Learning method of artificial intelligence model for determining travel sensibility feature of travel destination and electronic apparatus

Abstract

The present disclosure relates to a method for an electronic device according to one embodiment to train an artificial intelligence model for analyzing travel sentiment by travel destination, and an electronic device for performing the same. A method for an electronic device according to one embodiment to train an artificial intelligence model for analyzing travel sentiment by travel destination may include: a step of constructing a travel sentiment analysis artificial intelligence model that outputs design sentiment feature data based on design travel-specific dictionary data according to the correlation of design morphological keywords included in the design social data when design social data including text included in social content regarding the travel destination is input; and a step of training the travel sentiment analysis artificial intelligence model by inputting training social data by travel destination into the constructed travel sentiment analysis artificial intelligence model so that training sentiment feature data obtained from the travel sentiment analysis artificial intelligence model is matched with the user's travel tendencies.

Description

Technology Field

[0001] The present disclosure relates to a method and apparatus for determining travel sentiment characteristics by travel destination based on artificial intelligence. More specifically, it relates to a method and apparatus for identifying travel sentiment characteristics regarding a travel destination based on social big data regarding a travel destination. Background Technology

[0002] With the spread of solo and spontaneous travel cultures, the demand for travel recommendation services based on travel tendencies and preferences is increasing. While general travel recommendation technologies may be suitable for situations involving high traffic and travel data, the travel industry is characterized by a low volume of new travelers and low traffic.

[0003] Therefore, although technologies utilizing conventional artificial intelligence to recommend travel content have been disclosed, they are not suitable for the travel industry with low traffic due to the difficulty in securing training data for accurate travel content recommendations. Furthermore, most travel recommendation technologies fail to build sufficient training databases due to low traffic, resulting in limitations in recommending travel content suitable for user preferences.

[0004] In addition, conventional travel recommendation technologies utilizing artificial intelligence have the problem that the objectivity and reliability of the training data used are not guaranteed, and not only is it impossible to match the emotional characteristics of a travel destination with the user's travel tendencies, but there are also limitations in that they cannot reflect the phenomenon where the emotional characteristics of a travel destination change depending on the travel season, type of travel companion, etc.

[0005] Therefore, in the low-traffic travel industry, there is a need to analyze the sentimental characteristics of travel destinations through social content and to develop travel content recommendation technology suitable for users' travel preferences. The problem to be solved

[0006] According to one embodiment, a method for identifying travel sentiment characteristics by travel location and an electronic device for performing the same may be provided.

[0007] According to one embodiment, a method for training an artificial intelligence model for analyzing travel sentiment by travel destination and an electronic device for performing the same may be provided. means of solving the problem

[0008] According to one embodiment, as a technical means for achieving the technical task described above, a method for an electronic device to train an artificial intelligence model for travel sentiment analysis by travel destination may include: a step of constructing a travel sentiment analysis artificial intelligence model that outputs design sentiment feature data based on design travel-specific dictionary data according to the correlation of design morphological keywords included in the design social data when design social data including text included in social content regarding the travel destination is input; and a step of training the travel sentiment analysis artificial intelligence model by inputting training social data by travel destination into the constructed travel sentiment analysis artificial intelligence model so that training sentiment feature data obtained from the travel sentiment analysis artificial intelligence model is matched with the user's travel tendencies.

[0009] According to another embodiment as a technical means for achieving the technical task described above, an electronic device for training an artificial intelligence model for analyzing travel sentiment by travel destination comprises: a network interface; a memory for storing one or more instructions; and at least one processor for executing the one or more instructions; wherein the at least one processor, by executing the one or more instructions, constructs a travel sentiment analysis artificial intelligence model that outputs design sentiment feature data based on design travel-specific dictionary data according to the correlation of design morphological keywords included in the design social data when design social data including text included in social content regarding the travel destination is input, and trains the travel sentiment analysis artificial intelligence model so that the training sentiment feature data obtained from the travel sentiment analysis artificial intelligence model is matched with the user's travel tendency by inputting training social data by travel destination into the constructed travel sentiment analysis artificial intelligence model.

[0010] According to another embodiment as a technical means for achieving the technical task described above, a computer-readable recording medium may be provided that stores a program for performing a method for training an artificial intelligence model for travel sentiment analysis by travel destination, comprising: a step of constructing a travel sentiment analysis artificial intelligence model that outputs design sentiment feature data based on design travel-specific dictionary data according to the correlation of design morphological keywords included in the design social data when design social data including text included in social content regarding the travel destination is input; and a step of training the travel sentiment analysis artificial intelligence model by inputting training social data by travel destination into the constructed travel sentiment analysis artificial intelligence model so that training sentiment feature data obtained from the travel sentiment analysis artificial intelligence model is matched with the travel tendency of the user. Effects of the invention

[0011] According to one embodiment, by using an artificial intelligence model for travel sentiment analysis, the travel sentiment characteristics of a travel destination can be accurately identified using only social data posted on social media.

[0012] According to one embodiment, the accuracy of an artificial intelligence model for analyzing traveler travel trends in a low-traffic travel industry sector can be improved.

[0013] According to one embodiment, the travel sentiment of a travel destination can be accurately identified based on the construction of training data. Brief explanation of the drawing

[0014] FIG. 1 is a diagram schematically illustrating the process of identifying travel sentiment features by an electronic device for identifying travel sentiment of a travel destination according to one embodiment and a travel sentiment analysis system including the same. FIG. 2 is a flowchart of a method for an electronic device to identify travel sensations according to one embodiment. Figure 3 is a diagram illustrating the structure of an artificial intelligence model for analyzing travel sentiment according to one embodiment. FIG. 4 is a flowchart of a method for an electronic device according to one embodiment to train a travel sentiment analysis artificial intelligence model. FIG. 5 is a flowchart of a method for an electronic device according to one embodiment to build a travel sentiment analysis artificial intelligence model. FIG. 6 is a flowchart of a specific method for an electronic device according to one embodiment to build travel-specific prior data and a travel sentiment analysis artificial intelligence model. FIG. 7 is a flowchart of a specific method for an electronic device according to one embodiment to train a travel sentiment analysis artificial intelligence model. FIG. 8 is a block diagram of an electronic device according to one embodiment. FIG. 9 is a block diagram of an electronic device according to another embodiment. FIG. 10 is a block diagram of a server according to one embodiment. FIG. 11 is a flowchart illustrating a method for identifying travel sentiment characteristics by travel destination by having an electronic device and a server interact with each other according to one embodiment. FIG. 12 is a flowchart of a method in which an electronic device according to one embodiment provides content about a travel destination based on user information and emotional feature data. Specific details for implementing the invention

[0015] The terms used in this specification will be briefly explained, and the present disclosure will be described in detail.

[0016] The terms used in this disclosure have been selected to be as widely used and general as possible, taking into account their functions within this disclosure; however, these terms may vary depending on the intent of those skilled in the art, case law, the emergence of new technologies, etc. Additionally, in specific cases, terms have been selected at the applicant's discretion, and in such cases, their meanings will be described in detail in the relevant description of the invention. Therefore, terms used in this disclosure should be defined not merely by their names, but based on their meanings and the overall content of this disclosure.

[0017] When a part of a specification is described as "including" a certain component, this means that, unless specifically stated otherwise, it does not exclude other components but may include additional components. Furthermore, terms such as "...part" or "module" as used in the specification refer to a unit that processes at least one function or operation, and this may be implemented in hardware or software, or as a combination of hardware and software.

[0018] Embodiments of the present disclosure are described below with reference to the attached drawings so that those skilled in the art can easily implement them. However, the present disclosure may be embodied in various different forms and is not limited to the embodiments described herein. Furthermore, in order to clearly explain the present disclosure in the drawings, parts unrelated to the explanation have been omitted, and similar parts throughout the specification are denoted by similar reference numerals.

[0019] FIG. 1 is a diagram schematically illustrating the process of identifying travel sentiment features by an electronic device for identifying travel sentiment of a travel destination according to one embodiment and a travel sentiment analysis system including the same.

[0020] According to one embodiment, the travel sentiment analysis system (10) can obtain social data (102) from social content that mentions a travel destination, and input the obtained social data (102) into an artificial intelligence model after preprocessing (e.g., sentence summarization, morphological extraction) to output sentiment feature data (104). For example, the travel sentiment analysis system (10) can identify sentiment features of a travel destination to provide a user-customized travel recommendation service by analyzing travel sentiment by travel destination.

[0021] According to one embodiment, the electronic device (1000) can output sentiment feature data using a travel sentiment analysis artificial intelligence model (130), and the travel sentiment analysis artificial intelligence model (130) may include at least one of a word-to-vector model or a covert classification model. For example, the electronic device (1000) can extract valid keywords using a word-to-vector model, determine some of the output valid keywords as valid sentiment keywords based on user input, and output the valid sentiment keywords. Additionally, for example, the electronic device (1000) can output sentiment feature data (140) by inputting the output valid sentiment keywords into a covert classification model.

[0022] According to one embodiment, the travel sentiment analysis system (10) can perform morphological analysis of social data (102) using an OKT morphological analyzer. However, it is not limited thereto, and it is understood that various morphological analyzers such as Komoran, Hannanum, and Kkma morphological analyzers may be used depending on the case.

[0023] According to one embodiment, social data (102) is a post that mentions a travel destination among posts posted on a social network service (SNS), and may include information related to the travel destination or emotional elements felt at the travel destination (e.g., at least one of text, sentences, or keywords related to travel sentiment). According to one embodiment, emotional feature data (104) may include feature values ​​specific to each travel destination output from an artificial intelligence model to identify features related to the travel destination.

[0024] According to one embodiment, the travel sentiment analysis system (10) may include an electronic device (1000) and a server (2000). According to another embodiment, the travel sentiment analysis system (10) may include an electronic device (1000), a server (2000), and an electronic device (4000). Not limited to the examples described above, the travel sentiment analysis system (10) effectively analyzes travel sentiment by travel destination through an artificial intelligence model that analyzes valid sentiment keywords, and can share travel tendency content utilizing data regarding travel sentiment with another electronic device or a server connected to another electronic device.

[0025] Although the electronic device (1000) and the server (2000) are depicted separately in the present specification, it is understood that the entire process of identifying user travel tendencies, or at least part thereof, may be performed by the interoperability of at least one device among the electronic device (1000), the server (2000), or the electronic device (1000) and the server (2000), or by one or more devices. Furthermore, according to one embodiment, the electronic device (1000) may correspond to a server (2000) device used to provide a user-customized travel recommendation service or to determine emotional characteristics by travel destination.

[0026] According to one embodiment, the electronic device (1000) may be connected to a server (2000) via a network (3000). The electronic device (1000) may also be connected to another electronic device (4000) via the network (3000). For example, the travel sentiment analysis system (10) may recommend travel destinations by matching travel sentiment characteristics and travel tendency content for users who use a user-customized travel recommendation service. According to one embodiment, the electronic device (1000) may be linked with another electronic device (4000) to output various linked travel content based on identified travel sentiment characteristic data.

[0027] According to one embodiment, an electronic device (1000) may store an artificial intelligence model (130). Although not shown in FIG. 1, a server (2000) may also store an artificial intelligence model (130) used to determine travel tendency content. According to one embodiment, the artificial intelligence model (130) may include at least one of an STTI network, a PREDICT network (cluster identification network), a TRIFIT network, and a WORD V LINKING network, but is not limited thereto. According to one embodiment, the travel sentiment analysis artificial intelligence model (130) may include at least one neural network-based network that outputs or classifies a plurality of morphological keywords or vectors corresponding to keywords included in social data (102) based on travel-specific dictionary data and a cosine similarity labeling module.

[0028] According to one embodiment, when response data is input, the STTI network can output user travel tendency feature information (e.g., comprehensive travel tendency feature information, detailed travel tendency feature information) including feature values ​​for each user travel tendency variable, and the cluster identification network can include neural network-based networks that output label information of the cluster to which the user travel tendency feature information belongs when the user travel tendency feature information is input.

[0029] According to one embodiment, the TRIFIT network may include at least one neural network-based network that acquires social data including tourist destinations for place feature analysis and outputs sentiment feature data (104) regarding travel destinations when the acquired social data (102) is input. According to one embodiment, the WORD V LINKING network may include at least one neural network-based network that outputs user travel tendency feature information (e.g., comprehensive travel tendency feature information, detailed travel tendency feature information) including feature values ​​for user travel tendency variables when sentiment feature data (104) regarding travel destinations is input.

[0030] According to one embodiment, when response data for multiple queries, to which a predetermined number of response indicators is assigned according to a predetermined priority is obtained through a simulation provided to a user, the STTI network may output user travel tendency feature information including feature values ​​for each of the user's travel tendency variables (e.g., values ​​in the form of vectors output from a travel tendency analysis AI model). According to one embodiment, the STTI network may include neural network-based numerical computation networks that output detailed travel tendency feature information including feature values ​​for detailed travel tendency variables among the user's travel tendency variables, and comprehensive travel tendency feature information including feature values ​​for comprehensive travel tendency variables representing some of the detailed travel tendency variables.

[0031] According to one embodiment, when user travel tendency feature information output from an STTI network is input, a cluster identification network (e.g., a PREDICT model) may output user travel tendency cluster label information to identify the cluster to which the user travel tendency feature information belongs according to the response data among a plurality of clusters representing the user's travel tendency. According to one embodiment, the cluster identification network may be referred to as a PREDICT model. According to one embodiment, the cluster identification network may include neural network-based numerical computation networks that output cluster label information indicating whether the user travel tendency feature information output from the STTI network belongs to which cluster among a plurality of clusters.

[0032] According to one embodiment, user travel tendency characteristic information may include detailed travel tendency characteristic information containing characteristic values ​​for each detailed travel tendency variable and comprehensive travel tendency characteristic information containing characteristic values ​​for each comprehensive travel tendency variable. More specifically, among the travel tendency variables, the detailed travel tendency characteristic information may include characteristic values ​​for each detailed travel tendency variable (V11~Vnn), and among the travel tendency variables, the comprehensive travel tendency characteristic information may include characteristic values ​​for each comprehensive travel tendency variable (G1, G2, G3, G4, G5, G6) representing some detailed travel tendency variables.

[0033] According to one embodiment, the label information of the cluster may include comprehensive travel tendency cluster label information identifying one cluster among comprehensive travel tendency clusters regarding comprehensive travel tendency arranged in a multidimensional space defined based on comprehensive travel tendency variables, and specific detailed travel tendency cluster label information identifying one cluster among detailed travel tendency clusters regarding detailed travel tendency arranged in a multidimensional space defined based on specific detailed travel tendency variables. According to one embodiment, detailed travel tendency clusters may be generated for each comprehensive travel tendency cluster, but are not limited thereto.

[0034] According to one embodiment, the STTI network, cluster identification network, TRIFIT network, or WORD V LINKING model described herein may be a CNN (Convolutional Neural Network), DNN (Deep Neural Network), RNN (Recurrent Neural Network), RBM (Restricted Boltzmann Machine), DBN (Deep Belief Network), BRDNN (Bidirectional Recurrent Deep Neural Network), or Deep Q-Network, but is not limited thereto.

[0035] FIG. 2 is a flowchart of a method for an electronic device to identify travel sensations according to one embodiment.

[0036] In S210, the electronic device (1000) can obtain text included in social content regarding a travel destination as social data. According to one embodiment, the electronic device (1000) can obtain social data from social content posts mentioning a travel destination and from a list of national tourist destinations (e.g., F&B, tourist attractions, etc.).

[0037] In S220, the electronic device (1000) can obtain social summary data including certain sentences among the acquired social data. According to one embodiment, the electronic device (1000) can obtain social summary data by inputting the social data into a TextRank-based sentence summary model that outputs social summary data including certain sentences (preferably 3 to 4 sentences) representing travel destinations when social data containing a plurality of sentences is input.

[0038] In S230, the electronic device (1000) can obtain a plurality of morphological keywords by performing morphological analysis on the social summary data. According to one embodiment, the electronic device (1000) can obtain a plurality of morphological keywords included in the social summary data by performing morphological analysis on the social summary data using a morphological analyzer (e.g., OKT library).

[0039] In S240, the electronic device (1000) can obtain the sentiment feature data according to the plurality of morpheme keywords from a travel sentiment analysis artificial intelligence model that outputs sentiment feature data representing travel sentiment by travel destination when the plurality of morpheme keywords are input. According to one embodiment, the electronic device (1000) can obtain sentiment feature data determined based on the similarity score between each morpheme keyword and pre-constructed travel-specific dictionary data and the weight score of the artificial intelligence model by inputting the plurality of morpheme keywords included in the social data into the travel sentiment analysis artificial intelligence model.

[0040] Although not illustrated in FIG. 2, an electronic device (1000) according to one embodiment may obtain the place travel tendency characteristic data according to the sentiment characteristic data from a place travel tendency variable linking model (e.g., WORD V LINKING model) that outputs place travel tendency characteristic data including feature values ​​for each travel tendency variable of a user expected to travel to the travel destination when the sentiment characteristic data is input. Additionally, according to one embodiment, the electronic device (1000) may determine the place travel tendency characteristic data as sentiment characteristic data regarding the travel destination. For example, by determining the place travel tendency characteristic data as the sentiment characteristic data, the electronic device (1000) may provide content that recommends a travel destination including sentiment characteristic data that matches the user's tendency.

[0041] Hereinafter, the process of acquiring sentiment feature data of an electronic device (1000) will be described in detail. According to one embodiment, the electronic device (1000) may input a plurality of morphological keywords into a pre-trained travel sentiment analysis artificial intelligence model. According to one embodiment, the travel sentiment analysis artificial intelligence model may include at least one of a TextRank-based sentence summary model, a Word-to-Vector model, or a Covert classification model.

[0042] According to one embodiment, when the social data is input, the TextRank-based sentence summary model can output social summary data including predetermined sentences determined based on the rank of the sentences among the sentences included in the social data. For example, the TextRank-based sentence summary model can determine the rank of sentences containing travel-related keywords by analyzing the correlation and sentence structure between travel-related keywords included in the social data.

[0043] According to one embodiment, the Word to Vector model can output multiple valid sentiment keywords regarding the travel destination when the multiple morphological keywords are input. More specifically, the Word to Vector model according to one embodiment may include neural network-based numerical computation networks that acquire some of the multiple morphological keywords among the input multiple morphological keywords, wherein the input frequency for each travel destination is greater than or equal to a threshold frequency, as the multiple valid keywords.

[0044] In addition, according to one embodiment, the Word to Vector model can determine some of the filtered valid keywords as valid sentiment keywords by filtering only the keyword types related to travel sentiment among the plurality of valid keywords based on user input regarding the acquired valid keywords. According to another example, the electronic device (1000) may extract certain keywords based on the user input among the keywords output from the Word to Vector model.

[0045] According to one embodiment, the Covert classification model can output sentiment feature data regarding the travel destination when the plurality of valid sentiment keywords output from the Word-to-Vector model are input. More specifically, the Covert classification model according to one embodiment may include neural network-based numerical computation networks that output sentiment feature data including feature values ​​of the representative sentiment keywords based on the similarity scores of the input valid sentiment keywords and the representative sentiment keywords included in the travel-specific dictionary data.

[0046] Figure 3 is a diagram illustrating the structure of an artificial intelligence model for analyzing travel sentiment according to one embodiment.

[0047] According to one embodiment, the travel sentiment analysis artificial intelligence model (310) may include a TRIFIT network including at least one of a word-to-vector model (330) or a covert classification model (340). When a morphological keyword (312) is input, the travel sentiment analysis artificial intelligence model (310) may output sentiment feature data (352) representing the sentiment of the travel destination.

[0048] According to one embodiment, the input and output values ​​of each layer within the travel sentiment analysis artificial intelligence model (310) may be provided in a predetermined sequence form (or vector). Additionally, according to one embodiment, the input and output values ​​of each layer within the word-to-vector model (310) and the covert classification model (340) may be provided in a predetermined vector form.

[0049] According to one embodiment, the word-to-vector model (310) may include at least one of a valid keyword extraction model (332) or a valid sentiment keyword determination model (332), but is not limited thereto. According to another embodiment, the word-to-vector model (310) may include only the valid keyword extraction model (332), and the operation performed by the valid sentiment keyword determination model (332) may be performed by an electronic device (1000). For example, the electronic device (1000) may determine some of the valid keywords among the valid keywords output from the word-to-vector model (310) as valid sentiment keywords based on user input.

[0050] According to one embodiment, when a plurality of morphological keywords (312) are input, the valid keyword extraction model (332) can extract valid keywords based on frequency by travel destination. For example, the valid keyword extraction model (332) can extract some keywords with high usage frequency by travel destination among a plurality of morphological keywords (312) including particles that constitute a sentence.

[0051] According to one embodiment, the valid sentiment keyword determination model (332) can output valid sentiment keywords related to travel sentiment based on user input regarding valid keywords obtained from the valid keyword extraction model (332). For example, the valid sentiment keyword determination model (332) can determine and output only some keywords representing sentiment characteristics related to travel destinations as valid sentiment keywords by filtering only keyword types related to travel sentiment among a plurality of valid keywords based on user input.

[0052] According to one embodiment, the place feature identification model (342) can output sentiment feature data (352) containing feature values ​​of representative sentiment keywords for identifying the features of a travel place, based on the similarity score between the input valid sentiment keywords and the representative sentiment keywords (364) included in the pre-constructed travel-specialized dictionary data. For example, the place feature identification model (342) can output sentiment feature data (352) containing the feature value of the representative sentiment keyword with the highest similarity score based on the cosine similarity score between the input valid sentiment keywords and the representative sentiment keywords included in the travel-specialized dictionary data (360), and can modify and update the travel-specialized dictionary data (360) by determining the valid sentiment keywords as the derived sentiment keywords (364) of the representative sentiment keywords (362).

[0053] According to one embodiment, travel-specific dictionary data (360) may be generated for each representative sentiment keyword. According to one embodiment, the travel-specific dictionary data (360) may include derived sentiment keywords associated with the representative sentiment keywords. Additionally, according to one embodiment, the representative sentiment keywords (362) and the derived sentiment keywords (364) may include at least one travel destination and a weight based on a cosine similarity score. More specifically, cosine similarity may refer to the degree of similarity between vectors measured using the cosine value of the angle between two vectors in an inner product space.

[0054] According to one embodiment, travel-specific dictionary data (360) can be modified and updated based on the cosine similarity scores of the valid sentiment keywords and the representative sentiment keywords (362) obtained from the word-to-vector model (330). For example, among the valid sentiment keywords output from the word-to-vector model (330), valid sentiment keywords whose cosine similarity score with the representative sentiment keyword (362) is greater than or equal to a predetermined score can be modified and updated by deriving them as derived sentiment keywords (364) mapped to the representative sentiment keyword.

[0055] According to one embodiment, the electronic device (1000) can update the travel-specific dictionary data (360) by adding, changing, or deleting derived sentiment keywords (364) at a predetermined interval, or by modifying and updating the weights based on the cosine similarity scores of the derived sentiment keywords (364). For example, the electronic device (1000) can more accurately identify sentiment characteristics regarding travel destinations by updating the derived sentiment keywords (364) mapped to representative sentiment keywords (362) based on the update interval of social data collected for the training of the travel sentiment analysis artificial intelligence model (310).

[0056] FIG. 4 is a flowchart of a method for an electronic device according to one embodiment to train a travel sentiment analysis artificial intelligence model.

[0057] The method of training the travel sentiment analysis artificial intelligence model described in FIG. 4 may be performed by an electronic device (1000) or by a server (2000). According to another example, the method of training the travel sentiment analysis artificial intelligence model described in FIG. 3 may be performed by the connection of an electronic device (1000) or a server (2000).

[0058] In S410, when design social data including text included in social content regarding a travel destination is input, the electronic device (1000) can build a travel sentiment analysis artificial intelligence model that outputs design sentiment feature data based on design travel specialized dictionary data according to the correlation of design morphological keywords included in the design social data.

[0059] According to one embodiment, an electronic device (1000) may acquire design social data including predetermined design morpheme keywords for analyzing travel sentiment by travel destination, and may construct a travel sentiment analysis artificial intelligence model based on design travel-specific dictionary data such that design sentiment feature data is output when predetermined design morpheme keywords are input. According to one embodiment, the design morpheme keywords may include keywords related to travel sentiment that have a high usage frequency among the design social data by travel destination based on user input.

[0060] According to one embodiment, the operation of an electronic device (1000) constructing a travel sentiment analysis artificial intelligence model may include the operation of determining the input data and output of the travel sentiment analysis artificial intelligence model, the operation of defining a training data system, and the operation of defining the structure and variables of a neural network-based network within the travel sentiment analysis artificial intelligence model.

[0061] In S420, the electronic device (1000) can train the travel sentiment analysis AI model by inputting learning social data for each travel location into the constructed travel sentiment analysis AI model so that learning sentiment feature data obtained from the travel sentiment analysis AI model is output. According to one embodiment, when learning social data is input, the electronic device (1000) can train the travel sentiment analysis AI model so that the output value output from the travel sentiment analysis AI model can be matched with the user's travel tendency.

[0062] For example, the operation of an electronic device (1000) training a travel sentiment analysis artificial intelligence model may include modifying and updating weights within the model through machine learning operations to provide a user-customized travel recommendation service for each travel destination based on the structure and input / output relationship determined in the construction phase.

[0063] FIG. 5 is a flowchart of a method for an electronic device according to one embodiment to build a travel sentiment analysis artificial intelligence model.

[0064] In S510, the electronic device (1000) can obtain design social summary data including summary sentences obtained from the design social data by using a TextRank-based sentence summary model. In S520, the electronic device (1000) can obtain design morphological keywords from the design social summary data. According to one embodiment, the electronic device (1000) can obtain design morphological keywords by using a morphological analyzer (e.g., OKT library). Additionally, according to one embodiment, the TextRank-based sentence summary model may include neural network-based numerical computation networks that output training social summary data including a plurality of training morphological keywords regarding the travel destination when training social data is input.

[0065] In S530, the electronic device (1000) can construct the travel sentiment analysis artificial intelligence model such that when the design morphological keywords are input based on the design travel-specific dictionary data, the design sentiment feature data is output. According to one embodiment, the electronic device (1000) can construct the travel sentiment analysis artificial intelligence model such that the design sentiment feature data is output based on a similarity score between representative sentiment keywords included in the design travel-specific dictionary data and some valid sentiment keywords selected among the design morphological keywords.

[0066] FIG. 6 is a flowchart of a specific method for an electronic device according to one embodiment to build travel-specific prior data and a travel sentiment analysis artificial intelligence model.

[0067] In S610, the electronic device (1000) may construct a word-to-vector model that outputs a morpheme keyword vector based on the correlation between the morpheme keywords when the design morpheme keywords are input. According to one embodiment, the word-to-vector model may include neural network-based numerical computation networks that output a plurality of valid learning sentiment keywords regarding the travel destination when the plurality of learning morpheme keywords are input. According to another embodiment, the word-to-vector model may include neural network-based numerical computation networks that output a plurality of valid learning keywords regarding the travel destination in vector form when the plurality of learning morpheme keywords are input. Additionally, for example, the keyword types labeled in the morpheme keyword vector may include travel keyword types related to travel destinations and surrounding infrastructure information, and sentiment keyword types representing travel sentiment.

[0068] In S620, the electronic device (1000) can construct design travel specialized dictionary data by representative sentiment keywords representing some of the design valid sentiment keywords among the design valid sentiment keywords selected from the morphological keywords, based on the output value of the word-to-vector model. According to one embodiment, the design travel specialized dictionary data may include a plurality of derived sentiment keywords mapped to one representative sentiment keyword.

[0069] In S630, the electronic device (1000) can construct a Covert classification model that outputs design sentiment feature data including feature values ​​for at least one representative sentiment keyword among the representative sentiment keywords representing the design travel-specialized dictionary data, based on the similarity score between the derived sentiment keywords included in the design travel-specialized dictionary data and the input design valid sentiment keywords.

[0070] According to one embodiment, the Covert classification model may include neural network-based numerical computation networks that, when the plurality of valid learning sentiment keywords output from the Word-to-Vector model are input, output learning sentiment feature data regarding the travel place based on learning travel-specific dictionary data.

[0071] In S640, the electronic device (1000) can construct a neural network-based numerical computation network for at least one of the TextRank-based sentence summary model, Word-to-Vector model, and Covert classification model as the travel sentiment analysis artificial intelligence model.

[0072] FIG. 7 is a flowchart of a specific method for an electronic device according to one embodiment to train a travel sentiment analysis artificial intelligence model.

[0073] Although not illustrated in FIG. 7, an electronic device (1000) according to one embodiment may modify and update the weights of the TextRank-based sentence summary model so that the ratio in which a plurality of learning morphological keywords included in the learning social summary data are determined as the learning valid sentiment data output from the WordToVector model increases. For example, the electronic device (1000) may modify and update the weights of the TextRank-based sentence summary model to determine the input values ​​of the WordToVector model to output more learning valid sentiment keywords.

[0074] In S710, the electronic device (1000) can acquire learning morphological keywords. According to one embodiment, the electronic device (1000) can acquire learning social summary data by inputting learning social data into a TextRank-based sentence summary model, and can acquire learning morphological keywords by using a morphological analyzer with the acquired learning social summary data.

[0075] In S720, the electronic device (1000) can obtain valid sentiment keywords for learning from the word-to-vector model by inputting morphological keywords for learning into the word-to-vector model. According to one embodiment, the electronic device (1000) can improve the accuracy of identifying travel sentiment by travel location by filtering only valid sentiment keywords for learning related to travel locations or travel sentiments based on user input.

[0076] In S730, the electronic device (1000) can determine the valid sentiment keywords for learning, whose similarity score is greater than or equal to a threshold score, as the derived sentiment keywords for learning, based on the cosine similarity score between the valid sentiment keywords for learning and the representative sentiment keywords for learning included in the learning travel-specialized dictionary data. For example, the electronic device (1000) can identify sentiment characteristics for various keywords included in social content by determining the valid sentiment keywords for learning that are highly associated with the representative sentiment keywords for learning among the valid sentiment keywords for learning output from the word-to-vector model as the derived sentiment keywords.

[0077] In S740, the electronic device (1000) can modify and update the learning travel-specific dictionary data by mapping the determined learning derived sentiment keywords to the learning representative sentiment keywords. According to one embodiment, the electronic device (1000) can tag each of the learning derived sentiment keywords with a similarity score and travel location information with respect to the learning representative sentiment keywords, and the similarity score and travel location information can be modified and updated during the learning process.

[0078] In S750, the electronic device (1000) can train the travel sentiment analysis artificial intelligence model to output at least one learning sentiment feature data that matches the user's travel tendency content based on the learning travel-specific dictionary data. According to one embodiment, the electronic device (1000) can modify and update the weights of the Covert classification model so that the feature value of the learning representative sentiment keyword included in the sentiment feature data matches the feature value of the user's travel tendency variable.

[0079] FIG. 8 is a block diagram of an electronic device according to one embodiment.

[0080] FIG. 9 is a block diagram of an electronic device according to another embodiment.

[0081] As illustrated in FIG. 8, an electronic device (1000) according to one embodiment may include a processor (1300), a network interface (1500), and a memory (1700). However, not all of the illustrated components are essential components. The electronic device (1000) may be implemented with more components than those illustrated, or with fewer components.

[0082] For example, as illustrated in FIG. 9, the electronic device (1000) may further include a user input interface (1100), an output unit (1200), a sensing unit (1400), a network interface (1500), an A / V input unit (1600), and a memory (1700) in addition to a processor (1300), a network interface (1500), and a memory (1700).

[0083] The user input interface (1100) refers to a means for a user to input data for controlling an electronic device (1000). For example, the user input interface (1100) may include a key pad, a dome switch, a touch pad (contact capacitive method, pressure resistive method, infrared sensing method, surface ultrasonic conduction method, integral tension measurement method, piezo effect method, etc.), a jog wheel, a jog switch, etc., but is not limited thereto.

[0084] The user input interface (1100) can obtain user input for filtering valid sentiment keywords related to travel destinations or sentiments among the valid keywords output from the valid keyword extraction model. According to another example, it can obtain user input for tagging similarity scores and travel destination information for derived sentiment keywords included in travel-specific dictionary data. The output unit (1200) can output an audio signal, a video signal, or a vibration signal, and the output unit (1200) may include a display unit (1210), an audio output unit (1220), and a vibration motor (1230).

[0085] The display unit (1210) includes a screen for displaying information processed by the electronic device (1000). Additionally, the screen can output information regarding travel tendencies determined by the electronic device (1000), travel location characteristics, and travel tendency content according to the travel location. The sound output unit (1220) outputs audio data received from the network interface (1500) or stored in the memory (1700). Additionally, the sound output unit (1220) outputs sound signals related to functions performed by the electronic device (1000) (e.g., call signal reception sound, message reception sound, notification sound).

[0086] The processor (1300) typically controls the overall operation of the electronic device (1000). For example, the processor (1300) can generally control the user input interface (1100), output unit (1200), sensing unit (1400), network interface (1500), A / V input unit (1600), etc. by executing programs stored in memory (1700). Additionally, the processor (1300) can perform the functions of the electronic device (1000) described in FIGS. 1 to 7 by executing programs stored in memory (1700).

[0087] According to one embodiment, at least one processor (1300) can obtain text included in social content regarding the travel destination as social data by executing one or more instructions, obtain social summary data including predetermined sentences among the obtained social data, obtain a plurality of morphological keywords by performing morphological analysis on the social summary data, and obtain the sentiment feature data according to the plurality of morphological keywords from a travel sentiment analysis artificial intelligence model that outputs sentiment feature data representing travel sentiment by travel destination when the plurality of morphological keywords are input.

[0088] According to one embodiment, at least one processor (1300) can perform at least one or all of the travel sentiment feature identification processes for each travel location described in FIGS. 1 to 7.

[0089] According to one embodiment, at least one processor (1300) can construct a travel sentiment analysis AI model that outputs design sentiment feature data based on design travel-specific dictionary data according to the correlation of design morphological keywords included in the design social data when design social data including text included in social content regarding the travel destination is input, and can train the travel sentiment analysis AI model so that the training sentiment feature data obtained from the travel sentiment analysis AI model is matched with the user's travel tendency by inputting training social data for each travel destination into the constructed travel sentiment analysis AI model.

[0090] According to one embodiment, the processor (1300) may obtain information about an AI model for travel sentiment analysis that has already been learned from the server (2000), and may modify and update the weights regarding the layers and nodes within the AI ​​model for travel sentiment analysis and the connection strengths of the layers based on the learning data.

[0091] The sensing unit (1400) can detect the state of the electronic device (1000) or the state of the surroundings of the electronic device (1000) and transmit the detected information to the processor (1300). The sensing unit (1400) can sense specification information of the electronic device (1000), temperature, humidity, atmospheric pressure information, etc., regarding the space to be monitored.

[0092] For example, the sensing unit (1400) may include at least one of a magnetic sensor (1410), an acceleration sensor (1420), a temperature / humidity sensor (1430), an infrared sensor (1440), a gyroscope sensor (1450), a position sensor (e.g., GPS) (1460), a barometric pressure sensor (1470), a proximity sensor (1480), and an RGB sensor (illuminance sensor) (1490), but is not limited thereto. Since the function of each sensor can be intuitively inferred by a person skilled in the art from its name, a detailed description is omitted.

[0093] The network interface (1500) may include one or more components that enable the electronic device (1000) to communicate with another device (not shown) and a server (2000). The other device (not shown) may be a computing device such as the electronic device (1000) or a sensing device, but is not limited thereto. For example, the network interface (1500) may include a wireless communication interface (1510), a wired communication interface (1520), and a mobile communication unit (1530). The wireless communication interface (1510) may include a short-range wireless communication unit, a Bluetooth communication unit, a Bluetooth Low Energy (BLE) communication unit, a Near Field Communication unit, a Wi-Fi (WLAN) communication unit, a Zigbee communication unit, an infrared (IrDA, infrared Data Association) communication unit, a Wi-Fi Direct (WFD) communication unit, an ultra-wideband (UWB) communication unit, but is not limited thereto.

[0094] The wired communication interface (1520) may include at least one wired interface for exchanging data with an external device connected to the electronic device via wired communication. The mobile communication unit (1520) transmits and receives wireless signals with at least one of a base station, an external terminal, and a server on a mobile communication network. Here, the wireless signal may include various forms of data such as voice call signals, video call call signals, or text / multimedia message transmission and reception.

[0095] The A / V (Audio / Video) input unit (1600) is for inputting audio or video signals and may include a camera (1610) and a microphone (1620), etc. The camera (1610) can obtain image frames, such as still images or video, through an image sensor in video call mode or shooting mode. Images captured through the image sensor can be processed through a processor (1300) or a separate image processing unit (not shown).

[0096] The microphone (1620) receives an external acoustic signal and processes it into electrical voice data. For example, the microphone (1620) can receive an acoustic signal from an external device or a user. The microphone (1620) can receive voice input from a user. The microphone (1620) can use various noise removal algorithms to remove noise generated during the process of receiving the external acoustic signal.

[0097] The memory (1700) can store a program for processing and controlling the processor (1300), and can also store data that is input to or output from the electronic device (1000). Additionally, the memory (1700) can store information regarding the user travel tendency analysis artificial intelligence model, language model, neural network-based model, STTI network model, cluster identification network, TRIFIT network, and WORD V LINKING network model used by the electronic device (1000).

[0098] For example, the memory (1700) may store weight values ​​regarding layers, nodes, and connection strengths of at least one neural network model. Additionally, the electronic device (1000) may further store training data generated by the electronic device (1000) to train the neural network model. Additionally, the memory (1700) may further store information regarding the operating environment of cameras or servers connected to the electronic device.

[0099] The memory (1700) may include at least one type of storage medium among flash memory type, hard disk type, multimedia card micro type, card type memory (e.g., SD or XD memory, etc.), RAM (Random Access Memory), SRAM (Static Random Access Memory), ROM (Read-Only Memory), EEPROM (Electrically Erasable Programmable Read-Only Memory), PROM (Programmable Read-Only Memory), magnetic memory, magnetic disk, and optical disk.

[0100] Programs stored in memory (1700) can be classified into multiple modules according to their function, for example, UI module (1710), touch screen module (1720), notification module (1730), etc.

[0101] The UI module (1710) can provide the electronic device (1000) with a UI, GUI, etc. for analyzing travel trends, travel location features, and providing travel content. The touch screen module (1720) can detect touch gestures on the user's touch screen and transmit information regarding the touch gestures to the processor (1300). In some embodiments, the touch screen module (1720) can recognize and analyze touch codes. The touch screen module (1720) may be configured as separate hardware including a controller.

[0102] The notification module (1730) can generate a signal to notify of the occurrence of an event of the electronic device (1000). For example, it can provide a notification sound when user feedback regarding travel content provided by the electronic device (1000) is obtained, or when notification content regarding travel tendencies or travel sentiments is provided. Examples of events according to one embodiment include receiving a call signal, receiving a message, inputting a key signal, and schedule notification. The notification module (1730) may output a notification signal in the form of a video signal through the display unit (1210), may output a notification signal in the form of an audio signal through the sound output unit (1220), or may output a notification signal in the form of a vibration signal through the vibration motor (1230).

[0103] FIG. 10 is a block diagram of a server according to one embodiment.

[0104] According to one embodiment, the server (2000) may include a network interface (2100), a database (2200), and a processor (2300). The configuration of the server (2000) shown in FIG. 10 may correspond to the configuration of the electronic device (1000) described in FIG. 8 and FIG. 9. The network interface (2100) may correspond to the network interface (not shown) of the electronic device (1000) described above. According to one embodiment, the network interface (2100) may receive information about an artificial intelligence model learned by the electronic device, or information about a neural network model (e.g., weight values ​​regarding layers and the connection strength between layers). According to another embodiment, the network interface (2100) may transmit to the electronic device (1000) information about the layers of an artificial neural network and the nodes included in the layers, or weight values ​​regarding the connection strength of the layers within the neural network, as information about a travel sentiment analysis artificial intelligence model learned by the server.

[0105] In addition, according to one embodiment, the database (2200) may correspond to the memory described above in FIGS. 8 and 9. For example, the database (2200) may store information such as social data, valid sentiment keywords, sentiment feature data, and traveler tendency feature data obtained from the electronic device (1000).

[0106] According to one embodiment, the processor (2300) can control the overall operation of the server (2000). For example, the processor (2300) can perform all or at least part of the operation of identifying user travel preferences and user-customized travel recommendation services described in FIG. 1 to 7 by controlling the network interface (2100) and the database (2200) in conjunction with the electronic device (1000).

[0107] FIG. 11 is a flowchart illustrating a method for identifying travel sentiment characteristics by travel destination by having an electronic device and a server interact with each other according to one embodiment.

[0108] In S1102, the electronic device (1000) can train an artificial intelligence model for travel sentiment analysis. In S1104, the server (2000) can train a TextRank-based sentence summary model. In S1106, the electronic device (1000) can acquire social data from social content that mentions travel destinations. In S1108, the electronic device (1000) can transmit the acquired social data to the server (2000). According to one embodiment, when social data is acquired at a predetermined interval, the electronic device (1000) can sequentially transmit each of the acquired social data to the server (2000).

[0109] In S1110, the server (2000) can input social data into a TextRank-based sentence summary model. In S1112, the server (2000) can obtain social summary data by inputting social data into a TextRank-based sentence summary model. In S1114, the server (2000) can obtain multiple morphological keywords by using a morphological analyzer. In S1116, the server (2000) can transmit the obtained multiple morphological keywords to an electronic device (1000). In S1118, the electronic device (1000) can input multiple morphological keywords into a travel sentiment analysis artificial intelligence model. In S1120, the electronic device (1000) can obtain sentiment feature data by inputting multiple morphological keywords into a travel sentiment analysis artificial intelligence model. The travel sentiment identification method described in FIG. 11 can be performed by a travel sentiment analysis system (10).

[0110] FIG. 12 is a flowchart of a method in which an electronic device according to one embodiment provides content about a travel destination based on user information and emotional feature data.

[0111] In S1210, the electronic device (1000) can obtain user information based on user input. According to one embodiment, the electronic device (1000) outputs an interface for obtaining user information and can obtain user information including at least one of gender information, age information, phone number information, information confirming consent to personal information, and information confirming receipt of marketing information through the interface. The electronic device (1000) may also transmit the obtained user information to a server (2000).

[0112] In S1220, the electronic device (1000) can identify user travel tendency characteristic information based on the acquired user information. According to one embodiment, when user information is acquired, the electronic device (1000) can identify user travel tendency characteristic information for said user from user-specific user travel tendency characteristic information stored in advance in the server (2000).

[0113] In S1230, the electronic device (1000) can identify the output transformation feature information by inputting the previously identified sentiment feature data by travel location into a location travel tendency variable linking model. According to one embodiment, the electronic device (1000) can identify the transformation feature information by inputting the sentiment feature data output from a travel sentiment analysis artificial intelligence model into a travel tendency variable linking model (e.g., Word V Linking model) which outputs transformation feature information representing the user's travel tendency according to the travel sentiment feature by travel location when the sentiment feature data is input. For example, the transformation feature information may correspond to the user's travel tendency feature information according to the travel sentiment feature by travel location.

[0114] In S1240, the electronic device (1000) can identify at least one transformation feature information among the transformation feature information based on a priority according to the matching or similarity score of the transformation feature information and the user travel tendency feature information. According to one embodiment, the electronic device (1000) determines a priority based on the matching score or similarity score between the user travel tendency feature information according to the user information obtained in S1210 and the transformation feature information output from the travel tendency variable linkage model, and can identify at least one transformation feature information based on the determined priority.

[0115] In S1250, the electronic device (1000) can output content regarding a travel destination corresponding to at least one transformation feature information. According to one embodiment, the content regarding the travel destination may include a character image representing the user's travel tendency, a character name, a text sequence containing a plurality of sentence sequences to explain the travel tendency, and link information for viewing the travel tendency content. According to another embodiment, the content regarding the travel destination may include an image representing the characteristics of the travel destination, a representative place, a text sequence containing a plurality of sentence sequences to explain the travel sentiment, and link information for viewing the travel sentiment content. However, it is not limited to the examples described above, and the content regarding the travel destination may further include other information to represent the user's travel tendency or travel sentiment specific to each travel destination.

[0116] A method for identifying travel sentiments by travel destination and a method for an electronic device to train an artificial intelligence model for travel sentiment analysis according to the present disclosure may be implemented in the form of program instructions that can be executed through various computer means and recorded on a computer-readable medium. The computer-readable medium may include program instructions, data files, data structures, etc., either individually or in combination. The program instructions recorded on the medium may be those specifically designed and configured for the present invention, or they may be those known and available to those skilled in the art of computer software.

[0117] Examples of computer-readable recording media include magnetic media such as hard disks, floppy disks, and magnetic tapes; optical recording media such as CD-ROMs and DVDs; magneto-optical media such as floptical disks; and hardware devices specifically configured to store and execute program instructions, such as ROM, RAM, and flash memory. Examples of program instructions include machine code, such as that generated by a compiler, as well as high-level language code that can be executed by a computer using an interpreter, etc.

[0118] Although embodiments of the present invention have been described in detail above, the scope of the present invention is not limited thereto, and various modifications and improvements by those skilled in the art using the basic concept of the present invention as defined in the following claims also fall within the scope of the present invention.

Claims

Claim 1 A method for an electronic device to train an artificial intelligence model for analyzing travel sentiment by travel destination comprises: a step of constructing a travel sentiment analysis artificial intelligence model that outputs design sentiment feature data based on design travel-specific dictionary data according to the correlation of design morphological keywords included in the design social data when design social data including text included in social content regarding the travel destination is input; and a step of training the travel sentiment analysis artificial intelligence model by inputting training social data by travel destination into the constructed travel sentiment analysis artificial intelligence model so that the training sentiment feature data obtained from the travel sentiment analysis artificial intelligence model is matched with the user's travel tendencies; wherein the step of constructing the travel sentiment analysis artificial intelligence model comprises: a step of obtaining design social summary data including summary sentences obtained from the design social data by using a TextRank-based sentence summary model; a step of obtaining design morphological keywords from the design social summary data; and a step of constructing the travel sentiment analysis artificial intelligence model such that when the design morphological keywords are input based on the design travel-specific dictionary data, the design sentiment feature data is output. The step of constructing the travel sentiment analysis AI model comprises: a step of constructing a word-to-vector model that outputs a morpheme keyword vector based on the correlation between morpheme keywords when the design morpheme keywords are input; and a step of constructing design-specific travel dictionary data for each representative sentiment keyword representing some of the design valid sentiment keywords among the design valid sentiment keywords selected from the morpheme keywords, based on the output value of the word-to-vector model.A method comprising: a step of constructing a Covert classification model that, when the above-mentioned valid sentiment keywords for design are input, outputs design sentiment feature data including feature values ​​for at least one representative sentiment keyword among the representative sentiment keywords representing the above-mentioned travel-specialized dictionary data, based on a similarity score between the derived sentiment keywords included in the above-mentioned design travel-specialized dictionary data and the above-mentioned input valid sentiment keywords for design; and a step of constructing a neural network-based numerical computation network regarding at least one of the TextRank-based sentence summarization model, Word-to-Vector model, and Covert classification model as the above-mentioned travel sentiment analysis artificial intelligence model. Claim 2 delete Claim 3 delete Claim 4 A method according to claim 1, wherein the TextRank-based sentence summary model outputs learning social summary data including a plurality of learning morphological keywords regarding the travel destination when learning social data is input, comprising neural network-based numerical computation networks. Claim 5 In claim 4, the method comprises neural network-based numerical computation networks, wherein the word-to-vector model outputs a plurality of valid sentiment keywords for learning regarding the travel destination when the plurality of learning morphological keywords are input. Claim 6 In claim 5, the method comprises neural network-based numerical computation networks, wherein the Covert classification model outputs learning sentiment feature data regarding the travel place based on learning travel-specific dictionary data when the plurality of valid learning sentiment keywords output from the Word-to-Vector model are input. Claim 7 In claim 6, the step of training the travel sentiment analysis artificial intelligence model comprises: a step of modifying and updating the weights of the TextRank-based sentence summary model so that the ratio in which a plurality of training morphological keywords included in the training social summary data are determined as valid training sentiment data output from the Word2Vector model increases; a method. Claim 8 In claim 7, the step of training the travel sentiment analysis artificial intelligence model comprises: a step of obtaining the training morphological keywords; a step of obtaining training valid sentiment keywords from the Word-to-Vector model by inputting the training morphological keywords into the Word-to-Vector model; a step of determining the training valid sentiment keywords whose similarity score is greater than or equal to a threshold score as training derived sentiment keywords based on the cosine similarity score between the training valid sentiment keywords and the training representative sentiment keywords included in the training travel-specialized dictionary data; a step of modifying and updating the training travel-specialized dictionary data by mapping the determined training derived sentiment keywords to the training representative sentiment keywords; and a step of training the travel sentiment analysis artificial intelligence model to output at least one training sentiment feature data that matches the user's travel tendency content based on the training travel-specialized dictionary data. Claim 9 In claim 8, the step of obtaining the valid sentiment keywords for training further comprises the step of modifying and updating the weights of the Word-to-Vector model based on the cosine similarity scores of the representative sentiment keywords for training and the valid sentiment keywords for training. Claim 10 In claim 8, the step of training the travel sentiment analysis artificial intelligence model further comprises the step of modifying and updating the weights of the Covert classification model so that the feature value of the representative sentiment keyword for training included in the sentiment feature data is matched with the feature value of each user's travel tendency variable. Claim 11 An electronic device for training an artificial intelligence model for analyzing travel sentiment by travel destination, comprising: a network interface; a memory for storing one or more instructions; and at least one processor for executing the one or more instructions. The method comprises: a travel sentiment analysis AI model that outputs design sentiment feature data based on design travel-specific dictionary data according to the correlation of design morphological keywords included in the design social data when design social data including text included in social content regarding the travel destination is input by executing one or more instructions; and by inputting training social data for each travel destination into the constructed travel sentiment analysis AI model so that the training sentiment feature data obtained from the travel sentiment analysis AI model is matched with the user's travel tendencies. The method comprises: the A Word-to-Vector model is constructed, and based on the output value of the above Word-to-Vector model, design-specific travel dictionary data is constructed for each representative sentiment keyword representing some of the design-specific valid sentiment keywords among the design-specific valid sentiment keywords selected from the above morphological keywords; and when the above design-specific valid sentiment keywords are input, the derived sentiment keywords included in the design-specific travel dictionary data, andAn electronic device that constructs a Covert classification model that outputs design sentiment feature data including feature values ​​for at least one representative sentiment keyword among the representative sentiment keywords representing the design travel-specialized dictionary data, based on similarity scores between the input valid design sentiment keywords, and constructs a neural network-based numerical computation network regarding at least one of the TextRank-based sentence summarization model, Word-to-Vector model, and Covert classification model as the travel sentiment analysis artificial intelligence model. Claim 12 A method for an electronic device to train an artificial intelligence model for analyzing travel sentiment by travel destination comprises: a step of constructing a travel sentiment analysis artificial intelligence model that outputs design sentiment feature data based on design travel-specific dictionary data according to the correlation of design morphological keywords included in the design social data when design social data including text included in social content regarding the travel destination is input; and a step of training the travel sentiment analysis artificial intelligence model by inputting training social data by travel destination into the constructed travel sentiment analysis artificial intelligence model so that the training sentiment feature data obtained from the travel sentiment analysis artificial intelligence model is matched with the user's travel tendencies; wherein the step of constructing the travel sentiment analysis artificial intelligence model comprises: a step of obtaining design social summary data including summary sentences obtained from the design social data by using a TextRank-based sentence summary model; a step of obtaining design morphological keywords from the design social summary data; and a step of constructing the travel sentiment analysis artificial intelligence model such that when the design morphological keywords are input based on the design travel-specific dictionary data, the design sentiment feature data is output. The step of constructing the travel sentiment analysis AI model comprises: a step of constructing a word-to-vector model that outputs a morpheme keyword vector based on the correlation between morpheme keywords when the design morpheme keywords are input; and a step of constructing design-specific travel dictionary data for each representative sentiment keyword representing some of the design valid sentiment keywords among the design valid sentiment keywords selected from the morpheme keywords, based on the output value of the word-to-vector model.A computer-readable recording medium storing a program that enables the execution of a method comprising: a step of constructing a Covert classification model that, when the above-mentioned valid sentiment keywords for design are input, outputs design sentiment feature data including feature values ​​for at least one representative sentiment keyword among the representative sentiment keywords representing the above-mentioned travel-specialized dictionary data, based on a similarity score between the derived sentiment keywords included in the above-mentioned design travel-specialized dictionary data and the above-mentioned input valid sentiment keywords; and a step of constructing a neural network-based numerical computation network regarding at least one of the TextRank-based sentence summarization model, Word-to-Vector model, and Covert classification model as the above-mentioned travel sentiment analysis artificial intelligence model.

Images