Automatic review response system based on review data learning

Abstract

The present invention relates to an automatic review response system and, more specifically, to a system that automatically responds to a target review on the basis of learning results of individual hotel review data. The automatic review response system based on review data learning, according to the present invention, comprises: a collection unit for collecting review data from an OTA server; a storage unit for storing the review data in a database; an analysis unit for analyzing the review data by using an artificial intelligence engine; and a transfer unit for transferring the analysis results of the analysis unit to the OTA server and an operator terminal.

Description

Automatic review response system based on review data learning

[0001] The present invention relates to an automatic review response system, and more particularly, to a system that automatically responds to target reviews based on the results of review data learning of individual hotels.

[0002] Customer reviews of businesses that provide services to customers, such as restaurants, beauty salons, and hotels, contain a variety of insights.

[0003] Customer reviews reflect customers' opinions and feelings, and a survey conducted by the Korea Consumer Agency found that over 95% of respondents answered positively to the question of whether they check other buyers' reviews before purchasing goods or services.

[0004] However, many businesses are unable to respond appropriately to customer reviews due to a lack of resources such as manpower and available time. For example, large hotels respond to customer reviews by assigning specialized personnel, but small hotels or motels are unable to pay attention to customer reviews.

[0005] The present invention has been proposed to solve the above-mentioned problems, and its purpose is to provide an automatic review response system that can replace dedicated personnel in responding to customer reviews with an artificial intelligence-based automatic review analysis and response (writing a reply to a review), thereby reducing the workload of a goods / service provider, and suggesting to a manager a problem and solution (e.g., recommending a professional cleaning company when a cleanliness issue arises in a hotel room) based on analysis of customer reviews.

[0006] The present invention relates to an automatic review response system, and more particularly, to a system that automatically responds to target reviews based on the results of review data learning of individual hotels.

[0007] An automatic review response system based on review data learning according to the present invention includes a collection unit that collects review data from an OTA server, a storage unit that stores the review data in a database, an analysis unit that performs analysis on the review data using an artificial intelligence engine, and a transmission unit that transmits the analysis results of the analysis unit to an OTA server and an operator terminal.

[0008] The collection unit collects the above review data for each hotel by parsing and crawling the page specified through the URL of the HTML.

[0009] In addition to the above review data, the collection unit collects information on the characteristics and surroundings of each hotel.

[0010] The above analysis unit automatically writes a reply to the review data using the review data, characteristic information of each hotel, and surrounding information.

[0011] The above analysis unit uses the results of analyzing the review data left by other customers to generate a pre-notification message to be provided to customers who have not yet left a review.

[0012] The above analysis unit generates the advance guidance text using the characteristic information and surrounding information of the individual hotel.

[0013] The above transmission unit transmits information of an automatically created reply to the above OTA server.

[0014] The above transmission unit transmits recommended business information collected based on the analysis results of the review data to the operator terminal.

[0015] According to the present invention, review management, which had not been paid attention to due to lack of manpower and available time, absence of a person in charge of the work, etc., can be performed, thereby providing customers with higher quality service in terms of customer management.

[0016] According to the present invention, it is possible to derive points for improvement in hotel operation based on the analysis results of customer reviews, thereby enabling hotel operation to be carried out in a direction with high customer satisfaction.

[0017] The effects of the present invention are not limited to those mentioned above, and other effects not mentioned will be clearly understood by those skilled in the art from the description below.

[0018] Figure 1 illustrates an automatic review response system based on review data learning, an OTA server, and a hotel manager terminal according to an embodiment of the present invention.

[0019] FIG. 2 illustrates a part of the data collection and storage process of an automatic review response system based on review data learning according to an embodiment of the present invention.

[0020] FIG. 3 illustrates the storage unit, analysis unit, and transmission unit of an automatic review response system based on review data learning according to an embodiment of the present invention, and the information transmission and reception process of an OTA server and a hotel manager terminal.

[0021] Figure 4 illustrates an automatic review response method based on review data learning according to an embodiment of the present invention.

[0022] FIG. 5 is a block diagram illustrating a computer system for implementing a method according to an embodiment of the present invention.

[0023] Figure 6 illustrates a system for providing search results based on customer review data according to an embodiment of the present invention.

[0024] Figure 7 illustrates a method for providing search results based on customer review data according to an embodiment of the present invention.

[0025] Figure 8 illustrates a review writer's tendency analysis and facility recommendation system according to another embodiment of the present invention.

[0026] Figure 9 illustrates a method for analyzing the tendencies of a reviewer and recommending facilities according to another embodiment of the present invention.

[0027] The above-described objects and other objects, advantages and features of the present invention, and methods for achieving them will become clear with reference to the embodiments described in detail below together with the accompanying drawings.

[0028] However, the present invention is not limited to the embodiments disclosed below, but can be implemented in various different forms, and the following embodiments are merely provided to easily inform a person having ordinary skill in the art to which the present invention pertains of the purpose, structure, and effect of the invention, and the scope of the rights of the present invention is defined by the description of the claims.

[0029] Meanwhile, the terminology used in this specification is for the purpose of describing embodiments and is not intended to limit the present invention. In this specification, the singular also includes the plural unless specifically stated otherwise. As used herein, "comprises" and / or "comprising" do not exclude the presence or addition of one or more other components, steps, operations, and / or elements to the mentioned components, steps, operations, and / or elements.

[0030] According to an embodiment of the present invention, in order to support the operational efficiency of small and medium-sized accommodation facilities based on customer review data, problems contained in customer reviews are analyzed, replies to the reviews are automatically provided, and solutions for improving the analyzed problems are provided and applied to the operation of the accommodation facilities.

[0031] According to an embodiment of the present invention, it is possible to reduce the gap in service quality between large hotels and small and medium-sized hotels by automatically providing, based on learning, a customer management service called replying to reviews, which is available in large hotels but not in small and medium-sized hotels.

[0032] Figure 1 illustrates an automatic review response system based on review data learning, an OTA server, and a hotel manager terminal according to an embodiment of the present invention.

[0033] An automatic review response system (200) according to an embodiment of the present invention includes a collection unit (210) that collects review data from an OTA server (100), a storage unit (220) that stores the review data, an analysis unit (230) that learns and analyzes the review data, and a transmission unit (240) that transmits the analysis results to the OTA server (100) and a hotel manager terminal (300).

[0034] According to an embodiment of the present invention, review data of accommodation facilities is collected, analyzed and processed using artificial intelligence, and utilized for hotel operation, and replies to the reviews are automatically created and provided to customers.

[0035] The collection unit (210) collects review data for each hotel through page parsing and crawling on a page specified through an HTML URL.

[0036] The storage unit (220) stores review data collected for each hotel in a database.

[0037] The analysis unit (230) uses an artificial intelligence engine to perform metadata analysis on review data.

[0038] The transmission unit (240) transmits the analyzed data (analysis result) to the OTA server and the hotel operator, and the analysis result includes reply data corresponding to the review.

[0039] According to an embodiment of the present invention, an artificial intelligence engine learns review data of individual accommodations collected through a crawler, facility information such as facilities owned and surrounding commercial district characteristics, and utilizes this to write responses for each hotel. In addition, the analysis results of the review data are transmitted to the reservation office, manager, and dedicated company, thereby being utilized for follow-up actions on customer reviews.

[0040] Each hotel has its own unique characteristics (e.g., hotels near historical sites offer convenient transportation but are noisy; hotels near famous tourist attractions often attract tourists from outside the region and therefore require recommendations for nearby restaurants). Therefore, the review data generated varies. The analysis unit (230) reflects these unique characteristics of each hotel, analyzes the review data, and automatically generates replies to the reviews. The analysis unit (230) learns the review data and unique characteristics of each hotel, and generates and provides replies and recommended operation plan information tailored to the unique characteristics of the hotel.

[0041] FIG. 2 illustrates a part of the data collection and storage process of an automatic review response system based on review data learning according to an embodiment of the present invention.

[0042] The OTA server (100) includes multiple servers, namely, a first OTA server (100a), a second OTA server (100b), and an nth OTA server (100n).

[0043] The communication module (211) included in the collection unit (210) requests customer review data from the OTA server (100) and receives and collects the customer review data from the OTA server (100). At this time, in addition to the customer review data, the communication module (211) collects characteristic information of each hotel (whether there is a ground line nearby, whether it is located within a preset distance from a famous tourist attraction, whether a local festival is being held within a preset distance, etc.).

[0044] The communication module (211) included in the collection unit (210) transmits customer review data, characteristic information of individual hotels, and surrounding information to the data management module (221) of the storage unit (220).

[0045] The data management module (221) included in the storage unit (220) classifies the transmitted data and stores it in a database. The data classification includes price data, review data, festival data, weather data, and analyzed review data.

[0046] The analyzed review data is analysis data for customer reviews received from the analysis unit (230), and the data is also stored in a database and used for relearning and automatic reply writing.

[0047] FIG. 3 illustrates the storage unit, analysis unit, and transmission unit of an automatic review response system based on review data learning according to an embodiment of the present invention, and the information transmission and reception process of an OTA server and a hotel manager terminal.

[0048] The data management module (221) transmits customer review data to the analysis unit (230).

[0049] The analysis unit (230) includes a review data analysis module (231), a reply writing module (232), and a transmission module (233). Analysis data and replies, which are the analysis results performed by the analysis unit (230), are transmitted to the storage unit (220), and the storage unit (220) transmits them to the transmission unit (240).

[0050] The reply writing module (232) included in the analysis unit (230) comprehensively considers the characteristic information of each hotel and the analysis results received from the review data analysis module (231), and automatically writes text and emoticons to be attached as replies to the customer reviews.

[0051] For example, it is assumed that factors such as <formality>, <intimacy>, <empathy>, <problem solution>, and <reward provision> are taken into consideration when writing automatic replies.

[0052] For individual hotel A, it is assumed that it is classified as a guesthouse and is a facility mainly used by people in their early 20s, such as college students. For individual hotel B, it is assumed that it is a 5-star hotel and is a facility mainly used by people in their 40s or older.

[0053] The reply writing module (232) considers the <formality> factor, determines the reply writing stance so that the <intimacy> of the reply writing is greater than a preset value, and writes the reply text accordingly.

[0054] If the <Formality> factor index of individual hotels A and B is greater than the preset value, the reply writing stance is determined differently so that the difference in <Intimacy> of the reply is greater than the preset value. This can be determined by whether to use emoticons in the reply, or whether to use ~, !, ..., ^^, ㅠㅠ, etc.

[0055] If the <Formality> factor indices of individual hotels A and B are greater than a preset value, the reply writing stance is determined differently so that the difference in <Sympathy> of the replies is greater than a preset value. For example, let's assume that a user of individual hotel A left a review data with the gist of <The bunk bed was uncomfortable>, and a user of individual hotel B left a review data with the gist of <The mattress seemed a bit hard>. The reply writing stance is determined to determine the <Sympathy> level of the automatically written reply by comprehensively considering the customer's gender, age group, the degree of customer dissatisfaction included in the review data analysis results, and whether the reply is classified as an urgent corrective action target. For users who stayed on the second floor of a bunk bed at individual hotel A, we will decide on a reply writing stance to include a phrase expressing sympathy along the lines of <The ladder to the bunk bed is narrow, and you must have had difficulty going up and down multiple times>. For users who stayed on the first floor of a bunk bed, we will decide on a reply writing stance to include a phrase expressing sympathy along the lines of <Our bunk bed is made of wood, so even the slightest movement of the user on the second floor can cause a loud creaking sound, which must have been inconvenient>. For users of individual hotel B, we will decide on an automatic reply writing stance to include a phrase expressing sympathy along the lines of <Sympathy> along the lines of <Hotel B provides mattresses made of the highest quality ~ material to our guests, but in the case of ~, they may have felt somewhat firm. However, in recognition of your valuable feedback, we are considering replacing the mattress with a mattress from brand OO. If you would like to leave a comment in advance the next time you stay, we will help you reserve a room with a renewed mattress from brand OO>.The stance for writing automatic replies to the <Empathy> index is ultimately a factor that must be taken into consideration in terms of the individual hotel's capital power for the <Problem Resolution> and <Reward Granting> indices. Therefore, you can decide whether to actively empathize with the customer's complaints but only provide a consolation message, or to give more weight to active problem-solving methods even if it requires a certain amount of capital investment depending on the customer's level (customers who visit more than 4 nights a year and have a gold level or higher).

[0056] The reply writing module (232) determines an automatic reply writing stance based on the classification elements of complaint data among the customer review data received from the review data analysis module (231). For example, if the customer review data includes phrases such as <ㅠㅠ, ㅜㅜ, ~용, ~하다>, etc., the automatic reply writing stance is determined by applying an upward weight that is greater than a preset value to the <customer intimacy> index, so that an automatic reply is written that feels friendly rather than a stiff, businesslike tone. If the customer review data includes phrases such as <severely, rigorously, top manager, responsibility, compensation, indemnification>, the automatic reply writing stance is determined by applying a downward weight that is greater than a preset value to the <customer intimacy> index, so that an automatic reply is written in a professional and businesslike tone.

[0057] The reply writing module (232) may, in some cases, write a preemptive warning message before generating customer review data. For example, if construction is underway around the hotel and customers traveling by car are expected to experience inconvenience in entering and exiting the parking lot, it is possible to provide a preemptive warning message such as, "Due to the subway station construction that took place from ~month ~day, we ask if customers using vehicle OOOO experienced any inconvenience. If you experienced any inconvenience other than the inconvenience caused by the construction, please leave a message in the review data or at our hotel desk, and we will listen carefully to your concerns." The preemptive warning message may be provided to the customer's terminal via a push alarm, etc.

[0058] In other words, in addition to automatically responding to review data containing customer complaints, you can proactively provide preemptive guidance messages, including comforting and thank-you messages, in response to anticipated customer complaints.

[0059] At this time, the reply writing module (232) crawls and analyzes SNS review data and review data left by other customers in advance, and comprehensively considers the customer's gender, age group, whether the facility was used, and information on complaints from other customers received during the same stay period to provide a prior guidance message.

[0060] For example, if the complaint information of other customers during the same stay is information that the customer is expected to have experienced the same inconvenience (e.g., the complaint was about a long breakfast line and the customer used breakfast), then such information should be applied to the pre-notification text. However, if the complaint information of other customers during the same stay is information that the customer is expected to not have experienced the same inconvenience (e.g., the complaint was about a long breakfast line or the customer did not use breakfast), then such information should not be applied to the pre-notification text. As another example, even if the complaint information of other customers during the same stay is information that the customer is expected to not have experienced the same inconvenience (e.g., the complaint was about a long breakfast line or the customer did not use breakfast), then the customer may not have used breakfast due to reasons such as a long breakfast line based on the review data of other purchasers in advance. Therefore, based on the analysis results of existing review data, <OOO 고객님, 1박 2일의 일정은 편안하셨는지요. 저희 AAA 호텔에서는 저희 지역의 특산물인 BBB를 활용한 CCC 메뉴를 주 메뉴로 하는 조식 서비스를 제공 드리고 있습니다만, 금번에는 조식 서비스를 이용하지 않으셨던 것으로 확인됩니다. 대기 줄이 길다는 다른 고객님들의 불편사항 해결을 위해, 조식 서비스 동선을 개선하여 운영 중에 있사오니, 다음 번에 방문하시는 경우에는 저희 조식 서비스를 한번 이용해보시는 것을 추천 드립니다. 특별 프로모션으로 3개월 이내에 재방문하시어 조식 서비스 이용시, OO% 할인 혜택을 제공드리오니 참고 부탁 드립니다. 감사합니다.> You can decide the stance of writing an automatic reply by writing a pre-information message.

[0061] The transmission unit (240) includes a reply transmission module (241) and an analysis content transmission module (242). The reply transmission module (241) transmits a reply to a specific automatically generated review to the OTA server (100), so that the reply is posted on the review data. The analysis content transmission module (242) transmits the analysis results for the customer review data to the hotel manager terminal (300).

[0062] The hotel manager terminal (300) includes a recommended cleaning company judgment module (310) and a recommended FnB company judgment module (320).

[0063] When a problem with the cleanliness of a room occurs among the review data analysis results received from the delivery unit (240), the recommended cleaning company judgment module (310) is called, and based on the crawled data, recommended information on the cleaning company is generated and provided to the hotel manager.

[0064] Among the review data analysis results received from the delivery unit (240), if a problem occurs with food such as breakfast or room service or with service at a cafe, the recommended FnB business judgment module (320) is called, and based on the crawled data, recommended information on the FnB business is generated and provided to the hotel manager.

[0065] According to an embodiment of the present invention, a relay server is established to perform data crawling to generate information on the characteristics of individual hotels, customer review data, recommended cleaning service providers, and recommended FnB service providers. To collect data, securing a stable proxy is required. However, if collection exceeds a certain threshold for each site using a fixed IP address, the IP address may be banned. Since distributed IP addresses offer advantages in achieving the collection objective, an embodiment of the present invention utilizes a relay server formed by a group of devices with dynamic IP addresses during the crawling process.

[0066] According to an embodiment of the present invention, a process of creating a group of devices that perform data collection for each purpose of collecting characteristic information of individual hotels, collecting customer review data, generating recommended cleaning company information, and generating recommended FnB company information, and releasing and re-grouping the group is performed.

[0067] The management module manages the registration status for the relay server by using information about devices joining and leaving the group.

[0068] The management module monitors the communication load of the grouped devices and subgroups at least one device within the relay server to perform data collection according to the crawling request.

[0069] The management module performs sub-grouping of devices by considering the expected task completion time of devices that are currently performing data collection requests based on other crawling requests, and requests the introduction of additional devices to the sub-group if additional device specifications are required.

[0070] The management module considers the characteristics of the data collection request (urgency, time-series characteristics of the information collection target), integrates the first subgroup that is currently performing tasks according to the first data collection request, and the second subgroup that is to perform tasks according to the new second data collection request, controls the tasks according to the second data collection request to be intensively performed for a certain period of time, and then regroups them into the first and second subgroups to perform their respective existing tasks.

[0071] If the data to be collected by the first subgroup is below a preset level for real-time collection (since the request is for "daily cumulative data collection: cumulative data aggregation for one day from 22:00 to 24:00 every day," and the required operating time for data collection is 1 / 12 of a day), the first subgroup can be temporarily utilized for other data collection purposes. In this case, the first subgroup is incorporated into an integrated group that performs tasks according to the new second data collection request, performs second data collection during the preset period, and performs tasks according to the original first data collection request during other periods.

[0072] The management module operates a standby group when a new device is added to the group to manage the relay server.

[0073] The management module monitors the communication status of each device within the relay server and transmits a data collection request to designate a mobile device to perform the task. It also monitors the computational load required for monitoring and designation.

[0074] As the number of monitored devices increases, the management module determines the timing of permission for inclusion of new devices in a group by considering the computational load required for monitoring and designation.

[0075] When the computational load required for monitoring and specifying the communication status of devices within a relay server is assumed to be 1, and when n new devices are added, the computational load required for monitoring and specifying the communication status is determined to be 2, the increase in computational load is 100%.

[0076] At this time, considering the increase / decrease in the number of data collection requests currently being received, if n devices newly added should be allowed to be immediately incorporated into the group despite the increase in computational load, n devices are immediately incorporated into the group.

[0077] On the other hand, if it is determined that there is no need to incorporate n new mobile devices at this point in time, considering the increase / decrease in the number of data collection requests currently being received, a reserve group is operated / managed so that they can be incorporated into the relay server group immediately when necessary in the future.

[0078] It is desirable to efficiently perform management of the relay server by managing the groups in multiple layers within the relay server, such as a first group that prioritizes tasks for collecting data for the target website, a second group that reserves tasks by considering data collection request characteristics (amount of data to be collected, collection period, etc.) without prioritizing tasks, and a reserve group mentioned above when a new device is incorporated.

[0079] Figure 4 illustrates an automatic review response method based on review data learning according to an embodiment of the present invention.

[0080] An automatic review response method based on review data learning according to an embodiment of the present invention includes a step of collecting individual hotel characteristic information, surrounding information, and review data (S410), a step of analyzing the review data and writing an automatic reply (S420), and a step of transmitting automatic reply information and reflecting the review data in hotel operation (S430).

[0081] In step S410, the collection unit (210) collects review data for each hotel through page parsing and crawling from a page specified by an HTML URL. In step S410, information on the characteristics of each hotel is collected, including characteristics of each hotel (e.g., hotels near historical sites have convenient transportation but are noisy, hotels near famous tourist sites have many tourists from other regions and therefore need recommendations for nearby restaurants, etc.). In step S410, surrounding information is collected, including information on surrounding festivals and weather.

[0082] In step S420, an AI engine is used to perform metadata analysis on the review data. In step S420, the automated reply writing stance is determined based on the characteristics of each hotel, surrounding information, and the review data analysis results. Factors such as <formality>, <intimacy>, <empathy>, <problem resolution>, and <reward granting> are considered when determining the automated reply writing stance.

[0083] At step S420, the same user's reviews for Hotels A, B, and C are shared with multiple OTA servers. Based on the analysis of these reviews, a decision on whether the user is a black consumer is shared with multiple OTA servers and the hotel operator terminal. This decision determines the automatic response stance and determines whether to grant a reward.

[0084] At step S420, based on analysis of other customers' reviews, a preemptive message is generated and provided before the customer leaves a review. In addition to automatically responding to reviews containing customer complaints, the system proactively provides messages of consolation and gratitude for anticipated customer complaints. This message is generated by comprehensively considering factors such as the customer's gender, age group, facility usage, and complaints from other customers during the same stay.

[0085] At step S430, if a problem with the cleanliness of the room is found in the review data analysis, the recommended cleaning company judgment module is called and the results of the recommended cleaning company judgment are reflected in the hotel operation.

[0086] At step S430, if a problem occurs with food such as breakfast or room service or with the response at the cafe among the review data analysis results, the recommended FnB business judgment module is called and the results of the recommended FnB business judgment are reflected in the hotel operation.

[0087] FIG. 5 is a block diagram illustrating a computer system for implementing a method according to an embodiment of the present invention.

[0088] Referring to FIG. 5, a computer system (1300) may include at least one of a processor (1313), a memory (1330), an input interface device (1350), an output interface device (1360), and a storage device (1340) that communicate via a bus (1370). The computer system (1300) may also include a communication device (1320) coupled to a network. The processor (1310) may be a central processing unit (CPU), or a semiconductor device that executes instructions stored in the memory (1330) or the storage device (1340). The memory (1330) and the storage device (1340) may include various forms of volatile or non-volatile storage media. For example, the memory may include read-only memory (ROM) and random access memory (RAM). In embodiments of the present disclosure, the memory may be located internally or externally to the processor, and the memory may be connected to the processor via various known means. Memory is a variety of volatile or non-volatile storage media, and may include, for example, read-only memory (ROM) or random access memory (RAM).

[0089] Accordingly, embodiments of the present invention may be implemented as a computer-implemented method or as a non-transitory computer-readable medium storing computer-executable instructions. In one embodiment, when executed by a processor, the computer-readable instructions may perform a method according to at least one aspect of the present disclosure.

[0090] The communication device (1320) can transmit or receive wired or wireless signals.

[0091] In addition, the method according to the embodiment of the present invention may be implemented in the form of program commands that can be executed through various computer means and recorded on a computer-readable medium.

[0092] The computer-readable medium may include program commands, data files, data structures, etc., either singly or in combination. The program commands recorded on the computer-readable medium may be specially designed and configured for embodiments of the present invention, or may be known and usable by those skilled in the art of computer software. The computer-readable recording medium may include a hardware device configured to store and execute the program commands. For example, the computer-readable recording medium may be a magnetic medium such as a hard disk, a floppy disk, and a magnetic tape, an optical medium such as a CD-ROM or a DVD, a magneto-optical medium such as a floptical disk, a ROM, a RAM, a flash memory, etc. The program commands may include not only machine language codes such as those generated by a compiler, but also high-level language codes that can be executed by a computer through an interpreter, etc.

[0093] Although the 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 made by those skilled in the art using the basic concept of the present invention defined in the following claims also fall within the scope of the present invention.

[0094] Below, a system for providing search results based on customer review data according to another embodiment of the present invention is described.

[0095] Customers are increasingly booking accommodations through online travel agencies (OTAs). Customers use OTAs to search for accommodations of their choice, using basic information like price, location, and property name.

[0096] Since basic information alone cannot provide detailed information about the specific characteristics of an accommodation or the level of satisfaction with its services, customers often refer to reviews left by customers who have used the accommodation before them.

[0097] Conventional OTAs do not provide convenient features for referencing review information, such as review search, so customers spend considerable effort and time searching for accommodations they want.

[0098] The present invention has been proposed to solve the aforementioned problems, and proposes a search result provision system based on customer review data that can identify the characteristics of accommodations by building an AI model capable of analyzing the strengths, weaknesses, preferred facilities, and features of accommodations by performing accommodation review data learning, and recommend accommodations with similar characteristic information to the characteristics desired by customers who wish to use the accommodations.

[0099] A system for providing search results based on customer review data according to another embodiment of the present invention includes an AI analysis unit (610) that performs learning and analysis on customer review data, a storage unit (620) that stores the analysis results of the AI ​​analysis unit (610), and a service provision unit (630) that provides information on recommended facilities corresponding to request information of potential customers.

[0100] The AI ​​analysis unit (610) performs learning using review data on existing customers' accommodations stored in OTA, and performs analysis on the review data to generate analysis results.

[0101] The AI ​​analysis unit (610) learns the original review text included in the review data and generates characteristic information of the corresponding accommodation facility based on the analysis results.

[0102] The AI ​​analysis unit (610) uses the results of learning to classify review data by item, including the type of use, facilities, surrounding infrastructure, service, and cleanliness of the accommodation, and analyzes the sentiment of positive, negative, and neutral by item.

[0103] The AI ​​analysis unit (610) generates characteristic information of the accommodation facility using the focus information of the image information included in the review data and the object information included in the central area.

[0104] The AI ​​analysis unit (610) groups and analyzes multiple review data that have different evaluations of the same facility of the accommodation facility.

[0105] The storage unit (620) combines and stores basic information including the facility name, address, usage price, and facility type with analysis results.

[0106] The service provider (630) analyzes the text received from the potential customer to derive the request information and generates information on recommended facilities corresponding to the request information.

[0107] According to another embodiment of the present invention, a customer who wishes to use an accommodation through an OTA can conveniently search for and reserve an accommodation having the characteristics he or she desires by utilizing search results based on customer review data, in addition to utilizing basic information about the accommodation provided by the OTA, thereby overcoming the limitations of the search function of an OTA according to the prior art.

[0108] According to another embodiment of the present invention, an AI model trained using customer review data is utilized to identify accommodations' characteristic information, including customer satisfaction information and facility usage reviews. According to another embodiment of the present invention, when searching for and booking accommodations, customers can rationally make reservations by referencing various facility characteristic information and satisfaction information, overcoming the limitations of conventional techniques that rely solely on basic information such as the facility's name, rating, location, and price.

[0109] FIG. 6 illustrates a system for providing search results based on customer review data according to another embodiment of the present invention.

[0110] A system for providing search results based on customer review data according to another embodiment of the present invention includes an AI analysis unit (610) that learns and analyzes pre-stored review data, a storage unit (620) that stores analyzed review data, and a service provision unit (630) that recognizes searched sentences and provides search results.

[0111] The AI ​​analysis unit (610) learns and analyzes the original text of the previously stored customer review data to generate characteristic information of the corresponding accommodation. The AI ​​analysis unit (610) learns the collected customer reviews and performs a classification task based on the learned model. The AI ​​analysis unit (610) classifies the customer review data into categories including the type of accommodation use, the facilities of the accommodation, the surrounding infrastructure of the accommodation, service, and cleanliness, and analyzes the sentiment of positive, negative, and neutral for each category. For example, if the review data of customer B for Hotel A is <It was great for a family trip. In particular, the swimming pool was the best among the hotels I've been to. I'm very satisfied.>, it is classified into <1) type of use, family, emotional evaluation, positive 2) facilities, swimming pool, emotional evaluation, positive>.

[0112] If a customer attaches a photo to their review data, the AI ​​analysis unit (610) extracts facility information from the photo information, inferring that the facility has a high customer satisfaction rating. For example, if a customer attaches a photo of a heated pool, outdoor sauna, and water slide while playing in the water with an 8-year-old child, the AI ​​analysis unit (610) determines that the customer's satisfaction with the facility is high.

[0113] The AI ​​analysis unit (610) comprehensively analyzes reviews from customers who have given different evaluations of the same facility. For example, among other reviews of Hotel A, if the review data of Customer C who used Hotel A states, "The slide seems too steep for a 3-year-old child. However, the heated pool made it great for swimming in the evening," it is classified into <1) Usage type, family, emotional evaluation, positive 2) Facilities, swimming pool, emotional evaluation, neutral>. At this time, with regard to the facilities, the sentiment analysis results are negative for the slide, positive for the heated pool, and overall neutral.

[0114] The service provider (630) plans a family trip with a 3-year-old child and provides customer D, who wants to use the swimming pool, with review data from customers B and C who have previously used hotel A, and provides information on the facilities available, such as <satisfaction levels may vary depending on the age of the child using the swimming pool>.

[0115] In addition, the service provision department (630) can analyze the review data of customers B and C who previously used hotel A to induce reservations by customer D, and generate and provide the recommended information text, such as <Another customer who used the swimming pool with a 3-year-old child said that the slope of the slide was somewhat steep. Accordingly, we have additionally installed a children's slide suitable for children 5 years old and younger, so please use it a lot.>

[0116] The storage unit (620) combines and stores the analysis results of the AI ​​analysis unit (610) with basic information about the accommodation facility (basic information such as the name, address, average selling price, and type of facility of the accommodation facility). The service provision unit (630) receives the needs information of a prospective customer through a user interface, receives the data from the storage unit (620), and generates and provides a list of accommodation facilities requested by the customer. For example, the aforementioned review analysis information is stored together with basic information such as 'Hotel A', 'Gangwon-do', 'Hotel', and '120,000~150,000'.

[0117] The service provider (630) provides a list of accommodations that match the customer's search for accommodation information based on data stored in the database. For example, if a customer searches for "a hotel that's good for a family of three with a 5-year-old child and includes a swimming pool," the service provider can provide a list of accommodations that match the search criteria, including "Hotel A," and since the region, price range, and type of accommodation are not specified, the service provider can recommend accommodations that match the characteristics of each region, price range, and accommodation type.

[0118] When a customer uses a hotel they searched for and writes a new review, the review is sent back to the AI ​​analysis unit (610) and used for learning and updating the features of the hotel.

[0119] FIG. 7 illustrates a method for providing search results based on customer review data according to another embodiment of the present invention.

[0120] A method for providing search results based on customer review data according to another embodiment of the present invention includes a step (S710) of obtaining customer review information to identify facility characteristic information and customer satisfaction information, a step (S720) of storing facility analysis information in conjunction with basic information, and a step (S730) of generating and providing a list of facilities corresponding to a customer request.

[0121] In step S710, customer review data stored in OTAs is acquired and the original review data is learned. In step S710, customer review data is categorized into categories including accommodation usage, accommodation facilities, surrounding infrastructure, service, and cleanliness. Each category in the customer review data is analyzed for positive, negative, and neutral sentiment. In step S710, each category is analyzed for positive, negative, and neutral sentiment, and an overall sentiment analysis is performed. For example, if most categories are analyzed as positive, but one category has a significantly higher negative sentiment than the threshold, the overall sentiment is analyzed as negative. In step S710, customer review data is analyzed using a learning engine and, when identifying facility characteristics and customer satisfaction information, photos attached by customers are considered for analysis. For example, an object within the focus or central area of ​​a photo is determined to be positive or negative. If a customer using a swimming pool attaches a number of photos taken in the heated pool that is greater than the standard (e.g., more than twice as many) compared to other photos, the customer satisfaction score for the facility is determined by considering factors such as the emotional score analyzed from the facial expressions of the people in the photos and the inclusion of satisfaction-related words in the text.

[0122] In step S710, reviews with different ratings for the same facility are grouped and analyzed. Referring to the example above, if reviews for Hotel A—namely, those from Customer B and Customer C who used Hotel A—include reviews of the swimming pool, these reviews are grouped and analyzed. Customer B, a parent with an 8-year-old child, provides a positive review of the swimming pool, while Customer C, a parent with a 3-year-old child, provides positive reviews of some features of the pool and negative reviews of others. Even for the same facility, customer reviews can vary depending on who uses it, when they use it, and the conditions of the day. Therefore, step S710 comprehensively analyzes the review data and provides suggestions for improvement to the accommodation manager. Furthermore, step S710 manages the review data so that potential buyers (potential customers searching for accommodations) who will refer to these reviews are provided with the review data of Customers B and C who previously used Hotel A. Step S730 analyzes customer request information, generates and provides a list of accommodations that meet customer requests, and if Hotel A is included in the list, it provides information on the facilities it owns, such as <satisfaction levels may vary depending on the age of children using the swimming pool>, along with review data from customers B and C. In addition, if there is information on areas for improvement for the facility in order to induce reservations, it generates and provides advertising text that includes information on areas for improvement.

[0123] In step S720, the results of the analysis performed in step S710 and basic information about the accommodation facility (basic information such as the name, address, average selling price, and type of facility of the accommodation facility) are combined and stored.

[0124] At step S730, the potential customer's needs information is input through a user interface, and a list of recommended accommodations corresponding to the potential customer's request is generated and provided.

[0125] Below, a review writer's tendency analysis and facility recommendation system according to another embodiment of the present invention is described.

[0126] Conventional technology allows customers who book and use accommodations through online travel agencies (OTAs) to receive hotel recommendations from the OTAs based on various criteria. However, customers often lack the understanding of the specific reasons for a particular accommodation recommendation, leading to low trust and satisfaction with the recommendation service.

[0127] A review writer's tendency analysis and facility recommendation system according to another embodiment of the present invention is proposed to solve the above-mentioned problem, and the purpose is to provide a system that can increase customer satisfaction with facility use and reliability of recommendation services by identifying the customer's exact tendency and needs and using the quantified data to provide a recommendation service for facilities that can satisfy the said tendency and needs.

[0128] A review writer's tendency analysis and facility recommendation system according to another embodiment of the present invention includes an AI-based customer tendency analysis unit (910) that analyzes customer review data to derive customer tendency information, a storage unit (920) that links and stores customer tendency information and customer basic information, and a service provision unit (930) that identifies customer needs and provides facility recommendation information that takes customer tendency information into consideration.

[0129] The AI-based customer tendency analysis unit (910) analyzes customer review data stored in OTA and derives tendency information using the results of review data analysis according to preset classification.

[0130] The AI-based customer tendency analysis unit (910) considers the preset classification of the characteristics of the facility, assigns the number of mentions and positivity to the review data, and derives tendency information.

[0131] The storage unit (920) stores preference information in conjunction with basic information including residence and preferred region information.

[0132] The service provider (930) receives a text containing request information from a customer and analyzes the text to identify the customer's needs.

[0133] The service provider (930) uses the results of analyzing the customer's tendency information and review data of other customers with a similarity level above a preset level to generate facility recommendation information expected to meet the customer's needs.

[0134] The AI-based customer tendency analysis unit (910) derives customer tendency information by using customer consumption pattern information.

[0135] The service provider (930) generates facility recommendation information by considering consumption pattern information when text with the same dictionary meaning is used differently by each customer.

[0136] According to another embodiment of the present invention, rather than providing unclear recommended facility information, the customer can be provided with information on recommended facilities based on the inclinations derived from analysis of review data actually written by the customer.

[0137] Figure 8 illustrates a review writer's tendency analysis and facility recommendation system according to another embodiment of the present invention.

[0138] According to another embodiment of the present invention, a reviewer tendency analysis and facility recommendation system analyzes previously written review data to identify a customer's tendencies and recommends facilities that have received positive reviews from other customers with similar tendencies. The system identifies a customer's tendencies by analyzing various characteristics of the accommodation, such as "cleanliness," "price," "service," "facilities," "interior," and "breakfast," which are frequently or significantly mentioned by the customer. Based on an AI model for review classification, the system accurately identifies the customer's tendencies and needs, clusters customers with similar tendencies, and recommends accommodations expected to have high customer satisfaction, taking into account the customer's tendencies.

[0139] A review writer's tendency analysis and facility recommendation system according to another embodiment of the present invention includes an AI-based customer tendency analysis unit (910) that learns and analyzes review data written by customers, a storage unit (920) that stores the analyzed customer tendency information and basic information, and a service provision unit (930) that provides a list of recommended accommodations based on the customer data stored in the storage unit (920).

[0140] The AI-based customer tendency analysis unit (910) obtains and analyzes review data written by the customer to determine the tendency of the customer. The AI-based customer tendency analysis unit (910) classifies the reviews written by the customer into various characteristics of the accommodation, such as 'cleanliness', 'price', 'service', 'facilities', 'interior', and 'breakfast', and assigns scores based on the number of mentions and positivity to determine which characteristics of the hotel the customer considers most important. For example, if a customer writes three reviews including <This hotel has really good value for money. I was very satisfied>, <I was satisfied with the value for money and the friendliness of the staff, but the cleanliness was disappointing>, and <It's good value for money and clean. I had a good stay.>, the AI-based customer tendency analysis unit assigns 3 points to 'price', 2 points to 'cleanliness', and 1 point to 'service' to classify the customer tendency.

[0141] At this time, customer preference information is analyzed and segmented by category. For example, the dictionary definition of "cost-effectiveness" is "performance for the price," but it can be used differently depending on each customer's preference. For example, customer A primarily uses accommodations priced under 100,000 won and leaves a review. If the accommodation price itself falls below their standard (100,000 won), they will leave a review praising the accommodation as "good value for money." Another example is customer B, who primarily uses accommodations in the 500,000 won range. Even if the accommodation price is higher than average, their review data may indicate "good value for money" due to their high satisfaction with the facilities and services. In this way, the same term can be used differently based on each customer's consumption patterns. If a 500,000 won accommodation were recommended to customer A simply based on the "cost-effectiveness" category, using customer B's review data, customer A would likely find it difficult to trust the recommended service. Therefore, a preference analysis is performed on each accommodation category, taking into account each consumer's consumption patterns to generate a list of recommended accommodations.

[0142] The storage unit (920) stores the customer's preferences identified by the AI-based customer preference analysis unit, along with basic information about the customer (name, residence, preferred region, etc.). The customer's information is stored in conjunction with their ID, residential area, and preferred region (primarily traveled to).

[0143] The service provider (930) clusters other customers with similar preferences to the customer and generates and provides a list of recommended facilities that meet the customer's needs. The service provider (930) recommends accommodations for which other customers with similar preferences have left positive reviews. The service provider (930) integrates basic information, such as the customer's preferred region, to recommend accommodations under various categories, such as "Recommended New Areas" and "Recommended Hotels in Frequently Visited Areas."

[0144] When a customer uses an accommodation and writes a new review, the new review is sent back to the AI-based customer tendency analysis unit (910) and used for customer tendency analysis.

[0145] Figure 9 illustrates a method for analyzing the tendencies of a reviewer and recommending facilities according to another embodiment of the present invention.

[0146] A method for analyzing the tendencies of a reviewer and recommending facilities according to another embodiment of the present invention includes a step (S1010) of analyzing existing review data of the customer to identify the tendencies of the customer, a step (S1020) of linking the customer tendencies data with basic customer information and storing it, and a step (S1030) of creating and providing a list of accommodation facilities corresponding to a customer request.

[0147] In Step S1010, the customer review data is analyzed and categorized into various characteristics of the accommodation, such as "cleanliness," "price," "service," "facilities," "interior," and "breakfast." Each characteristic is then assigned a number of mentions and a positive rating. Through this process, the customer determines which aspects of the accommodation are most important to them.

[0148] Step S1020 stores customer preference data by linking it with customer basic information (name, residence, preferred region, etc.).

[0149] Step S1030 clusters other customers with similar tendencies to the customer in question, and generates and provides a list of recommendations for facilities that meet the customer's needs. For example, suppose that the analysis of review data for customer A indicates that the customer tends to have less congestion during breakfast hours, prefers rice noodle menus for breakfast, and is highly satisfied with outdoor swimming pool facilities with heated pools. Step S1030 generates and provides a list of recommended accommodations based on the analysis of review data from other customers regarding congestion during breakfast hours, breakfast menus, and outdoor swimming pool facilities, taking the customer's tendencies into account.

Claims

1. A collection unit that collects review data from the OTA server; A storage unit that stores the above review data in a database; An analysis unit that performs analysis on the above review data using an artificial intelligence engine; A transmission unit that transmits the analysis results of the above analysis unit to the OTA server and operator terminal. An automatic review response system based on review data learning that includes .

2. In paragraph 1, The above collection unit collects the review data for each hotel through page parsing and crawling on the page specified through the URL of HTML. An automatic review response system based on in-review data learning.

3. In paragraph 1, The above collection unit collects, in addition to the above review data, information on the characteristics of each hotel and surrounding information. An automatic review response system based on in-review data learning.

4. In paragraph 3, The above analysis unit automatically writes a reply to the review data using the review data, characteristic information of each hotel, and surrounding information. An automatic review response system based on in-review data learning.

5. In paragraph 3, The above analysis section uses the results of analyzing the review data left by other customers to generate a pre-notification message to be provided to customers who have not yet left a review. An automatic review response system based on in-review data learning.

6. In paragraph 5, The above analysis unit generates the advance guidance text by using the characteristic information and surrounding information of the individual hotel. An automatic review response system based on in-review data learning.

7. In paragraph 1, The above transmission unit transmits information of the automatically created reply to the OTA server. An automatic review response system based on in-review data learning.

8. In paragraph 1, The above transmission unit transmits the recommended business information collected based on the analysis results of the review data to the operator terminal. An automatic review response system based on in-review data learning.

Images