related word dictionary creation unit 12 may include in the ESG-related word dictionary data only those similar words found for a given representative word whose similarity to that representative word is among the top X words (where X is, for example, a value specified by the user).

[0045] Furthermore, if the ESG-related word dictionary creation unit 12 does not find any similar words (registered similar words) already included in the ESG-related word dictionary data stored in the ESG-related word dictionary storage unit 11 as a result of the similar word search in step S3, it may delete such registered similar words from the ESG-related word dictionary data. This makes it possible to delete outdated and unused similar words from the ESG-related word dictionary data.

[0046] (Step S5) If the ESG-related word dictionary creation process is complete, the process shown in Figure 4 is terminated. If the ESG-related word dictionary creation process is to continue, the process returns to Step S2.

[0047] According to the ESG-related word dictionary creation method described above, new ESG-related words (similar words) are automatically added to the ESG-related word dictionary data. This automatically updates the ESG-related word dictionary data, saving users the trouble of adding training data, and enabling the calculation of ESG scores using accurate ESG-related word dictionary data.

[0048] [ESG Score Calculation Method] The ESG score calculation method according to this embodiment will be explained with reference to Figure 5. Figure 5 is a flowchart showing an example of the procedure for the ESG score calculation method according to this embodiment. Note that in some cases, a subscript such as "a" may be written as "_a".

[0049] (Step S11) The document acquisition unit 13 acquires publicly available corporate documents of companies subject to ESG score calculation from the publicly available corporate document database 2. For example, the document acquisition unit 13 acquires securities reports of companies subject to ESG score calculation from the publicly available corporate document database 2.

[0050] The document acquisition unit 13 counts the total number of publicly available documents (total number of documents) of all companies subject to ESG score calculation, acquired from the publicly available corporate document database 2.

[0051] (Step S12) The ESG score calculation unit 14 performs morphological analysis on the publicly available corporate documents of the companies to be calculated for ESG scores, which have been acquired by the document acquisition unit 13, and extracts noun words from those publicly available corporate documents. The ESG-related word dictionary creation unit 12 counts the total number of noun words (total number of nouns) extracted from the publicly available corporate documents for each company to be calculated for ESG scores.

[0052] Next, the ESG score calculation unit 14 extracts words (ESG-related words) that match the ESG-related words (representative words, similar words) in the ESG-related word dictionary data stored in the ESG-related word dictionary storage unit 11, from among the words extracted from the company's public documents, for each company to be included in the ESG score calculation unit 14. The ESG score calculation unit 14 then counts the number of times each ESG-related word (representative word, similar word) in the ESG-related word dictionary data stored in the ESG-related word dictionary storage unit 11 has been extracted from the company's public documents, for each company to be included in the ESG score calculation unit 14.

[0053] The ESG score calculation unit 14 counts the total number of publicly available documents containing the relevant ESG-related word i (number of documents containing ESG-related word i) from the publicly available documents of all companies subject to ESG score calculation, which are acquired by the document acquisition unit 13 from the publicly available documents database 2.

[0054] (Step S13) The ESG score calculation unit 14 calculates an ESG score for each company to be scored, using information such as words extracted from publicly available company documents and ESG-related words. In this embodiment, the ESG score calculation unit 14 calculates an ESG score for each company to be scored using the following formula (1).

[0055]

number

[0056] N is the total number of noun words extracted from the publicly available documents of the companies included in the score calculation (total noun count). M is the total number of ESG-related words (representative words, similar words) registered in the ESG-related word dictionary data stored in the ESG-related word dictionary storage unit 11 (number of registered words). The number of registered words is the sum of the number of representative words and the number of similar words registered in the ESG-related word dictionary data.

[0057] c_i is the number of occurrences (frequency of occurrence) of the i-th ESG-related word, out of the M ESG-related words (representative words, similar words) registered in the ESG-related word dictionary data stored in the ESG-related word dictionary storage unit 11, extracted from the publicly available corporate documents of the companies for which the score is calculated. r_i is the similarity between the i-th ESG-related word and the representative word. This similarity is obtained from the ESG-related word dictionary data stored in the ESG-related word dictionary storage unit 11.

[0058] idf_i is the idf (inverse document frequency) value for the i-th ESG-related word among the M ESG-related words (representative words, similar words) registered in the ESG-related word dictionary data stored in the ESG-related word dictionary storage unit 11. idf_i is calculated by the following equation (2).

[0059]

number

[0060] The total number of documents is the total number of publicly available company documents of all companies subject to ESG score calculation, which were acquired by the document acquisition unit 13 from the publicly available company document database 2. The number of documents containing the ESG-related word i is the total number of publicly available documents from all companies subject to ESG score calculation, obtained by the document acquisition unit 13 from the publicly available corporate document database 2, that contain the i-th ESG-related word among the M ESG-related words (representative words, similar words) registered in the ESG-related word dictionary data stored in the ESG-related word dictionary storage unit 11.

[0061] According to equation (1) above, an ESG score is calculated based on the ratio of ESG-related words to the total number of nouns contained in a company's publicly available documents (e.g., securities reports). Because this ESG score is based on the proportion of ESG-related words in the publicly available documents, it can avoid the influence of the amount of text in the documents. This has the effect of suppressing the discrepancy between the ESG score and the actual ESG efforts of the company, and improving the accuracy of company evaluations regarding ESG.

[0062] Furthermore, according to equation (1) above, the similarity between the representative word and similar words is used as a weighting coefficient when calculating the proportion of ESG-related words. This takes into account the ESG relevance of similar words to the representative word, resulting in an improved accuracy of the ESG score.

[0063] Furthermore, according to equation (1) above, the IDF value (reverse document frequency) of ESG-related words is used as a weighting coefficient when calculating the proportion of ESG-related words. This takes into account the rarity of ESG-related words, resulting in an improved accuracy of the ESG score.

[0064] Furthermore, since ESG scores are calculated from publicly available company documents, it has the effect of being able to evaluate all companies fairly. Moreover, since ESG scores are calculated from publicly available company documents belonging to the same category for each company (e.g., securities reports), it is possible to suppress differences in evaluation characteristics due to differences in document format and information type between companies, thereby improving the accuracy of ESG company evaluations.

[0065] The ESG score calculation unit 14 calculates the breakdown of each ESG item (E (Environment), S (Social), G (Governance)) of the ESG score for each company for which a score is calculated.

[0066] The ESG score calculation unit 14 calculates the E score for each company subject to score calculation using the following equations (3) and (4). The E score is the company's evaluation score for E (environment) in the ESG score.

[0067]

number

[0068]

number

[0069] Elabel is the total number of E-flags "1" assigned to ESG-related words extracted from the publicly available documents of the companies subject to score calculation, out of the M ESG-related words (representative words, similar words) registered in the ESG-related word dictionary data stored in the ESG-related word dictionary storage unit 11. Slabel is the total number of S flags "1" assigned to ESG-related words extracted from the publicly available documents of the companies subject to score calculation, out of the M ESG-related words (representative words, similar words) registered in the ESG-related word dictionary data stored in the ESG-related word dictionary storage unit 11. Glabel is the total number of G flags "1" assigned to ESG-related words extracted from the publicly available documents of the companies whose scores are calculated, out of the M ESG-related words (representative words, similar words) registered in the ESG-related word dictionary data stored in the ESG-related word dictionary storage unit 11.

[0070] The ESG score calculation unit 14 calculates the S score for each company subject to score calculation using the following equations (5) and (6). The S score is the company's evaluation score for S (social) in the ESG score.

[0071]

number

[0072]

number

[0073] The ESG score calculation unit 14 calculates the G score for each company subject to score calculation using the following equations (7) and (8). The G score is the company's evaluation score for G (corporate governance) in the ESG score.

[0074]

number

[0075]

number

[0076] For each company included in the score calculation, the sum of the E score, S score, and G score equals the ESG score (E score + S score + G score = ESG score).

[0077] The information processing device 1 uses, for example, each company's annual securities report to calculate the ESG score and its breakdown (E score, S score, G score) for each company for each year. These calculation results are stored in, for example, a tabular file and provided to the user. This allows for a company evaluation not only on the overall ESG, but also on the three ESG items: "environment," "social," and "corporate governance."

[0078] Figure 6 shows an example of the morphological analysis results of a securities report according to this embodiment. Figure 6 shows the analysis results, such as words extracted from a securities report published by a listed company and their frequency of occurrence, obtained by performing morphological analysis on the securities report. The ESG score calculation unit 14 extracts words (ESG-related words) that exist in both the analysis results in Figure 6 and the ESG-related word dictionary data stored in the ESG-related word dictionary storage unit 11, along with their frequency of occurrence, similarity, and flags (E flag, S flag, G flag). An example of these extraction results is shown in Figure 7.

[0079] In Figure 7, the frequency of occurrence is c_i in equations (1), (3), (5), and (7) above. The similarity is r_i in equations (1), (3), (5), and (7) above. The sum of the values ​​for each flag (E flag, S flag, G flag) is the value for each label (E label, S label, G label) in equations (4), (6), and (8) above.

[0080] This embodiment offers the advantage of improving the accuracy of ESG-related corporate evaluations.

[0081] In the embodiment described above, the information processing device 1 is configured as an example shown in Figure 2, but the system is not limited to this. For example, the ESG-related word dictionary creation unit 12 may be implemented as a separate device. Furthermore, in the embodiments described above, a publicly released document from a company (publicly available company document) was given as an example of a company document containing company information. However, company documents are not limited to publicly available company documents. Company documents may also be confidential documents that have not been made public by the company. In addition, the company information contained in a company document may not be information that has been explicitly made public by the company.

[0082] Furthermore, this will enable improvements in overall service quality, for example, in ESG-related corporate evaluation systems, thereby contributing to Goal 9 of the United Nations-led Sustainable Development Goals (SDGs): "Build resilient infrastructure, promote sustainable industrialization and foster innovation."

[0083] Although embodiments of the present invention have been described in detail above with reference to the drawings, the specific configuration is not limited to these embodiments, and design modifications and the like are also included within the scope of the gist of the present invention.

[0084] For example, the embodiments described above can also be applied to corporate evaluations of other items besides ESG.

[0085] Alternatively, computer programs for realizing the functions of each of the above-mentioned devices may be recorded on a computer-readable recording medium, and the programs recorded on this recording medium may be loaded into a computer system and executed. Note that the term "computer system" here may include hardware such as an operating system and peripheral devices. Furthermore, "computer system" shall also include the homepage provisioning environment (or display environment) if a WWW system is being used. Furthermore, "computer-readable recording media" refers to writable non-volatile memory such as flexible disks, magneto-optical disks, ROMs, and flash memory, portable media such as DVDs (Digital Versatile Discs), and storage devices such as hard disks built into computer systems.

[0086] Furthermore, "computer-readable recording media" also includes volatile memory (such as DRAM (Dynamic Random Access Memory)) within computer systems that act as servers or clients when programs are transmitted via networks such as the Internet or communication lines such as telephone lines, which retain programs for a certain period of time. Furthermore, the above program may be transmitted from a computer system that stores the program in a memory device or the like to another computer system via a transmission medium or by transmission waves within the transmission medium. Here, the "transmission medium" used to transmit the program refers to a medium that has the function of transmitting information, such as a network (communication network) like the Internet or a communication line (communication line) like a telephone line. Furthermore, the above program may be intended to implement some of the functions described above. It may also be a so-called differential file (differential program) that can implement the aforementioned functions in combination with programs already recorded in the computer system. [Explanation of symbols]

[0087] 1... Information processing device, 2... Corporate public document database, 3... ESG-related document database, 11... ESG-related word dictionary storage unit, 12... ESG-related word dictionary creation unit, 13... Document acquisition unit, 14... ESG score calculation unit

Claims

1. In an information processing device that calculates an ESG score, which is a corporate evaluation score related to ESG (Environment, Social, and Governance), An ESG-related word dictionary storage unit stores ESG-related word dictionary data in which ESG-related words are registered, A document acquisition unit that acquires corporate documents, which are documents containing corporate information, An ESG score calculation unit calculates the proportion of ESG-related words contained in corporate documents based on the aforementioned ESG-related word dictionary data, and calculates an ESG score based on the proportion of ESG-related words. The system includes an ESG-related word dictionary creation unit that obtains ESG-related words from a document database, which is publicly available via a communication network and contains documents that include at least ESG-related text, to be added to the ESG-related word dictionary data stored in the ESG-related word dictionary storage unit, The ESG-related word dictionary creation unit searches the document database for ESG-related words, and if it does not find any registered ESG-related words already included in the ESG-related word dictionary data stored in the ESG-related word dictionary storage unit, it deletes the undiscovered registered ESG-related words from the ESG-related word dictionary data stored in the ESG-related word dictionary storage unit. Information processing device.

2. The timing for updating the ESG-related word dictionary data stored in the ESG-related word dictionary storage unit is the timing when an existing document stored in the document database is deleted from the document database. The information processing apparatus according to claim 1.

3. The aforementioned corporate documents are documents released by the company. The information processing apparatus according to claim 1 or 2.

4. The aforementioned ESG-related word dictionary data includes representative words related to ESG, similar words similar to the representative words, and the degree of similarity between the representative words and the similar words. The ESG score calculation unit uses the similarity as a weighting coefficient when calculating the proportion of ESG-related words. The information processing apparatus according to any one of claims 1 to 3.

5. The ESG score calculation unit calculates the IDF (inverse document frequency) value of ESG-related words and uses the IDF value of ESG-related words as a weighting coefficient when calculating the proportion of ESG-related words. The information processing apparatus according to any one of claims 1 to 4.

6. The aforementioned ESG-related word dictionary data includes a flag indicating which of the three ESG categories (environment, social, and corporate governance) an ESG-related word belongs to. The ESG score calculation unit calculates the proportion of each ESG item related to ESG-related words contained in the corporate document, and calculates the breakdown of each ESG item in the ESG score based on the proportion of each ESG item. The information processing apparatus according to any one of claims 1 to 5.

7. The aforementioned document acquisition unit acquires corporate documents belonging to the same category for each company. The ESG score calculation unit calculates the ESG score for each company from corporate documents belonging to the same category. The information processing apparatus according to any one of claims 1 to 6.

8. The ESG-related word dictionary creation unit obtains similar words to representative words related to ESG from the document database. The information processing apparatus according to any one of claims 1 to 7.

9. The aforementioned ESG-related word dictionary creation unit assigns the same flag to similar words to representative ESG-related words as the representative word, as a flag to indicate the item to which an ESG-related word belongs among the three ESG items (environment, social, and corporate governance). The information processing apparatus according to claim 8.

10. An information processing method for calculating an ESG score, which is a corporate evaluation score related to ESG (Environment, Social, and Governance), The information processing device performs an ESG-related word dictionary storage step, which stores ESG-related word dictionary data in which ESG-related words are registered, The aforementioned information processing device includes a document acquisition step in which it acquires a corporate document which is a document containing corporate information, The information processing device performs an ESG score calculation step, which involves calculating the proportion of ESG-related words contained in a corporate document based on the ESG-related word dictionary data, and calculating an ESG score based on the proportion of ESG-related words. The process includes: creating an ESG-related word dictionary by obtaining ESG-related words to be added to the ESG-related word dictionary data from a document database that is made publicly available via a communication network and stores documents containing at least ESG-related text; The step of creating the ESG-related word dictionary involves searching the document database for ESG-related words, and if no registered ESG-related words already included in the ESG-related word dictionary data are found, the undiscovered registered ESG-related words are deleted from the ESG-related word dictionary data. Information processing methods.

11. A computer that calculates an ESG score, which is a company's evaluation score related to ESG (Environment, Social, and Governance), An ESG-related word dictionary storage step stores ESG-related word dictionary data in which ESG-related words are registered, The document acquisition step involves acquiring corporate documents, which are documents containing corporate information, and An ESG score calculation step involves calculating the proportion of ESG-related words contained in a corporate document based on the aforementioned ESG-related word dictionary data, and calculating an ESG score based on the proportion of ESG-related words. The process involves creating an ESG-related word dictionary by obtaining ESG-related words to be added to the ESG-related word dictionary data from a document database that is publicly available via a communication network and contains documents that include at least ESG-related text. The step of creating the ESG-related word dictionary involves searching the document database for ESG-related words, and if no registered ESG-related words already included in the ESG-related word dictionary data are found, the undiscovered registered ESG-related words are deleted from the ESG-related word dictionary data. Computer program.

Images