Assistance system, assistance method, and assistance program

Abstract

This assistance system comprises a processor, wherein the processor: acquires a reception history including an inquiry from a customer and an answer to the inquiry; acquires, for a plurality of topics, feature words defined in advance as words that are characteristic to each topic; classifies the reception history into any of the plurality of topics on the basis of the feature words; and outputs information corresponding to a reception history classification result obtained by classifying the reception history.

Description

Support system, support method, and support program

[0001] The present invention relates to an assistance system, an assistance method, and an assistance program.

[0002] Patent Document 1 describes "a FAQ registration support method that supports the decision on whether to register new FAQs (frequently asked questions) using machine learning, a method for automatic learning from FAQs, and a computer system."

[0003] JP 2019-36210 A

[0004] Customer needs are contained in response history, especially in the content of customer inquiries. Effectively utilizing this information to meet those needs will differentiate a company from its competitors. However, response history contains a huge amount of data, making it difficult to manually review and analyze the content comprehensively. Therefore, there is a need for data analysis technology to capture the overall picture of customer needs and support decision-making regarding how to respond to response history.

[0005] In response to this, Patent Document 1 discloses a technology that uses machine learning to support the creation of FAQs. In Patent Document 1, inquiries are grouped using clustering. However, when unsupervised learning is used, it is not possible to group response histories by intended items. On the other hand, when supervised learning is used, it is necessary to assign labels, which requires a lot of work, and it is difficult to define classes in the first place.

[0006] The present disclosure has been made in consideration of these circumstances, and aims to provide a system, method, and program that, when assisting in decision-making regarding how to respond to response history, can group items by intended items compared to when unsupervised learning is used, and can eliminate the complexity of the work compared to when supervised learning is used.

[0007] The support system according to the first aspect of the present disclosure includes a processor that acquires a response history including inquiries from customers and responses to the inquiries, acquires predefined feature words for each of a plurality of topics as words characteristic of each topic, classifies the response history into one of the plurality of topics based on the feature words, and outputs information according to the response history classification result obtained by classifying the response history.

[0008] A support system according to a second aspect of the present disclosure is the support system according to the first aspect, wherein the processor performs a Pareto analysis on the response history classification results and visualizes the analysis results of the Pareto analysis.

[0009] A support system according to a third aspect of the present disclosure is the support system according to the first or second aspect, in which the processor performs trend analysis of the response history classification results based on the trends of the inquiries or the responses, and outputs information according to the analysis results of the trend analysis.

[0010] A support system according to a fourth aspect of the present disclosure is the support system according to the third aspect, wherein the processor visualizes the results of the trend analysis in the form of a scatter plot including parameters of at least one of the response time from the inquiry to the response and the first-line response rate.

[0011] A support system according to a fifth aspect of the present disclosure is the support system according to the third or fourth aspect, wherein the processor selects a target topic to be turned into an FAQ from among the plurality of topics based on the results of the trend analysis, and outputs information according to the target topic.

[0012] A support system according to a sixth aspect of the present disclosure is a support system according to any one of the first to fifth aspects, in which the processor acquires existing FAQs that have already been converted into FAQs, classifies the existing FAQs into one of the plurality of topics using a classification model generated when classifying the response history, and outputs information according to the FAQ classification result into which the existing FAQs have been classified.

[0013] A support system according to a seventh aspect of the present disclosure is a support system according to the sixth aspect, wherein the processor performs a Pareto analysis on each of the response history classification results and the FAQ classification results, compares the response history analysis results obtained by Pareto analysis of the response history classification results with the FAQ analysis results obtained by Pareto analysis of the FAQ classification results, and outputs information according to the comparison results.

[0014] An assistance system according to an eighth aspect of the present disclosure is the assistance system according to the seventh aspect, wherein the processor determines a priority order among the topics to be turned into FAQs for the plurality of topics based on the comparison results, and outputs information according to the priority order among the topics.

[0015] A support system according to a ninth aspect of the present disclosure is the support system according to the seventh or eighth aspect, wherein the processor generates a plurality of similar groups for at least one of the plurality of topics based on the similarity of the response history and the existing FAQs, determines a priority within the topics to be turned into FAQs based on the number of response history items and the number of existing FAQs in each of the plurality of similar groups, and outputs information according to the priority within the topic.

[0016] A support system according to a tenth aspect of the present disclosure is a support system according to any one of the first to ninth aspects, wherein the processor subdivides at least one of the plurality of topics into a plurality of detailed topics based on the similarity of the response histories, and outputs information according to the plurality of detailed topics.

[0017] An assistance method according to an eleventh aspect of the present disclosure includes a computer acquiring a response history including inquiries from customers and responses to the inquiries, acquiring, for each of a plurality of topics, characteristic words that are predefined as words characteristic of each topic, classifying the response history into one of the plurality of topics based on the characteristic words, and outputting information according to the response history classification result into which the response history has been classified.

[0018] An assistance program relating to a twelfth aspect of the present disclosure causes a computer to perform the following processes: acquiring a response history including inquiries from customers and responses to the inquiries; acquiring, for each of a plurality of topics, pre-defined feature words that are characteristic of each topic; classifying the response history into one of the plurality of topics based on the feature words; and outputting information according to the response history classification results obtained by classifying the response history.

[0019] According to the support system, support method, and support program disclosed herein, when supporting decision-making regarding how to respond to response history, it is possible to group by intended items compared to when unsupervised learning is used, and it is possible to eliminate the complexity of the work compared to when supervised learning is used.

[0020] 1 is a diagram showing an example of an outline of decision-making using the support system 100 according to the present embodiment; FIG. 1 is a diagram showing an example of a hardware configuration of the support system 100 according to the present embodiment; FIG. 2 is a diagram showing an example of a functional configuration of the support system 100 according to the present embodiment; FIG. 3 is a diagram showing an example of a response history; FIG. 4 is a diagram showing an example of a feature word; FIG. 5 is a diagram showing an example of an existing FAQ; FIG. 6 is a diagram showing a processing flow executed by the support system 100 according to the first embodiment; FIG. 7 is a diagram showing details of a response history classification process; FIG. 8 is a diagram showing a first example of visualization by the support system 100 according to the first embodiment; FIG. 9 is a diagram showing a second example of visualization by the support system 100 according to the first embodiment; FIG. 10 is a diagram showing a third example of visualization by the support system 100 according to the first embodiment; FIG. 11 is a diagram showing details of a target topic selection process; FIG. 12 is a diagram showing a first example of visualization by the support system 100 according to the second embodiment; FIG. 13 is a diagram showing a second example of visualization by the support system 100 according to the second embodiment; FIG. 14 is a diagram showing a processing flow executed by the support system 100 according to the third embodiment; FIG. 15 is a diagram showing details of an existing FAQ classification process; FIG. 16 is a diagram showing details of a topic comparison and analysis process; FIG. 17 is a diagram showing a first example of visualization by the support system 100 according to the third embodiment. FIG. 10 is a diagram showing a second example of visualization by the support system 100 according to the third embodiment. FIG. 11 is a diagram showing details of similar group generation processing. FIG. 12 is a diagram showing a third example of visualization by the support system 100 according to the third embodiment. FIG. 13 is a diagram showing a fourth example of visualization by the support system 100 according to the present embodiment. FIG. 14 is a diagram showing a processing flow executed by the support system 100 according to the fourth embodiment. FIG. 15 is a diagram showing details of detailed topic classification processing. FIG. 16 is a diagram showing a first example of visualization by the support system 100 according to the fourth embodiment. FIG. 17 is a diagram showing a processing flow executed by the support system 100 according to the fifth embodiment. FIG. 18 is a diagram showing a processing flow executed by the support system 100 according to the sixth embodiment.

[0021] An example of an embodiment of the present disclosure will be described below with reference to the drawings. In each drawing, the same or equivalent components and parts are denoted by the same reference numerals. Furthermore, the dimensional proportions in the drawings are exaggerated for the sake of explanation and may differ from the actual proportions.

[0022] 1 is a diagram showing an example of an outline of decision-making using a support system 100 according to this embodiment. Note that this diagram comprehensively shows all processes that can be executed by the support system 100. However, the support system 100 does not necessarily need to execute all of the processes shown in this diagram, and may execute at least some of the processes shown in this diagram.

[0023] The support system 100 may execute at least one of the following processes: a response history classification process shown in step S10, a target topic selection process shown in step S20, an FAQ classification process shown in step S30, a topic comparison and analysis process shown in step S40, a similar group generation process shown in step S50, and a detailed topic classification process shown in step S60. Each process will be described in detail below.

[0024] As a result, the support system 100 may support the user in making a decision to turn the questions into FAQs in order of priority. The support system 100 may also support the user in making a decision to check the content and consider countermeasures. FAQ stands for Frequently Asked Questions, and is a collection of frequently asked questions about products and services and their answers.

[0025] 2 is a diagram showing an example of the hardware configuration of the assistance system 100 according to this embodiment. The assistance system 100 may be a computer. The assistance system 100 includes a processor 101, a ROM (Read Only Memory) 102, a RAM (Random Access Memory) 103, a storage 104, a communication interface 105, and a user interface 106. These components are connected to each other via a bus 109 so as to be able to communicate with each other.

[0026] The processor 101 executes various programs and controls each component. Here, the processor 101 is assumed to be a CPU (Central Processing Unit). The ROM 102 stores various programs and various data. The RAM 103 temporarily stores programs or data as a working area. The storage 104 is configured with an HDD (Hard Disk Drive) or an SSD (Solid State Drive) and stores various programs including an operating system and various data.

[0027] In the assistance system 100 according to this embodiment, an assistance program is stored in the ROM 102 or the storage 104. The processor 101 reads the assistance program from the ROM 102 or the storage 104 and deploys it in the RAM 103 as a working area, whereby the computer controls each component and executes various arithmetic processing.

[0028] The communication interface 105 is an interface through which the assistance system 100 communicates with other devices. The user interface 106 is an input / output interface through which the assistance system 100 exchanges information with a user. The user interface 106 may include input devices such as a mouse, keyboard, touch panel, and microphone, and output devices such as a monitor and speaker.

[0029] 3 is a diagram showing an example of the functional configuration of the assistance system 100 according to this embodiment. The assistance system 100 includes an input / output unit 110, a storage unit 130, a data processing unit 150, and a display unit 170. The processor 101 may read out an assistance program from the ROM 102 or the storage 104 and load it into the RAM 103, thereby implementing each functional unit in the computer.

[0030] The input / output unit 110 inputs various data from the outside and outputs various data to the outside.

[0031] The storage unit 130 stores input data necessary for processing and also stores output data corresponding to the processing results. The storage unit 130 may store, as input data, data related to at least one of response history, parameters, a technical term dictionary, a synonym dictionary, unnecessary words, characteristic words, and existing FAQs. The storage unit 130 may also store, as output data, data related to at least one of classification results, analysis results, classification models, comparison results, and similar groups.

[0032] The data processing unit 150 performs data processing using input data. The data processing unit 150 may include at least one of an analysis target extraction unit 151, a morphological analysis unit 152, a noise removal unit 153, a synonym processing unit 154, an unnecessary word removal unit 155, a classification unit 156, a Pareto analysis unit 157, a visualization unit 158, a trend analysis unit 159, a target topic selection unit 160, a classification model application unit 161, a topic comparison analysis unit 162, a similar group generation unit 163, and a detailed topic classification unit 164.

[0033] The display unit 170 displays various data. The display unit 170 may include at least one of the display unit 170, a parameter display unit 171, a response history display unit 172, a characteristic word display unit 173, a classification result display unit 174, an analysis result display unit 175, an existing FAQ display unit 176, a target topic display unit 177, a comparison result display unit 178, a similar group display unit 179, and a detailed topic display unit 180.

[0034] The following describes input data used for data processing by the support system 100. The response history is data including inquiries from customers and responses to those inquiries. Fig. 4 is a diagram showing an example of the response history. The response history may include items such as reception number, reception date and time, inquiry content, response content, first-line response, response lead time, and information source.

[0035] The reception number is an identification number that identifies the response history. The reception numbers may be assigned in ascending order starting from the earliest reception date and time. The inquiry content is an item that indicates the content of the inquiry from the customer related to a specific product or service. The response content is an item that indicates the content of the response to the inquiry from the customer.

[0036] The first-line response is an item that indicates whether the problem was resolved on the first contact. The first-line response may be marked with "O" if the problem was resolved without escalation or return contact, or "X" if the problem was not resolved. The first-line response is also called primary resolution. The response lead time is an item that indicates the response time from the inquiry to the response. The information source is an item that indicates the source that served as the basis for the response.

[0037] The response history that may include such items may be data collected and accumulated at a contact center such as a call center. Note that the response history is data that has not yet been converted into FAQs, and can be said to be data that will become the basis for FAQs.

[0038] The parameters are data for extracting a target product or service from the response history. The parameters may be data such as the product name or product model of a photoelectric sensor, for example.

[0039] The technical term dictionary and the synonym dictionary are, as the name suggests, dictionary data that lists technical terms and synonyms. The unnecessary words are data that define words that are unnecessary for data processing. The unnecessary words may be data such as greetings, particles, and conjunctions.

[0040] The feature words are data containing predefined feature words for each of multiple topics. Fig. 5 shows an example of feature words. The feature words may include items such as a topic number, a topic name, and a feature word.

[0041] The topic number is an identification number that identifies a topic. The topic name is an item that indicates the name of the topic. The topic name may indicate a theme, subject, or field, such as "product selection" or "settings."

[0042] A characteristic word is an item that indicates a word characteristic of a topic and a predefined word. For example, if the topic name is "product selection," the characteristic words may be "selection," "substitution," "discontinuation," etc. Also, if the topic name is "settings," the characteristic words may be "adjustment," "range," "day of the week," "parameter," etc. In other words, characteristic words can be said to be predefined words that are likely to appear when the subject matter is the theme, subject, or field indicated by the topic name.

[0043] Such feature words may be defined by the user. In this case, the user may define the feature words from scratch, or the user may define the feature words by selecting words that have been mechanically extracted based on probability.

[0044] The existing FAQ is data of an existing FAQ that has already been converted into a FAQ. Fig. 6 is a diagram showing an example of an existing FAQ. The existing FAQ may include items such as an FAQ number, a question, an answer, a product classification, and a creation date.

[0045] The FAQ number is an identification number that identifies the FAQ. The questions and answers are items that indicate the questions and answers that have been made into FAQs. The product classification is an item that indicates the classification of the product or service to which the FAQ relates. The creation date is an item that indicates the date on which the FAQ was made into FAQs.

[0046] By turning response history into FAQs in this way, customers can solve their own questions, reducing the number of similar inquiries. This reduces the burden on operators and improves work efficiency. However, creating FAQs indiscriminately is not a good idea. If there are too many FAQs, customers will not be able to find the information they are looking for. Furthermore, FAQs need to be updated regularly, which can be costly.

[0047] Therefore, it is important to create a FAQ containing just the right amount of useful information from the accumulated response history. The support system 100 according to this embodiment supports decision-making regarding how to respond to such response history. Below, the processing executed by the support system 100 according to this embodiment will be described in detail using six embodiments as examples.

[0048] 7 is a diagram showing a processing flow executed by the support system 100 according to the first embodiment. In the first embodiment, the support system 100 executes only the response history classification processing shown in step S10.

[0049] 8 is a diagram showing details of the response history classification process. In step S11, the processor 101 reads each data. At this time, the processor 101 reads at least the response history and characteristic word data. That is, the processor 101 acquires the response history including inquiries from customers and responses to those inquiries, and acquires characteristic words predefined as characteristic words for each of multiple topics. In addition, the processor 101 may also read data on parameters, a technical term dictionary, a synonym dictionary, and unnecessary words.

[0050] In step S12, the processor 101 extracts an analysis target as the analysis target extraction unit 151. For example, the processor 101 extracts an analysis target from the response history using parameters. As an example, the processor 101 may extract, as the analysis target, a response history related to a photoelectric sensor from a mixture of response histories related to various products and services.

[0051] In step S13, the processor 101 performs morphological analysis as the morphological analysis unit 152. For example, the processor 101 uses a technical terminology dictionary to divide the response history extracted as the analysis target into morphemes, which are the smallest units of language, and analyzes the part of speech of each morpheme, etc.

[0052] In step S14, the processor 101 removes noise as the noise removal unit 153. For example, the processor 101 removes, as noise, those of the divided morphemes that do not have meaning on their own (numbers, symbols, punctuation marks, etc.).

[0053] In step S15, the processor 101 executes synonym processing as the synonym processing unit 154. For example, the processor 101 uses a synonym dictionary to standardize the expressions of words that are treated as synonyms.

[0054] In step S16, the processor 101 functions as the unnecessary word removal unit 155 to remove unnecessary words. For example, the processor 101 uses unnecessary words to remove greetings, particles, conjunctions, and the like. The processor 101 may preprocess the response history to be analyzed in this manner, for example. This allows the processor 101 to improve the accuracy of classification.

[0055] In step S17, the processor 101, functioning as the classification unit 156, classifies the response history into one of multiple topics based on the feature words. For example, the processor 101 may calculate the probability that the response history belongs to multiple topics for each reception number. As an example, the processor 101 may calculate the probability that the response history with reception number = 00001 belongs to each of topic numbers = 0 to 7. Then, for example, if the probability of belonging to topic number = 2 is the highest, the processor 101 may classify the response history with reception number = 00001 into topic number = 2. The topic classified in this way becomes the belonging topic. The processor 101 may sequentially perform a similar classification process for each reception number. In this way, the processor 101 can generate a classification model.

[0056] In step S18, the processor 101, as the Pareto analysis unit 157, performs a Pareto analysis on the response history classification results obtained by classifying the response histories. For example, the processor 101 may count the number of response histories for each topic. Next, the processor 101 may sort the topics in descending order of the number of cases. Furthermore, the processor 101 may calculate the ratio for each topic when the total number of cases is 100. Then, the processor 101 may calculate the cumulative ratio in descending order of the number of cases.

[0057] In step S19, the processor 101, functioning as the visualization unit 158, visualizes the above-described processing results in various ways. FIG. 9 is a diagram showing a first example of visualization by the support system 100 according to the first embodiment. As shown in the figure, the processor 101 may visualize the response history and the classification results in a table format in association with each other. In this case, the classification results may include the associated topics and the probability of each topic belonging to the topic.

[0058] 10 is a diagram showing a second example of visualization by the support system 100 according to the first embodiment. In this diagram, the horizontal axis represents the topic number. In addition, in this diagram, the left vertical axis represents the number of cases, and a bar graph represents the number of cases. In addition, in this diagram, the right vertical axis represents the cumulative ratio, and a line graph represents the cumulative ratio. As shown in this diagram, the processor 101 may visualize the analysis results of the Pareto analysis.

[0059] FIG. 11 is a diagram showing a third example of visualization by the support system 100 according to the first embodiment. This diagram shows an example in which words used in classified response histories for each of a plurality of topics are represented as a word cloud. A word cloud is a visualization of frequently used words, extracted from text data such as sentences and color-coded according to the number of times the words appear. As shown in this diagram, the processor 101 may visualize the words used in the response histories in a manner according to the number of times they appear. In this way, the processor 101 can output information according to the response history classification results obtained by classifying the response histories.

[0060] 12 is a diagram showing a processing flow executed by the support system 100 according to the second embodiment. In the second embodiment, the support system 100 executes a target topic selection process shown in step S20 in addition to the response history classification process shown in step S10. The response history classification process may be the same as in the first embodiment, and therefore a description thereof will be omitted here.

[0061] 13 is a diagram showing details of the target topic selection process. In step S21, the processor 101 acquires a response history classification result obtained by classifying the response history. For example, the processor 101 may acquire data in which the response history and the classification result are associated with each other, as shown in FIG.

[0062] In step S22, the processor 101, as the trend analysis unit 159, performs trend analysis on the response history classification results based on the trends of inquiries or responses. For example, the processor 101 may calculate the average response time by averaging the response lead time for response histories categorized into the same belonging topic. The processor 101 may also calculate the first-line response rate by tallying the success or failure of first-line responses for response histories categorized into the same belonging topic.

[0063] The processor 101 may perform such trend analysis for each of a plurality of topics. In the above description, an example in which trend analysis is performed on a topic-by-topic basis has been shown, but the present invention is not limited to this. The processor 101 can also perform trend analysis of the response history based on the elapsed time since the sale of a product or service. In addition, the processor 101 can also perform trend analysis of the response history in comparison with sales data. In this case, classification by product group, inquiry classification, inquiry category, or the like may be used.

[0064] In step S23, the processor 101, as the target topic selection unit 160, selects a target topic to be turned into an FAQ from among the multiple topics based on the results of the trend analysis. For example, the processor 101 may select, as the target topic, a topic whose average response time is less than a predetermined threshold from among the multiple topics. Alternatively, or in addition, the processor 101 may select, as the target topic, a topic whose first-line response rate is equal to or greater than a predetermined threshold from among the multiple topics.

[0065] In step S24, the processor 101, functioning as the visualization unit 158, visualizes the above-described processing results in various ways. For example, the processor 101 may display the topic number of the topic selected as the target topic. In this way, the processor 101 can output information according to the target topic.

[0066] 14 is a diagram showing a first example of visualization by the support system 100 according to the second embodiment. In this diagram, the horizontal axis represents the first-line response rate, and the vertical axis represents the average response time. As shown in this diagram, the processor 101 may visualize the results of the trend analysis in the form of a scatter plot whose parameters include at least one of the response time from inquiry to response and the first-line response rate.

[0067] 15 is a diagram showing a second example of visualization by the support system 100 according to the second embodiment. In this diagram, the horizontal axis represents the year and month, and the vertical axis represents the number of inquiries. In this way, the processor 101 can output information according to the results of the trend analysis.

[0068] 16 is a diagram showing a processing flow executed by a support system 100 according to a third embodiment. In the third embodiment, the support system 100 executes an existing FAQ classification process shown in step S30, a topic comparison and analysis process shown in step S40, and a similar group generation process shown in step S50, in addition to the response history classification process shown in step S10 and the target topic selection process shown in step S20. The response history classification process and the target topic selection process may be the same as those in the second embodiment, and therefore will not be described here.

[0069] 17 is a diagram showing details of the existing FAQ classification process. In step S31, the processor 101 reads each piece of data. At this time, the processor 101 reads at least data on existing FAQs and classification models. That is, the processor 101 may acquire existing FAQs that have already been converted into FAQs and classification models generated when classifying response histories. In addition, the processor 101 may read data on technical terms, synonyms, and unnecessary words.

[0070] Steps S32 to S35 may be the same as steps S13 to S16 except that the target changes from response history to existing FAQs, and therefore a description thereof will be omitted here.

[0071] In step S36, the processor 101, as the classification model application unit 161, applies the classification model generated in step S17 to the existing FAQ. As a result, the processor 101 classifies the existing FAQ into one of a plurality of topics using the classification model generated when classifying the response history.

[0072] Steps S37 and S38 may be similar to steps S18 and S19, except that the target changes from the response history classification result obtained by classifying the response history to the FAQ classification result obtained by classifying the existing FAQ, and therefore a description thereof will be omitted here. In this way, for example, the processor 101 can output information according to the FAQ classification result obtained by classifying the existing FAQ.

[0073] 18 is a diagram showing details of the topic comparison analysis process. In step S41, the processor 101 obtains each analysis result obtained by performing a Pareto analysis on each of the response history classification result and the FAQ classification result.

[0074] In step S42, the processor 101, as the topic comparison and analysis unit 162, compares and analyzes the response history analysis result obtained by performing a Pareto analysis on the response history classification result and the FAQ analysis result obtained by performing a Pareto analysis on the FAQ classification result. For example, the processor 101 may calculate the total number and difference between the number of response histories and existing FAQs for each topic.

[0075] In step S43, the processor 101 determines the priority order among the multiple topics to be converted into FAQs based on the comparison result. For example, the processor 101 may determine the priority order so that the priority is higher in descending order of the difference. Note that, although the above description has been given as an example of determining the priority order based on the difference, the present invention is not limited to this. For example, the processor 101 may determine the priority order using the FAQ usage record.

[0076] In step S44, the processor 101, functioning as the visualization unit 158, visualizes the above-described processing results in various ways. FIG. 19 is a diagram showing a first example of visualization by the support system 100 according to the third embodiment. As shown in this diagram, the processor 101 may visualize, in a table format, the number of response history entries, the number of existing FAQ entries, the total, the difference, and the priority for each topic. In this way, the processor 101 can output information according to the priority between topics, for example.

[0077] FIG. 20 is a diagram showing a second example of visualization by the support system 100 according to the third embodiment. In this diagram, the horizontal axis represents the topic number. In addition, in this diagram, the left vertical axis represents the total, and the bar graph represents the number of cases. Note that the dotted bar graph represents the number of existing FAQs, and the bar graph without dots represents the number of response history cases, so that the breakdown can be seen. In addition, in this diagram, the right vertical axis represents the difference, and the line graph represents this difference. In this way, for example, the processor 101 can output information according to the comparison result obtained by comparing the response history analysis result and the FAQ analysis result.

[0078] 21 is a diagram showing details of the similar group generation process. In step S51, the processor 101 acquires the classification results of the response history and the existing FAQ, that is, the response history classification result and the FAQ classification result.

[0079] In step S52, the processor 101, as the similarity group generation unit 163, generates multiple similarity groups for at least one of the multiple topics based on the similarity of the response history and the existing FAQ. For example, when a target topic has been selected by the target topic selection process, the processor 101 may select the topic selected as the target topic from the multiple topics. Then, the processor 101 may generate multiple similarity groups for the topic based on, for example, the Levenshtein distance, which calculates the similarity between character strings.

[0080] In step S53, the processor 101 determines the priority within the topic to be turned into an FAQ based on the number of response history entries and the number of existing FAQ entries in each of the multiple similar groups. For example, the processor 101 may determine the priority so that for a group with no existing FAQ entries, the priority is given to the group with the largest number of response history entries.

[0081] In step S54, the processor 101, functioning as the visualization unit 158, visualizes the above-described processing results in various forms. Fig. 22 is a diagram showing a third example of visualization by the support system 100 according to the third embodiment. As shown in this figure, the processor 101 may visualize the reception number, FAQ number, and group number in a table format in association with each other.

[0082] Here, the group number is an identification number that identifies a similar group. That is, the processor 101 may display which response history belongs to which similar group and which existing FAQ belongs to which similar group.

[0083] FIG. 23 is a diagram showing a fourth example of visualization by the support system 100 according to this embodiment. While FIG. 22 shows details of similar groups, FIG. 23 shows a summary of similar groups. The group size indicates the size of the similar group, and here indicates the total number of existing FAQs and response history items belonging to the same similar group. Mean and Std indicate the average and standard deviation, and here indicate values ​​internally processed when calculating the Levenshtein distance. In this way, the processor 101 can output information according to the priority within a topic, for example.

[0084] <Fourth embodiment> Figure 24 is a diagram showing a processing flow executed by the support system 100 according to the fourth embodiment. In the fourth embodiment, the support system 100 executes a detailed topic classification process shown in step S60 in addition to the response history classification process shown in step S10 and the target topic selection process shown in step S20. The response history classification process and the target topic selection process may be similar to those in the second embodiment, and therefore will not be described here.

[0085] 25 is a diagram showing details of the detailed topic classification process. In step S61, the processor 101 acquires a response history classification result obtained by classifying the response history. For example, the processor 101 may acquire data in which the response history and the classification result are associated with each other, as shown in FIG.

[0086] In step S62, the processor 101, as the detailed topic classification unit 164, classifies at least one of the multiple topics into multiple detailed topics based on the similarity of the response histories. At this time, for example, when a target topic has been selected by the target topic selection process, the processor 101 may select a topic from the multiple topics that was not selected as the target topic. Then, the processor 101 may classify the topic into multiple detailed topics using an algorithm such as clustering.

[0087] In step S63, the processor 101, functioning as the visualization unit 158, visualizes the above-described processing results in various ways. FIG. 26 is a diagram showing a first example of visualization by the support system 100 according to the fourth embodiment. As shown in this figure, the processor 101 may visualize the detailed topic number and the number of handling history records for each of a plurality of detailed topics. In response to this, the user may review the handling history records categorized into the detailed topics and input the content and product analysis / failure-related items.

[0088] 27 is a diagram showing a processing flow executed by the support system 100 according to the fifth embodiment. In the fifth embodiment, the support system 100 may be the same as that of the third embodiment except that the target topic selection process shown in step S20 is not executed, and therefore a description thereof will be omitted here.

[0089] 28 is a diagram showing a processing flow executed by the support system 100 according to the sixth embodiment. In the sixth embodiment, the support system 100 may be the same as that of the fourth embodiment except that it does not execute the target topic selection process shown in step S20, and therefore a description thereof will be omitted here.

[0090] In this way, the support system 100 of this embodiment acquires a response history including inquiries from customers and responses to the inquiries, acquires pre-defined characteristic words for each of multiple topics as words that are characteristic of each topic, classifies the response history into one of multiple topics based on the characteristic words, and outputs information according to the response history classification results obtained by classifying the response history.

[0091] As a result, the support system 100 according to this embodiment classifies call histories into one of multiple topics using semi-supervised learning with feature words, enabling more targeted grouping than when unsupervised learning is used. In other words, the support system 100 according to this embodiment utilizes a priori knowledge from a contact center or the like to classify customer needs into targeted categories and outputs information based on the results, thereby providing more useful information for decision-making. In this case, the user only needs to set feature words, eliminating the need for extensive labeling. Therefore, the support system 100 according to this embodiment can reduce the complexity of the work compared to when supervised learning is used.

[0092] Furthermore, according to the support system 100 of this embodiment, when converting to FAQs, information is output that takes into consideration not only the response history but also the type and volume of existing FAQs, so that information that is useful for converting high-priority topics into FAQs can be provided.

[0093] Furthermore, with the assistance system 100 according to this embodiment, topics that require a response other than creating an FAQ can be broken down into detailed topics, making it possible to provide information that makes it easy to understand the content and plan countermeasures. In this case, because a priori knowledge is used, the time required to check the content can be significantly reduced.

[0094] Additionally, by using the support system 100 according to this embodiment, the user can grasp the magnitude of needs for each category. Furthermore, the user can determine whether or not a FAQ should be created for each category. Furthermore, since information about similar groups generated by mixing response history and existing FAQs is output, the user can grasp the response history content for which a new FAQ should be created without performing lengthy processing. Additionally, the user can also check the overlap of existing FAQs.

[0095] The above-described processing can also be realized by a dedicated hardware circuit. In this case, the processing may be executed by a single piece of hardware or by multiple pieces of hardware.

[0096] In the above description, the term "processor" refers to a processor in a broad sense, and includes general-purpose processors (e.g., CPU: Central Processing Unit, etc.) and dedicated processors (e.g., GPU: Graphics Processing Unit, ASIC: Application Specific Integrated Circuit, FPGA: Field Programmable Gate Array, programmable logic device, etc.).

[0097] Furthermore, the processor operations described above may not only be performed by a single processor, but may also be performed by multiple processors located at physically separate locations working together. Furthermore, the order of the processor operations is not limited to the order described above and may be changed as appropriate.

[0098] The above-mentioned program may be provided by a computer-readable non-transitory recording medium such as a USB (Universal Serial Bus) memory, a flexible disk, or a CD-ROM (Compact Disc Read Only Memory), or may be provided online via a network such as the Internet. In this case, the program recorded on the computer-readable non-transitory recording medium is typically transferred and stored in a memory or storage device. The program may be provided as standalone application software, or may be incorporated into the software of each device as a function of the device.

[0099] The above-described program can be provided as a program product. The program product includes any type of product for providing the program. For example, the program product includes a program provided over a network such as the Internet, and a non-transitory computer-readable recording medium such as a CD-ROM or DVD on which the program is stored.

[0100] The present disclosure is not limited to the above, and it goes without saying that various modifications can be made without departing from the spirit of the present disclosure.

[0101] REFERENCE SIGNS LIST 100 Support system 101 Processor 102 ROM 103 RAM 104 Storage 105 Communication interface 106 User interface 109 Bus 110 Input / output unit 130 Memory unit 150 Data processing unit 170 Display unit

Claims

1. A support system comprising a processor that acquires a response history including inquiries from customers and responses to the inquiries, acquires pre-defined characteristic words for each of a plurality of topics as words characteristic of each topic, classifies the response history into one of the plurality of topics based on the characteristic words, and outputs information according to the response history classification results obtained by classifying the response history.

2. The support system according to claim 1, wherein the processor performs a Pareto analysis on the response history classification results, and visualizes the analysis results of the Pareto analysis.

3. The support system according to claim 1 or 2, wherein the processor performs a trend analysis of the response history classification results based on the trends of the inquiries or the responses, and outputs information according to the results of the trend analysis.

4. The support system according to claim 3, wherein the processor visualizes the results of the trend analysis in the form of a scatter plot including parameters of at least one of the response time from the inquiry to the response and the first-line response rate.

5. The support system according to claim 3, wherein the processor selects a target topic to be turned into an FAQ from among the plurality of topics based on the results of the trend analysis, and outputs information according to the target topic.

6. The support system of claim 1, wherein the processor: acquires existing FAQs that have already been converted into FAQs; classifies the existing FAQs into one of the plurality of topics using a classification model generated when classifying the response history; and outputs information according to the FAQ classification results obtained by classifying the existing FAQs.

7. The support system according to claim 6, wherein the processor performs a Pareto analysis on each of the response history classification results and the FAQ classification results, compares the response history analysis results obtained by Pareto analysis of the response history classification results with the FAQ analysis results obtained by Pareto analysis of the FAQ classification results, and outputs information according to the comparison results.

8. The support system according to claim 7, wherein the processor determines a priority order among the plurality of topics to be turned into FAQs based on the comparison result, and outputs information according to the priority order among the topics.

9. The support system described in claim 7, wherein the processor generates a plurality of similar groups for at least one of the plurality of topics based on the similarity of the response history and the existing FAQs, determines a priority within the topic to be turned into an FAQ based on the number of response history items and the number of existing FAQs in each of the plurality of similar groups, and outputs information according to the priority within the topic.

10. The support system according to claim 1 or 2, wherein the processor further classifies at least one of the plurality of topics into a plurality of detailed topics based on the similarity of the response history, and outputs information according to the plurality of detailed topics.

11. A support method comprising: a computer acquiring a response history including inquiries from customers and responses to said inquiries; acquiring, for each of a plurality of topics, characteristic words that have been predefined as words that are characteristic of each topic; classifying said response history into one of said plurality of topics based on said characteristic words; and outputting information according to the response history classification results into which said response history has been classified.

12. An assistance program that causes a computer to perform the following processes: acquiring a response history including inquiries from customers and responses to said inquiries; acquiring, for each of a plurality of topics, characteristic words that have been predefined as words that are characteristic of each topic; classifying said response history into one of said plurality of topics based on said characteristic words; and outputting information according to the response history classification results obtained by classifying said response history.

Images