A multi-channel hotel data intelligent merging method and device

Abstract

The present application discloses a multi-channel hotel data intelligent merging method and device, the method includes: collecting hotel data from a plurality of preset channels for pre-processing, obtaining hotel data for each urban area; identifying hotel data belonging to the same hotel brand from the hotel data for each urban area, obtaining multiple groups of hotel data to be analyzed for each urban area; obtaining multi-dimensional feature data of each hotel in each group of hotel data to be analyzed, calculating the similarity of multi-dimensional feature data of each group of hotel data to be analyzed; obtaining the hotel name and address of each hotel in each group of hotel data to be analyzed, generating a semantic understanding result of each group of hotel data to be analyzed; determining the final merging score of each group of hotel data to be analyzed according to the similarity of multi-dimensional feature data and the semantic understanding result; merging each group of hotel data to be analyzed when the final merging score is greater than or equal to a preset threshold. Therefore, the accuracy of hotel data merging can be improved by adopting the present application.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to a method and device for intelligently merging multi-channel hotel data. Background Art

[0002] In the modern tourism industry, online travel agency (OTA) platforms provide users with comprehensive hotel information by aggregating hotel data from different channels. These channels may include but are not limited to official websites, third-party booking systems, social media platforms, etc. Since hotels may use different names, descriptions, prices, and pictures on different channels, the information of the same hotel may differ on different platforms. This inconsistency in information may cause users to be confused and even suspect that the platform is engaging in price fraud or big data fraud, which may affect users' trust and the platform's reputation.

[0003] In the related art, detailed hotel information, including but not limited to name, price, location, facilities, etc., is collected from multiple channels through API interfaces or web crawler technology. The collected data is preprocessed to remove duplicates and inconsistent information. The preprocessing operation specifically uses a string matching algorithm. Existing string matching cannot accurately identify all data entries belonging to the same hotel, especially when the hotel name has slight changes in different channels, thereby reducing the accuracy of hotel data merging. Summary of the invention

[0004] The embodiments of the present application provide a multi-channel hotel data intelligent merging method and device. In order to have a basic understanding of some aspects of the disclosed embodiments, a simple summary is given below. This summary is not a general review, nor is it intended to identify key / important components or describe the scope of protection of these embodiments. Its only purpose is to present some concepts in a simple form as a preface to the detailed description that follows.

[0005] In a first aspect, an embodiment of the present application provides a multi-channel hotel data intelligent merging method, the method comprising:

[0006] Trigger the data collection task regularly to collect hotel data from multiple preset channels for pre-processing to obtain hotel data for each city area;

[0007] Based on the predefined hotel brand list, hotel data belonging to the same hotel brand are identified from the hotel data of each city area, so as to obtain multiple groups of hotel data to be analyzed in each city area;

[0008] Obtain multi-dimensional feature data of each hotel in each group of hotel data to be analyzed, and calculate the similarity of the multi-dimensional feature data of each group of hotel data to be analyzed based on the multi-dimensional feature data;

[0009] Obtain the hotel name and address of each hotel in each group of hotel data to be analyzed, and perform semantic understanding based on the hotel name and address of each hotel and through a large language model to obtain the semantic understanding results of each group of hotel data to be analyzed;

[0010] Determine the final combined score of each group of hotel data to be analyzed based on the similarity of multi-dimensional feature data and semantic understanding results;

[0011] When the final combined score is greater than or equal to the preset threshold, each group of hotel data to be analyzed will be merged.

[0012] Optionally, hotel data is collected from multiple preset channels for preprocessing to obtain hotel data for each city area, including:

[0013] In response to a data collection task trigger instruction, hotel data is collected from a plurality of preset channels;

[0014] Remove duplicate records, correct format errors, and standardize text data for the collected hotel data to obtain the target hotel data;

[0015] Extract the address information of each hotel in the target hotel data;

[0016] Use geocoding services to convert the address information of each hotel into longitude and latitude coordinates to obtain the longitude and latitude of each hotel;

[0017] According to the longitude and latitude of each hotel, the target hotel data is classified into the corresponding city or district classification to obtain the hotel data of each urban area.

[0018] Optionally, the multi-dimensional feature data of each hotel in each group of hotel data to be analyzed includes latitude and longitude coordinates, front desk telephone number, hotel facilities and user reviews;

[0019] Based on the multi-dimensional feature data, the similarity of the multi-dimensional feature data of each group of hotel data to be analyzed is calculated, including:

[0020] Based on the longitude and latitude coordinates of each hotel in each group of hotel data to be analyzed, the longitude and latitude similarities between the hotels are calculated to obtain multiple longitude and latitude similarities of each hotel, and the multiple longitude and latitude similarities are summed and averaged to obtain the target longitude and latitude similarity;

[0021] Based on the front desk telephone numbers of each hotel in each group of hotel data to be analyzed, the similarity of the front desk telephone numbers between the hotels is calculated to obtain the similarities of multiple front desk telephone numbers of each hotel, and the similarities of multiple front desk telephone numbers are summed and averaged to obtain the target front desk telephone number similarity;

[0022] Based on the hotel facilities of each hotel in each group of hotel data to be analyzed, the hotel facility similarities between the hotels are calculated to obtain multiple hotel facility similarities of each hotel, and the multiple hotel facility similarities are summed and averaged to obtain the target hotel facility similarity;

[0023] Based on the user reviews of each hotel in each group of hotel data to be analyzed, the user review similarity between the hotels is calculated to obtain multiple user review similarities of each hotel, and the multiple user review similarities are summed and averaged to obtain the target user review similarity;

[0024] The target longitude and latitude similarity, target front desk telephone number similarity, target hotel facility similarity, and target user evaluation similarity are used as the multi-dimensional feature data similarity of each group of hotel data to be analyzed.

[0025] Optionally, the calculation formula for the similarity of longitude and latitude of any two hotels is:

[0026]

[0027] Among them, d is the similarity of longitude and latitude, is the absolute value of the latitude difference between the two points, = - , and are the dimension values ​​of the two points respectively, is the difference in longitude between the two points, = , and are the longitude values ​​of the two points, is the radius of the Earth, is the inverse sine function, is a sine function, is the cosine function;

[0028] The formula for calculating the similarity of the front desk telephone numbers of any two hotels is:

[0029]

[0030] in, is the front desk phone number similarity, To calculate the edit distance between two phone numbers, and Two front desk telephone numbers; To take the larger value function;

[0031] The formula for calculating the similarity of hotel facilities between any two hotels is:

[0032]

[0033] in, is the similarity of hotel facilities, A and B are the facility sets of the two hotels respectively;

[0034] The calculation formula for the similarity of user reviews of any two hotels is:

[0035]

[0036] in, Evaluate similarity for users. and are the average sentiment scores of user reviews of the first hotel and the average sentiment scores of user reviews of the second hotel, respectively. The sentiment score is a value between 0 and 1, where 0 represents a very negative sentiment and 1 represents a very positive sentiment. is the maximum sentiment score.

[0037] Optionally, based on the hotel name and address of each hotel, semantic understanding is performed through a large language model to obtain semantic understanding results for each group of hotel data to be analyzed, including:

[0038] Obtaining a preset prompt information set for guiding the large language model to analyze whether the hotel data are identical, the prompt information set including a first mapping relationship between a hotel brand and a hotel name set in the hotel industry, and a second mapping relationship between a hotel brand and a hotel address set;

[0039] Input the hotel name and address of each hotel, the first mapping relationship, the second mapping relationship, and the text expression into the large language model; wherein the large language model is ChatGPT, and the text expression is used to instruct the large language model to determine whether the hotel name and address of each hotel are semantically the same based on the first mapping relationship and the second mapping relationship;

[0040] Output the semantic understanding results of each hotel in each group of hotel data to be analyzed.

[0041] Optionally, generate a prompt information set by following the steps below, including:

[0042] From a preset hotel database, historical hotel data collected for preset different hotel brands is obtained according to a preset period;

[0043] For each piece of historical hotel data, mark the historical hotel name and historical hotel address in the historical hotel data to obtain each marked historical hotel name and historical hotel address;

[0044] Get the hotel brand corresponding to each marked historical hotel name and historical hotel address;

[0045] Input each annotated historical hotel name and historical hotel address into the preset ChatGPT model;

[0046] Output the hotel name set and hotel address set corresponding to each annotated historical hotel name and historical hotel address, which are used to guide the ChatGPT model to analyze whether the hotel data is the same;

[0047] The mapping relationship between each hotel brand and the corresponding hotel name set is stored to obtain a first mapping relationship between the hotel brand and the hotel name set in the hotel industry;

[0048] The mapping relationship between each hotel brand and the corresponding hotel address set is stored to obtain a second mapping relationship between the hotel brand and the hotel address set in the hotel industry;

[0049] The first mapping relationship and the second mapping relationship are used as a prompt information set.

[0050] Optionally, the multi-dimensional feature data similarity includes target latitude and longitude similarity, target front desk telephone number similarity, target hotel facility similarity, and target user evaluation similarity;

[0051] According to the multi-dimensional feature data similarity and semantic understanding results, the final combined score of each group of hotel data to be analyzed is determined, including:

[0052] According to the semantic understanding results, the weight values ​​corresponding to the target latitude and longitude similarity, the target front desk phone number similarity, the target hotel facility similarity and the target user evaluation similarity are determined;

[0053] The target latitude and longitude similarity, target front desk phone number similarity, target hotel facility similarity, and target user evaluation similarity are respectively multiplied by their corresponding weight values;

[0054] The multiple similarities obtained by product are summed up as the final combined score of each group of hotel data to be analyzed.

[0055] Optionally, based on the semantic understanding result, weight values ​​corresponding to the target latitude and longitude similarity, the target front desk telephone number similarity, the target hotel facility similarity, and the target user evaluation similarity are determined, including:

[0056] Quantify the semantic understanding results into numerical form to obtain a confidence score;

[0057] Get the default weights corresponding to the target latitude and longitude similarity, target front desk phone number similarity, target hotel facility similarity, and target user evaluation similarity;

[0058] Inputting the confidence score and the default weight into a pre-trained weight value optimization model to dynamically adjust the default weight according to the confidence score; the pre-trained weight value optimization model is a mathematical model that learns the relationship between the semantic understanding result and the optimal weight;

[0059] Output the weight values ​​corresponding to the target latitude and longitude similarity, target front desk phone number similarity, target hotel facility similarity, and target user evaluation similarity.

[0060] Optionally, generate a pre-trained weight optimization model by following these steps, including:

[0061] Collect historical multi-channel hotel data merging cases;

[0062] Extract historical latitude and longitude similarity, historical front desk phone number similarity, historical hotel facility similarity, and historical user review similarity from each case;

[0063] Assign initial weights to the historical latitude and longitude similarity, historical front desk telephone number similarity, historical hotel facility similarity, and historical user evaluation similarity;

[0064] Generate a historical confidence score corresponding to the semantic understanding result of each case;

[0065] Use random neural networks to create a weight value optimization model;

[0066] Input the initial weight and historical confidence score into the weight value optimization model to learn the relationship between the semantic understanding result and the optimal weight, and output the cost value;

[0067] When the cost value reaches the minimum, a pre-trained weight value optimization model is generated.

[0068] In a second aspect, an embodiment of the present application provides a multi-channel hotel data intelligent merging device, the device comprising:

[0069] The hotel data preprocessing module is used to trigger the data collection task at a fixed time to collect hotel data from multiple preset channels for preprocessing to obtain hotel data for each city area;

[0070] A hotel data identification module is used to identify hotel data belonging to the same hotel brand from the hotel data of each city area based on a predefined hotel brand list, and obtain multiple groups of hotel data to be analyzed in each city area;

[0071] A multi-dimensional feature data similarity calculation module is used to obtain the multi-dimensional feature data of each hotel in each group of hotel data to be analyzed, and calculate the multi-dimensional feature data similarity of each group of hotel data to be analyzed based on the multi-dimensional feature data;

[0072] The semantic understanding module is used to obtain the hotel name and address of each hotel in each group of hotel data to be analyzed, and based on the hotel name and address of each hotel, perform semantic understanding through a large language model to obtain the semantic understanding results of each group of hotel data to be analyzed;

[0073] The final combined score determination module is used to determine the final combined score of each group of hotel data to be analyzed based on the multi-dimensional feature data similarity and semantic understanding results;

[0074] The data merging module is used to merge each group of hotel data to be analyzed when the final merged score is greater than or equal to a preset threshold.

[0075] The technical solution provided by the embodiments of the present application may have the following beneficial effects:

[0076] In the embodiment of the present application, on the one hand, by calculating the similarity of multi-dimensional feature data and applying the semantic understanding results, the accuracy of data merging is improved, ensuring that the merged hotel data is consistent and accurate in all dimensions, reducing the erroneous information and user confusion caused by data inconsistency, and improving the reliability and quality of the data; on the other hand, by comprehensively considering the similarity of multi-dimensional feature data and the semantic understanding results, the system can make more intelligent data merging decisions. This decision-making method is not only based on quantitative similarity calculations, but also incorporates qualitative semantic analysis, making the merging decision more comprehensive and accurate. When the final merged score reaches the preset threshold, the system will automatically perform the merge operation, reducing manual intervention and improving the efficiency and intelligence level of the decision-making process.

[0077] It should be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0078] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0079] Figure 1 This is a method flow diagram of a multi-channel hotel data intelligent merging method provided by an embodiment of the present application;

[0080] Figure 2 This is a data representation of hotel data for each city area provided by an embodiment of the present application;

[0081] Figure 3 This application provides a data representation intent of the similarity of multi-dimensional feature data of each group of hotel data to be analyzed;

[0082] Figure 4It is a data representation intent of a first mapping relationship provided by this application;

[0083] Figure 5 It is a data representation of a second mapping relationship provided by the present application;

[0084] Figure 6 It is a flowchart of a weight value optimization model training method provided by this application;

[0085] Figure 7 It is a structural diagram of a multi-channel hotel data intelligent merging device provided by the present application;

[0086] Figure 8 It is a structural schematic diagram of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0087] The following description and the accompanying drawings sufficiently illustrate specific embodiments of the present application to enable those skilled in the art to practice them.

[0088] It should be clear that the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in the field without creative work are within the scope of protection of the present application.

[0089] When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present application. Instead, they are only examples of devices and methods consistent with some aspects of the present application as detailed in the attached claims.

[0090] In the description of the present application, it should be understood that the terms "first", "second", etc. are used for descriptive purposes only and should not be understood as indicating or implying relative importance. For those of ordinary skill in the art, the specific meanings of the above terms in the present application can be understood according to the specific circumstances. In addition, in the description of the present application, unless otherwise specified, "multiple" refers to two or more. "And / or" describes the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. The character " / " generally indicates that the objects associated before and after are in an "or" relationship.

[0091] Currently, detailed hotel information, including but not limited to name, price, location, facilities, etc., is collected from multiple channels through API interfaces or web crawler technology. The collected data is preprocessed to remove duplicates and inconsistent information. The preprocessing operation specifically uses a string matching algorithm.

[0092] The inventors realize that existing string matching cannot accurately identify all data entries belonging to the same hotel, especially when the hotel name has slight changes in different channels, thereby reducing the accuracy of hotel data merging.

[0093] In order to solve the above problems, the present application provides a multi-channel hotel data intelligent merging method and device to solve the problems existing in the above-mentioned related technical problems. In the embodiment of the present application, on the one hand, through the application of the similarity calculation of multi-dimensional feature data and the semantic understanding results, the accuracy of data merging is improved, ensuring that the merged hotel data is consistent and accurate in all dimensions, reducing the erroneous information and user confusion caused by data inconsistency, and improving the reliability and quality of the data; on the other hand, by comprehensively considering the similarity of multi-dimensional feature data and the semantic understanding results, the system can make a more intelligent data merging decision. This decision-making method is not only based on quantitative similarity calculation, but also incorporates qualitative semantic analysis, making the merging decision more comprehensive and accurate. When the final merging score reaches the preset threshold, the system will automatically perform the merging operation, reducing manual intervention and improving the efficiency and intelligence level of the decision-making process. The following is a detailed description using an exemplary embodiment.

[0094] The following will be combined with the attached Figure 1 -Attached Figure 6 , a multi-channel hotel data intelligent merging method provided by an embodiment of the present application is introduced in detail. The method can be implemented by relying on a computer program and can be run on a multi-channel hotel data intelligent merging device based on a von Neumann system. The computer program can be integrated into an application or run as an independent tool application.

[0095] See also Figure 1 , which is a flowchart of a multi-channel hotel data intelligent merging method provided in an embodiment of the present application. Figure 1 As shown, the method of the embodiment of the present application may include the following steps:

[0096] S101, triggering a data collection task at a fixed time to collect hotel data from a plurality of preset channels for preprocessing to obtain hotel data for each city area;

[0097] Among them, the timed trigger is to automatically start the data collection process at a predetermined time interval (such as every hour, every day); the preset multiple channels refer to a list of data sources determined in advance, which may include different websites, API interfaces, databases, etc., which provide hotel-related data. Hotel data refers to various types of information related to hotel business, such as hotel name, address, price, user evaluation, facility description, geographical location (latitude and longitude), etc. The hotel data of each urban area refers to the hotel data classified and sorted by geographical location (city or region).

[0098] In some embodiments of the present application, the specific processing process of collecting hotel data from preset multiple channels for pre-processing to obtain hotel data for each urban area includes: responding to a data collection task trigger instruction, collecting hotel data from preset multiple channels; removing duplicate records, correcting format errors, and standardizing text data for the collected hotel data to obtain target hotel data; extracting the address information of each hotel in the target hotel data; using a geocoding service to convert the address information of each hotel into longitude and latitude coordinates to obtain the longitude and latitude of each hotel; classifying the target hotel data into corresponding city or district and county classifications according to the longitude and latitude of each hotel to obtain hotel data for each urban area.

[0099] Among them, responding to the data collection task trigger instruction refers to the process in which the system starts to perform data collection after receiving a specific data collection task trigger signal. Target hotel data: refers to hotel data that meets specific quality standards and format requirements after being processed in the preprocessing step. Geocoding service is a service that converts address information into geographic coordinates (latitude and longitude). Through geocoding, addresses in text form can be converted into points that can be located on a map. Latitude and longitude coordinates are the location of a point on the surface of the earth, which are determined by two coordinate values: longitude (east-west direction) and latitude (north-south direction).

[0100] In the embodiment of the present application, by responding to the data collection task trigger instruction, hotel data is collected from multiple preset channels, and meticulous preprocessing is performed, including removing duplicate records, correcting format errors, and standardizing text data, so that accurate and consistent target hotel data can be obtained. Further extracting the address information from these data, and using the geocoding service to convert the address into longitude and latitude coordinates, and finally classifying the data into the corresponding city or district according to the geographical location, this series of operations greatly improves the accuracy, availability and operability of hotel data.

[0101] For example, suppose an online travel platform (OTA) needs to collect hotel data from various cities around the world from multiple channels. The engineers of the OTA platform can set up a data collection system that automatically performs data collection tasks at 2 a.m. every day. The preset channels include major hotel booking websites (such as Booking.com, Expedia), the official API of the hotel group, and third-party travel information platforms. The collected data is automatically imported into the data warehouse for data cleaning and classification by city area to obtain hotel data for each city area, such as Figure 2 shown.

[0102] Specifically, data cleaning includes: removing duplicate hotel entries, correcting format errors (such as dates and phone numbers), converting all text data to lowercase, unifying currencies and units of measurement, and converting address data into longitude and latitude coordinates through geocoding services.

[0103] S102, based on a predefined hotel brand list, identifying hotel data belonging to the same hotel brand from the hotel data of each city area, and obtaining multiple groups of hotel data to be analyzed in each city area;

[0104] The predefined hotel brand list is a pre-created and maintained list containing the hotel brand names known in the industry. The hotel data to be analyzed refers to the hotel data sets that have been collected, pre-processed, and grouped according to brand or other attributes, ready for further analysis.

[0105] In the embodiment of the present application, for example, the data team has created a list of hundreds of well-known hotel brands, including international chain brands and some regional brands. Based on this list, the hotel data belonging to the same brand can be identified from the hotel data of each urban area by keyword matching, for example, by searching for brand names such as "Hotel A", "Hotel B", and "Hotel C", and classifying the corresponding hotel data together. The identified hotel data of the same brand is divided into a group of data to be analyzed.

[0106] For example, in the data for the Shanghai area, all data for "Hotel A" will be grouped into one group, and all data for "Hotel B" will be grouped into another group.

[0107] S103, obtaining multi-dimensional feature data of each hotel in each group of hotel data to be analyzed, and calculating the similarity of the multi-dimensional feature data of each group of hotel data to be analyzed based on the multi-dimensional feature data;

[0108] Among them, multi-dimensional feature data refers to a data set extracted from hotel data that contains multiple attributes or parameters. These attributes can describe different aspects or characteristics of the hotel, such as latitude and longitude, price, user ratings, facilities, etc.

[0109] Among them, the multi-dimensional feature data of each hotel in each group of hotel data to be analyzed includes latitude and longitude coordinates, front desk telephone number, hotel facilities and user reviews.

[0110] In some embodiments of the present application, based on the multi-dimensional feature data, the specific process of calculating the similarity of the multi-dimensional feature data of each group of hotel data to be analyzed includes: based on the longitude and latitude coordinates of each hotel in each group of hotel data to be analyzed, calculating the longitude and latitude similarities between each hotel, obtaining multiple longitude and latitude similarities of each hotel, summing up and averaging the multiple longitude and latitude similarities, and obtaining the target longitude and latitude similarity; based on the front desk telephone number of each hotel in each group of hotel data to be analyzed, calculating the similarity of the front desk telephone numbers between each hotel, obtaining multiple front desk telephone number similarities of each hotel, summing up and averaging the multiple front desk telephone number similarities, and obtaining the target front desk telephone number similarity; based on each group The hotel facilities of each hotel in the hotel data to be analyzed are calculated, and the hotel facility similarities between each hotel are obtained to obtain multiple hotel facility similarities of each hotel. The multiple hotel facility similarities are summed and averaged to obtain the target hotel facility similarity; based on the user reviews of each hotel in each group of hotel data to be analyzed, the user review similarities between each hotel are calculated to obtain multiple user review similarities of each hotel. The multiple user review similarities are summed and averaged to obtain the target user review similarity; the target latitude and longitude similarity, the target front desk telephone number similarity, the target hotel facility similarity, and the target user review similarity are used as the multi-dimensional feature data similarity of each group of hotel data to be analyzed, for example Figure 3 shown.

[0111] Specifically, the calculation formula for the similarity of longitude and latitude of any two hotels is:

[0112]

[0113] Among them, d is the similarity of longitude and latitude, is the absolute value of the latitude difference between the two points, = - , and are the dimension values ​​of the two points respectively, is the difference in longitude between the two points, = , and are the longitude values ​​of the two points, is the radius of the Earth, is the inverse sine function, is a sine function, is the cosine function.

[0114] Specifically, the calculation formula for the similarity of the front desk telephone numbers of any two hotels is:

[0115]

[0116] in, is the front desk phone number similarity, To calculate the edit distance between two phone numbers, and Two front desk telephone numbers; To take the larger value function.

[0117] Specifically, the calculation formula for the similarity of hotel facilities between any two hotels is:

[0118]

[0119] in, is the similarity of hotel facilities, A and B are the facility sets of the two hotels respectively.

[0120] Specifically, the calculation formula for the similarity of user reviews of any two hotels is:

[0121]

[0122] in, Evaluate similarity for users. and are the average sentiment scores of user reviews of the first hotel and the average sentiment scores of user reviews of the second hotel, respectively. The sentiment score is a value between 0 and 1, where 0 represents a very negative sentiment and 1 represents a very positive sentiment. is the maximum sentiment score.

[0123] S104, obtaining the hotel name and address of each hotel in each group of hotel data to be analyzed, and performing semantic understanding based on the hotel name and address of each hotel and using a large language model to obtain a semantic understanding result of each group of hotel data to be analyzed;

[0124] Among them, the hotel name refers to the official name of the hotel, which is usually used to identify and distinguish different hotels. It can reflect the brand, positioning and style of the hotel. For example, the name of a hotel chain may include the brand name and the name of a branch. The address refers to the specific geographical location of the hotel, usually including information such as city, region, street and zip code. The Large Language Model (LLM) is a natural language processing model based on deep learning that can understand and generate natural language text. Semantic understanding refers to a deep understanding of the content of the text, beyond the literal meaning, to identify the intent, emotion and contextual relationship in the text.

[0125] In some embodiments of the present application, based on the hotel name and address of each hotel, and performing semantic understanding through a large language model, the specific process of obtaining the semantic understanding results of each group of hotel data to be analyzed includes: obtaining a preset prompt information set for guiding the large language model to analyze whether the hotel data are the same, the prompt information set including a first mapping relationship between a hotel brand and a hotel name set in the hotel industry, and a second mapping relationship between a hotel brand and a hotel address set; inputting the hotel name and address of each hotel, the first mapping relationship, the second mapping relationship, and the text expression into the large language model; wherein the large language model is ChatGPT, and the text expression is used to instruct the large language model to determine whether the hotel name and address of each hotel are semantically the same based on the first mapping relationship and the second mapping relationship; and outputting the semantic understanding results of each hotel in each group of hotel data to be analyzed.

[0126] Among them, the prompt information set is a set of preset data used to guide the large language model on how to identify and process specific information during analysis. In this application, the prompt information set contains the mapping relationship between the hotel brand and the hotel name and address, helping the model understand that different expressions may point to the same entity. The first mapping relationship refers to the correspondence between the hotel brand and the hotel name set, which helps the model identify the same brand that may be represented behind different names. The second mapping relationship refers to the correspondence between the hotel brand and the hotel address set. Text expressions are instructions or queries used to guide large language models to perform specific tasks. ChatGPT is a large language model that can understand and generate natural language text.

[0127] In an embodiment of the present application, by using a pre-set prompt information set, including the mapping relationship between the hotel brand and the name and address, and inputting this information and the hotel name and address into the large language model ChatGPT, the semantic understanding ability of hotel data can be significantly improved. This method enables the model to more accurately identify and merge different data entries pointing to the same hotel, thereby reducing data redundancy and improving data accuracy and consistency. The output semantic understanding results can be used to optimize data management processes, improve user experience, and provide deeper insights and analysis for hotel business.

[0128] In an embodiment of the present application, a prompt information set is generated according to the following steps, and the specific process includes: from a preset hotel database, historical hotel data collected for preset different hotel brands is obtained according to a preset period; for each historical hotel data, the historical hotel name and historical hotel address existing in the historical hotel data are marked to obtain each marked historical hotel name and historical hotel address; the hotel brand corresponding to each marked historical hotel name and historical hotel address is obtained; each marked historical hotel name and historical hotel address is input into a preset ChatGPT model; a hotel name set and a hotel address set corresponding to each marked historical hotel name and historical hotel address are output for guiding the ChatGPT model to analyze whether the hotel data are the same; a mapping relationship between each hotel brand and the corresponding hotel name set is stored to obtain a first mapping relationship between a hotel brand in the hotel industry and a hotel name set; a mapping relationship between each hotel brand and the corresponding hotel address set is stored to obtain a second mapping relationship between a hotel brand in the hotel industry and a hotel address set; the first mapping relationship and the second mapping relationship are used as a prompt information set.

[0129] Among them, the mapping relationship between the hotel brand and the hotel name set (i.e., the first mapping relationship) is, for example Figure 4 As shown. The mapping relationship between hotel brands and hotel address sets (i.e., the second mapping relationship), for example Figure 5 shown.

[0130] S105, determining the final combined score of each group of hotel data to be analyzed based on the multi-dimensional feature data similarity and semantic understanding results;

[0131] Among them, the multi-dimensional feature data similarity includes the target latitude and longitude similarity, the target front desk telephone number similarity, the target hotel facility similarity and the target user evaluation similarity.

[0132] In some embodiments of the present application, the specific process of determining the final combined score of each group of hotel data to be analyzed based on the similarity of multi-dimensional feature data and the results of semantic understanding includes: determining the weight values ​​corresponding to the target longitude and latitude similarity, the target front desk telephone number similarity, the target hotel facility similarity, and the target user evaluation similarity based on the semantic understanding results; respectively multiplying the target longitude and latitude similarity, the target front desk telephone number similarity, the target hotel facility similarity, and the target user evaluation similarity with their respective corresponding weight values; and summing the multiple similarities obtained by the product as the final combined score of each group of hotel data to be analyzed.

[0133] Specifically, the specific process of determining the weight values ​​corresponding to the target longitude and latitude similarity, the target front desk telephone number similarity, the target hotel facility similarity and the target user evaluation similarity based on the semantic understanding results includes: quantifying the semantic understanding results into numerical form to obtain a confidence score; obtaining the default weights corresponding to the target longitude and latitude similarity, the target front desk telephone number similarity, the target hotel facility similarity and the target user evaluation similarity; inputting the confidence score and the default weight into a pre-trained weight value optimization model to dynamically adjust the default weight according to the confidence score; the pre-trained weight value optimization model is a mathematical model for learning the relationship between the semantic understanding results and the optimal weight; and outputting the weight values ​​corresponding to the target longitude and latitude similarity, the target front desk telephone number similarity, the target hotel facility similarity and the target user evaluation similarity.

[0134] Among them, the confidence score is a quantitative representation of the semantic understanding result, which reflects the model's certainty or trust in its output results. In this application, this score can indicate the similarity or matching degree of the hotel name and address identified by the model. The default weight is a weight value pre-assigned to different feature similarities (such as latitude and longitude, phone number, facilities, user reviews) for similarity calculation without additional information. The weight value optimization model is a mathematical model that dynamically adjusts the default weight by learning the relationship between the semantic understanding results and the optimal weight.

[0135] In the embodiment of the present application, by quantifying the semantic understanding results into confidence scores and combining the default weights of the feature data of each dimension, the weights can be dynamically adjusted using a pre-trained weight value optimization model. This model can learn and optimize the weight distribution according to the confidence scores, thereby more accurately reflecting the importance of different features in the merging decision. This method improves the accuracy and flexibility of hotel data merging, enables merging decisions to adapt to different data characteristics and the complexity of semantic understanding, and ultimately outputs weight values ​​that are more in line with the actual situation, enhancing the intelligence and automation level of data processing.

[0136] For example Figure 6As shown, a pre-trained weight value optimization model is generated according to the following steps, including: collecting historical multi-channel hotel data merging cases; extracting historical longitude and latitude similarity, historical front desk telephone number similarity, historical hotel facility similarity, and historical user evaluation similarity from each case; assigning initial weights to the historical longitude and latitude similarity, historical front desk telephone number similarity, historical hotel facility similarity, and historical user evaluation similarity; generating a historical confidence score corresponding to the semantic understanding result of each case; using a random neural network to create a weight value optimization model; inputting the initial weight and the historical confidence score into the weight value optimization model to learn the relationship between the semantic understanding result and the optimal weight, and outputting a cost value; when the cost value reaches the minimum, generating a pre-trained weight value optimization model.

[0137] In an embodiment of the present application, by collecting historical multi-channel hotel data merging cases and extracting various historical similarity data therefrom, initial weights can be assigned to these similarity indicators. Combining the semantic understanding results of each case and the corresponding historical confidence scores, a weight value optimization model can be created using a random neural network. This model dynamically adjusts the weights to output the most optimized cost value by learning the relationship between the semantic understanding results and the optimal weights. When this cost value reaches a minimum, a pre-trained weight value optimization model is obtained, which can accurately guide future hotel data merging tasks and improve the accuracy and efficiency of merging decisions.

[0138] S106, when the final combined score is greater than or equal to the preset threshold, each group of hotel data to be analyzed is combined.

[0139] The final merge score refers to a total score calculated based on the similarity of multiple dimensions, which is used to determine whether to merge a group of hotel data to be analyzed. The preset threshold is a pre-set score limit used to determine whether the final merge score is high enough to determine whether to perform the merge operation.

[0140] In some embodiments, a merge score threshold is set based on historical data and business needs, for example, 0.8, which means that the data will be merged only when the merge score reaches or exceeds 0.8. The calculated similarity score is compared with the preset threshold. If the final merge score is greater than or equal to 0.8, the hotel data are considered similar and should be merged. For hotel data whose merge score exceeds the threshold, a merge operation is performed to create a comprehensive hotel data record, retain the most comprehensive and up-to-date information, and update the database. The merged data is used to improve the accuracy of search results, optimize user experience, and support data analysis and business decision-making on the platform.

[0141] In the embodiment of the present application, on the one hand, by calculating the similarity of multi-dimensional feature data and applying the semantic understanding results, the accuracy of data merging is improved, ensuring that the merged hotel data is consistent and accurate in all dimensions, reducing the erroneous information and user confusion caused by data inconsistency, and improving the reliability and quality of the data; on the other hand, by comprehensively considering the similarity of multi-dimensional feature data and the semantic understanding results, the system can make more intelligent data merging decisions. This decision-making method is not only based on quantitative similarity calculations, but also incorporates qualitative semantic analysis, making the merging decision more comprehensive and accurate. When the final merged score reaches the preset threshold, the system will automatically perform the merge operation, reducing manual intervention and improving the efficiency and intelligence level of the decision-making process.

[0142] The following is an embodiment of the device of the present application, which can be used to execute the embodiment of the method of the present application. For details not disclosed in the embodiment of the device of the present application, please refer to the embodiment of the method of the present application.

[0143] See also Figure 7 , which shows a schematic diagram of the structure of a multi-channel hotel data intelligent merging device provided by an exemplary embodiment of the present application. The multi-channel hotel data intelligent merging device can be implemented as all or part of an electronic device through software, hardware or a combination of both. The device 1 includes a hotel data preprocessing module 10, a hotel data identification module 20, a multi-dimensional feature data similarity calculation module 30, a semantic understanding module 40, a final merging score determination module 50, and a data merging module 60.

[0144] The hotel data preprocessing module 10 is used to trigger the data collection task at a fixed time, so as to collect hotel data from a plurality of preset channels for preprocessing, and obtain hotel data of each city area;

[0145] The hotel data identification module 20 is used to identify hotel data belonging to the same hotel brand from the hotel data of each city area based on a predefined hotel brand list, and obtain multiple groups of hotel data to be analyzed in each city area;

[0146] A multi-dimensional feature data similarity calculation module 30 is used to obtain the multi-dimensional feature data of each hotel in each group of hotel data to be analyzed, and calculate the multi-dimensional feature data similarity of each group of hotel data to be analyzed based on the multi-dimensional feature data;

[0147] The semantic understanding module 40 is used to obtain the hotel name and address of each hotel in each group of hotel data to be analyzed, and perform semantic understanding based on the hotel name and address of each hotel and through the large language model to obtain the semantic understanding result of each group of hotel data to be analyzed;

[0148] A final combined score determination module 50 is used to determine the final combined score of each group of hotel data to be analyzed based on the multi-dimensional feature data similarity and semantic understanding results;

[0149] The data merging module 60 is used to merge each group of hotel data to be analyzed when the final merging score is greater than or equal to a preset threshold.

[0150] It should be noted that the multi-channel hotel data intelligent merging device provided in the above embodiment only uses the division of the above functional modules as an example when executing the multi-channel hotel data intelligent merging method. In actual applications, the above functional distribution can be completed by different functional modules as needed, that is, the internal structure of the device is divided into different functional modules to complete all or part of the functions described above. In addition, the multi-channel hotel data intelligent merging device provided in the above embodiment and the multi-channel hotel data intelligent merging method embodiment belong to the same concept, and the embodiment and implementation process thereof are detailed in the method embodiment, which will not be repeated here.

[0151] The serial numbers of the above-mentioned embodiments of the present application are for description only and do not represent the advantages or disadvantages of the embodiments.

[0152] In the embodiment of the present application, on the one hand, by calculating the similarity of multi-dimensional feature data and applying the semantic understanding results, the accuracy of data merging is improved, ensuring that the merged hotel data is consistent and accurate in all dimensions, reducing the erroneous information and user confusion caused by data inconsistency, and improving the reliability and quality of the data; on the other hand, by comprehensively considering the similarity of multi-dimensional feature data and the semantic understanding results, the system can make more intelligent data merging decisions. This decision-making method is not only based on quantitative similarity calculations, but also incorporates qualitative semantic analysis, making the merging decision more comprehensive and accurate. When the final merged score reaches the preset threshold, the system will automatically perform the merge operation, reducing manual intervention and improving the efficiency and intelligence level of the decision-making process.

[0153] The present application also provides a computer-readable medium having program instructions stored thereon, and when the program instructions are executed by a processor, the multi-channel hotel data intelligent merging method provided by the above-mentioned various method embodiments is implemented.

[0154] The present application also provides a computer program product comprising instructions, which, when executed on a computer, enables the computer to execute the multi-channel hotel data intelligent merging method of each of the above method embodiments.

[0155] See also Figure 8 , is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. Figure 8As shown, the electronic device 1000 may include: at least one processor 1001 , at least one network interface 1004 , a user interface 1003 , a memory 1005 , and at least one communication bus 1002 .

[0156] The communication bus 1002 is used to realize the connection and communication between these components.

[0157] The user interface 1003 may include a display screen (Display) and a camera (Camera). Optionally, the user interface 1003 may also include a standard wired interface and a wireless interface.

[0158] The network interface 1004 may optionally include a standard wired interface or a wireless interface (such as a WI-FI interface).

[0159] Among them, the processor 1001 may include one or more processing cores. The processor 1001 uses various interfaces and lines to connect various parts of the entire electronic device 1000, and executes various functions and processes data of the electronic device 1000 by running or executing instructions, programs, code sets or instruction sets stored in the memory 1005, and calling data stored in the memory 1005. Optionally, the processor 1001 can be implemented in at least one hardware form of digital signal processing (Digital Signal Processing, DSP), field programmable gate array (Field-Programmable Gate Array, FPGA), and programmable logic array (Programmable Logic Array, PLA). The processor 1001 can integrate one or a combination of a central processing unit (Central Processing Unit, CPU), a graphics processing unit (Graphics Processing Unit, GPU) and a modem. Among them, the CPU mainly processes the operating system, user interface and application programs; the GPU is responsible for rendering and drawing the content to be displayed on the display screen; the modem is used to process wireless communications. It can be understood that the above-mentioned modem may not be integrated into the processor 1001, and it can be implemented separately through a chip.

[0160] Among them, the memory 1005 may include a random access memory (Random Access Memory, RAM) and may also include a read-only memory (Read-Only Memory). Optionally, the memory 1005 includes a non-transitory computer-readable storage medium. The memory 1005 can be used to store instructions, programs, codes, code sets or instruction sets. The memory 1005 may include a program storage area and a data storage area, wherein the program storage area may store instructions for implementing an operating system, instructions for at least one function (such as a touch function, a sound playback function, an image playback function, etc.), instructions for implementing the above-mentioned various method embodiments, etc.; the data storage area may store data involved in the above-mentioned various method embodiments, etc. The memory 1005 may also be optionally at least one storage system located away from the aforementioned processor 1001. As Figure 8 As shown, the memory 1005 as a computer storage medium may include an operating system, a network communication module, a user interface module, and a multi-channel hotel data intelligent merging application.

[0161] exist Figure 8 In the electronic device 1000 shown, the user interface 1003 is mainly used to provide an input interface for the user and obtain the data input by the user; and the processor 1001 can be used to call the multi-channel hotel data intelligent merging application stored in the memory 1005, and specifically perform the following operations:

[0162] Trigger the data collection task regularly to collect hotel data from multiple preset channels for pre-processing to obtain hotel data for each city area;

[0163] Based on the predefined hotel brand list, hotel data belonging to the same hotel brand are identified from the hotel data of each city area, so as to obtain multiple groups of hotel data to be analyzed in each city area;

[0164] Obtain multi-dimensional feature data of each hotel in each group of hotel data to be analyzed, and calculate the similarity of the multi-dimensional feature data of each group of hotel data to be analyzed based on the multi-dimensional feature data;

[0165] Obtain the hotel name and address of each hotel in each group of hotel data to be analyzed, and perform semantic understanding based on the hotel name and address of each hotel and through a large language model to obtain the semantic understanding results of each group of hotel data to be analyzed;

[0166] Determine the final combined score of each group of hotel data to be analyzed based on the similarity of multi-dimensional feature data and semantic understanding results;

[0167] When the final combined score is greater than or equal to the preset threshold, each group of hotel data to be analyzed will be merged.

[0168] In one embodiment, when the processor 1001 collects hotel data from a plurality of preset channels for preprocessing to obtain hotel data for each city area, the processor 1001 specifically performs the following operations:

[0169] In response to a data collection task trigger instruction, hotel data is collected from a plurality of preset channels;

[0170] Remove duplicate records, correct format errors, and standardize text data for the collected hotel data to obtain the target hotel data;

[0171] Extract the address information of each hotel in the target hotel data;

[0172] Use geocoding services to convert the address information of each hotel into longitude and latitude coordinates to obtain the longitude and latitude of each hotel;

[0173] According to the longitude and latitude of each hotel, the target hotel data is classified into the corresponding city or district classification to obtain the hotel data of each urban area.

[0174] In one embodiment, when the processor 1001 calculates the similarity of the multi-dimensional feature data of each group of hotel data to be analyzed based on the multi-dimensional feature data, the processor 1001 specifically performs the following operations:

[0175] Based on the longitude and latitude coordinates of each hotel in each group of hotel data to be analyzed, the longitude and latitude similarities between the hotels are calculated to obtain multiple longitude and latitude similarities of each hotel, and the multiple longitude and latitude similarities are summed and averaged to obtain the target longitude and latitude similarity;

[0176] Based on the front desk telephone numbers of each hotel in each group of hotel data to be analyzed, the similarity of the front desk telephone numbers between the hotels is calculated to obtain the similarities of multiple front desk telephone numbers of each hotel, and the similarities of multiple front desk telephone numbers are summed and averaged to obtain the target front desk telephone number similarity;

[0177] Based on the hotel facilities of each hotel in each group of hotel data to be analyzed, the hotel facility similarities between the hotels are calculated to obtain multiple hotel facility similarities of each hotel, and the multiple hotel facility similarities are summed and averaged to obtain the target hotel facility similarity;

[0178] Based on the user reviews of each hotel in each group of hotel data to be analyzed, the user review similarity between the hotels is calculated to obtain multiple user review similarities of each hotel, and the multiple user review similarities are summed and averaged to obtain the target user review similarity;

[0179] The target longitude and latitude similarity, target front desk telephone number similarity, target hotel facility similarity, and target user evaluation similarity are used as the multi-dimensional feature data similarity of each group of hotel data to be analyzed.

[0180] In one embodiment, when the processor 1001 performs semantic understanding based on the hotel name and address of each hotel and through the large language model to obtain the semantic understanding result of each group of hotel data to be analyzed, the processor 1001 specifically performs the following operations:

[0181] Obtaining a preset prompt information set for guiding the large language model to analyze whether the hotel data are identical, the prompt information set including a first mapping relationship between a hotel brand and a hotel name set in the hotel industry, and a second mapping relationship between a hotel brand and a hotel address set;

[0182] Input the hotel name and address of each hotel, the first mapping relationship, the second mapping relationship, and the text expression into the large language model; wherein the large language model is ChatGPT, and the text expression is used to instruct the large language model to determine whether the hotel name and address of each hotel are semantically the same based on the first mapping relationship and the second mapping relationship;

[0183] Output the semantic understanding results of each hotel in each group of hotel data to be analyzed.

[0184] In one embodiment, when executing the generation of the prompt information set, the processor 1001 specifically performs the following operations:

[0185] From a preset hotel database, historical hotel data collected for preset different hotel brands is obtained according to a preset period;

[0186] For each piece of historical hotel data, mark the historical hotel name and historical hotel address in the historical hotel data to obtain each marked historical hotel name and historical hotel address;

[0187] Get the hotel brand corresponding to each marked historical hotel name and historical hotel address;

[0188] Input each annotated historical hotel name and historical hotel address into the preset ChatGPT model;

[0189] Output the hotel name set and hotel address set corresponding to each annotated historical hotel name and historical hotel address, which are used to guide the ChatGPT model to analyze whether the hotel data is the same;

[0190] The mapping relationship between each hotel brand and the corresponding hotel name set is stored to obtain a first mapping relationship between the hotel brand and the hotel name set in the hotel industry;

[0191] The mapping relationship between each hotel brand and the corresponding hotel address set is stored to obtain a second mapping relationship between the hotel brand and the hotel address set in the hotel industry;

[0192] The first mapping relationship and the second mapping relationship are used as a prompt information set.

[0193] In one embodiment, when the processor 1001 determines the final combined score of each group of hotel data to be analyzed according to the multi-dimensional feature data similarity and semantic understanding results, the processor 1001 specifically performs the following operations:

[0194] According to the semantic understanding results, the weight values ​​corresponding to the target latitude and longitude similarity, the target front desk phone number similarity, the target hotel facility similarity and the target user evaluation similarity are determined;

[0195] The target latitude and longitude similarity, target front desk phone number similarity, target hotel facility similarity, and target user evaluation similarity are respectively multiplied by their corresponding weight values;

[0196] The multiple similarities obtained by product are summed up as the final combined score of each group of hotel data to be analyzed.

[0197] In the embodiment of the present application, on the one hand, by calculating the similarity of multi-dimensional feature data and applying the semantic understanding results, the accuracy of data merging is improved, ensuring that the merged hotel data is consistent and accurate in all dimensions, reducing the erroneous information and user confusion caused by data inconsistency, and improving the reliability and quality of the data; on the other hand, by comprehensively considering the similarity of multi-dimensional feature data and the semantic understanding results, the system can make more intelligent data merging decisions. This decision-making method is not only based on quantitative similarity calculations, but also incorporates qualitative semantic analysis, making the merging decision more comprehensive and accurate. When the final merged score reaches the preset threshold, the system will automatically perform the merge operation, reducing manual intervention and improving the efficiency and intelligence level of the decision-making process.

[0198] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program, and the program for intelligently merging hotel data from multiple channels can be stored in a computer-readable storage medium, and when the program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, the storage medium of the program for intelligently merging hotel data from multiple channels can be a disk, an optical disk, a read-only storage memory, or a random access memory, etc.

[0199] The above disclosure is only the preferred embodiment of the present application, which certainly cannot be used to limit the scope of rights of the present application. Therefore, equivalent changes made according to the claims of the present application are still within the scope covered by the present application.

Claims

1. A multi-channel hotel data intelligent merging method, characterized in that: The method comprises: Trigger the data collection task regularly to collect hotel data from multiple preset channels for pre-processing to obtain hotel data for each city area; Based on a predefined hotel brand list, identifying hotel data belonging to the same hotel brand from the hotel data of each urban area, and obtaining multiple groups of hotel data to be analyzed in each urban area; Acquire multi-dimensional feature data of each hotel in each group of hotel data to be analyzed, and calculate the similarity of the multi-dimensional feature data of each group of hotel data to be analyzed based on the multi-dimensional feature data; Obtain the hotel name and address of each hotel in each group of hotel data to be analyzed, and based on the hotel name and address of each hotel, perform semantic understanding through a large language model to obtain the semantic understanding result of each group of hotel data to be analyzed; wherein, The semantic understanding result of each group of hotel data to be analyzed is obtained by performing semantic understanding based on the hotel name and address of each hotel through a large language model, including: Obtaining a preset prompt information set for guiding the large language model to analyze whether the hotel data are identical, wherein the prompt information set includes a first mapping relationship between a hotel brand and a hotel name set in the hotel industry, and a second mapping relationship between a hotel brand and a hotel address set; Inputting the hotel names and addresses of the hotels, the first mapping relationship, the second mapping relationship, and a text expression into the large language model; wherein the large language model is ChatGPT, and the text expression is used to instruct the large language model to determine whether the hotel names and addresses of the hotels are semantically identical based on the first mapping relationship and the second mapping relationship; Outputting the semantic understanding results of each hotel in each group of hotel data to be analyzed; Determining a final combined score of each group of hotel data to be analyzed according to the multi-dimensional feature data similarity and the semantic understanding result; When the final combined score is greater than or equal to a preset threshold, each group of hotel data to be analyzed is combined.

2. The method according to claim 1, characterized in that Hotel data is collected from multiple preset channels and pre-processed to obtain hotel data for each city area, including: In response to a data collection task trigger instruction, hotel data is collected from a plurality of preset channels; Remove duplicate records, correct format errors, and standardize text data for the collected hotel data to obtain the target hotel data; Extracting the address information of each hotel in the target hotel data; Using a geocoding service, converting the address information of each hotel into longitude and latitude coordinates to obtain the longitude and latitude of each hotel; According to the longitude and latitude of each hotel, the target hotel data is classified into corresponding city or district classification to obtain hotel data of each urban area.

3. The method according to claim 1, characterized in that The multi-dimensional feature data of each hotel in each group of hotel data to be analyzed includes latitude and longitude coordinates, front desk telephone number, hotel facilities and user reviews; The calculating the similarity of the multi-dimensional feature data of each group of hotel data to be analyzed based on the multi-dimensional feature data includes: Based on the longitude and latitude coordinates of each hotel in each group of hotel data to be analyzed, calculate the longitude and latitude similarities between the hotels to obtain multiple longitude and latitude similarities of each hotel, sum and average the multiple longitude and latitude similarities to obtain the target longitude and latitude similarity; Based on the front desk telephone numbers of each hotel in each group of hotel data to be analyzed, calculate the similarity of the front desk telephone numbers between the hotels to obtain multiple front desk telephone number similarities of each hotel, sum and average the multiple front desk telephone number similarities to obtain the target front desk telephone number similarity; Based on the hotel facilities of each hotel in each group of hotel data to be analyzed, calculate the hotel facility similarities between the hotels to obtain multiple hotel facility similarities of each hotel, sum and average the multiple hotel facility similarities to obtain the target hotel facility similarity; Based on the user evaluations of each hotel in each group of hotel data to be analyzed, the user evaluation similarities between the hotels are calculated to obtain multiple user evaluation similarities of the hotels, and the multiple user evaluation similarities are summed and averaged to obtain the target user evaluation similarity; The target longitude and latitude similarity, the target front desk telephone number similarity, the target hotel facility similarity, and the target user evaluation similarity are used as the multi-dimensional feature data similarity of each group of hotel data to be analyzed.

4. The method according to claim 3, characterized in that The calculation formula for the similarity of the longitude and latitude of any two hotels is: Among them, d is the similarity of longitude and latitude, is the absolute value of the latitude difference between the two points, = - , and are the dimension values ​​of the two points respectively, is the difference in longitude between the two points, = , and are the longitude values ​​of the two points, is the radius of the Earth, is the inverse sine function, is a sine function, is the cosine function; The calculation formula for the similarity of the front desk telephone numbers of any two hotels is: in, is the front desk phone number similarity, To calculate the edit distance between two phone numbers, and Two front desk telephone numbers; To take the larger value function; The calculation formula for the similarity of the hotel facilities of any two hotels is: in, is the similarity of hotel facilities, A and B are the facility sets of the two hotels respectively; The calculation formula for the similarity of the user reviews of any two hotels is: in, Evaluate similarity for users. and are the average sentiment scores of user reviews of the first hotel and the average sentiment scores of user reviews of the second hotel, respectively. The sentiment score is a value between 0 and 1, where 0 represents a very negative sentiment and 1 represents a very positive sentiment. is the maximum sentiment score.

5. The method according to claim 1, characterized in that Follow these steps to generate a prompt information set, including: From a preset hotel database, historical hotel data collected for preset different hotel brands is obtained according to a preset period; For each piece of historical hotel data, mark the historical hotel name and historical hotel address in the historical hotel data to obtain each marked historical hotel name and historical hotel address; Get the hotel brand corresponding to each marked historical hotel name and historical hotel address; Input each of the annotated historical hotel names and historical hotel addresses into a preset ChatGPT model; Output the hotel name set and hotel address set corresponding to each annotated historical hotel name and historical hotel address for guiding the ChatGPT model to analyze whether the hotel data is the same; Storing the mapping relationship between each hotel brand and the corresponding hotel name set, and obtaining a first mapping relationship between the hotel brand and the hotel name set in the hotel industry; Storing the mapping relationship between each hotel brand and the corresponding hotel address set, and obtaining a second mapping relationship between the hotel brand and the hotel address set in the hotel industry; The first mapping relationship and the second mapping relationship are used as a prompt information set.

6. The method according to claim 1, characterized in that The multi-dimensional feature data similarity includes target latitude and longitude similarity, target front desk telephone number similarity, target hotel facility similarity and target user evaluation similarity; Determining the final combined score of each group of hotel data to be analyzed according to the multi-dimensional feature data similarity and the semantic understanding result includes: According to the semantic understanding result, determine the weight values ​​corresponding to the target latitude and longitude similarity, the target front desk telephone number similarity, the target hotel facility similarity and the target user evaluation similarity; The target latitude and longitude similarity, the target front desk telephone number similarity, the target hotel facility similarity and the target user evaluation similarity are respectively multiplied by their corresponding weight values; The multiple similarities obtained by product are summed up as the final combined score of each group of hotel data to be analyzed.

7. The method according to claim 6, characterized in that Determining the weight values ​​corresponding to the target latitude and longitude similarity, the target front desk telephone number similarity, the target hotel facility similarity, and the target user evaluation similarity based on the semantic understanding result includes: quantifying the semantic understanding result into a numerical form to obtain a confidence score; Obtaining the default weights corresponding to the target latitude and longitude similarity, the target front desk telephone number similarity, the target hotel facility similarity, and the target user evaluation similarity; Inputting the confidence score and the default weight into a pre-trained weight value optimization model to dynamically adjust the default weight according to the confidence score; the pre-trained weight value optimization model is a mathematical model for learning the relationship between semantic understanding results and optimal weights; The weight values ​​corresponding to the target latitude and longitude similarity, the target front desk telephone number similarity, the target hotel facility similarity, and the target user evaluation similarity are output.

8. The method according to claim 7, characterized in that Follow these steps to generate a pre-trained weight optimization model, including: Collect historical multi-channel hotel data merging cases; Extract historical latitude and longitude similarity, historical front desk phone number similarity, historical hotel facility similarity, and historical user review similarity from each case; Assigning initial weights to the historical latitude and longitude similarities, historical front desk telephone number similarities, historical hotel facility similarities, and historical user evaluation similarities; Generate a historical confidence score corresponding to the semantic understanding result of each case; Use random neural networks to create a weight value optimization model; Inputting the initial weight and the historical confidence score into the weight value optimization model to learn the relationship between the semantic understanding result and the optimal weight, and outputting the cost value; When the cost value reaches the minimum, a pre-trained weight value optimization model is generated.

9. A multi-channel hotel data intelligent merging device implemented by the method described in any one of claims 1 to 8, characterized in that: The device comprises: The hotel data preprocessing module is used to trigger the data collection task at a fixed time to collect hotel data from multiple preset channels for preprocessing to obtain hotel data for each city area; A hotel data identification module, for identifying hotel data belonging to the same hotel brand from the hotel data of each city area based on a predefined hotel brand list, and obtaining multiple groups of hotel data to be analyzed in each city area; A multi-dimensional feature data similarity calculation module is used to obtain the multi-dimensional feature data of each hotel in each group of hotel data to be analyzed, and calculate the multi-dimensional feature data similarity of each group of hotel data to be analyzed based on the multi-dimensional feature data; A semantic understanding module, used for obtaining the hotel name and address of each hotel in each group of hotel data to be analyzed, and performing semantic understanding based on the hotel name and address of each hotel and through a large language model to obtain a semantic understanding result of each group of hotel data to be analyzed; A final combined score determination module, used to determine the final combined score of each group of hotel data to be analyzed according to the multi-dimensional feature data similarity and the semantic understanding result; The data merging module is used to merge each group of hotel data to be analyzed when the final merging score is greater than or equal to a preset threshold.

Images