Adverb dictionary update device and adverb dictionary update method

Abstract

To provide an information processor capable of solving the disadvantage of conventional information processors that any adverb dictionary could not be updated over time.SOLUTION: The information processor includes; a text data adverb extraction unit 330 that extracts adverbs acquired from external text data and stores the same in an adverb set storage unit; and an unregistered adverb degree value assignment unit 350 that reads a list of extracted adverbs in the adverb set storage unit and an adverb dictionary that has a table of registered adverbs and their degree values and updates the adverb dictionary by assigning degree value to unregistered adverbs among the extracted adverbs.SELECTED DRAWING: Figure 3

Description

[Technical field]

[0001] The present invention relates to an adverb dictionary updating device and an adverb dictionary updating method that are used for processing the contents of text data with a focus on emotions. [Background technology]

[0002] Traditionally, services and products cannot be developed without evaluation from the user's perspective (quality at the time of use, NPS). In addition, there is a demand for collecting detailed customer feedback through social networking services (SNS), which can collect the opinions of many more people than through surveys or review sites.

[0003] For example, an information processing device is disclosed that can collect preference information of users who use an Internet service to which content including at least images (including photographs (still images) and videos consisting of a series of still images) and text is posted (see Patent Document 1). Patent Document 1 (paragraph 0031) states that "the preference tendency analysis unit 205 creates and stores in advance a table for determining the degree of preference for a large number of adverbs, as exemplified in FIG. 9, and estimates the viewer's own subjective degree of preference by consulting the adverbs extracted by the language analysis unit 203." [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Patent Publication No. 2021-092931 Summary of the Invention [Problem to be solved by the invention]

[0005] As mentioned above, there has been research into calculating the degree of an adverb using an emotion dictionary (also called an adverb dictionary). However, an adverb dictionary created at a certain point in time (era) will add new words and change its degree value as time passes (the flow of time). For this reason, a method is desired that can update the adverb dictionary as time passes. Patent Document 1 states that "the table should be added to and updated from time to time" (paragraph 0031), but does not disclose a specific method for updating the emotion dictionary.

[0006] The present invention has been made in consideration of the above circumstances, and has an object to propose a method for updating an adverb dictionary over time. [Means for solving the problem]

[0007] In order to solve the above problem, an adverb dictionary updating device of one embodiment of the present invention is an adverb dictionary updating device that updates an adverb dictionary used for emotion degree estimation, and includes: a text data acquisition unit that acquires text data including a target of emotion degree estimation and stores it in a text set storage unit; a text data adverb extraction unit that extracts adverbs related to each text data from the text set storage unit and stores them in the adverb set storage unit; and an unregistered adverb degree value assignment unit that reads a list of extracted adverbs from the adverb set storage unit and an adverb dictionary having registered adverbs and degree values ​​as a table from the adverb dictionary storage unit, and assigns degree values ​​to unregistered adverbs among the extracted adverbs by referring to the degree values ​​of the registered adverbs, thereby updating the adverb dictionary and storing the adverb dictionary in the adverb dictionary storage unit. Effect of the Invention

[0008] At least one aspect of the present invention can support updating and maintenance of an emotion dictionary. Problems, configurations and effects other than those described above will become apparent from the following description of the embodiments. [Brief description of the drawings]

[0009] [Figure 1] FIG. 1 is a diagram illustrating an outline of an emotion object degree estimation method. [Diagram 2] FIG. 1 is a diagram showing an example of an emotion level estimation method having an adverb dictionary update function according to a first embodiment of the present invention. [Diagram 3] 1 is a block diagram showing an example of the internal configuration of an adverb dictionary updating device that realizes an emotion degree estimation method according to a first embodiment of the present invention. [Figure 4] 5 is a flowchart showing an example of a processing procedure of an unregistered adverb degree value assigning unit in the first exemplary embodiment of the present invention. [Diagram 5] FIG. 11 is a block diagram showing an example of the internal configuration of an adverb dictionary updating device that realizes an emotion degree estimation method according to a second embodiment of the present invention. [Figure 6] FIG. 13 is a diagram showing an example of a user interface screen for checking the degree value of an adverb in each embodiment of the present invention. [Figure 7] 13 is a diagram for explaining a user confirmation function of an adverb dictionary in each embodiment of the present invention. FIG. [Figure 8] FIG. 13 is a diagram showing an example of a user interface screen of an adverb dictionary maintenance tool according to each embodiment of the present invention. [Figure 9] 2 is a block diagram showing an example of a hardware configuration of a control system of the adverb dictionary updating device according to each embodiment of the present invention. FIG. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0010] Hereinafter, examples of modes for carrying out the present invention (hereinafter referred to as "embodiments") will be described with reference to the accompanying drawings. In this specification and the accompanying drawings, identical components or components having substantially the same functions are designated by the same reference numerals, and duplicated descriptions will be omitted.

[0011] <First embodiment> First, an information processing apparatus according to a first embodiment of the present invention will be described with reference to FIGS. 1 to 4. FIG.

[0012] Fig. 1 is a diagram showing an overview of an emotion degree estimation method. An emotion degree estimation method 30 shown in Fig. 1 estimates the emotion degree from input information 10, and outputs the estimation results as output information 40. The emotion degree is information indicating the three major categories of emotion (favorable / positive, neutral, and malicious / negative) as well as the degree (strength) of each emotion. A target sentence (text data) to be analyzed for emotion analysis device analysis result 11 is input as input information 10 to the emotion degree estimation method 30.

[0013] The emotion analysis device analysis result 11 is information that links information on a target sentence (text data) such as SNS data or business data, output from an emotion analysis device (not shown), with the result of emotion analysis on the target sentence. For example, in Fig. 1, text data 12 "Product A was very good" is a target sentence to be emotion analyzed, and an example is shown in which emotion information 13, keyword information 14, and dependency information 15 are output as an example with the target sentence as input.

[0014] Emotion information 13 is information resulting from judging whether the target sentence has a positive, negative, or neutral emotion. In Fig. 1, emotion information 13 for text data 12 is judged to be "Positive". The keyword information 14 is a list of keywords included in the target sentence. For example, the keywords are classified as "A: proper noun", "product: noun", and "good: adjective". The dependency information 15 is information that lists the dependency relationships of the target sentence. For example, it is "dependency: A, dependency: product". These three pieces of information are examples of information used in the present invention and are excerpts of the results of emotion analysis. The emotion analysis device also outputs various other pieces of information through natural language processing.

[0015] The emotion degree estimation method 30 is a processing unit that estimates the emotion degree of a target sentence. The emotion degree estimation method 30 extracts adverbs 31 as emotion expressions from the target sentence (S1), and calculates the emotion degree of the target sentence based on the extracted adverbs 31 by referring to an adverb dictionary 32 (S2). Adverbs are generally parts of speech that modify the nature or state of things. The adverb dictionary 32 is a table that has records of combinations of adverbs and degree values ​​as information representing the degree of the adverb. The emotion degree estimation method 30 then outputs emotion information 41 and emotion degree 42 for the target sentence. The emotion information 41 is emotion information 13. For example, FIG. 1 shows an example in which "very" is extracted as the adverb 31, and "Positive" is output as the emotion information 41 and "1.8" is output as the emotion degree 42.

[0016] The emotion degree estimation method 30 outputs output information 40 as a result of estimation processing on the input information 10. The output information 40 includes emotion information 41 and emotion degree 42 output from the emotion degree estimation method 30. In Fig. 1, an example of the output information 40 in which the emotion information 41 for the text data 12 is "Positive" and the emotion degree 42 is "1.8" is shown.

[0017] In the emotion degree estimation method, an adverb dictionary (emotion dictionary) is created and a formula for calculating the degree is defined. In this embodiment, a system of adverbs expressing degree is organized, and degree values ​​are set for representative adverbs based on human subjectivity.

[0018] For example, in an adverb dictionary, the degree of an adverb is set between "0" and "2." "1" is the reference value. The reference value may be interpreted as a standard value. For example, the adverb "quite" is considered to have a high degree of difficulty from a human subjective point of view, so a degree value of 1.8 is set within the degree value range of 0-2. An adverb dictionary may be created based on the subjective opinions of multiple people through a questionnaire or the like. In this case, the degree of each adverb is obtained from multiple people through the questionnaire, and the degree value is calculated statistically, such as the average, variance, or median. In this way, combinations of adverbs and degree values ​​are defined, and a standard adverb dictionary is created.

[0019] As examples of degree values ​​of adverbs, for example, as shown in FIG. 6 described later, for high-degree adverbs (very, extremely, etc.), the degree value can be set to "1.8", for medium-degree adverbs (quite, quite, etc.), the degree value can be set to "1.5", and for low-degree adverbs (somewhat, a little, etc.), the degree value can be set to "0.8".

[0020] In degree calculation (S2) in the emotion degree estimation method 30, the emotion degree is calculated from the major emotion classification and the degree value of the adverb. For example, the following two formulas can be used as the degree calculation formula.

[0021] Degree calculation formula 1 (multiplication formula) = Emotion classification (+1.0, 0, -1.0) × Π adverb degree value Degree calculation formula 2 (additive formula) = Emotion category (+1.0, 0, -1.0) + Σ(adverb degree value -1.0)

[0022] Here, "+1.0" is set for "favorable," "0" for "neutral," and "-1.0" for "unfavorable." Π in degree calculation formula 1 and Σ in degree calculation formula 2 indicate that, for example, when the sentence to be analyzed contains multiple adverbs, the number of adverbs and their degree values ​​are multiplied or added.

[0023] Fig. 2 is a diagram showing an example of an emotion degree estimation method having an adverb dictionary update function according to this embodiment. The emotion degree estimation method 200 shown in Fig. 2 is different from the emotion degree estimation method 30 shown in Fig. 1 in that it has an adverb dictionary 32 update function. That is, the emotion degree estimation method 200 has, as an adverb dictionary 32 update function, a function (S3) of automatically assigning a degree value to an unregistered adverb and a function (S4) of correcting the degree value.

[0024] When a user U of the emotion analysis device requests the provision (execution) of this service, text data 12 (target sentence) to be analyzed in the emotion analysis device analysis result 11 is input to the emotion degree estimation method 200. When an adverb 31 extracted from the text data is an unregistered adverb, the emotion degree estimation method 200 automatically assigns a degree value to the unregistered adverb 31 (S3) and updates the adverb dictionary 32. In degree calculation (S2), the emotion degree of the text data 12 is estimated by referring to the updated adverb dictionary 32. For example, in the automatic degree value assignment, when a large number of adverbs are added, a degree value is estimated according to the similarity with existing adverbs and assigned to the added adverb.

[0025] Furthermore, the emotion degree estimation method 200 has a function (S4) that allows the user U to check the degree values ​​of the adverbs registered in the adverb dictionary 32 and correct the degree values. In the manual degree value correction, the automatically assigned degree values ​​can be appropriately corrected manually. However, the degree value correction function may be configured to automatically correct the degree values.

[0026] Next, the configuration of an adverb dictionary updating device that realizes the emotion degree estimation method 200 will be described with reference to FIG. Fig. 3 is a block diagram showing an example of the internal configuration of an adverb dictionary updating device that realizes the emotion degree estimation method 200. The adverb dictionary updating device 300 shown in Fig. 3 includes a text data acquisition section 310, a text set storage section 320, a text data adverb extraction section 330, an adverb set storage section 340 related to text data (an example of an adverb set storage section), and an unregistered adverb degree value assignment section 350. The adverb dictionary updating device 300 also includes an adverb dictionary storage section 360, a text data emotion degree value calculation section 370, and an emotion degree value set storage section 380 related to text data (an example of an emotion degree value set storage section). The adverb dictionary updating device 300 also includes an emotion determination section 325, and a text data emotion determination result storage section 326.

[0027] The text data acquisition unit 310 transmits a query to a text medium 305 provided outside the adverb dictionary updating device 300, and acquires text data corresponding to the query from the text medium 305. In this embodiment, the emotion analysis device analysis result 11 shown in FIG. 1 is used for the text medium 305. The query is an extraction condition for text data (specification of search target, range, etc.), such as "Company H's washing machine" or "date". The text data acquisition unit 310 stores the acquired text data in the text set storage unit 320.

[0028] The text data adverb extraction unit 330 extracts adverbs related to each text data stored in the text set storage unit 320 and stores them in the text data adverb set storage unit 340.

[0029] The unregistered adverb degree value assigning unit 350 reads a list of extracted adverbs from the adverb set storage unit 340 related to the text data, and an adverb dictionary (adverb dictionary 32 in FIG. 2) having registered adverbs and degree values ​​as a table from the adverb dictionary storage unit 360. Then, the unregistered adverb degree value assigning unit 350 assigns degree values ​​to adverbs that are not registered in the adverb dictionary 32 among the extracted adverbs, thereby updating the adverb dictionary 32, and stores the adverb dictionary 32 in the adverb dictionary storage unit 360.

[0030] The emotion determination section 325 reads each piece of text data from the text set storage section 320. Then, the emotion determination section 325 determines the emotion of each piece of text data, and stores a combination of each piece of text data and the emotion determination result in the text data emotion determination result storage section 326.

[0031] As can be understood from the above description, the emotion determination unit 325 and the text data emotion determination result storage unit 326 correspond to the input side of the emotion degree estimation method 200 shown in Fig. 2. For example, the text data emotion determination result storage unit 326 may be considered to correspond to the emotion analysis device analysis result 11 in Fig. 1.

[0032] Text data emotion degree value calculation section 370 reads each piece of text data from text set storage section 320, emotion determination results for each piece of text data from text data emotion determination result storage section 326, extracted adverbs from adverb set related to text data storage section 340, and adverb dictionary 32 from adverb dictionary storage section 360. Then, text data emotion degree value calculation section 370 calculates the emotion degree value of each piece of text data, and stores each piece of text data, the extracted adverbs, and the emotion degree value as a table in text data emotion degree value set storage section 380.

[0033] Next, the procedure of the process of the unregistered adverb degree value assigning unit 350 will be described with reference to FIG. FIG. 4 is a flowchart showing an example of a processing procedure of the unregistered adverb degree value assigning unit 350. As shown in FIG.

[0034] First, the unregistered adverb degree value assigning unit 350 reads an extracted adverb from the adverb set storage unit 340 related to the text data (S401). The unregistered adverb degree value assigning unit 350 also reads an adverb and a degree value as the adverb dictionary 32 from the adverb dictionary storage unit 360 (S402).

[0035] Next, the unregistered adverb degree value assigning unit 350 executes a loop process A including the processes from step S404 to step S409 for all the extracted adverbs (S403).

[0036] In the loop process A, the unregistered adverb degree value assigning unit 350 first determines whether or not the extracted adverb is in the adverb dictionary 32 (S404). If it is determined that all the extracted adverbs are in the adverb dictionary 32 (YES determination in S404), the unregistered adverb degree value assigning unit 350 ends the processing of this flowchart.

[0037] On the other hand, if it is determined that the extracted adverb is not in the adverb dictionary 32 (NO determination in S404), the unregistered adverb degree value assigning unit 350 initializes the degree value of the corresponding extracted adverb (unregistered adverb) to "1" (reference value) (S405). As described above, the degree value "1" indicates that the degree of the adverb is medium. This applies when the degree value of the adverb is expressed in the range of 0 to 2, and does not apply when the degree value is expressed in another numerical range.

[0038] Next, the unregistered adverb degree value assigning unit 350 calculates the similarity between the corresponding extracted adverb (unregistered adverb) and all adverbs in the adverb dictionary 32 (S406). The method of calculating the similarity of the adverb is not specified. For example, the similarity of the adverb may be calculated using a distributed representation of the adverb and its cosine similarity, or may be calculated using hierarchical closeness in the thesaurus.

[0039] A word's distributed representation is a method of converting a single word into a vector with low dimensionality (for fruits, size, sweetness, yellowness (in the case of three dimensions), etc.), and by representing a word as a distributed representation, the meaning of the word can be expressed from the distance and positional relationship between vectors. Methods such as one-hot vectors, Word2Vec, fastText, and BERT (Bidirectional Encoder Representations from Transformers) have been proposed as methods of representing words as vectors.

[0040] A thesaurus is a dictionary (similar to a thesaurus) that organizes words based on their semantic relationships. More specifically, a thesaurus is a system of words that is organized with a focus on the "relationships between words," such as the relationship between superordinate and subordinate concepts of words (concepts they refer to), inclusion relationships, the "whole and part" relationship, or the relationship between synonyms in the same category. It is said that the distance between vectors based on distributed representations reflects historical changes (the passage of time) in the meaning and degree of adverbs more than the hierarchical closeness at higher levels in a thesaurus.

[0041] Next, the unregistered adverb degree value assigning unit 350 judges whether the similarity between the extracted adverb and the most similar adverb is equal to or greater than a preset threshold (S407). If it is judged that the similarity between the extracted adverb and the most similar adverb is less than the threshold (NO judgment in S407), the unregistered adverb degree value assigning unit 350 proceeds to the process of step S409.

[0042] On the other hand, if it is determined that the similarity between the extracted adverb and the most similar adverb is equal to or greater than the threshold value (YES judgment in S407), the unregistered adverb degree value assignment unit 350 updates the degree value of the extracted adverb set in step S405 (in this example, "1") to the degree value of the most similar adverb (S408).

[0043] In step S407, the similarity of the adverb is compared with a threshold value, but the threshold value may not be set or may be set to 0. In this case, the determination process in step S407 can be omitted.

[0044] Next, the unregistered adverb degree value assigning unit 350 registers the extracted adverb and the updated degree value in the adverb dictionary 32 (S409).

[0045] Then, when the loop process A has been completed for all extracted adverbs, that is, after degree values ​​have been assigned to all unregistered adverbs, the process of this flowchart ends.

[0046] As described above, the adverb dictionary updating device according to the first embodiment is an adverb dictionary updating device that updates an adverb dictionary used for emotion degree estimation, and includes a text data acquiring unit that acquires text data including a target of emotion degree estimation and stores it in a text set storage unit, a text data adverb extraction unit that extracts adverbs related to each text data from the text set storage unit and stores them in an adverb set storage unit (adverb set storage unit related to text data 340), and an unregistered adverb degree value assigning unit that reads a list of extracted adverbs from the adverb set storage unit and an adverb dictionary having registered adverbs and degree values ​​as a table from the adverb dictionary storage unit, refers to the degree values ​​of the registered adverbs, assigns degree values ​​to unregistered adverbs among the extracted adverbs, updates the adverb dictionary, and stores the adverb dictionary in the adverb dictionary storage unit.

[0047] The present embodiment having the above-mentioned configuration can support the updating and maintenance of the adverb dictionary. In this embodiment, for example, when a new adverb is extracted from text data resulting from analysis by the sentiment analysis device, the degree value of the adverb can be automatically assigned with reference to the degree values ​​of registered adverbs. In addition, the adverb dictionary can be updated manually or automatically in response to changes in the degree values ​​of the adverbs. Furthermore, when configured to automatically assign degree values ​​to unregistered adverbs, the cost of manual maintenance of the adverb dictionary can be reduced.

[0048] In this way, by using an adverb dictionary that has been appropriately updated over time by the adverb dictionary updating device of this embodiment, it is possible to identify customer opinions about companies, products, etc. from SNS, which contains a mixture of various information, and achieve customer satisfaction based on the strength of the sentiment analysis results (favorable, neutral, unfavorable).

[0049] <Second embodiment> The second embodiment is an example in which a method of calculating the similarity of adverbs that takes into account the passage of time of words is assumed for the adverb dictionary updating device 300 (see FIG. 3) of the first embodiment. Specifically, in this embodiment, a distributed representation model is updated in accordance with the passage of time as a method of calculating the similarity of adverbs.

[0050] Fig. 5 is a block diagram showing an example of an internal configuration of an adverb dictionary updating device that realizes an emotion degree estimation method according to a second embodiment of the present invention. The adverb dictionary updating device 300A shown in Fig. 5 is different from the adverb dictionary updating device 300 shown in Fig. 3 in that it includes a distributed representation model 510 and a distributed representation model updating unit 520.

[0051] The distributed representation model 510 is a natural language processing model that realizes the distributed representation of adverbs. The unregistered adverb degree value assigning unit 350 uses the distributed representation model 510 to distribute the adverbs extracted from the text data and the adverbs registered in the adverb dictionary, and calculates the similarity of the adverbs.

[0052] The distributed representation model update unit 520 has a function of updating the distributed representation model 510 over time. The distributed representation model update unit 520 may be configured to manually reconstruct the distributed representation model 510, or may be configured to automatically replace the distributed representation model 510 with the latest distributed representation model.

[0053] As an example, the distributed representation model may be obtained by the distributed representation model update unit 520 from an open model that is publicly available on the Internet, etc. Alternatively, the user may create the distributed representation model by collecting a large-scale text data set that was created at the time when the user wants to create the distributed representation model.

[0054] Next, a user interface for confirming and correcting the degree value of an adverb in each of the above-mentioned embodiments will be described with reference to FIGS.

[0055] 6 is a diagram showing an example of a user interface screen for checking the degree value of an adverb in each embodiment. When a user selects a mode for checking the degree value of an adverb on a menu screen (not shown), the unregistered adverb degree value assigning unit 350 performs processing for displaying a user interface screen 600 for checking the degree value of an adverb. The user interface screen 600 includes a degree value display area 610, a correction button 620, and an end button 630.

[0056] The degree value display area 610 is an area that displays an adverb dictionary, that is, registered adverbs and their degree values. In the degree value display area 610, the combination of a newly registered adverb and a degree value can be visualized by displaying the combination of the newly registered adverb and the degree value in color. The colored adverbs indicate adverbs that have been assigned a degree value but have not been confirmed by the user. The display method is not limited to color display, and any display that can distinguish the newly registered combination of adverb and degree value from others, such as gray display or pop-up display, may be used. When there are many combinations of adverbs and degree values, the hidden combinations of adverbs and degree values ​​can be moved and displayed by moving the slide bar up and down.

[0057] When the user presses the edit button 620, a screen for an adverb dictionary maintenance tool 800 shown in FIG. 8 is displayed, and the degree value of the adverb displayed in color in the degree value display area 610 can be edited.

[0058] When the user presses the end button 630, the user interface screen 600 disappears and the mode for checking the degree value of the adverb ends. By detecting that the end button 630 has been pressed, the unregistered adverb degree value assigning unit 350 determines that the user has checked the degree value of the adverb.

[0059] If the degree value is not to be modified, the user interface screen 600 disappears by directly pressing the exit button 630 without pressing the modify button 620. In the example of Fig. 6, the color display of the newly registered adverb (yatara, chokochoko) disappears the next time the user interface screen 600 is displayed.

[0060] FIG. 7 is a diagram explaining the user confirmation function of the adverb dictionary in each embodiment. A dictionary table 700 of an adverb dictionary (for example, the adverb dictionary 32 in FIG. 2) has the items of "adverb," ​​"degree value," and "check history." An item indicating "check history" (True / False flag) is added to the dictionary table 700. "True" indicates that the user has confirmed the degree value assigned to the adverb, and "False" indicates that the user has not yet confirmed the degree value assigned to the adverb. When the degree value has not been confirmed (False), the user confirms and judges the degree value of the adverb.

[0061] Fig. 7 shows an example (within a thick frame) in which the degree value "1.5" assigned to the adverb "yatara" and the degree value "0.8" assigned to the adverb "chocochoco" have not yet been confirmed. In Fig. 6 described above, when the end button 630 is pressed, the unregistered adverb degree value assigning unit 350 determines that confirmation or correction of the degree values ​​has been completed, and switches the "check history" in the dictionary table 700 in Fig. 7 from "False" to "True."

[0062] Here, the updating of the adverb dictionary and the handling of the True / False flag when updating the distributed representation model 510 shown in FIG. 5 will be described. (1) The adverb dictionary is not updated when the distributed representation model is updated. In this case, the True / False flag does not change. Therefore, the degree value is calculated using the new distributed representation model only for adverbs newly registered after the update. (2) When the degree values ​​of the adverb dictionary are updated using a new distributed representation model, the True / False flag is changed. As described above, a reference adverb dictionary (a set of adverbs and degree values) is prepared, and the degree values ​​of the adverbs are updated based on the new distributed representation model. At the timing of this dictionary update, the reference adverb dictionary may be manually reviewed. In that case, the degree values ​​of the adverbs other than those reviewed manually are updated based on the new distributed representation model. The True / False flag of the adverb whose degree value has been updated is changed to "False."

[0063] 8 is a diagram showing an example of a user interface screen of an adverb dictionary maintenance tool 800 according to each embodiment. The adverb dictionary maintenance tool 800 is a user interface for correcting automatically assigned degree values.

[0064] The adverb dictionary improvement tool 800 includes an adverb display field 810, a degree value input field 820, a decision button 830, an adverb degree value display area 840, and a check box 850. In the example of FIG. 8, the adverb degree value display area 840 displays "not at all" with a degree value of 0.01, "slightly" with a degree value of 0.30, "fairly" with a degree value of 0.70, "gradually" with a degree value of 1.00, "hurriedly" with a degree value of 1.20, "excessively" with a degree value of 1.50, and "very" with a degree value of 1.80 on a number line. In this way, in the adverb degree value display area 840, a plurality of adverbs and degree values ​​are arranged in a line according to the degree values. By arranging a plurality of adverbs and degree values ​​in a line, the user can compare a target adverb with other adverbs, and it becomes easier to input the degree value of the target adverb.

[0065] Here, an example is shown in which a plurality of adverbs and degree values ​​are arranged in a line according to the degree values, but the present invention is not limited to this example. For example, it is sufficient if the relationship between the plurality of adverbs and the degree values ​​is easily confirmed by arranging the plurality of adverbs and the degree values ​​according to the degree values. As an example, it is possible to arrange the plurality of adverbs and the degree values ​​in a ring or semicircle according to the degree values. Also, in the adverb degree value display area 840, the adverbs may be displayed in order according to the degree values, and the degree values ​​(numeric values) may not be displayed. Furthermore, the adverb degree value display area 840 may be deleted from the adverb dictionary development tool 800.

[0066] "Immediately" is displayed in adverb display field 810, and "1.2" is displayed in degree value input field 820. When enter button 830 is pressed here, the degree value "1.2" for "immediately" is confirmed, and "immediately" and the degree value "1.2" are registered in the adverb dictionary.

[0067] Furthermore, when a check mark is entered in the "No entry only" checkbox 850, adverbs with no degree value entered (check history is in "False" state) are displayed in the adverb display field 810. For example, in the examples of Figs. 6 and 7, "Yatarai" and "Chokochoko" are displayed. For example, the adverb to be displayed can be selected by operating the pull-down menu button in the adverb display field 810, and the degree value is displayed in the degree value input field 820 accordingly. In the example of Fig. 8, a degree value of "1.50" has already been assigned to "Yatarai."

[0068] 8, this embodiment is provided with a correction interface (adverb dictionary maintenance tool 800) that allows a user to correct adverbs and degree values ​​in an adverb dictionary by a user operation. The correction interface has an adverb display area (adverb display field 810) that displays a target adverb, a degree value input area (degree value input field 820) for inputting a degree value corresponding to the target adverb, and a decision button.

[0069] In this embodiment equipped with such an adverb dictionary maintenance tool 800, the degree value of an adverb can be updated at any timing, taking into consideration the passage of words over time. It is also possible to register new adverbs that are not registered in the adverb dictionary, and to change the automatically assigned degree value. By unchecking the "Unentered only" checkbox 850, it is possible to display (confirm) registered adverbs and modify their degree values.

[0070] [Hardware configuration of the control system of the adverb dictionary update device] Here, the hardware configuration of the control system of the adverb dictionary updating device according to each embodiment of the present invention will be described with reference to FIG.

[0071] Fig. 9 is a block diagram showing an example of the hardware configuration of a control system of an adverb dictionary updating device. A calculator 900 shown in Fig. 9 is an example of hardware used as a computer. In the adverb dictionary updating devices 300, 300A according to this embodiment, the calculator 900 (computer) executes a program to realize an adverb dictionary updating method performed by the respective functional blocks shown in Fig. 3 in cooperation with each other.

[0072] The computer 900 includes a CPU (Central Processing Unit) 901, a ROM (Read Only Memory) 902, and a RAM (Random Access Memory) 903, all connected to a system bus. The computer 900 further includes a display unit 904, an operation unit 905, a non-volatile storage 906, and a communication interface 907.

[0073] The CPU 901 reads out program code of software for realizing each function according to this embodiment from the ROM 902, loads it into the RAM 903, and executes it. Variables, parameters, etc. generated during the calculation processing of the CPU 901 are temporarily written into the RAM 903, and these variables, parameters, etc. are read out by the CPU 901 as appropriate. The functions of each functional block in the adverb dictionary updating device 300, 300A are realized by the CPU 901 executing the program code read out from the ROM 902. However, other processors such as an MPU (Micro Processing Unit) may be used instead of the CPU 901.

[0074] The functions of text data acquisition unit 310, text data adverb extraction unit 330, unregistered adverb degree value assignment unit 350, and text data emotion degree value calculation unit 370 of the above-mentioned adverb dictionary updating devices 300, 300A (FIGS. 3 and 5) are realized by CPU 901 executing a program.

[0075] The display unit 904 is a monitor such as a liquid crystal display, and displays a GUI screen, results of processing performed by the CPU 901, etc. The operation unit 905 generates an input signal according to a user's operation and outputs it to the CPU 901. For example, a mouse, a keyboard, etc. are used for the operation unit 905, and the user can input information and instructions by operating the operation unit 905. The display unit 904 and the operation unit 905 may be integrated into a touch panel. The display unit 904 and the operation unit 905 are used to display and operate a user interface screen 600 (FIG. 6) for checking the degree values ​​of adverbs, an adverb dictionary development tool 800 (FIG. 8), etc.

[0076] The non-volatile storage 906 is an example of a recording medium, and is capable of storing data used by a program, data obtained by executing a program, and the like. The text set storage unit 320, the adverb set storage unit 340 related to text data, the adverb dictionary storage unit 360, and the emotion degree value set storage unit 380 related to text data of the above-mentioned adverb dictionary updating device 300, 300A (FIGS. 3 and 5) are configured using the non-volatile storage 906. In addition, an OS (Operating System) and a program executed by the CPU 901 may be recorded in the non-volatile storage 906. As the non-volatile storage 906, a hard disk drive (HDD), a solid state drive (SSD), a disk medium using light or magnetism, a semiconductor memory card, or the like is used. Note that a cloud environment can also be utilized to store the acquired text data and data obtained by processing the text data.

[0077] For example, a NIC (Network Interface Card) is used as the communication interface 907. The communication interface 907 is configured to be capable of transmitting and receiving various data to and from an external device via a communication network such as a LAN or the Internet to which a terminal is connected, or a dedicated line, etc. The function of the text data acquisition unit 310 of the above-mentioned adverb dictionary updating device 300, 300A is realized by using the communication interface 907.

[0078] <Modification> The present invention is not limited to the above-described embodiment, and various other applications and modifications are possible without departing from the gist of the present invention described in the claims. For example, the above-described embodiment is a detailed and specific description of the configuration in order to easily explain the present invention, and is not necessarily limited to the configuration including all of the components described. In addition, it is also possible to add, replace, or delete other components to a part of the configuration of each embodiment.

[0079] In addition, the above-mentioned configurations, functions, processing units, etc. may be realized in part or in whole by hardware, for example, by designing them as integrated circuits. As the hardware, a broad processor device such as an FPGA (Field Programmable Gate Array) or an ASIC (Application Specific Integrated Circuit) may be used.

[0080] Furthermore, each component of the adverb dictionary updating device according to the above-mentioned embodiment may be implemented in any hardware as long as each hardware can transmit and receive information to and from each other via a network. Furthermore, the processing performed by a certain processing unit may be realized by a single piece of hardware, or may be realized by distributed processing using multiple pieces of hardware.

[0081] In the above-described embodiment, the control lines and information lines are those that are considered necessary for the explanation, and not all control lines and information lines in the product are necessarily shown. In reality, it may be considered that almost all components are connected to each other. [Explanation of symbols]

[0082] 11...Emotion analysis device analysis result, 200...Emotion degree estimation method, 300, 300A...Adverb dictionary update device, 305...Text media, 310...Text data acquisition unit, 320...Text set storage unit, 330...Text data adverb extraction unit, 340...Adverb set storage unit, 350...Unregistered adverb degree value assignment unit, 360...Adverb dictionary storage unit, 370...Text data emotion degree value calculation unit, 380...Emotion degree value set storage unit, 510...Distributed representation model, 520...Distributed representation model update unit, 600...User interface screen for checking degree value, 700...Dictionary table, 800...Adverb dictionary maintenance tool

Claims

1. An adverb dictionary updating device for updating an adverb dictionary used for emotion degree estimation, comprising: a text data acquisition unit that acquires text data including a target of the emotion degree estimation and stores the text data in a text set storage unit; a text data adverb extraction unit that extracts adverbs related to each text data from the text set storage unit and stores the adverbs in an adverb set storage unit; an unregistered adverb degree value assigning unit that reads a list of extracted adverbs from the adverb set storage unit and an adverb dictionary having registered adverbs and degree values ​​as a table from an adverb dictionary storage unit, assigns degree values ​​to unregistered adverbs among the extracted adverbs by referring to the degree values ​​of the registered adverbs, updates the adverb dictionary, and stores the updated adverb dictionary in the adverb dictionary storage unit. Adverb dictionary update device.

2. The unregistered adverb degree value assigning unit calculates the similarity between the unregistered adverb and all the registered adverbs by using a distributed representation, and assigns the degree value of the registered adverb with the highest similarity to the degree value of the unregistered adverb.

2. The adverb dictionary updating device according to claim 1.

3. The unregistered adverb degree value assigning unit compares the similarity between the unregistered adverb and the most similar registered adverb with a predetermined threshold, and if the similarity is equal to or greater than the threshold, assigns the degree value of the most similar registered adverb to the degree value of the unregistered adverb, and if the similarity is less than the threshold, assigns a predetermined reference value to the degree value of the unregistered adverb.

3. The adverb dictionary updating device according to claim 2.

4. A distributed representation model that realizes distributed representation of adverbs; a distributed representation model update unit that updates the distributed representation model.

3. An adverb dictionary updating device according to claim 1 or 2.

5. a modification interface for modifying the adverbs and the degree values ​​in the adverb dictionary by a user's operation; The correction interface has an adverb display area for displaying a target adverb, a degree value input area for inputting a degree value corresponding to the target adverb, and a decision button.

4. An adverb dictionary updating device according to claim 1.

6. The correction interface further includes an area in which a plurality of adverbs and degree values ​​are arranged according to the degree values.

6. An adverb dictionary updating device according to claim 5.

7. a text data emotion degree calculation unit that reads each piece of text data from the text set storage unit, an extracted adverb from the adverb set storage unit, and the adverb dictionary from the adverb dictionary storage unit, calculates an emotion degree value for each piece of text data, and stores each piece of text data, the extracted adverb, and the emotion degree value in an emotion degree value set storage unit.

2. The adverb dictionary updating device according to claim 1.

8. An adverb dictionary updating method for updating an adverb dictionary used in emotion degree estimation, comprising: A process of acquiring text data including a target of the emotion degree estimation and storing the data in a text set storage unit; A process of extracting adverbs relating to each piece of text data from the text set storage unit and storing the adverb set storage unit; a process of reading a list of extracted adverbs from the adverb set storage unit and an adverb dictionary having registered adverbs and degree values ​​as a table from an adverb dictionary storage unit, referencing the degree values ​​of the registered adverbs to assign degree values ​​to unregistered adverbs among the extracted adverbs, updating the adverb dictionary, and storing the updated adverb dictionary in the adverb dictionary storage unit; a process of reading each piece of text data from said text set storage unit, an extracted adverb from said adverb set storage unit, and said adverb dictionary from said adverb dictionary storage unit, calculating an emotion degree value for each piece of text data, and storing each piece of text data, the extracted adverb, and the emotion degree value in an emotion degree value set storage unit. How to update the adverb dictionary.

Images