Text processing device, text processing method, and program

Abstract

To provide a text processing apparatus, or the like, capable of obtaining a result of classifying words according to a request from a user while reducing calculation cost.SOLUTION: A text processing apparatus 10 includes: a display unit 140 which displays a range, in a text, of a target word which is classified into a target category out of multiple categories, of words included in the text, by using each of degrees of probability that each of the words included in the text belongs to the multiple categories; a modification receiving unit 150 which receives an instruction to modify the degree of probability of an adjustment category which is different from the target category, out of the multiple categories; a modification unit 160 which modifies the degree of probability of the adjustment category in accordance with the received modification instruction; and a classification unit 130 which re-classifies the words into one of the multiple categories using each of the degrees of probability of the multiple categories other than the adjustment category and the modified degree of probability of the adjustment category.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] This disclosure relates to a technique for processing text.

Background Art

[0002] When manually extracting phrases of a predetermined category from text, it is laborious. And as the amount of text increases, the labor required for extraction also increases. A technique for classifying words into categories is disclosed, for example, in Patent Document 1.

[0003] Patent Document 1 describes an expression classification device that performs relearning of an inference model for classifying a plurality of named expressions extracted from document information into categories. The named expressions in Patent Document 1 are a general term for words representing specific things. The named expression classification device in Patent Document 1 corrects the weights of the feature amounts used by the inference model in classification based on the correct / error information received from the user, and performs relearning of the inference model based on the corrected weights.

Prior Art Documents

Patent Documents

[0004]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0005] In the technique of Patent Document 1, when changing the classification result, relearning of the inference model is required. Relearning of the inference model requires computational cost. In the technique of Patent Document 1, the computational cost during relearning of the inference model cannot be reduced.

[0006] One object of this disclosure is to provide a text processing device or the like that can obtain a classification result of phrases according to user desires while reducing computational cost.

Means for Solving the Problems

[0007] A text processing device according to one aspect of the present disclosure includes: display means for displaying the range in the text of target phrases that have been classified into a target category from among the plurality of categories using each of the degrees of probability that the phrase belongs to a plurality of categories; change receiving means for receiving instructions to change the degree of probability of adjustment categories that are different from the target category from among the plurality of categories; change means for changing the degree of probability of the adjustment category in accordance with the received change instructions; and classification means for reclassifying the phrase into any of the plurality of categories using each of the degrees of probability of the plurality of categories other than the adjustment category and the changed degree of probability of the adjustment category.

[0008] A text processing method according to one aspect of the present disclosure includes: displaying the range in the text of target terms that are classified into a target category among the multiple categories, using each of the degrees of probability that the terms belong to a plurality of categories; receiving instructions to change the degree of probability of adjustment categories that are different from the target category among the multiple categories; changing the degree of probability of the adjustment categories in accordance with the received instructions to change; and reclassifying the terms into any of the multiple categories using each of the degrees of probability of the other categories and the changed degree of probability of the adjustment categories.

[0009] A program according to one aspect of the present disclosure causes a computer to perform the following: a display process that displays the range in the text of target terms that have been classified into a target category among the plurality of categories using each of the degrees of probability that the terms belong to a plurality of categories; a change acceptance process that accepts instructions to change the degree of probability of adjustment categories that are different from the target category among the plurality of categories; a change process that changes the degree of probability of the adjustment categories in accordance with the received change instructions; and a classification process that reclassifies the terms into any of the plurality of categories using each of the degrees of probability of the plurality of categories other than the adjustment categories and the changed degree of probability of the adjustment categories. [Effects of the Invention]

[0010] This disclosure has the effect of reducing computational costs while obtaining results for classifying words and phrases according to the user's requirements. [Brief explanation of the drawing]

[0011] [Figure 1] Figure 1 is a block diagram showing an example of the configuration of a text processing device according to the first embodiment of this disclosure. [Figure 2] Figure 2 is a flowchart illustrating an example of the operation of a text processing device according to the first embodiment of this disclosure. [Figure 3] Figure 3 is a block diagram showing an example of the configuration of a text processing device according to a second embodiment of the present disclosure. [Figure 4] Figure 4 is a flowchart illustrating an example of the operation of a text processing device according to the second and third embodiments of this disclosure. [Figure 5] Figure 5 is a flowchart illustrating an example of the operation of a text processing device according to the second and third embodiments of this disclosure. [Figure 6] Figure 6 is a block diagram showing an example of the configuration of a text processing device according to the third embodiment of this disclosure. [Figure 7]Figure 7 is a flowchart illustrating an example of the operation of a text processing device according to the third embodiment of this disclosure for determining the initial value of a coefficient. [Figure 8] Figure 8 is a diagram showing an example of a computer hardware configuration that can realize a text processing device according to the embodiment of this disclosure. [Modes for carrying out the invention]

[0012] Embodiments of this disclosure will be described in detail below with reference to the drawings.

[0013] <First Embodiment> First, a first embodiment of this disclosure will be described in detail with reference to the drawings.

[0014] <Structure> Figure 1 is a block diagram showing an example of the configuration of a text processing device according to the first embodiment of the present disclosure. In the example shown in Figure 1, the text processing device 10 of this embodiment includes a display unit 140, a change reception unit 150, a change unit 160, and a classification unit 130.

[0015] The display unit 140 displays the range in the text of target terms that have been classified into the target category from among the multiple categories, using each of the degrees of probability that the terms belong to one of the multiple categories. The change reception unit 150 receives instructions to change the degree of probability of adjustment categories that are different from the target category from among the multiple categories. The change unit 160 changes the degree of probability of the adjustment categories according to the received change instructions. The classification unit 130 reclassifies the terms into one of the multiple categories using each of the degrees of probability of the multiple categories other than the adjustment categories and the changed degree of probability of the adjustment categories.

[0016] In this embodiment, the text is pre-segmented into phrases using any of the existing language recognition technologies. And for each of the plurality of categories, the degree of possibility that each phrase belongs to the category has been pre-determined using the existing determination method.

[0017] The above-mentioned phrases are, for example, words. The phrase may be a collection of multiple words. The collection of multiple words may be, for example, one sentence. The collection of multiple words may also be a part of one sentence. The collection of multiple words may be a collection of multiple sentences. The phrase is not limited to these examples. The phrase is, for example, a meaningful part in the text.

[0018] Each of the plurality of categories is, for example, a group of phrases of the same type. The type of phrase may include, for example, at least any one of place names, personal names, organization names, and job titles, etc. The type of phrase may include the starting word of a place name and the continuing word of a place name. The starting word of a place name represents the part that identifies the region existing at the beginning of a compound word containing a part representing a region. The continuing word of a place name represents the part other than the part that identifies the region existing at the beginning of a compound word containing a part representing a region (for example, honorifics or job titles, etc.). For example, when the place name is "AA Park", the starting word of the place name is the part "AA", and the continuing word of the place name is "Park". The type of phrase may include the starting word of a personal name and the continuing word of a personal name. The starting word of a personal name represents the part that identifies the person existing at the beginning of a compound word containing a part representing a person. The continuing word of a person represents the part other than the part that identifies the person existing at the beginning of a compound word containing a part representing a person (for example, honorifics or job titles, etc.).

[0019] The type of phrase may be, for example, at least any one of a request, a claim, and an inquiry, etc.

[0020] The degree of possibility that a phrase belongs to a category is represented, for example, by a numerical value of the probability that the phrase belongs to the category. The degree of possibility that a phrase belongs to a category may be, for example, a score indicating the magnitude of the probability that the phrase belongs to the category.

[0021] The likelihood that a word or phrase belongs to a category is predetermined by the classification unit 130, for example, using one of the existing determination methods as described above. In this case, the classification unit 130 may, for example, receive information about words or phrases contained in a text as input and use a determination model that determines the likelihood that the received words or phrases belong to a category to determine the likelihood that the words or phrases belong to a category for each of multiple categories.

[0022] The words are then classified into one of several categories using a predetermined degree of likelihood that the word belongs to a category. For example, the words are classified into the category to which they are most likely to belong, as indicated by the determined degree of likelihood.

[0023] <Display section 140> The display unit 140 displays the range of text corresponding to the words classified into the target category. The target category may be predetermined. The display unit 140 displays text in which the format of the part of the words classified into the target category is set to a different format from the other parts. In this case, the format may be at least one of the following: the color of the characters, the font, and the background color.

[0024] <Change Request Department 150> The change reception unit 150 receives a change instruction that includes the content of the change regarding the degree of possibility of the adjustment category among several categories. As described later, the display unit 140 may display a user interface (e.g., a slider bar) for specifying the content of the change on a screen that displays, for example, the text range of a phrase classified into the target category. The area where the user interface for specifying the content of the change is displayed may be set as an input area that accepts input from the user. The display area of ​​the displayed user interface may display a display (e.g., a scale) that shows the values ​​of multiple coefficients. The change reception unit 150 receives the input indicating the content of the change made by the user to the display of the displayed user interface (e.g., the input area described above) as a change instruction. Specifically, the change reception unit 150 accepts the specification of a position for the input area as input. The change reception unit 150 then identifies the value associated with the position in the input area, which is identified by the received position specification, as the coefficient value. In this case, the coefficient value corresponding to the value shown in the displayed user interface is pre-associated with the pixels included in the input area. The nature of the change may be represented, for example, by a coefficient value multiplied by a value representing the degree of possibility of the adjustment category.

[0025] <Changed section 160> The modification unit 160 modifies the degree of probability of the adjustment category for each phrase according to the content of the change to the degree of probability of the adjustment category as represented by the received modification instructions. If the content of the change is represented by a coefficient value multiplied by the value representing the degree of probability of the adjustment category, the modification unit 160 modifies the value representing the degree of probability of the adjustment category by multiplying the value representing the degree of probability of the adjustment category by the value of that coefficient.

[0026] <Classification section 130> The classification unit 130 reclassifies each phrase into the category to which it is most likely to belong, using the probability levels of multiple categories, including the probability of the adjusted category being changed. In other words, the classification unit 130 selects the category that is most likely to belong among the probability levels of multiple categories, including the probability of the adjusted category being changed, and reclassifies the phrase into the selected category. The classification unit 130 reclassifies each phrase contained in the text into the category to which it is most likely to belong.

[0027] <Operation> Figure 2 is a flowchart illustrating an example of the operation of the text processing device 10 according to this embodiment. In the example shown in Figure 2, first, the display unit 140 displays the range of words classified into the target category from among multiple categories within the divided text (step S11). The display unit 140 displays the range of words classified into the target category on, for example, the display of the text processing device 10, or on the display of a terminal device that is communicably connected to the text processing device 10 and used by the user. As described above, the display unit 140 may display the text in a format in which the format of the words classified into the target category differs from the format of the other parts. In this case, the words classified into the target category may be displayed in a legible format.

[0028] Next, the change reception unit 150 receives the change instruction (step S12). The change unit 160 changes the degree of possibility of the adjustment category according to the change instruction received by the change reception unit 150 (step S13). Then, the classification unit 130 reclassifies the phrase into categories using the changed degree of possibility of multiple categories, including the degree of possibility of the adjustment category (step S14).

[0029] <Effects> This embodiment has the effect of reducing computational costs while obtaining the results of word classification according to the user's requests. The reason for this is that the change reception unit 150 receives input for the display of the results of the word classification into category determined for each of the multiple categories, and the changes are instructions This is because it accepts the change instructions. The change unit 160 then changes the likelihood that the phrase belongs to an adjustment category according to the received change instructions. Furthermore, the classification unit 130 uses the likelihood that the phrase belongs to a category for each of the multiple categories after the likelihood of the adjustment category has been changed to classify the phrase into one of the multiple categories. This eliminates the need to re-determine the likelihood that the phrase belongs to a category for each of the multiple categories. Therefore, computational costs are reduced. Also, as described above, the classification unit 130 uses the likelihood that the phrase belongs to a category for each of the multiple categories, including the likelihood of the adjustment category which has been changed according to the received change instructions, to reclassify the phrase into a category. Therefore, the result of classifying the phrase according to the user's request can be obtained.

[0030] <Second Embodiment> Next, a second embodiment of this disclosure will be described in detail with reference to the drawings.

[0031] <Structure> Figure 3 is a block diagram showing an example of the configuration of the text processing device 100 according to this embodiment. In the example shown in Figure 3, the text processing device 100 of this embodiment includes a target reception unit 110, a division unit 120, a classification unit 130, a display unit 140, a change reception unit 150, a change unit 160, and an output unit 170. The classification unit 130, display unit 140, change reception unit 150, and change unit 160 of this embodiment each have the same functions as the classification unit 130, display unit 140, change reception unit 150, and change unit 160 of the first embodiment. The classification unit 130, display unit 140, change reception unit 150, and change unit 160 of this embodiment each perform the same operations as the classification unit 130, display unit 140, change reception unit 150, and change unit 160 of the first embodiment. The following will mainly describe the differences between this embodiment and the first embodiment.

[0032] <Target reception department 110> The target reception unit 110 receives the text to be processed. The target reception unit 110 may receive the text to be processed from, for example, a server device, another information processing device, or a terminal device used by a user. The target reception unit 110 sends the received text to the splitting unit 120.

[0033] <Divided part 120> The splitting unit 120 receives the text to be processed from the target receiving unit 110. The splitting unit 120 divides the received text into words and phrases using existing language processing techniques. The splitting unit 120 generates information representing the words and phrases obtained by dividing the text. The splitting unit 120 sends the received text and the information representing the words and phrases obtained by dividing the text to the classification unit 130.

[0034] The information representing the words and phrases obtained by dividing the text is information that is appropriately defined so as to identify each word and phrase contained in the text. The information representing the words and phrases may include, for example, information indicating the boundaries of words and phrases in the text. The information representing the words and phrases may also be information that can identify the string representing the word and phrase and the order in which that word and phrase appear in the text. The information representing the words and phrases may also be data in which the strings representing the words and phrases are arranged in the order in which they appear in the text. The information representing the words and phrases may include, for example, a combination of the string representing the word and phrase and a word and phrase identifier (hereinafter also referred to as a word and phrase identifier), and information that represents the relationship between the word and phrase identifier and the order in which that word and phrase appear in the text. The word and phrase identifier may be information that can identify that word and phrase within the text that contains that word and phrase. The word and phrase identifier may also indicate the order in which that word and phrase appear in the text. For example, the division unit 120 assigns word and phrase identifiers to the words and phrases obtained by dividing the text.

[0035] <Classification section 130> The classification unit 130 receives text and information representing the words obtained by dividing the text from the division unit 120. The classification unit 130 uses a determination model generated to determine the degree to which a word is likely to belong to each of a predetermined set of categories, to determine the degree to which a word is likely to belong to each of the predetermined set of categories. The determination model is a determination that is pre-generated by learning using one of the existing learning methods to receive a word and determine the degree to which the received word is likely to belong to each of the predetermined set of categories. The classification unit 130 determines the degree to which a word is likely to belong to each of the predetermined set of categories, for each of the words obtained by dividing the text.

[0036] The classification unit 130 first classifies a phrase into one of several categories, using the degree of likelihood that the phrase belongs to each of the multiple categories. The classification unit 130 classifies each of the phrases into which the text has been divided into one of several categories. The classification unit 130 generates information about the category to which the phrase has been classified. The identifiers for each of the multiple categories (hereinafter referred to as category identifiers) may be predetermined. The information about the category to which the phrase has been classified is, for example, information that includes a combination of the phrase identifier of each phrase contained in the text and the category identifier of the category to which the phrase has been classified.

[0037] The classification unit 130 sends the text, information about the words and phrases into which the text has been divided, and information about the categories into which the words and phrases have been classified to the display unit 140.

[0038] Furthermore, the classification unit 130 sends to the modification unit 160 the degree to which each word or phrase in the text is likely to belong to the adjustment category.

[0039] The classification unit 130 receives from the modification unit 160 the degree of probability that each of the words in the text, as modified by the modification unit 160, belongs to an adjustment category. When the classification unit 130 receives the degree of probability that a word belongs to an adjustment category from the modification unit 160, it reclassifies the word into a category using the degree of probability that the word belongs to a category for multiple categories, including the degree of probability that the adjustment category has been modified. The degree of probability that the adjustment category has been modified represents the degree of probability that the word belongs to an adjustment category as modified by the modification unit 160. When reclassifying a word into a category, the classification unit 130 uses the degree of probability that the word belongs to an adjustment category received from the modification unit 160 instead of the determined degree of probability that the word belongs to an adjustment category.

[0040] The classification unit 130 may receive a completion instruction from the modification unit 160 (more specifically, from the modification reception unit 150 via the modification unit 160). Upon receiving a completion instruction, the classification unit 130 sends the text, information about the words and phrases contained in the text, and information about the category to which the words and phrases are classified to the output unit 170.

[0041] <Display section 140> The display unit 140 receives text, information on the words and phrases into which the text has been divided, and information on the categories into which the words and phrases have been classified, from the classification unit 130. The display unit 140 displays the range in the text of the words and phrases classified into the target category in the same manner as the display unit 140 of the first embodiment.

[0042] <Change Request Department 150> The change reception unit 150 receives a change instruction, including the content of the change to the degree of possibility of an adjustment category among multiple categories, in the same manner as the change reception unit 150 of the first embodiment, after the display unit 140 has displayed the range in text of the words classified into the target category. The change reception unit 150 sends the received change instruction to the change unit 160. The change reception unit 150 may receive a change instruction one or more times. Each time the change reception unit 150 receives a change instruction, it sends the received change instruction to the change unit 160.

[0043] The change reception unit 150 may receive an instruction indicating that there are no changes, in other words, an instruction indicating that the changes have been completed (hereinafter also referred to as a completion instruction). The completion instruction may also be an output instruction. When a completion instruction is received, the change reception unit 150 sends the completion instruction to the classification unit 130, for example, via the change unit 160. Specifically, the change reception unit 150 sends the completion instruction to the change unit 160. The change unit 160 then sends the received completion instruction to the classification unit 130.

[0044] <Changed section 160> The modification unit 160 receives from the classification unit 130 the degree of probability that each word or phrase in the text belongs to an adjustment category. The modification unit 160 further receives a modification instruction from the modification reception unit 150. As described above, the modification instruction includes, for example, a coefficient value that is multiplied by the degree of probability value of the adjustment category.

[0045] The modification unit 160 modifies the degree to which each phrase in the text is likely to belong to the adjustment category, in the same manner as the modification unit 160 of the first embodiment, according to the received modification instructions. Specifically, as described above, the modification unit 160 takes the product of the value of the degree to which each phrase in the text is likely to belong to the adjustment category and the value of the coefficient included in the modification instructions as the value of the degree to which that phrase is likely to belong to the adjustment category. The modification unit 160 sends the modified degree to which each phrase in the text is likely to belong to the adjustment category to the classification unit 130.

[0046] Furthermore, the modification unit 160 may receive a completion instruction from the modification reception unit 150. The modification unit 160 sends the received completion instruction to the classification unit 130.

[0047] <Output section 170> The output unit 170 receives text, information about the words and phrases contained in the text, and information about the categories to which the words and phrases are classified, from the classification unit 130.

[0048] The output unit 170 outputs the range in which words classified into the target category exist. The output unit 170 may output the text in a way that allows identification of the range of words classified into the target category within the text. In this case, the output unit 170 may output the text in a way that does not allow identification of the content of the words classified into the target category. The output unit 170 may output the text in a way that allows identification of the range of words classified into the target category within the text and the content of the words classified into the target category. The output unit 170 may output only the range of words classified into the target category within the text. In this case, the output unit 170 does not need to output the parts of the text other than the words classified into the target category.

[0049] Specifically, the output unit 170 displays text in which, for example, the format of the part of the word classified into the target category is set to a different format from the format of the other parts. In this case, the format is, for example, at least one of the following: the color of the characters, the font, and the background color. Note that the format of the part of the word classified into the target category set by the output unit 170 may be the same as the format of the part of the word classified into the target category set by the display unit 140. The format of the part of the word classified into the target category set by the output unit 170 does not need to be the same as the format of the part of the word classified into the target category set by the display unit 140.

[0050] The output unit 170 may display text in a way that makes the content of the words classified into the target category invisible, but allows the range of the words classified into the target category to be visible. An example of such display is blacking out, where the range of words to be hidden is filled in black. The output unit 170 may also display text in which the words classified into the target category are replaced with black rectangles. The color of the rectangles that replace the words classified into the target category is not limited to black.

[0051] Other examples of output from the output unit 170 will be described in detail later.

[0052] <Operation> Next, the operation of the text processing device 100 according to the second embodiment of this disclosure will be described in detail with reference to the drawings.

[0053] Figures 4 and 5 are flowcharts illustrating an example of the operation of the text processing device 100 according to this embodiment. In the example shown in Figure 4, first, the target receiving unit 110 receives the target text, which is the text to be processed (step S101). Next, the splitting unit 120 splits the target text into words and phrases (step S102). Then, the classification unit 130 determines the degree of probability that each word or phrase belongs to a category for multiple categories (step S103). Furthermore, the classification unit 130 uses the estimated degree of probability to classify the words and phrases into categories (step S104). Next, the text processing device 100 performs the operation shown in Figure 5.

[0054] In the example shown in Figure 5, the display unit 140 displays the range of words belonging to the target category among multiple categories (step S105). In addition to the range of words, the display unit 140 also displays a user interface, such as a slider bar for inputting instructions for changes.

[0055] The change reception unit 150 receives the change instruction (i.e., the change instruction described above) entered by the user using the user interface described above (step S106).

[0056] If a change instruction is received (YES in step S107), the change unit 160 changes the likelihood of the adjustment category according to the change instruction (step S108). In this explanation, the likelihood of the adjustment category represents the likelihood that a phrase belongs to the adjustment category. In step S108, the change unit 160 changes the likelihood of each phrase in the text received for processing according to the change instruction.

[0057] The classification unit 130 reclassifies the words into categories using the degree of probability for each of the multiple categories (step S109). Specifically, for each word in the text to be processed, the classification unit 130 selects the category to which the word belongs from among the multiple categories, which indicates the highest probability that the word belongs to that category. The classification unit 130 classifies the word into the selected category. Then, the operation of the text processing device 100 returns to step S105. In this way, the text processing device 100 repeats the operations from step S105 to step S109 until no further instructions for changes are input.

[0058] If a change instruction is not accepted (NO in step S107), the output unit 170 outputs the range in which the words classified into the target category exist (step S110). For example, if a completion instruction is accepted instead of a change instruction, the change acceptance unit 150 sends the accepted completion instruction to the classification unit 130, for example, via the change unit 160. When the classification unit 130 receives the completion instruction, it sends the text, the information of the words contained in the text, and the information of the category in which the words are classified to the output unit 170. The output unit 170 outputs the range in which the words classified into the target category exist.

[0059] <Effects> This embodiment has the same effects as the first embodiment. The reason for this is the same as the reason for the effects of the first embodiment.

[0060] Furthermore, generally speaking, when classifying a word or phrase into one of several categories, there is a trade-off between the rate of misclassification and the rate of unclassified entries. In other words, reducing misclassification tends to increase unclassified entries, and conversely, reducing unclassified entries tends to increase misclassification. Misclassification of a word or phrase occurs when a word or phrase that does not actually belong to a category is incorrectly classified into that category. Unclassified entries occur when a word or phrase that actually belongs to a category is not classified into that category. In this embodiment, the user can adjust the state of misclassification and unclassified entries to a desired state by, for example, using a user interface while visually observing the classification results.

[0061] <Other examples of output> The following describes other examples of output from the output unit 170.

[0062] For example, if the target category is a phrase that is to be concealed, the output unit 170 may replace the phrase classified as the target category with another string (hereinafter also referred to as a replacement string). The output unit 170 may then output text in which the part of the phrase classified as the target category has been replaced with the replacement string. The output unit 170 may also display text in which the phrase classified as the target category has been replaced with the replacement string, and the format of the replaced part of the text is set to a different format from the format of the part other than the replacement string.

[0063] The replacement string may, for example, be a string consisting only of whitespace characters. The replacement string may be whitespace characters with at least one of the background color and / or text color set to appear as if the words are blacked out. The whitespace characters may be space characters. The whitespace characters may be control characters that set the position of characters in text that appear to have spacing between them.

[0064] There is no limit to the number of target categories. Multiple target categories may be set. The format of the words classified into each target category may differ for each category.

[0065] The replacement strings mentioned above may be predetermined strings for each target category. For example, if the target category is a place name, the replacement string may be a place name that has been pre-set as the replacement place name (e.g., City A). If the target category is a person's name, the replacement string may be a person's name that has been pre-set as the replacement person name (e.g., Mr. B). If the target category is a country name, the replacement string may be a country name that has been pre-set as the replacement country name (e.g., Country C). If the target category is an address, the replacement string may be an address that has been pre-set as the replacement address.

[0066] Furthermore, the output unit 170 may be configured to replace identical phrases with the same replacement string and different phrases with different replacement strings. In this case, multiple replacement strings ordered together may be pre-set for a single target category. In this case, the output unit 170 assigns, for example, a different replacement string selected from the multiple replacement strings set for the target category to each of the multiple types of phrases classified into the target category. The output unit 170 then replaces the phrases classified into the target category in the text with the replacement strings assigned to those phrases. If there are multiple target categories, multiple replacement strings may be pre-set for each of one or more of the target categories. Multiple target categories may include target categories in which classified phrases are replaced with the same replacement string.

[0067] For example, if the target category is requests, the output unit 170 may be configured to output only the words classified into the target category from the text. The output unit 170 may also be configured to output the text in a format different from the format of the other parts of the text, specifically the parts of the text that are classified into the target category.

[0068] <First modified example of the second embodiment> The display unit 140 may display a user interface for selecting a target category from multiple categories. Such a user interface may include, for example, the name of the category and a checkbox for specifying the category as the target category.

[0069] For example, the change reception unit 150 may accept the designation of a target category through a user interface for selecting a target category from multiple categories. The change reception unit 150 may accept the designation of multiple target categories. The change reception unit 150 sends information about the target category identified by the accepted designation to the classification unit 130, for example via the change unit 160.

[0070] The classification unit 130 identifies the target category based on the target category information received from the change acceptance unit 150 via the change unit 160, and uses that category as the new target category.

[0071] <Second modified example of the second embodiment> The following sections provide further explanations of examples of target texts and target categories.

[0072] The target text may be a portion of text in electronic data that can be recognized by a computer. For example, the target text may be a portion of text in a file format used by email client software, a word processor, or other applications.

[0073] The target text may be, for example, text (e.g., the text itself or the text portion of data containing text) that has been entered into an organization's (e.g., government ministry, local government, or company) server via a network. In this case, the target categories may include, for example, requests, questions, information provision (e.g., information provision regarding product defects, etc.), and accusations (e.g., accusations of fraud or suspected crimes).

[0074] The target text may be, for example, text posted on social networking services. In this case, the target categories may be, for example, evaluations, requests, questions, information provision (e.g., information on product defects, etc.), requests (e.g., requests for assistance during disasters), or information on abnormal situations (e.g., information on accidents and disasters, etc.).

[0075] The target texts are not limited to the examples above. The target categories are also not limited to the examples above.

[0076] <Third modified example of the second embodiment> The nature of the change may be represented, for example, by a value indicating the likelihood of the adjustment category being changed. The nature of the change may be represented by a value that falls within the range of the value representing the likelihood of the adjustment category being changed (i.e., a value between the minimum and maximum values). If the range of the value representing the likelihood of the adjustment category being changed (i.e., the minimum and maximum values) is known, the nature of the change may be represented by the ratio of the difference between the likelihood of the minimum value and the likelihood of the adjustment category being changed to the difference between the minimum and maximum values.

[0077] If the content of the change is represented by a value indicating the likelihood of the adjustment category being changed, the change unit 160 changes the value representing the likelihood of the adjustment category to the value represented by the content of the change. If the content of the change is represented by the ratio of the difference between the minimum value and the likelihood of the adjustment category after the change to the difference between the minimum value and the maximum value, the change unit 160 calculates the value represented by the ratio between the minimum value and the maximum value, which is represented by the content of the change. Then, the change unit 160 sets the calculated value to the likelihood of the adjustment category after the change.

[0078] Furthermore, if the change reception unit 150 is configured to accept the degree of possibility of the adjustment category after the change, the classification unit 130 does not need to send to the change unit 160 the degree of possibility that each word in the text belongs to the adjustment category. And the change unit 160 does not need to receive from the classification unit 130 the degree of possibility that each word in the text belongs to the adjustment category.

[0079] <Third Embodiment> Next, a third embodiment of the present disclosure will be described in detail with reference to the drawings.

[0080] <Structure> Figure 5 is a block diagram showing an example of the configuration of a text processing device 101 according to a third embodiment of the present disclosure. In the example shown in Figure 5, the text processing device 101 of this embodiment includes a target receiving unit 110, a splitting unit 120, a classification unit 130, a display unit 140, a change receiving unit 150, a change unit 160, an output unit 170, a result receiving unit 181, a result storage unit 182, and an initial value determination unit 183. The target receiving unit 110, splitting unit 120, classification unit 130, display unit 140, change receiving unit 150, change unit 160, and output unit 170 of this embodiment are the same as the target receiving unit 110, splitting unit 120, classification unit 130, display unit 140, change receiving unit 150, change unit 160, and output unit 170 of the second embodiment, except for the differences described below.

[0081] <Results Reception Department 181> The result receiving unit 181 receives the results of classifying words into one of several categories that have been recorded in advance (hereinafter also referred to as the results of word classification and the records of word classification). The results of word classification are data that represent the result of classifying words obtained by text segmentation for one or more texts into multiple categories. The results of word classification may also be data obtained by manually classifying words into categories. The results of word classification may also be data obtained by classifying words into categories by a computer operating according to some algorithm. The results of word classification may also be data obtained by manually correcting the results of classifying words into categories by a computer operating according to some algorithm. The results of word classification are data that includes, for example, information on the words from which the text has been segmented and information on the categories to which the words have been classified. The results of word classification may further include text.

[0082] The result receiving unit 181 stores the classification results of the received words in the result storage unit 182.

[0083] <Result storage section 182> The result storage unit 182 stores the results of word classification stored by the result receiving unit 181.

[0084] <Initial value determination unit 183> The initial value determination unit 183 reads the results of word classification from the result storage unit 182. The results of word classification represent a record of the classification of a word into one of several categories that has been obtained in advance. Using the read results of word classification (i.e., the record), the initial value determination unit 183 determines the initial value of the coefficient so that the results of word classification into the target category, using the initial value of the coefficient for the degree of probability of the adjustment category, will be closer to the results of word classification into the target category in this record. The classification of a word into one of several categories, using the initial value of the coefficient for the degree of probability of the adjustment category, represents the classification of a word where the product of the value representing the determined degree of probability of the adjustment category and the initial value of the coefficient is used as the value of the degree of probability of the adjustment category.

[0085] Specifically, first, the initial value determination unit 183 determines, in the same manner as the classification unit 130, the likelihood that the words included in the classification result of the read words belong to a category for multiple categories. The initial value determination unit 183 may send the words included in the classification result of the read words to the classification unit 130. Then, the initial value determination unit 183 may receive from the classification unit 130 the likelihood that the words included in the classification result of the read words belong to a category for multiple categories, as determined by the classification unit 130.

[0086] In this case, the classification unit 130 receives a word or phrase (specifically, data representing the word or phrase) from the initial value determination unit 183. The classification unit 130 then determines the likelihood that the received word or phrase belongs to one of the multiple categories. The classification unit 130 then sends the determined likelihood of the received word or phrase belonging to one of the multiple categories back to the initial value determination unit 183.

[0087] The initial value determination unit 183 then determines the initial value of the coefficient so that the result of classifying words into the target category, using the initial value of the coefficient for the degree of probability of the adjustment category, is brought closer to the result of classifying words into the target category in the result of classifying the read words. The initial value determination unit 183 may, for example, determine the initial value of the coefficient so that the sum of the number of words misclassified into the target category and the number of words not classified into the target category is minimized. Words misclassified into the target category are words that are not classified into the target category in the result of classifying the read words, but are classified into the target category in the result of classification using the degree of probability. Words not classified into the target category are words that are classified into the target category in the result of classifying the read words, but are not classified into the target category in the result of classification using the degree of probability. The classification using the degree of probability represents the classification of words into categories using the degree of probability of multiple categories, with the product of the degree of probability of the adjustment category and the initial value of the coefficient being included instead of the degree of probability of the adjustment category. The initial value determination unit 183 may count the number of words in this case by treating words consisting of the same string as one word, regardless of their frequency of appearance in the text. The initial value determination unit 183 may also count the number of words in this case by assigning a weight to words consisting of the same string according to their frequency of appearance in the text. The counting method may be predetermined.

[0088] The initial value determination unit 183 may determine the initial value of the coefficient using any of the existing calculation methods. For example, the initial value determination unit 183 may change the value of the coefficient from the minimum value to the maximum value of the coefficient at intervals of a predetermined value, and perform classification of words into categories, and calculate the sum of the number of words misclassified into the target category and the number of words not classified into the target category. The initial value determination unit 183 may then identify the minimum value of the sum of the number of words misclassified into the target category and the number of words not classified into the target category, and identify the value of the coefficient when the identified minimum value is calculated. The initial value determination unit 183 uses the identified value of the coefficient as the initial value of the coefficient. If multiple coefficient values ​​are identified for the minimum value of the sum of the number of words misclassified into the target category and the number of words not classified into the target category, the initial value determination unit 183 determines one value from the multiple coefficient values ​​as the initial value of the coefficient according to a predetermined method. In this case, the method could be, for example, identifying the minimum value of multiple coefficients, or identifying the maximum value of multiple coefficients. Other methods may be used to determine the initial value of the coefficients from the values ​​of multiple coefficients.

[0089] The initial value determination unit 183 sends the determined initial values ​​of the coefficients to the modification unit 160.

[0090] <Changed section 160> The modification unit 160 receives from the classification unit 130 the degree to which each word or phrase in the text is likely to belong to the adjustment category.

[0091] Furthermore, the modification unit 160 receives the initial values ​​of the coefficients from the initial value determination unit 183.

[0092] The modification unit 160 uses the initial value of the received coefficient to change the degree of probability that each word in the text belongs to the adjustment category. Specifically, for example, the modification unit 160 takes the product of the value of the degree of probability that each word in the text belongs to the adjustment category and the initial value of the coefficient as the value of the degree of probability that the word belongs to the adjustment category. The modification unit 160 sends the modified degree of probability that each word in the text belongs to the adjustment category to the classification unit 130.

[0093] The modification 160 of this embodiment is the same as the modification 160 of the second embodiment, except for the points described above (for example, receiving an initial value for the coefficient and using the received initial value for the coefficient to change the degree of possibility of the adjustment category).

[0094] <Classification section 130> In this embodiment, the classification unit 130 first sends to the modification unit 160 the degree to which each word or phrase contained in the text is likely to belong to the adjustment category.

[0095] The classification unit 130 receives from the modification unit 160 the degree of probability that each word in the text belongs to an adjustment category, which has been modified by the modification unit 160 using the initial values ​​of the coefficients. Upon receiving the degree of probability that each word belongs to an adjustment category from the modification unit 160, the classification unit 130 uses the degree of probability that each word belongs to a category for multiple categories, including the modified degree of probability of the adjustment category, to classify the words into categories.

[0096] The classification unit 130 then sends the text, information about the words into which the text has been divided, and information about the categories into which the words have been classified using the initial values ​​of the coefficients to the display unit 140.

[0097] The classification unit 130 of this embodiment is the same as the classification unit 130 of the second embodiment, except for the points described above (for example, the fact that it first classifies words using the degree of probability for multiple categories, including the degree of probability of adjustment categories that have been modified using the initial value of the coefficient).

[0098] Other components of the text processing device 101 in this embodiment are the same as the components of the text processing device 100 in the second embodiment that are assigned the same names and reference numerals.

[0099] <Operation> Next, the operation of the text processing device 101 according to this embodiment will be described in detail with reference to the drawings.

[0100] Figure 7 is a flowchart illustrating an example of the operation of the text processing device 101 according to this embodiment to determine the initial value of the coefficient.

[0101] In the example shown in Figure 7, first, the result receiving unit 181 receives the results of the word classification (step S201).

[0102] Next, the initial value determination unit 183 determines the likelihood that the words included in the received classification result belong to a category for multiple categories (step S202). The initial value determination unit 183 may use the classification unit 130 to determine the likelihood that the words included in the received classification result belong to a category for multiple categories. In other words, the initial value determination unit 183 may receive from the classification unit 130 the likelihood that the words included in the received classification result belong to a category for multiple categories, as determined by the classification unit 130.

[0103] Then, the initial value determination unit 183 determines the initial values ​​of the coefficients so that the classification result using the degree of change made with the initial values ​​of the coefficients approaches the received classification result (step S203).

[0104] The operation of the text processing device 101 of this embodiment to output the range of words classified into the target category is the same as the operation of the text processing device 100 of the second embodiment shown in Figures 4 and 5, except for the operation in step S104. In step S104, the text processing device 101 of this embodiment classifies words into categories using the degree of probability modified using the initial value of the coefficient, rather than the degree of probability determined.

[0105] <Effects> The embodiment described above has the same effects as the first embodiment. The reason for this is the same as the reason for the effects of the first embodiment.

[0106] <Other Embodiments> The text processing device according to the above embodiment can be implemented by a computer that includes a memory into which a program read from a storage medium is loaded, and a processor that executes the program. The text processing device according to the above embodiment can also be implemented by dedicated hardware. The text processing device according to the above embodiment can also be implemented by a combination of the aforementioned computer and dedicated hardware.

[0107] Figure 8 is a diagram showing an example of the hardware configuration of a computer 1000 that can implement a text processing device according to the embodiment of this disclosure. In the example shown in Figure 8, the computer 1000 includes a processor 1001, a memory 1002, a storage device 1003, and an I / O (Input / Output) interface 1004. The computer 1000 can also access a storage medium 1005. The memory 1002 and the storage device 1003 are, for example, storage devices such as RAM (Random Access Memory) and hard disks. The storage medium 1005 is, for example, a storage device such as RAM and hard disks, ROM (Read Only Memory), and a portable storage medium. The storage device 1003 may also be the storage medium 1005. The processor 1001 can read and write data and programs to the memory 1002 and the storage device 1003. The processor 1001 can access other devices, for example, via the I / O interface 1004. The processor 1001 can access the storage medium 1005. The storage medium 1005 stores a program that causes the computer 1000 to operate as a text processing device according to the embodiment of this disclosure.

[0108] The processor 1001 loads into the memory 1002 a program stored in the storage medium 1005 that causes the computer 1000 to operate as a text processing device according to the embodiment of this disclosure. The processor 1001 then executes the program loaded into the memory 1002, causing the computer 1000 to operate as a text processing device according to the embodiment of this disclosure.

[0109] The target reception unit 110, division unit 120, classification unit 130, display unit 140, change reception unit 150, change unit 160, output unit 170, result reception unit 181, and initial value determination unit 183 can be implemented, for example, by a processor 1001 that executes a program loaded into memory 1002. The result storage unit 182 can be implemented by a storage device 1003 such as memory 1002 or a hard disk drive included in the computer 1000. Alternatively, some or all of the target reception unit 110, division unit 120, classification unit 130, display unit 140, change reception unit 150, change unit 160, output unit 170, result reception unit 181, result storage unit 182, and initial value determination unit 183 can also be implemented by dedicated circuits that realize the functions of each unit.

[0110] Furthermore, some or all of the above embodiments may also be described as follows, but are not limited to the following.

[0111] (Note 1) A display means for displaying the range in the text of target terms that are classified into a target category among the multiple categories, using each of the degrees of probability that the terms belong to multiple categories, among the terms contained in the text; A change acceptance means that accepts instructions for changes to the degree of possibility of adjustment categories that are different from the target category among the aforementioned multiple categories, A modification means for modifying the degree of possibility of the adjustment category in accordance with the received modification instructions, A classification means for reclassifying the phrase into one of the plurality of categories using the degree of probability of each of the plurality of categories other than the adjustment category and the degree of probability of the modified adjustment category, A text processing device equipped with the following features.

[0112] (Note 2) The change receiving means receives a change instruction indicating a value of a coefficient multiplied by the determined degree of possibility of the adjustment category, The modification means sets the product of the degree of possibility of the determined adjustment category and the coefficient indicated by the received modification instruction to be the degree of possibility of the modified adjustment category. The classification means reclassifies the phrase into the category with the highest degree of probability among the degree of probability of each of the multiple categories other than the adjustment category and the degree of probability of the modified adjustment category. The text processing device described in Appendix 1.

[0113] (Note 3) The aforementioned adjustment category is a category of words or phrases that do not belong to any of the categories other than the aforementioned adjustment category. A text processing device as described in Appendix 1 or 2.

[0114] (Note 4) Determination means for determining the degree of probability that a word or phrase in the text belongs to each of the multiple categories, using a determination model that has been trained to determine the degree of probability that a word or phrase belongs to each of the multiple categories. Furthermore, The classification means classifies the phrase into one of the plurality of categories using the degree of probability of the phrase belonging to each of the plurality of categories and the initial value of the coefficient of the degree of probability of the adjustment category. The text processing device described in Appendix 2.

[0115] (Note 5) Initial value determination means that determines the initial value so as to bring the classification of the word to the target category using the initial value of the coefficient for the degree of probability of the adjustment category to the result of the classification of the word to the target category included in the record of the classification of the word to the target category included in the record, using a record of the classification of the word to one of the plurality of categories obtained in advance. A text processing device as described in Appendix 4, comprising:

[0116] (Note 6) Output means for outputting the text in which the range in which the phrases that have been reclassified into the aforementioned target category exist is concealed. A text processing device according to appendix 1 or 2, further comprising the above.

[0117] (Note 7) The aforementioned categories represent types of words or phrases that include at least requests or complaints. The aforementioned target category is the aforementioned request or complaint, Output means for outputting the range in which the aforementioned words classified into the aforementioned target category exist. A text processing device according to appendix 1 or 2, further comprising the above.

[0118] (Note 8) The display means displays the text and an input area in which the value of the coefficient can be specified. The change acceptance means determines the value of the coefficient from a position specified in the input area. The text processing device described in Appendix 2.

[0119] (Note 9) The scope of the target phrases in the text that are classified into the target category among the multiple categories, using each of the degrees of probability that the phrase belongs to multiple categories, is displayed in the text. We accept instructions to change the degree of possibility of adjustment categories that are different from the target category among the aforementioned multiple categories. In accordance with the instructions for the aforementioned changes received, the degree of possibility for the aforementioned adjustment category is changed, Using the degree of probability of each of the aforementioned categories other than the aforementioned adjustment category, and the degree of probability of the modified adjustment category, the phrase is reclassified into one of the aforementioned categories. Text processing methods.

[0120] (Note 10) The instruction for the change indicates the value of a coefficient multiplied by the determined degree of possibility of the adjustment category, The product of the degree of probability of the determined adjustment category and the coefficient indicated by the received instruction for change is set as the degree of probability of the adjustment category being changed. The phrase is reclassified into the category with the highest degree of probability among the degree of probability of each of the multiple categories other than the adjustment category and the degree of probability of the changed adjustment category. The text processing method described in Appendix 9.

[0121] (Note 11) The aforementioned adjustment category is a category of words or phrases that do not belong to any of the categories other than the aforementioned adjustment category. The text processing method described in Appendix 9 or 10.

[0122] (Note 12) The degree to which a word or phrase in the text belongs to each of the multiple categories is determined using a decision model that has been trained to determine the degree to which a word or phrase belongs to each of the multiple categories. The terms are classified into one of the aforementioned categories using the degree of probability of each of the aforementioned categories that the terms belong to and the initial value of the coefficient of the degree of probability of the adjustment category. The text processing method described in Appendix 10.

[0123] (Note 13) The initial value is determined using a record of the classification of a word or phrase into one of the aforementioned multiple categories obtained in advance, so as to bring the classification of the word or phrase into the aforementioned target category, using the initial value of the coefficient for the degree of probability of the adjustment category, closer to the result of the classification of the word or phrase into the aforementioned target category included in the record. The text processing method described in Appendix 12.

[0124] (Note 14) Output the text in which the range in which the phrases that have been reclassified into the aforementioned target category exist is concealed. The text processing method described in Appendix 9 or 10.

[0125] (Note 15) The aforementioned categories represent types of words or phrases that include at least requests or complaints. The aforementioned target category is the aforementioned request or complaint, Output the range in which the aforementioned words classified into the aforementioned target category exist. The text processing method described in Appendix 9 or 10.

[0126] (Note 16) The above text and an input area where the value of the coefficient can be specified are displayed. The change acceptance means determines the value of the coefficient from a position specified in the input area. The text processing method described in Appendix 10.

[0127] (Note 17) A display process that displays the range in the text of target words that have been classified into a target category from among the multiple categories, using each of the degrees of probability that the word belongs to one of the multiple categories, A change request process that accepts instructions for changes to the degree of possibility of adjustment categories that are different from the target category among the aforementioned multiple categories, A change process that modifies the degree of possibility of the adjustment category in accordance with the received change instructions, A classification process that reclassifies the phrase into one of the aforementioned categories using the degree of probability of each of the aforementioned categories other than the aforementioned adjustment category and the degree of probability of the modified adjustment category, A program that causes a computer to execute something.

[0128] (Note 18) The change request process receives a change instruction indicating a coefficient value that is multiplied by the degree of possibility determined for the adjustment category, The modification process sets the product of the degree of possibility of the determined adjustment category and the coefficient indicated by the received modification instruction as the degree of possibility of the modified adjustment category. The classification process reclassifies the phrase into the category with the highest degree of probability among the degree of probability of each of the multiple categories other than the adjustment category and the degree of probability of the changed adjustment category. The program described in Appendix 17.

[0129] (Note 19) The aforementioned adjustment category is a category of words or phrases that do not belong to any of the categories other than the aforementioned adjustment category. The program described in Appendix 17 or 18.

[0130] (Note 20) A decision process that determines the degree of probability that a word or phrase in the text belongs to each of the multiple categories, using a decision model that has been trained to determine the degree of probability that a word or phrase belongs to each of the multiple categories. Further, have the computer execute it, The classification process classifies the phrase into one of the aforementioned categories, using the degree of probability of the phrase belonging to each of the aforementioned categories and the initial value of the coefficient of the degree of probability of the adjustment category. The program described in Appendix 18.

[0131] (Note 21) Initial value determination process: Using a record of the classification of a word or phrase into one of the aforementioned multiple categories obtained in advance, the initial value is determined so as to bring the classification of the word or phrase into the aforementioned target category, using the initial value of the coefficient for the degree of probability of the adjustment category, closer to the result of the classification of the word or phrase into the aforementioned target category included in the record. The program described in Appendix 20 is further executed by the computer.

[0132] (Note 22) Output process that outputs the text in which the range in which the phrases that have been reclassified into the aforementioned target category exist is concealed. A program described in Appendix 17 or 18 that further causes a computer to execute.

[0133] (Note 23) The aforementioned categories represent types of words or phrases that include at least requests or complaints. The aforementioned target category is the aforementioned request or complaint, Output process that outputs the range in which the aforementioned words classified into the aforementioned target category exist. A program described in Appendix 17 or 18 that further causes a computer to execute.

[0134] (Note 24) The display process displays the text and an input area where the value of the coefficient can be specified. The aforementioned change acceptance process determines the value of the coefficient from the position specified in the input area. The program described in Appendix 18.

[0135] Although the present disclosure has been described above with reference to embodiments, the present disclosure is not limited to the embodiments described above. Various modifications to the structure and details of the present disclosure can be understood by those skilled in the art within the scope of the present disclosure. [Explanation of symbols]

[0136] 10 Text Processing Device 100 Text Processing Devices 101 Text Processing Device 110 Target Reception Department 120 Division 130 Classification Department 140 Display section 150 Change Request Department 160 Changes 170 Output section 181 Results Reception Department 182 Result storage section 183 Initial Value Determination Unit 1000 computers 1001 Processor 1002 memory 1003 Storage device 1004 I / O Interface 1005 Storage medium

Claims

1. A display means for displaying the range in the text of target terms that are classified into a target category among the multiple categories, using each of the degrees of probability that the terms belong to multiple categories, among the terms contained in the text; A change acceptance means that accepts instructions for changes to the degree of possibility of adjustment categories that are different from the target category among the aforementioned multiple categories, A modification means for modifying the degree of possibility of the adjustment category in accordance with the received modification instructions, A classification means for reclassifying the phrase into one of the plurality of categories using the degree of probability of each of the plurality of categories other than the adjustment category and the degree of probability of the modified adjustment category, A text processing device equipped with the following features.

2. The change receiving means receives a change instruction indicating a value of a coefficient multiplied by the determined degree of possibility of the adjustment category, The modification means sets the product of the degree of possibility of the determined adjustment category and the coefficient indicated by the received modification instruction to be the degree of possibility of the modified adjustment category. The classification means reclassifies the phrase into the category with the highest degree of probability among the degree of probability of each of the multiple categories other than the adjustment category and the degree of probability of the modified adjustment category. The text processing device according to claim 1.

3. The aforementioned adjustment category is a category of words or phrases that do not belong to any of the categories other than the aforementioned adjustment category. The text processing device according to claim 1 or 2.

4. Determination means for determining the degree of probability that a word or phrase in the text belongs to each of the multiple categories, using a determination model that has been trained to determine the degree of probability that a word or phrase belongs to each of the multiple categories. Furthermore, The classification means classifies the phrase into one of the plurality of categories using the degree of probability of the phrase belonging to each of the plurality of categories and the initial value of the coefficient of the degree of probability of the adjustment category. The text processing device according to claim 2.

5. Initial value determination means that determines the initial value so as to bring the classification of the word to the target category using the initial value of the coefficient for the degree of probability of the adjustment category to the result of the classification of the word to the target category included in the record of the classification of the word to the target category included in the record, using a record of the classification of the word to one of the plurality of categories obtained in advance. The text processing device according to claim 4, comprising:

6. Output means for outputting the text in which the range in which the phrases that have been reclassified into the aforementioned target category exist is concealed. The text processing device according to claim 1 or 2, further comprising:

7. The aforementioned categories represent types of words or phrases that include at least requests or complaints. The aforementioned target category is the aforementioned request or complaint, Output means for outputting the range in which the aforementioned words classified into the aforementioned target category exist. The text processing device according to claim 1 or 2, further comprising:

8. The display means displays the text and an input area in which the value of the coefficient can be specified. The change acceptance means determines the value of the coefficient from a position specified in the input area. The text processing device according to claim 2.

9. A computer, The scope of the target phrases in the text that are classified into the target category among the multiple categories, using each of the degrees of probability that the phrase belongs to multiple categories, is displayed in the text. We accept instructions to change the degree of possibility of adjustment categories that are different from the target category among the aforementioned multiple categories. In accordance with the instructions for the aforementioned changes received, the degree of possibility for the aforementioned adjustment category is changed, Using the degree of probability of each of the aforementioned categories other than the aforementioned adjustment category, and the degree of probability of the modified adjustment category, the phrase is reclassified into one of the aforementioned categories. Text processing methods.

10. A display process that displays the range in the text of target words that have been classified into a target category from among the multiple categories, using each of the degrees of probability that the word belongs to one of the multiple categories, A change request process that accepts instructions for changes to the degree of possibility of adjustment categories that are different from the target category among the aforementioned multiple categories, A change process that modifies the degree of possibility of the adjustment category in accordance with the received change instructions, A classification process that reclassifies the phrase into one of the aforementioned categories using the degree of probability of each of the aforementioned categories other than the aforementioned adjustment category and the degree of probability of the modified adjustment category, A program that causes a computer to execute something.

Images