Operation method of electronic device for providing, in text, analysis result of emotion states identified from multiple pieces of review data and answer corresponding to emotion state identified from review

Abstract

An operation method of an electronic device is disclosed. An operation method of an electronic device according to the present disclosure comprises the steps of: obtaining review information targeting a product or service of a seller; obtaining a numerical emotion value obtained by converting an emotion state identified from the obtained review information into a numerical value; generating graph information in which the numerical emotion value is visualized; and generating first text for describing emotion states identified from multiple pieces of review information on the basis of the generated graph information and the numerical emotion value.

Description

Method of operation of an electronic device that provides an analysis result of the sentiment state of multiple review data and a response corresponding to the sentiment state of the review as text

[0001] The present disclosure relates to a method of operation of an electronic device, and more specifically, to a method of operation of an electronic device that obtains a numerical emotional value representing the emotional state of acquired review information, generates graph data in which the acquired emotional value is visualized, and generates text describing the emotional state of the review information according to the graph data and the emotional value.

[0002] Due to the continuous increase in online shopping services and delivery platforms, as well as the ongoing development of wireless internet services manifested through smartphones and tablet PCs, accessibility related to the sale of individual services or products is improving. This implies that a relatively large amount of review data regarding such services or products is also accumulating. Generally, service or product providers manually review every single piece of review data to provide responses or analyze the reviews.

[0003] Accordingly, there is an increasing need for technology that automatically analyzes multiple review data, and in particular, technology that automatically analyzes the emotional state of service beneficiaries as revealed in the review data.

[0004] The purpose of the present disclosure is to provide a method of operation for an electronic device that automatically analyzes reviews by a service beneficiary by generating text that analyzes the sentiment state of a plurality of review information.

[0005] In addition, the present disclosure aims to provide a method of operation for an electronic device that automatically generates a response corresponding to the sentiment state of received review information.

[0006] The purposes of the present disclosure are not limited to those mentioned above, and other purposes and advantages of the present disclosure not mentioned may be understood from the following description and will be more clearly understood from the embodiments of the present disclosure. Furthermore, it will be readily apparent that the purposes and advantages of the present disclosure can be realized by the means and combinations thereof set forth in the claims.

[0007] A method of operating an electronic device according to one embodiment of the present disclosure includes the steps of: obtaining review information targeting a seller's product or service; obtaining an emotional value in which the emotional state of the obtained review information is quantified; generating graph information in which the emotional value is visualized; and generating a first text for describing the emotional state of a plurality of review information based on the generated graph information and the emotional value.

[0008] At this time, the method of operation of the electronic device may include the step of generating a second text corresponding to the seller's response to the review information according to the acquired sentiment value.

[0009] Alternatively, the step of obtaining an emotional value in which the emotional state of the review information is quantified may include the step of extracting at least one keyword related to the emotional state from the text included in the review information, and the step of inputting the text and the keyword into a first artificial intelligence model that quantifies the emotional state of a word within the text into an emotional value to obtain an emotional value for each keyword, wherein the first artificial intelligence model may be characterized by being trained to output the emotional value as a positive number if the keyword is in a positive emotional state and to output the emotional value as a negative number if the keyword is in a negative emotional state.

[0010] In this case, the step of generating the graph information may include: classifying a plurality of keywords into one of a positive emotional state keyword and a negative emotional state keyword according to the acquired keyword-specific emotional values; obtaining the sum of the emotional values ​​of each keyword classified as a positive emotional state keyword as a first emotional value and obtaining the sum of the emotional values ​​of each keyword classified as a negative emotional state keyword as a second emotional value; and generating at least one graph for comparing the first emotional value and the second emotional value with each other.

[0011] Meanwhile, the step of generating the first text above may generate the first text by applying a pre-set text template according to the range of sentiment values ​​of a plurality of review information.

[0012] Alternatively, the step of generating the first text may generate the first text by inputting the emotional value of each of the plurality of review information into a second artificial intelligence model that outputs the analysis result of the emotional state of the plurality of review information.

[0013] Additionally, the method of operating the electronic device may include a step of calculating a suitability for the review information by comparing the review information with at least one past review information that is in the same category as the review information, and the step of obtaining an emotional value in which the emotional state of the review information is quantified may be to obtain the emotional value only when the calculated suitability is greater than or equal to a threshold value.

[0014] Meanwhile, the step of generating the first text may generate the first text for each unit period based on a plurality of review information obtained for each pre-set unit period, and

[0015] The method of operation of the above electronic device may include the step of generating a third text to explain the change over time of the first text generated for each unit period.

[0016] Through the present disclosure, a service or product provider can efficiently collect and analyze review data regarding the provided service or product, and can identify the sentiment state of the review data through sentiment analysis and automatically generate a response suitable thereto.

[0017] FIG. 1 is a diagram illustrating the operation of an electronic device generating text based on a plurality of review information,

[0018] FIG. 2 is a drawing for explaining the configuration of an electronic device according to one embodiment of the present disclosure,

[0019] FIG. 3 is a flowchart for explaining the operation of an electronic device according to one embodiment of the present disclosure,

[0020] FIG. 4 is a flowchart illustrating the operation of an electronic device according to one embodiment of the present disclosure acquiring sentiment values ​​for each keyword constituting text included in review information based on an artificial intelligence model.

[0021] FIG. 5 is a flowchart illustrating the operation of an electronic device according to one embodiment of the present disclosure classifying keywords according to sentiment values ​​and generating a graph that compares the sum of the sentiment values ​​of the classified keywords.

[0022] FIG. 6 is a flowchart illustrating the operation of an electronic device according to one embodiment of the present disclosure to generate a third text to explain a change over time of a first text,

[0023] FIG. 7 is a drawing for explaining that an electronic device according to one embodiment of the present disclosure is connected to a separate external device.

[0024] Before specifically describing the present disclosure, the method of description in the specification and drawings is described.

[0025] First, the terms used in this specification and claims have been selected based on general terms considering their functions in the various embodiments of this disclosure. However, these terms may vary depending on the intent of those skilled in the art, legal or technical interpretations, and the emergence of new technologies. Additionally, some terms have been arbitrarily selected by the applicant. Such terms may be interpreted according to the meanings defined in this specification; in the absence of specific definitions, they may be interpreted based on the overall content of this specification and common technical knowledge in the relevant field.

[0026] In addition, the same reference numbers or symbols described in each drawing attached to this specification represent parts or components that perform substantially the same function. For convenience of explanation and understanding, the same reference numbers or symbols are used to describe different embodiments. That is, even if components having the same reference number are all depicted in multiple drawings, the multiple drawings do not imply a single embodiment.

[0027] Additionally, in this specification and claims, terms including ordinal numbers, such as "first," "second," etc., may be used to distinguish between components. These ordinal numbers are used to distinguish identical or similar components from one another, and the meaning of the terms should not be limited by the use of such ordinal numbers. For example, the order of use or arrangement of components combined with such ordinal numbers should not be restricted by the number. If necessary, each ordinal number may be used interchangeably.

[0028] In this specification, singular expressions include plural expressions unless the context clearly indicates otherwise. In this application, terms such as "comprising" or "consisting of" are intended to specify the existence of the features, numbers, steps, actions, components, parts, or combinations thereof described in the specification, and should be understood as not precluding the existence or addition of one or more other features, numbers, steps, actions, components, parts, or combinations thereof.

[0029] In the embodiments of the present disclosure, terms such as "module," "unit," "part," etc. are used to refer to a component that performs at least one function or operation, and such component may be implemented in hardware or software, or in a combination of hardware and software. Additionally, a plurality of "modules," "units," "parts," etc. may be integrated into at least one module or chip and implemented as at least one processor, except where each needs to be implemented in specific individual hardware.

[0030] Furthermore, in the embodiments of the present disclosure, when a part is described as being connected to another part, this includes not only a direct connection but also an indirect connection through another medium. Additionally, the meaning that a part includes a certain component implies that, unless specifically stated otherwise, it does not exclude other components but may include additional components.

[0031] Hereinafter, various embodiments of the present disclosure will be described in detail with reference to the attached drawings.

[0032] FIG. 1 is a diagram illustrating the operation of an electronic device generating text based on multiple review information.

[0033] According to Fig. 1, the electronic device is configured to generate text related to the input review information when review information for a specific product or service is input.

[0034] At this time, the generated text may correspond to the result of analyzing whether the user's emotional state, which can be confirmed in the review information, is negative or positive (= first text). Alternatively, the text generated through the electronic device (100) may correspond to text related to the seller's response to the input review information (= second text) and text related to the statistical analysis of the review information (= third text).

[0035] Figure 2, according to this, is a drawing for explaining the configuration of an electronic device according to one embodiment of the present disclosure.

[0036] According to FIG. 2, the electronic device (100) may include a memory (110), a communication unit (120), and a processor (130).

[0037] The memory (110) is configured to store at least one instruction or data related to an operating system (OS) for controlling the overall operation of the components of the electronic device (100) and the components of the electronic device (100).

[0038] The memory (110) may include non-volatile memory such as ROM or flash memory, and may include volatile memory such as DRAM. Additionally, the memory (110) may include at least one storage medium capable of storing data permanently or semi-permanently, such as a flash memory device, a hard disk drive (HDD), a solid state drive (SSD), a DVD, or a laser disc.

[0039] The memory (110) can store review information obtained by the electronic device (100) to compare the review information with past review information, and may include at least one instruction to perform an analysis of multiple review information based on the review information.

[0040] Alternatively, the memory (110) may include at least one artificial intelligence model for generating text corresponding to an emotional state identified in the review information, a response to the review information, etc., based on the review information.

[0041] The communication unit (120) is configured for the electronic device (100) to communicate with an external device to obtain review information. The communication unit (120) may include circuits, modules, chips, etc., for performing communication using various wired or wireless communication methods. The communication unit (120) may also be connected to external devices, such as servers and terminal devices, through various networks.

[0042] Depending on the area or scale, a network may be a Personal Area Network (PAN), Local Area Network (LAN), Wide Area Network (WAN), etc., and depending on the openness of the network, it may be an Intranet, Extranet, or Internet, etc.

[0043] The communication unit (120) can be connected to external devices through various wireless communication methods such as LTE (long-term evolution), LTE-A (LTE Advance), 5G (5th Generation) mobile communication, CDMA (code division multiple access), WCDMA (wideband CDMA), UMTS (universal mobile telecommunications system), WiBro (Wireless Broadband), GSM (Global System for Mobile Communications), DMA (Time Division Multiple Access), WiFi (Wi-Fi), WiFi Direct, Bluetooth, BLE (Bluetooth Low Energy), NFC (near field communication), Zigbee, LoRa, etc.

[0044] Additionally, the communication unit (120) may be connected to external devices via wired communication methods such as Ethernet, an optical network, USB (Universal Serial Bus), and Thunderbolt.

[0045] In addition, the communication unit (120) may be configured to utilize various communication methods / technologies that are newly devised in the future.

[0046] In addition, the communication unit (120) is not limited to performing a communication connection in one way, but can perform communication connections in multiple ways, such as a wireless method and a wired method.

[0047] The processor (130) is configured to control the electronic device (100) overall. Specifically, the processor (130) can perform operations according to various embodiments of the present disclosure by being connected to the memory (110) and executing at least one instruction stored in the memory (110).

[0048] For example, the processor (130) can control one or any combination of other components of the electronic device (100) and can perform operations or data processing related to communication.

[0049] When a method according to one embodiment of the present disclosure includes a plurality of operations, the plurality of operations may be performed by a single processor or by a plurality of processors. For example, when a first operation, a second operation, and a third operation are performed by a method according to one embodiment, the first operation, the second operation, and the third operation may all be performed by a first processor, or the first operation and the second operation may be performed by a first processor (e.g., a general-purpose processor) and the third operation may be performed by a second processor (e.g., an artificial intelligence dedicated processor).

[0050] The processor (130) may be implemented as a single-core processor including one core, or as one or more multicore processors including multiple cores (e.g., homogeneous multicore or heterogeneous multicore). When the processor is implemented as a multicore processor, each of the multiple cores included in the multicore processor may include internal processor memory such as on-chip memory, and a common cache shared by the multiple cores may be included in the multicore processor. Additionally, each of the multiple cores included in the multicore processor (or some of the multiple cores) may independently read and execute program instructions for implementing a method according to one embodiment of the present disclosure, or all (or some) of the multiple cores may be linked together to read and execute program instructions for implementing a method according to one embodiment of the present disclosure.

[0051] In this case, if the method according to one embodiment of the present disclosure includes a plurality of operations, the plurality of operations may be performed by one of the plurality of cores included in a multi-core processor, or may be performed by a plurality of cores. For example, when a first operation, a second operation, and a third operation are performed by the method according to one embodiment, the first operation, the second operation, and the third operation may all be performed by a first core included in a multi-core processor, or the first operation and the second operation may be performed by a first core included in a multi-core processor and the third operation may be performed by a second core included in a multi-core processor.

[0052] The processor (130) may include a general-purpose processor such as a CPU, AP, DSP (Digital Signal Processor), a graphics-dedicated processor such as a GPU, VPU (Vision Processing Unit), or an artificial intelligence-dedicated processor such as an NPU. An artificial intelligence-dedicated processor may be designed with a hardware structure specialized for training or utilizing a specific artificial intelligence model.

[0053] FIG. 3 is a flowchart for explaining the operation of an electronic device according to one embodiment of the present disclosure.

[0054] Referring to FIG. 3, the electronic device (100) can obtain review information targeting the seller's goods or services (S310).

[0055] Specifically, the electronic device (100) can obtain review information related to goods or services provided by a seller in real time from multiple terminal devices corresponding to buyers. Alternatively, the review information provided by multiple terminal devices corresponding to buyers may first be collected through a separate external server, and the electronic device (100) may obtain the review information by receiving the collected review information from the external server.

[0056] When review information is obtained, the electronic device (100) can obtain an emotional value in which the emotional state of the obtained review information is quantified (S320).

[0057] The sentiment score is calculated based on the content of the review included in the review information evaluated by the review provider (buyer) regarding the product or service, and can be scored according to the emotional state indicated by the review content.

[0058] Specifically, sentiment scores can be quantified based on whether the text identified from review information is positive or negative, and the degree of positivity or negativity.

[0059] For example, the emotional score can be calculated as -1 when the degree of negativity is at its maximum and as 1 when the degree of positivity is at its maximum. Accordingly, the emotional score can be calculated within the range of -1 to 1. That is, the emotional score can be calculated as a negative value closer to -1 the more negative it is, and as a positive value closer to 1 the more positive it is.

[0060] Here, the sentiment value can be calculated for each specific keyword included in the review information, and the sentiment value can be calculated as the sum of the calculated keywords. Specifically, the electronic device (100) can identify at least one keyword that affects the sentiment state in the review content, and the degree of influence on the sentiment state for each identified keyword can be calculated as the sentiment value. In this case, the electronic device (100) can calculate the sum of the sentiment values ​​(: degree of influence) calculated for each keyword as the sentiment value of the corresponding review information. A specific method for identifying keywords that affect the sentiment state will be described later through FIG. 4.

[0061] Additionally, the electronic device (100) may classify texts of similar content into at least one evaluation item (e.g., speed of service provision, quality of product or service, response to problem occurrence, etc.) for a product or service based on texts included in review information, and obtain an evaluation item-specific sentiment score as the sum of sentiment scores calculated for each classified text.

[0062] Meanwhile, the electronic device (100) can obtain an emotional value by selecting only the review information that matches the seller's product or service from the acquired review information. That is, the category or index for the review information is set for the seller's product or service, but the content included in the actual review information may correspond to review information for a product that is completely different from the seller's product (or service).

[0063] Accordingly, the electronic device (100) can calculate the suitability by comparing the currently acquired review information with at least one past review information corresponding to a product or service of the same category among the past acquired review information. At this time, the electronic device (100) can acquire an evaluation score only for review information where the suitability is greater than or equal to a threshold value.

[0064] Specifically, the electronic device (100) can calculate a suitability by comparing past review information and text constituting each of the review information. For example, the electronic device (100) can obtain a first vector corresponding to the past text and a second vector corresponding to the target text by embedding the past text constituting the past review information and the target text constituting the review information, respectively. Then, the electronic device (100) can calculate the cosine similarity between the first vector and the second vector as a suitability for the review information.

[0065] In this case, if there is only one piece of past review information to be compared, the cosine similarity between the two vectors can be calculated as the fit, and if there are multiple pieces of past review information, the average of the cosine similarity of each of the multiple past reviews to the review information can be calculated as the fit.

[0066] And, the electronic device (100) can generate graph information in which emotional values ​​are visualized (S330).

[0067] In one embodiment, the electronic device (100) may generate a ratio graph to check the ratio of the sentiment values ​​of positively identified keywords and the ratio of the sentiment values ​​of negatively identified keywords relative to the sentiment values ​​of review information, a bar graph to compare sentiment values ​​between keywords of the same type (positive keywords or negative keywords), a bar graph showing sentiment values ​​by evaluation item, or a radar graph, but is not limited thereto, and the electronic device (100) may visualize sentiment values ​​by generating graph information of various forms.

[0068] Based on the generated graph information and sentiment values, the electronic device (100) can generate a first text to explain the sentiment status of multiple review information (S340).

[0069] For example, the electronic device (100) can generate a first text describing the sentiment state of multiple review information, such as “The ○○ product has more positive evaluations overall” or “The overall negative evaluation score is 5% higher,” based on a ratio graph calculated for multiple review information.

[0070] Alternatively, the electronic device (100) may generate a first text such as ‘Customers are praising the taste of the food and the remaining items are above average, but there are complaints about the delivery time’ based on the emotional value of each evaluation item represented by a radar graph for multiple review information.

[0071] Thus, the electronic device (100) can generate a first text, which is the result of interpreting the emotional state of multiple review information, based on the emotional value of the text constituting the review information and graph information based on the emotional value.

[0072] Meanwhile, when the electronic device (100) obtains an emotional value for the aforementioned review information (step S320), the electronic device (100) can quantify the emotional state of each keyword constituting the review information based on an artificial intelligence model trained to output the emotional state of a specific word in the text as an emotional value.

[0073] In this regard, FIG. 4 is a flowchart illustrating the operation of an electronic device according to one embodiment of the present disclosure to obtain sentiment values ​​for each keyword constituting text included in review information based on an artificial intelligence model.

[0074] According to FIG. 4, the electronic device (100) can extract keywords from text included in review information (S410).

[0075] Specifically, the electronic device (100) can extract at least one keyword related to an emotional state from text included in review information based on various known types of keyword extraction models.

[0076] For example, the electronic device (100) can extract at least one keyword related to emotion based on the importance of words constituting a sentence through the combination of a keyword extraction model such as RAKE (Rapid Automatic Keyword Extraction), TextRank, TF-IDF, LDA (Latent Dirichlet Allocation) and a pre-configured word dictionary related to emotion (e.g., positive words - satisfaction, good, fast, negative words - however, inconvenient, scratch, salty, bland, etc.).

[0077] Alternatively, since a keyword extraction model based on a natural language processing model such as BERT (Bidirectional Encoder Representations from Transformers) is trained to analyze the context of the text and identify emotions and important words without utilizing a pre-set word dictionary, the electronic device (100) may extract words related to emotions within the text as keywords based on the keyword extraction model.

[0078] And, the electronic device (100) can input text and keywords into the first artificial intelligence model to obtain an emotion value for each keyword (S420).

[0079] At this time, the first artificial intelligence model may correspond to an sentiment analysis model based on BERT, RoBERTa (Robustly optimized BERT approach), Transformer-based XLNet, GPT, etc.

[0080] The first artificial intelligence model of the present disclosure, according to this, is trained with various text data constituting review information and keywords related to emotional states, and can determine and score what emotional state a keyword has in context within the text and to what degree it indicates an emotional state. For example, regarding "this food is not good" and "this food is terrible," the first artificial intelligence model may be trained to identify both keywords "not good" and "terrible" as negative keywords, but to calculate the degree of negativity for each keyword differently. For example, for the keyword "not good," the first artificial intelligence model may calculate a score of -0.6 as the emotional score, and for the keyword "terrible," a score of -0.95 as the emotional value.

[0081] In conclusion, the electronic device (100) can extract at least one keyword that affects the emotional state from the text included in the review information, and input the extracted keyword into the first artificial intelligence model to obtain an emotional value, which is a score for the emotional state, for each keyword.

[0082] Furthermore, in generating graph information in which emotional values ​​are visualized (step S330), the electronic device (100) may generate graph information to determine whether the text included in the review information has a positive tendency or a negative tendency based on the emotional values ​​for each acquired keyword.

[0083] In this case, the electronic device (100) can classify keywords into positive keywords and negative keywords, and generate graph information for comparing emotional values ​​that match each of the positive keywords and negative keywords.

[0084] In this regard, FIG. 5 is a flowchart illustrating the operation of an electronic device (100) according to one embodiment of the present disclosure classifying keywords according to emotional values ​​and generating a graph that compares the sum of the emotional values ​​of the classified keywords.

[0085] Referring to FIG. 5, the electronic device (100) can classify multiple keywords into keywords of a positive emotional state and keywords of a negative emotional state.

[0086] In this case, keywords can be classified based on the sentiment value, which is the output of the first artificial intelligence model, for each keyword. That is, keywords for which a negative sentiment value is output through the first artificial intelligence model can be classified as keywords of a negative emotional state, and similarly, keywords for which a positive sentiment value is output can be classified as keywords of a positive emotional state.

[0087] And, the electronic device (100) can obtain a first emotional value for a keyword of a positive emotional state (S520).

[0088] In this case, the first emotional value may correspond to the sum of the emotional values ​​of each keyword classified as a keyword of a positive emotional state.

[0089] For example, keywords such as 'delicious', 'fast', 'satisfying', and 'special' can be classified as keywords representing a positive emotional state, and the sum of the emotional values ​​of each keyword, '0.5', '0.6', '0.5', and '0.8', may correspond to the first emotional value (= '2.4').

[0090] Likewise, the electronic device (100) can obtain a second emotional value for a keyword of a negative emotional state (S530).

[0091] That is, the second emotional value can be obtained as the sum of the emotional values ​​of each keyword classified as a keyword of a negative emotional state, in the same way that the first emotional value was obtained.

[0092] And, the electronic device (100) can generate a graph to compare the first emotional value and the second emotional value with each other (S540).

[0093] For example, the electronic device (100) may generate a ratio graph representing the proportion that each of the first emotional value or the second emotional value occupies in the sum of the first emotional value and the second emotional value, or the electronic device (100) may simply generate a graph such as a bar graph representing the magnitude of the first emotional value and the second emotional value.

[0094] Meanwhile, when the electronic device (100) generates a first text to explain the sentiment state of the aforementioned multiple review information (step S340), the electronic device (100) may generate the first text by applying a pre-set text template.

[0095] Specifically, the electronic device (100) can generate a first template to which a text template is applied based on the sentiment value of each of the multiple review information and the graph information corresponding thereto.

[0096] For example, the electronic device (100) may have multiple sections set according to the range of sentiment values ​​of multiple review information, and may apply a template that analyzes the entire multiple review information according to the ratio of review information included in each section.

[0097] In one embodiment, it is assumed that a first range (0.6 or more), a second range (0.2 or more and less than 0.6), a third range (-0.2 or more and less than 0.2), a fourth range (-0.6 or more and less than -0.2), and a fifth range (-0.6 or less) are pre-set based on the sentiment value of the review information.

[0098] At this time, each of the first to fifth sections may correspond to an emotional value that is very positive (= first section), positive (= second section), neutral (= third section), negative (= fourth section), and very negative (= fifth section).

[0099] In this case, the electronic device (100) can classify each of the multiple review information into sections based on the sentiment value, and can generate a first text by applying a pre-set text template to the section containing the review information in the largest proportion.

[0100] Specifically, when the ratio of review information included in the second section among multiple review information is the highest, the electronic device (100) can generate a first text, 'A significant number of reviews are positive. Service users are generally satisfied with the service,' to which a text template matching the second section is applied.

[0101] Here, the electronic device (100) may also generate a first text by further including the analysis results of graph information related to each evaluation item of the review information classified into sections where a text template is applied (: second section).

[0102] For example, the electronic device (100) can generate a first text containing content related to the most influential evaluation item based on a radar graph for each evaluation item of the review information classified into a second section.

[0103] In one embodiment, the electronic device (100) can generate a first text to which a text template related to evaluation items is additionally applied, to the first text of the above-described ‘A significant number of reviews are positive. Service users are generally satisfied with the service.’, as a first text of ‘A significant number of reviews are positive. Service users are generally satisfied with the service. Specifically, service users showed great satisfaction with the fast service provision.’

[0104] In this case, in the positive or very positive range (Section 1, Section 2), the evaluation item with the highest positive value among the calculated emotional values ​​for each evaluation item may be the most influential evaluation item. Conversely, in the negative or very negative range (Section 4, Section 5), the evaluation item with the lowest negative value among the calculated emotional values ​​for each item may be the most influential evaluation item, and in the neutral range (Section 3), the evaluation item with the value closest to zero among the calculated emotional values ​​for each item may be the most influential evaluation item.

[0105] Additionally, the electronic device (100) may generate a first text by additionally applying a text template of the section containing the largest proportion of review information among the multiple sections, as well as a text template of the section containing the next largest proportion of review information.

[0106] Specifically, if the second section is the section with the highest proportion and the fourth section (: negative) is the next most frequent section, the electronic device (100) can output the first text based on the sentiment values ​​and radar graph for the second section as well as the sentiment values ​​and radar graph for the fourth section, as follows: 'The majority of customers are generally positive about the product. Customers are particularly positive about the product's performance and provide feedback that they are satisfied with daily use. However, some customers showed a negative reaction pointing out issues with the product's durability. In this regard, many opinions were provided that improvements are needed.'

[0107] In this case, it can be seen that the first text was generated by applying not only the text template for the second section and the text template for the fourth section, but also the text template related to the evaluation items of each section.

[0108] Alternatively, the electronic device (100) may generate a first text based on a separate artificial intelligence model. Specifically, the electronic device (100) may include a second artificial intelligence model that outputs the results of an analysis of the emotional state of review information as text, and may generate a first text by inputting the emotional value of each of the multiple review information into the second artificial intelligence model.

[0109] The second artificial intelligence model of the present disclosure according to this may correspond to generative artificial intelligence models such as ChatGPT and Gemini, which are based on a Large Language Model (LLM).

[0110] Generative AI models are artificial intelligence models trained to generate new data based on large-scale data such as text, audio, and images. Specifically, generative AI models can be trained on vast amounts of raw data, which is the same type of data typically built to be generated. When given an arbitrary input, a generative AI model outputs a response (text, audio, image, etc.) that is statistically likely to be related to that input.

[0111] Accordingly, the second artificial intelligence model is trained based on text data corresponding to the results of analyzing the emotional state of each of the multiple review information and the emotional value or graph information regarding it, and when the emotional value or graph information of each of the multiple review information is input, it can generate a first text to explain the emotional state of the multiple review information.

[0112] In addition, the second artificial intelligence model may be trained by receiving all sentiment values ​​and graph information for each of the multiple review information, thereby generating the first text to further include content related to the aforementioned evaluation items.

[0113] According to various embodiments, the electronic device (100) can generate a second text corresponding to an answer for each review information according to the acquired emotional value after obtaining an emotional value that quantifies the emotional state of the review information.

[0114] For example, the electronic device (100) may generate a second text containing content expressing gratitude for the review information provided by the customer when the emotional value corresponding to a positive reaction (e.g., emotional value is 0.2 or higher), a second text containing empathy and feedback regarding additional opinions from the customer regarding the product or service when the emotional value corresponding to a neutral reaction (e.g., emotional value is -0.2 or higher but less than 0.2), and a second text containing content expressing an apology, willingness to resolve the problem, and responsibility regarding the problem pointed out by the customer when the emotional value corresponding to a negative reaction (e.g., emotional value is less than -0.2).

[0115] To this end, the electronic device (100) may utilize the aforementioned second artificial intelligence model, or may generate the second text through a separate generative artificial intelligence model based on the aforementioned Large Language Model (LLM).

[0116] In this case, a second artificial intelligence model or a separate artificial intelligence model is trained with text data corresponding to answers to multiple review information and sentiment values ​​calculated for the review information, and can output a second text corresponding to an answer based on the sentiment value of the input review information.

[0117] Meanwhile, the electronic device (100) may also generate a separate text for analyzing the first text generated for specific time intervals.

[0118] In this regard, FIG. 6 is a flowchart illustrating the operation of an electronic device (100) according to one embodiment of the present disclosure to generate a third text to explain a change over time of a first text.

[0119] As illustrated in FIG. 6, the electronic device (100) can classify review information by unit period (S610). Specifically, the electronic device (100) can classify each review information by unit period according to the time at which each of the multiple review information is acquired. Here, the unit period may correspond to a period such as one day, one week, one month, one quarter, one year, etc.

[0120] For example, the electronic device (100) can classify review information obtained in the first week of July, review information obtained in the second week of July, and review information obtained in the third week of July differently according to a unit interval set to one week.

[0121] And, the electronic device (100) can generate a first text for each unit period (S620). In this case, review information separated by unit period can be used to generate the first text. That is, the electronic device (100) can generate a first text corresponding to the first week of July using only the review information obtained in the first week of July, and in the same way, can generate a first text corresponding to the second week of July and a first text corresponding to the third week of July.

[0122] And, the electronic device (100) can generate a third text to explain the change over time of the first text generated for each unit period (S630).

[0123] The third text may correspond to text intended to explain the results of an analysis of changes in the sentiment levels of reviews over time in the first text, or changes over time in the ratios of each segment included in the first text.

[0124] To generate a third text, a third artificial intelligence model may be included that analyzes changes in the first text over time and generates a text corresponding to them.

[0125] In this case, the third artificial intelligence model may correspond to an artificial intelligence model based on the aforementioned generative artificial intelligence model and time series analysis model.

[0126] Specifically, the third artificial intelligence model can identify time changes such as the first week, the second week, and the third week, and can be trained to generate a third text to explain how the first text generated according to each unit of time changes as time passes.

[0127] For example, when a third artificial intelligence model is input with multiple first texts generated for a unit period of one week, it can output a third text to explain changes in patterns of emotional states appearing in review information, such as, 'During the month of last July, there were generally many positive reviews in the first week, but as the third week began, there was a tendency for negative reviews to increase. In particular, complaints regarding product quality appeared frequently after the middle of the third week.'

[0128] Alternatively, the third artificial intelligence model may output a third text to explain the change over time in the ratio of review information included in each segment identified from the first text, such as, 'Over the past six months, the ratio of negative reviews was initially 20%, but increased to 50% over time. This change began in earnest from the third month.'

[0129] FIG. 7 is a drawing for explaining that an electronic device according to one embodiment of the present disclosure is connected to a separate external device.

[0130] As illustrated in FIG. 7, the electronic device (100) may be connected to at least one external device (200) through a communication unit. In this case, the electronic device (100) may obtain review information about a seller's product or service through the external device (200).

[0131] Accordingly, the electronic device (100) may correspond to a seller's terminal device or a server device connected to the seller's terminal device. For example, the electronic device (100) may correspond to a terminal device such as a desktop PC, laptop PC, tablet PC, or smartphone, or a server device including at least one computer.

[0132] The external device (200) is configured to provide review information of a buyer who has purchased a seller's product or service to the electronic device (100). In this case, the external device (200) may correspond to each buyer's terminal device and directly provide the review information entered through user input to the electronic device (100), or it may correspond to a server device connected to each buyer's terminal device and receive the review information of each buyer from the terminal device and provide the received review information to the electronic device (100).

[0133] The external device (200) may also correspond to a server device including at least one computer or a terminal device for each buyer. For example, the external device (200) may correspond to a terminal device such as a desktop PC, a laptop PC, a tablet PC, or a smartphone.

[0134] Alternatively, the electronic device (100) may correspond to a POS device, etc., for collecting review information received from the payment device, connected to a payment device (e.g., an ordering tablet, kiosk, etc., provided per table) in a space where a seller provides goods or services.

[0135] Meanwhile, the various embodiments described above may be implemented by combining two or more embodiments, provided that they do not conflict or contradict each other.

[0136] Meanwhile, the various embodiments described above may be implemented in a recording medium readable by a computer or a similar device using software, hardware, or a combination thereof.

[0137] According to hardware implementation, the embodiments described in this disclosure may be implemented using at least one of ASICs (Application Specific Integrated Circuits), DSPs (digital signal processors), DSPDs (digital signal processing devices), PLDs (programmable logic devices), FPGAs (field programmable gate arrays), processors, controllers, microcontrollers, microprocessors, and other electrical units for performing functions.

[0138] In some cases, the embodiments described herein may be implemented as the processor itself. In a software implementation, embodiments such as the procedures and functions described herein may be implemented as separate software modules. Each of the aforementioned software modules may perform one or more functions and operations described herein.

[0139] Meanwhile, computer instructions or computer programs for performing processing operations in electronic devices, such as robots and servers, according to the various embodiments of the present disclosure described above, may be stored on a non-transitory computer-readable medium. When such computer instructions or computer programs stored on the non-transitory computer-readable medium are executed by the processor of a specific device, the specific device described above performs the processing operations in the electronic device according to the various embodiments described above.

[0140] A non-transient computer-readable medium refers to a medium that stores data semi-permanently and can be read by a device, unlike media that store data for a short period of time such as registers, caches, and memory. Specific examples of non-transient computer-readable media include CDs, DVDs, hard disks, Blu-ray discs, USBs, memory cards, and ROMs.

[0141] Although preferred embodiments of the present disclosure have been illustrated and described above, the present disclosure is not limited to the specific embodiments described above. It is understood that various modifications can be made by those skilled in the art without departing from the essence of the present disclosure as claimed in the claims, and such modifications should not be understood individually from the technical spirit or perspective of the present disclosure.

Claims

1. In a method of operating an electronic device, A step of obtaining review information targeting a seller's products or services; A step of obtaining a numerical emotional value representing the emotional state of the aforementioned acquired review information; A step of generating graph information in which the above-mentioned emotion values ​​are visualized; and A method of operating an electronic device comprising the step of generating a first text to describe the emotional state of a plurality of review information based on the generated graph information and the emotional value.

2. In Claim 1, The method of operation of the above electronic device is, A method of operating an electronic device comprising the step of generating a second text corresponding to the seller's response to the review information according to the aforementioned acquired sentiment value.

3. In Claim 1, The step of obtaining a numerical emotional value representing the emotional state of the above review information is: A step of extracting at least one keyword related to an emotional state from the text included in the above review information; and The method includes the step of inputting the text and the keyword into a first artificial intelligence model that quantifies the emotional state of words within the text into an emotional value to obtain an emotional value for each keyword; The above-mentioned first artificial intelligence model is, A method of operation of an electronic device characterized by being trained to output the emotional value as a positive number if the above keyword is in a positive emotional state, and to output the emotional value as a negative number if the above keyword is in a negative emotional state.

4. In Claim 3, The step of generating the above graph information is, A step of classifying multiple keywords into either keywords of a positive emotional state or keywords of a negative emotional state according to the emotional values ​​obtained for each keyword above; A step of obtaining a first emotional value by summing the emotional values ​​of each keyword classified as a keyword of the above-mentioned positive emotional state, and obtaining a second emotional value by summing the emotional values ​​of each keyword classified as a keyword of the above-mentioned negative emotional state; and A method of operating an electronic device comprising the step of generating at least one graph for comparing the first emotional value and the second emotional value with each other.

5. In Claim 1, The step of generating the first text above is, A method of operation of an electronic device for generating the first text by applying a pre-set text template according to the range of sentiment values ​​of multiple review information.

6. In Claim 1, The step of generating the first text above is, A method of operation of an electronic device, wherein the emotional value of each of the plurality of review information is input into a second artificial intelligence model that outputs the analysis result of the emotional state of the plurality of review information to generate the first text.

7. In Claim 1, The method of operation of the above electronic device is, The method includes the step of calculating the suitability of the review information by comparing the review information with at least one past review information that is in the same category as the review information; The step of obtaining a numerical emotional value representing the emotional state of the above review information is: A method of operation of an electronic device for obtaining the above-mentioned appraisal value only when the above-mentioned suitability is greater than or equal to a threshold value.

8. In Claim 1, The step of generating the first text above is, Based on multiple review information obtained for each preset unit period, the first text is generated for each unit period, and The method of operation of the above electronic device is, A method of operating an electronic device comprising the step of generating a third text to explain the change over time of the first text generated for each unit period.

Images