Tourism recommendation method and system based on large model

Abstract

The invention relates to the technical field of tourism recommendation, and discloses a tourism recommendation method and system based on a large model, and the method comprises the steps: collecting the historical tourism information of a user; determining a recommendation season of the tourism recommendation place according to the travel times in each season; judging whether to delete the tourist recommendation site according to the fact that the recommendation season is in the slack season or the peak season and the tourist site of the user is in the slack season or the peak season; and preliminarily judging whether to delete the tourist recommendation place according to the average single tourist attraction quantity of the user and the tourist attraction quantity of the tourist recommendation place, calculating the range of the tourist attraction quantity according to the average single tourist attraction quantity of the user, and deleting the recommendation place according to the range of the tourist attraction quantity and the tourist attraction quantity of the tourist recommendation place. According to the method, multiple aspects such as personal preference of the user, seasonal factors and tourist flow of a tourism place can be comprehensively considered, accurate and personalized tourism recommendation is provided, and the effect of a recommendation system and user satisfaction are remarkably improved.

Description

Technical Field

[0001] The present invention relates to the technical field of tourism recommendation, and more particularly, to a tourism recommendation method and system based on a large model. Background Art

[0002] With the rapid development of the tourism industry, the global tourism market is constantly expanding, and the tourism needs of users are becoming increasingly diverse and personalized. More and more tourists are starting to rely on intelligent recommendation systems to help them plan their trips. These systems can provide customized travel suggestions by analyzing large amounts of data and using complex algorithms.

[0003] However, although traditional tourism recommendation methods can already provide suggestions to a certain extent based on users' historical behaviors and preferences, these methods usually rely on some basic matching techniques, such as users' preference settings, the geographical locations of tourist destinations, or the recommendations of popular scenic spots. Although such recommendation methods have improved the user experience to a certain extent, they ignore many key factors, such as users' travel habits, changes in travel seasons, tourist flows in different seasons, and the congestion levels of scenic spots. These factors often determine whether a travel destination is suitable for the needs of tourists. Therefore, traditional recommendation systems may face problems such as low accuracy and poor user experience.

[0004] Therefore, it is necessary to provide a tourism recommendation method and system based on a large model to solve the problem that traditional recommendation methods often cannot effectively meet the diverse needs of tourists. Summary of the Invention

[0005] In view of this, the present invention proposes a tourism recommendation method and system based on a large model, aiming to solve the problem that traditional recommendation methods often cannot effectively meet the diverse needs of tourists.

[0006] On the one hand, the present invention proposes a tourism recommendation method based on a large model, including:

[0007] Collecting users' historical tourism information; wherein, the historical tourism information includes users' residential addresses, users' visited locations, users' single-trip distances, the number of tourist attractions visited by users per trip, and users' travel seasons;

[0008] Obtaining multiple tourism recommendation locations based on the users' residential addresses and single-trip distances, counting the users' travel seasons, counting the number of trips made by users in each season, and determining the recommended seasons of the tourism recommendation locations based on the number of trips made in each season;

[0009] Collect the number of tourists at the travel recommended locations in each season, and collect the number of tourists at the locations where the user has traveled in each season. Determine whether the travel recommended locations are in the off-season or peak season in the recommended season based on the number of tourists at the travel recommended locations in each season. Determine whether the locations where the user has traveled are in the off-season or peak season when the user travels based on the number of tourists at the locations where the user has traveled in each season. Determine whether to delete the travel recommended locations based on whether the recommended season is in the off-season or peak season and whether the locations where the user has traveled are in the off-season or peak season;

[0010] Calculate the average number of tourist attractions per single trip of the user according to the number of tourist attractions of the user in a single trip. Collect the number of tourist attractions at the travel recommended locations. Initially determine whether to delete the travel recommended locations based on the average number of tourist attractions per single trip of the user and the number of tourist attractions at the travel recommended locations. If the initial judgment result is that deletion is required, calculate the range of the number of tourist attractions according to the average number of tourist attractions per single trip of the user. Delete the recommended locations based on the range of the number of tourist attractions and the number of tourist attractions at the travel recommended locations to obtain the final travel recommended locations.

[0011] Further, when obtaining multiple travel recommended locations based on the user's residential address and the user's single-trip distance, it includes:

[0012] Set a distance fluctuation value, obtain the maximum value of the user's single-trip distance, and calculate the upper limit value of the user's single-trip distance; wherein, the upper limit value of the user's single-trip distance is the sum of the maximum value of the user's single-trip distance and the distance fluctuation value;

[0013] Draw a circular area on the map with the user's residential address as the center and the upper limit value of the user's single-trip distance as the radius, count the cities within the circular area and construct a city sequence, remove the locations where the user has traveled from the city sequence, and use the city sequence after removing the locations where the user has traveled as the travel recommended location sequence; wherein, the cities within the travel recommended location sequence are the travel recommended locations.

[0014] Further, when statistically analyzing the user's travel seasons, counting the number of trips of the user in each season, and determining the recommended season of the travel recommended locations according to the number of trips in each season, it includes:

[0015] Count the total number of trips of the user and the number of trips of the user in each season;

[0016] Calculate the travel proportion of the number of trips of the user in each season to the total number of trips of the user;

[0017] Determine the recommended season of the travel recommended locations according to the travel proportion.

[0018] Further, when determining the recommended season of the tourism recommended location according to the travel proportion, it includes:

[0019] If the travel proportion in a certain season is zero, it is determined not to recommend the tourism recommended location to the user in that season;

[0020] If the travel proportion in a certain season is greater than zero, it is determined to recommend the tourism recommended location to the user in that season.

[0021] Further, when determining whether the tourism recommended location is in the off-season or peak season in the recommended season according to the number of tourists in each season of the tourism recommended location, and determining whether the user's already visited locations are in the off-season or peak season when the user travels according to the number of tourists in each season of the user's already visited locations, it includes:

[0022] Count the number of tourists in each season of the tourism recommended location, calculate the first proportion of the number of tourists in each season of the tourism recommended location to the total number of tourists in the whole year, and sort the first proportion in descending order;

[0023] Count the number of tourists in each season of the user's already visited locations, calculate the second proportion of the number of tourists in each season of the user's already visited locations to the total number of tourists in the whole year, and sort the second proportion in descending order;

[0024] Among them, the two seasons with the highest first proportion are set as the peak season, and the two seasons with the lowest first proportion are set as the off-season; the two seasons with the highest second proportion are set as the peak season, and the two seasons with the lowest second proportion are set as the off-season.

[0025] Further, when determining whether to delete the tourism recommended location according to whether the recommended season is in the off-season or peak season and whether the user's already visited locations are in the off-season or peak season, it includes:

[0026] If the user travels at the user's already visited locations are all in the off-season and the recommended season is in the off-season, the tourism recommended location is not deleted;

[0027] If the user travels at the user's already visited locations are all in the peak season and the recommended season is in the peak season, the tourism recommended location is not deleted;

[0028] Otherwise, the tourism recommended location is deleted.

[0029] Further, when calculating the average number of scenic spots per single trip of the user according to the number of scenic spots of the user's single trip, collecting the number of scenic spots of the tourism recommended location, and preliminarily determining whether to delete or retain the tourism recommended location according to the average number of scenic spots per single trip of the user and the number of scenic spots of the tourism recommended location, it includes:

[0030] If the number of scenic spots at the tourist recommended location is equal to the average number of scenic spots per single tourist trip of the user, it is preliminarily determined not to delete the tourist recommended location;

[0031] If the number of scenic spots at the tourist recommended location is not equal to the average number of scenic spots per single tourist trip of the user, it is preliminarily determined to delete the tourist recommended location.

[0032] Furthermore, if the preliminary judgment result is that deletion is required, calculate the range of the number of scenic spots according to the average number of scenic spots per single tourist trip of the user. When obtaining the final tourist recommended location by deleting the recommended location according to the range of the number of scenic spots and the number of scenic spots at the tourist recommended location, it includes:

[0033] Calculate the range of the number of scenic spots according to the average number of scenic spots per single tourist trip of the user. If the number of scenic spots at the tourist recommended location is within the range of the number of scenic spots, it is determined not to delete the tourist recommended location;

[0034] If the number of scenic spots at the tourist recommended location is not within the range of the number of scenic spots, it is determined to delete the tourist recommended location, and the tourist city recommended location in the tourist recommended location sequence after deleting the tourist recommended location is used as the final tourist recommended location.

[0035] Furthermore, when calculating the range of the number of scenic spots according to the average number of scenic spots per single tourist trip of the user, it includes:

[0036] Set the fluctuation value of the number of scenic spots, and calculate the range of the number of scenic spots through the following formula:

[0037] Range of the number of scenic spots = [average number of scenic spots per single tourist trip of the user - fluctuation value of the number of scenic spots, average number of scenic spots per single tourist trip of the user + fluctuation value of the number of scenic spots].

[0038] Compared with the prior art, the beneficial effects of the present invention are as follows: Firstly, by collecting the user's historical travel information, the present invention can deeply understand the user's travel habits and preferences, ensuring that the recommended content highly matches the user's needs. Secondly, the method recommends the most suitable travel destinations for the user according to the user's place of residence and travel distance, combined with seasonal factors, further improving the accuracy and personalization of the recommendation. In addition, by counting the number of tourists at the travel recommended destinations in different seasons, it can dynamically determine whether the destination is in the peak season or off-season in a specific season, avoiding recommending overly crowded tourist destinations and enhancing the user experience. Moreover, the method also takes into account the number of tourist attractions in the user's single travel, and reasonably determines whether to delete attractions by comparing with the number of attractions at the recommended destinations, ensuring that the recommended travel destinations not only conform to the user's travel habits but also maximize the satisfaction of their needs. This multi-dimensional analysis method can comprehensively consider multiple aspects such as user personal preferences, seasonal factors, and tourist flow at travel destinations, providing accurate and personalized travel recommendations, significantly enhancing the effectiveness of the recommendation system and user satisfaction.

[0039] On the other hand, the present application also provides a travel recommendation system based on a large model, including:

[0040] A collection module, configured to collect the user's historical travel information; wherein, the historical travel information includes the user's residential address, the user's traveled locations, the user's single-trip travel distance, the number of tourist attractions in the user's single travel, and the user's travel season;

[0041] A preliminary recommendation determination module, configured to obtain multiple travel recommended destinations according to the user's residential address and the user's single-trip travel distance, count the number of trips made by the user in each season, and determine the recommended season of the travel recommended destinations according to the number of trips made in each season;

[0042] A recommendation judgment module, configured to collect the number of tourists at the travel recommended destinations in each season, and collect the number of tourists at the user's traveled locations in each season, determine whether the travel recommended destinations are in the off-season or peak season in the recommended season according to the number of tourists at the travel recommended destinations in each season, determine whether the user's traveled locations are in the off-season or peak season when the user travels according to the number of tourists at the user's traveled locations in each season, and determine whether to delete the travel recommended destinations according to whether the recommended season is in the off-season or peak season and whether the user's traveled locations are in the off-season or peak season;

[0043] The final recommendation determination module is configured to calculate the average number of tourist attractions per single trip of the user according to the number of tourist attractions at the user's single trip, collect the number of tourist attractions at the tourist recommendation locations, initially determine whether to delete the tourist recommendation locations according to the average number of tourist attractions per single trip of the user and the number of tourist attractions at the tourist recommendation locations. If the initial determination result is that deletion is required, calculate the range of the number of tourist attractions according to the average number of tourist attractions per single trip of the user, and delete the recommendation locations according to the range of the number of tourist attractions and the number of tourist attractions at the tourist recommendation locations to obtain the final tourist recommendation locations.

[0044] It can be understood that the tourism recommendation method and system based on the large model provided by this application have the same beneficial effects and will not be elaborated here. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] By reading the following detailed description of the preferred embodiments, various other advantages and benefits will become clear to those of ordinary skill in the art. The drawings are only for the purpose of showing the preferred embodiments and are not considered as limiting the present invention. Moreover, throughout the drawings, the same reference numerals are used to represent the same components. In the drawings:

[0046] Figure 1 is a flowchart of the tourism recommendation method based on the large model provided by the embodiment of the present invention;

[0047] Figure 2 is a functional block diagram of the tourism recommendation system based on the large model provided by the embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0048] Hereinafter, exemplary embodiments of the present disclosure will be described in more detail with reference to the drawings. Although the exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided so that the present disclosure can be more thoroughly understood and the scope of the present disclosure can be fully communicated to those skilled in the art. It should be noted that, without conflict, the embodiments in the present invention and the features in the embodiments can be combined with each other. Hereinafter, the present invention will be described in detail with reference to the drawings and in conjunction with the embodiments.

[0049] In some embodiments of the present application, referring to Figure 1 as shown, this embodiment provides a tourism recommendation method based on a large model, including the following steps:

[0050] S100. Collect user historical travel information; wherein, the historical travel information includes the user's residential address, the user's traveled locations, the user's single-trip distance, the number of tourist attractions per single trip of the user, and the user's travel season;

[0051] S200, obtaining a plurality of recommended tourist spots according to the user's residential address and the user's single travel distance, counting the user's travel seasons, counting the number of times the user travels in each season, and determining the recommended season of the recommended tourist spots according to the number of times the user travels in each season;

[0052] S300, collecting the number of tourists in each season at the recommended tourist spot, and collecting the number of tourists in each season at the tourist spot that the user has visited, judging whether the recommended tourist spot is in the off-season or peak season in the recommended season according to the number of tourists in each season at the tourist spot that the user has visited, judging whether the tourist spot that the user has visited is in the off-season or peak season when the user travels according to the number of tourists in each season at the tourist spot that the user has visited, and judging whether to delete the recommended tourist spot according to whether the recommended season is in the off-season or peak season and whether the tourist spot that the user has visited is in the off-season or peak season;

[0053] S400, calculating the average number of tourist attractions per visit of the user based on the number of tourist attractions per visit of the user, collecting the number of attractions of the recommended tourist locations, and preliminarily determining whether to delete the recommended tourist locations based on the average number of tourist attractions per visit of the user and the number of attractions of the recommended tourist locations; if the preliminary determination result is that deletion is required, calculating the range of the number of attractions based on the average number of tourist attractions per visit of the user, deleting the recommended locations based on the range of the number of attractions and the number of attractions of the recommended tourist locations, and obtaining the final recommended tourist locations.

[0054] It is understandable that the present invention can firstly gain an in-depth understanding of the user's travel habits and preferences by collecting the user's historical travel information, and ensure that the recommended content is highly consistent with the user's needs. Secondly, the method recommends the most suitable tourist attractions to the user based on the user's place of residence and travel distance, combined with seasonal factors, and further improves the accuracy and personalization of the recommendation. In addition, by counting the number of tourists in the recommended tourist attractions in different seasons, it is possible to dynamically determine whether the location is in the peak season or off-season in a specific season, avoid recommending overcrowded tourist attractions, and improve the user experience. Furthermore, the method also takes into account the number of single tourist attractions of the user, and by comparing it with the number of attractions in the recommended locations, it is reasonably determined whether the attractions need to be deleted, ensuring that the recommended tourist attractions are in line with the user's travel habits and can also maximize their needs. This multi-dimensional analysis method can comprehensively consider multiple aspects such as user personal preferences, seasonal factors, and tourist traffic in tourist destinations, and provide accurate and personalized travel recommendations, which significantly improves the effect of the recommendation system and user satisfaction.

[0055] In some embodiments of the present application, obtaining a plurality of recommended tourist spots according to the user's residential address and the user's single travel distance includes:

[0056] Set a distance fluctuation value, obtain the maximum value of the user's single-trip distance, and calculate the upper limit value of the user's single-trip distance; wherein, the upper limit value of the user's single-trip distance is the sum of the maximum value of the user's single-trip distance and the distance fluctuation value;

[0057] Taking the user's residential address as the center and the upper limit value of the user's single-trip distance as the radius, draw a circular area on the map, count the cities within the circular area and construct a city sequence, remove the user's already visited tourist locations from the city sequence, and use the city sequence after removing the user's already visited tourist locations as the tourist recommendation location sequence; wherein, the cities within the tourist recommendation location sequence are the tourist recommendation locations.

[0058] It can be understood that the present invention can effectively improve the accuracy and personalization of recommendations. By setting a distance fluctuation value and calculating the upper limit of the user's single trip, the recommended range can be dynamically adjusted to better meet the different travel needs of users. Using the user's residential address as the center and combining the upper limit distance to delimit the recommended area can quickly screen out the cities closer to the user, and further remove the already visited locations to ensure that the recommended results are more in line with the destinations that the user has not visited. This method not only avoids repeated recommendations but also helps users discover new tourist destinations and improves the user experience. At the same time, through the statistics and screening of the cities within the area, the travel range and comfort of the user can be considered to provide a more reasonable and personalized travel recommendation plan.

[0059] In some embodiments of the present application, when statistically analyzing the user's travel seasons, counting the number of times the user travels in each season, and determining the recommended season of the tourist recommendation location according to the number of times of travel in each season, it includes:

[0060] Count the total number of times the user travels and the number of times the user travels in each season;

[0061] Calculate the travel proportion of the number of times the user travels in each season to the total number of times the user travels;

[0062] Determine the recommended season of the tourist recommendation location according to the travel proportion.

[0063] In some embodiments of the present application, when determining the recommended season of the tourist recommendation location according to the travel proportion, it includes:

[0064] If the travel proportion of a certain season is zero, it is determined that the tourist recommendation location is not recommended to the user in that season;

[0065] If the travel proportion of a certain season is greater than zero, it is determined that the tourist recommendation location is recommended to the user in that season.

[0066] It can be understood that by calculating the travel proportion of users in each season, the present invention can more accurately reflect the travel habits and seasonal preferences of users. By counting the total number of trips of users and the number of trips in each season, it is possible to clearly understand the travel frequency of users in different seasons, thereby judging their preferred travel seasons, and effectively avoiding recommending travel destinations that do not match the user's preferences. For example, if the travel proportion of a user in a certain season is zero, avoid recommending this travel destination in that season and avoid recommending seasons that the user is not interested in. Conversely, if the travel proportion in a certain season is relatively high, relevant travel destinations will be recommended in that season to ensure that the recommendation is closer to the user's needs. This dynamic adjustment based on user behavior analysis can provide more personalized and accurate travel recommendations, significantly improving user satisfaction and experience.

[0067] In some embodiments of the present application, when judging whether the travel recommendation destination is in the off-season or peak season in the recommended season according to the number of tourists at the travel recommendation destination in each season, and judging whether the user's already traveled destination is in the off-season or peak season when the user travels according to the number of tourists at the user's already traveled destination in each season, it includes:

[0068] Count the number of tourists at the travel recommendation destination in each season, calculate the first proportion of the number of tourists in each season of the travel recommendation destination to the total number of tourists in the whole year, and sort the first proportion in descending order;

[0069] Count the number of tourists at the user's already traveled destination in each season, calculate the second proportion of the number of tourists in each season of the user's already traveled destination to the total number of tourists in the whole year, and sort the second proportion in descending order;

[0070] Among them, the two seasons with the highest first proportion are set as peak seasons, and the two seasons with the lowest first proportion are set as off-seasons; the two seasons with the highest second proportion are set as peak seasons, and the two seasons with the lowest second proportion are set as off-seasons.

[0071] In some embodiments of the present application, when judging whether to delete the travel recommendation destination according to whether the recommended season is in the off-season or peak season and whether the user's already traveled destination is in the off-season or peak season, it includes:

[0072] If the user travels at the user's already traveled destination during off-seasons, and the recommended season is an off-season, then do not delete the travel recommendation destination;

[0073] If the user travels at the user's already traveled destination during peak seasons, and the recommended season is a peak season, then do not delete the travel recommendation destination;

[0074] Otherwise, delete the travel recommendation destination.

[0075] It can be understood that the present invention greatly improves the accuracy of travel recommendations and the user experience. By counting the tourist flow at travel recommendation locations in different seasons and combining the proportion of the number of tourists, it is possible to determine the peak and off-peak seasons of each season, and then reasonably adjust the recommendation strategy. At the same time, through the seasonal analysis of the locations where the user has traveled, it is possible to understand the user's travel habits and judge whether they prefer peak-season or off-peak-season travel methods. Combining these two types of information, the recommendation results can be dynamically adjusted according to the user's seasonal travel habits at the locations where they have traveled and the seasonal status of the recommendation locations, ensuring that the recommended travel locations not only meet the user's seasonal preferences but also avoid recommending travel destinations that do not meet their expectations. If the user's travel experiences are mainly concentrated in the off-peak or peak season, it can be automatically determined whether the recommended travel locations are suitable, avoiding ineffective recommendations and enhancing the personalization of the recommendations. Through this multi-dimensional seasonal analysis, the recommendation system can provide travel locations that are more in line with the individual travel needs of users, thereby enhancing user satisfaction and the travel experience.

[0076] In some embodiments of the present application, when calculating the average number of tourist attractions per single trip of the user according to the number of tourist attractions at the user's single trip, collecting the number of tourist attractions at the travel recommendation location, and initially determining whether to delete the travel recommendation location based on the average number of tourist attractions per single trip of the user and the number of tourist attractions at the travel recommendation location, it includes:

[0077] If the number of tourist attractions at the travel recommendation location is equal to the average number of tourist attractions per single trip of the user, it is initially determined not to delete the travel recommendation location;

[0078] If the number of tourist attractions at the travel recommendation location is not equal to the average number of tourist attractions per single trip of the user, it is initially determined to delete the travel recommendation location.

[0079] It can be understood that the present invention can effectively optimize travel recommendations according to the user's travel habits. By calculating the average number of tourist attractions per single trip of the user, it is possible to understand how many tourist attractions the user usually visits during each trip, and then compare it with the number of tourist attractions at the recommendation location to determine whether it is necessary to delete travel recommendation locations. If the number of tourist attractions at the travel recommendation location matches the average number of tourist attractions per single trip of the user, it is determined that no deletion is required. On the contrary, if the number of tourist attractions does not match, it will be initially determined whether deletion is necessary, avoiding recommending too many or too few tourist attractions, ensuring that the recommended travel itinerary not only meets the user's needs but also does not result in excessive or insufficient travel arrangements. This method optimizes the number of recommended tourist attractions by docking with the user's travel habits, making the recommendation results more in line with the user's actual needs and enhancing the comfort and satisfaction of the travel experience.

[0080] In some embodiments of the present application, when the preliminary judgment result is that deletion is required, calculating the range of the number of scenic spots according to the average number of scenic spots per single trip of the user, and deleting the recommended locations according to the range of the number of scenic spots and the number of scenic spots of the travel recommendation locations to obtain the final travel recommendation locations includes:

[0081] Calculating the range of the number of scenic spots according to the average number of scenic spots per single trip of the user. If the number of scenic spots of the travel recommendation location is within the range of the number of scenic spots, it is determined not to delete the travel recommendation location;

[0082] If the number of scenic spots of the travel recommendation location is not within the range of the number of scenic spots, it is determined to delete the travel recommendation location, and the travel city recommendation location in the travel recommendation location sequence after deleting the travel recommendation location is used as the final travel recommendation location.

[0083] In some embodiments of the present application, when calculating the range of the number of scenic spots according to the average number of scenic spots per single trip of the user, it includes:

[0084] Setting the fluctuation value of the number of scenic spots, and calculating the range of the number of scenic spots through the following formula:

[0085] Range of the number of scenic spots = [average number of scenic spots per single trip of the user - fluctuation value of the number of scenic spots, average number of scenic spots per single trip of the user + fluctuation value of the number of scenic spots].

[0086] It can be understood that the present invention can accurately adjust the number of recommended scenic spots to better conform to the travel habits of users. By setting the fluctuation value of the number of scenic spots and calculating the range of the number of scenic spots, a suitable range of the number of scenic spots for users can be flexibly defined, avoiding recommending too many or too few scenic spots. If the number of scenic spots of the travel recommendation location is within this range, it will not be deleted, ensuring that the number of recommended scenic spots is consistent with the travel habits of users. If the number of scenic spots exceeds the range, the location will be deleted, and the final travel recommendation list will be generated based on the recommended locations after deletion. This method can effectively control the number of recommended scenic spots while not losing personalized recommendations, improve the accuracy of recommendations and user satisfaction, ensure that the travel itinerary is neither cumbersome nor monotonous, and meet the needs of users to the greatest extent.

[0087] On the other hand, as shown in Figure 2 the present application also provides a travel recommendation system based on a large model for applying the above-mentioned travel recommendation method based on a large model, including:

[0088] A collection module configured to collect historical travel information of users; wherein, the historical travel information includes the user's residential address, the locations where the user has traveled, the single-trip distance of the user, the average number of scenic spots per single trip of the user, and the travel season of the user;

[0089] The preliminary recommendation determination module is configured to obtain multiple tourist recommendation locations based on the user's residential address and the user's single-trip distance, count the user's travel seasons, count the number of trips made by the user in each season, and determine the recommended seasons of the tourist recommendation locations according to the number of trips made in each season;

[0090] The recommendation judgment module is configured to collect the number of tourists at the tourist recommendation locations in each season, and collect the number of tourists at the user's already-visited tourist locations in each season. Determine whether the tourist recommendation location is in the off-season or peak season during the recommended season according to the number of tourists at the tourist recommendation location in each season, determine whether the user's already-visited tourist location is in the off-season or peak season when the user travels according to the number of tourists at the user's already-visited tourist location in each season, and determine whether to delete the tourist recommendation location according to whether the recommended season is in the off-season or peak season and whether the user's already-visited tourist location is in the off-season or peak season;

[0091] The final recommendation determination module is configured to calculate the user's average number of tourist attractions per single trip according to the number of tourist attractions of the user's single trip, collect the number of tourist attractions of the tourist recommendation location, and preliminarily determine whether to delete the tourist recommendation location according to the user's average number of tourist attractions per single trip and the number of tourist attractions of the tourist recommendation location. If the preliminary judgment result is that deletion is required, calculate the range of the number of tourist attractions according to the user's average number of tourist attractions per single trip, and delete the recommended location according to the range of the number of tourist attractions and the number of tourist attractions of the tourist recommendation location to obtain the final tourist recommendation location.

[0092] It can be understood that through modular design, the present invention precisely realizes personalized travel recommendations and has significant advantages. First of all, the collection module can comprehensively obtain the user's historical travel information, including residential address, already-visited tourist locations, travel distance, number of tourist attractions, and travel seasons, providing an accurate user portrait for the system. This information lays a solid data foundation for subsequent recommendations. The preliminary recommendation determination module can quickly screen out multiple tourist recommendation locations on the map by analyzing the user's travel habits, combined with the place of residence and travel distance, and determine the most suitable recommended season based on seasonal statistics, making the recommendation more in line with the user's travel needs. The recommendation judgment module further intelligently judges whether the recommended location is in the peak season or off-season by analyzing the tourist flow of the tourist recommendation location and combining the seasonal data of the user's already-visited tourist locations. This seasonal analysis can enhance the user's travel experience. Finally, the final recommendation determination module dynamically adjusts the number of recommended tourist attractions by comparing the user's scenic spot preferences with the number of scenic spots at the recommended location, avoiding too many or too few recommendations, making the travel itinerary not only in line with the user's habits but also avoiding irrationality in scenic spot arrangements. Generally speaking, this system can provide accurate and personalized travel recommendations for users through multi-level and refined analysis, greatly improving the accuracy of the recommendation system and user satisfaction.

[0093] Specifically, when playing in a scenic area in the city, the present invention performs data prediction through the following steps and finally completes the recommendation of the scenic area route.

[0094] Step 1: Generate passenger flow prediction data by using the constructed fusion prediction algorithm. Construction process of the fusion prediction algorithm: In the data feature extraction stage, historical passenger flow data (basic data), weather (such as sunny / rainy, wind speed, etc.), temperature, air quality, season, holidays, scenic area activities, surrounding hotels, restaurants, number of business districts, congestion status of weekly report sections, and ticket price features are extracted as inputs. Passenger flow prediction is performed through time series prediction algorithms TimeGPT, N-HiTS, and LGBM. Feature fusion is performed on the three types of prediction results, and the prediction algorithm of the second layer is trained to perform prediction again. The prediction results of each recommendation algorithm are used as features, and the statistically obtained scenic area passenger flow data is used as the training sample to form the training set of the second layer prediction model. By capturing the complementary information between them, the overall prediction accuracy is improved. The second layer prediction model selects the SVM classification algorithm, compares it with the historical real data set, selects the probability threshold, and updates it through random search to prevent the algorithm from falling into local optimality and make the result converge to meet the prediction frequency of hourly output. This prediction algorithm can be output at time granularities of hour, day, week, and month to meet the needs of different users or application scenarios.

[0095] Step 2: Route Recommendation Based on Knowledge Graph. Based on the operator's Jiutian large model platform, use the large model to generate a scenic spot knowledge graph. The specific process is as follows: Set a predefined formatted scenic spot dictionary template, input the scenic spot dictionary template and the scenic spot name into the large model, and use the context example selection strategy to provide multiple examples to the model to guide it to output a formatted scenic spot dictionary description that matches the examples. This description follows the predefined template and contains detailed information about the scenic spot. Subsequently, according to the entity attributes related to the scenic spot dictionary template, extract the corresponding description information and store it in the database table for later use. Among them, the entity attributes are predicted passenger flow, estimated play time, scenic spot score, scenic spot name, geographical location, opening hours, ticket price, scenic spot features, etc. At the same time, with the help of the offline map API, obtain the transportation methods, travel expenses, distances, and required times between all scenic spots, and then organize and store this information in the scenic spot relationship table of the database. Each row of the record reflects the travel method, travel expenses, distance, and time consumption between two scenic spots. When constructing the scenic spot knowledge graph, read the data one by one from the scenic spot relationship table in the database, and recursively construct a knowledge graph covering all scenic spots in the city according to the structure of "scenic spot - relationship - scenic spot". At the same time, retrieve and add the detailed information of the corresponding scenic spots from the scenic spot dictionary table in the database to the corresponding nodes of the graph, map the scenic spot knowledge graph to the feature vector space, and obtain the feature representation of the scenic spot attribute context. In addition, adopt the supervised learning method to train the large model using the labeled dataset. Finally, realize the automatic construction of the knowledge graph based on the large model.

[0096] Calculation of Scenic Spot Play Time: Based on the operator's signaling data, analyze the residence duration of scenic spot visitors in the scenic area to obtain the corresponding relationship table between the number of visitors and the play time. Obtain the hourly passenger flow prediction data, weather, temperature, air quality, holidays, seasons, scenic area activities, historical play time distribution, etc. of each scenic area in a day as training samples, and based on the ensemble learning algorithm, output the estimated play time of the scenic area at a certain future moment. Compared with the scenic area play duration data obtained through Internet data in the past, the play duration of the scenic area predicted by this method is more in line with the local actual situation and is updated in real time, solving the problems of missing and large deviation in the Internet play duration data of the scenic area.

[0097] Calculation of Scenic Spot Score: Obtain the interactive data of tourists in the scenic area from Internet data such as travel websites and APPs, such as travel notes and comments. Based on multiple value information of the scenic area, we calculate to obtain the scenic spot score and store this index in an independent table, and at the same time ensure that this table is associated with the scenic spot information table. We regard the scenic spot score as an important attribute information. Then, we compare the newly generated score with the data of the corresponding attribute in the original database. If there is a significant difference between the two, we will update the new valid information to the corresponding attribute of the database table. Scenic Spot Si The score of Score(S i ) is calculated as follows:

[0098] Score(S i ) = ω1 × pvalue(S i ) + ω2 × prate(S i );

[0099] In the above formula, Score(S i ) represents the score of scenic spot S i , prate(S i ) represents the rating of scenic spot S i , pvalue(S i ) represents the value degree of scenic spot S i ; ω1 and ω2 are adjustment parameters, and ω1 + ω2 = 1;

[0100] The value degree pvalue(S i ) of scenic spot S i is defined as:

[0101]

[0102] In the above formula, S i-1 represents the previous scenic spot of scenic spot S i in the route, S i+1 represents the next scenic spot of scenic spot S i in the route, T(S i ) represents the playing duration of scenic spot S i , T(S i , S i-1 ) represents the travel time from scenic spot S i-1 to scenic spot S i , T(S i , S i+1 ) represents the travel time from scenic spot S i to S i+1 , M(S i ) represents the playing cost of scenic spot S i , M(S i , S i-1 ) represents the travel cost from scenic spot S i-1 to scenic spot S i , M(S i , S i+1 ) represents the travel cost from scenic spot S i to S i+1 .

[0103] Using web crawler technology, collect evaluation information of each scenic spot, and store these evaluations in independent database tables according to the affiliated scenic spots respectively, establishing a logical connection with the relevant information tables of the scenic spots. These evaluations are passed as input data to the large model. The model uses its semantic understanding ability to perform binary classification on the evaluations, distinguishing positive evaluations from negative evaluations. The classification results are directly fed back to the corresponding evaluation records, approaching the real experience of users.

[0104] The number of repeat visits to a scenic area reflects the overall rating of the scenic area. Based on the operator's location capabilities, data mining is carried out on tourists who have visited the scenic area multiple times, and then the number of tourists who have visited a certain scenic area more than twice within half a year is analyzed, and the proportion of this part of tourists in all tourists within half a year is calculated to dynamically adjust the overall rating of the scenic spot.

[0105] Scenic spot S i 's rating prate(S i ) is defined as:

[0106]

[0107] Among them, N t is the number of positive evaluations captured, N f is the number of negative evaluations captured; U i is the number of people in the i-th play frequency interval of the scenic area, and m is the play frequency.

[0108] Calculation of route score. Finally, the score G(R i ) of a route R i is:

[0109]

[0110] Among them, n is the number of scenic spots included in route R i , score(S i ) represents the score of scenic spot S i ; j represents the j-th scenic spot.

[0111] Step 3: Optimize and recommend the scenic spot visit order by restricting the constraint conditions.

[0112] The coordinate information L(lat, lon) of the user's starting point location, as well as the given user time constraint, and at the same time, the estimated play time information of each scenic spot has also been marked. A local path network is formed between the scenic spots, and the numbers marked on the path edges represent the passing distance between two scenic spots. On the basis of adding the play time of scenic spot tourists, use the location attributes of scenic spots in the knowledge graph to calculate the traffic time between scenic spots. The traffic time T(S x ,S y ) required from scenic spot S1 to scenic spot S2 is calculated as follows:

[0113] T(S x ,S y ) = D(S x ,S y ) / V;

[0114] Among them, D(S x ,S y ) is the driving distance between scenic spot x and scenic spot y; V is the traveling speed between two points. In this article, during the calculation of traffic time, different speeds are selected according to the length of the one-way driving distance of tourists: walking is selected when it is less than 1.5 km, cycling is selected when it is greater than 1.5 km and less than 4 km, and driving is selected when it is greater than 4 km. By crawling the price data of Meituan, Hello Bicycle, and taxi, the passing cost between scenic spots is obtained. Finally, a travel route sequence is obtained.

[0115] Default route recommendation: Based on the analysis of user location signaling (historical play trajectories, time, frequently visited scenic spots, etc.), Internet access logs (behavior characteristics, etc.) and other characteristics (mobile phone brands, communications, consumption, etc.), different groups such as office workers, the elderly, and business people are characterized, and the shortest travel route in terms of time is recommended by default; for groups with low consumption ability and college students, the travel route with the least cost is recommended by default; for short-term travelers and self-driving tour groups, the shortest travel route in terms of distance is recommended by default. This intelligent decision-making method can avoid the blindness and inefficiency in traditional recommendation methods and improve the accuracy and effectiveness of recommendations.

[0116] Personalized route recommendation: Based on the demands of tourists such as preferences and schedule arrangements, using the knowledge theory of knowledge graphs, an automatic travel route schedule plan is generated for tourists' travel. At the same time, tourists can mark and modify according to their own needs and can customize the travel route according to their needs. The specific steps are as follows: clarify multiple goals in path planning, such as restricted play time, scenic spot play scores, costs, etc., and define corresponding functions for each goal. These functions will be used as the input of the multi-objective optimization algorithm; use the dynamic path planning algorithm to construct an initial path under the initial environmental information, and this path will be used as the starting point of the multi-objective optimization algorithm; input the initial path and multi-objective functions into the multi-objective optimization algorithm, and the algorithm will search for the optimal solution set that satisfies all objective functions; select one or more optimal solutions from the Pareto front.

[0117] Suppose each scenic spot is represented as S1, S2, S3, ……, S n in descending order of scores. The visit time for each scenic spot is obtained from the scenic spot knowledge graph as T(S1), T(S2), ……, T(S n ), and the time on the road between every two scenic spots is T(S1, S2), T(S2, S3), ……, T(S n-1 , S n), if the tour is carried out in the order of S1, S2, ……, S n , the required time is:

[0118] T = T(S1) + T(S1, S2) + T(S2) +...... + T(S n-1 , S n ) + T(S n );

[0119] Shortest time route recommendation: Arrange and combine the recommended scenic spots within different time periods, obtain multiple tour routes according to the arrangement and combination results, calculate the commuting time between two scenic spots based on the scenic spot knowledge graph, as well as the playing duration of each scenic area, and add up the commuting times of all arranged and combined scenic spots and the estimated playing times of the scenic areas to obtain the total commuting time, and select the route and scenic area with the shortest time as the recommendation:

[0120] T(S1) + T(S1, S2) + T(S2) +...... + T(S n-1 , S n ) + T(S n ) ≤ T1 - T0;

[0121] Among them, T0 is the starting time of the trip defined by the user, and T1 is the ending time of the trip defined by the user.

[0122] Least cost route recommendation: By collecting and analyzing the ticket prices and transportation costs of each scenic spot, estimate the costs of each arranged and combined route, and select the route within the user's budget as the recommendation:

[0123] M(S1) + M(S1, S2) + M(S2) +...... + M(S n-1 , S n ) + M(S n ) ≤ M;

[0124] Highest route score recommendation: By analyzing data such as passenger flow prediction, stay duration estimation, travel time between scenic areas, scenic area ratings, tourist reviews, and historical visit volumes of each scenic spot, score each route based on the above evaluation dimensions:

[0125] G(R i ) ≥ G;

[0126] According to the preferences of tourists, using the semantic understanding function of the large language model, screen out the scenic spots that are most relevant to the tourists' preferences and needs from the arrangement and combination of scenic spots, and then generate a special tourist route using the intelligent tourist route recommendation scheme. At the same time, tourists can modify the route to meet their expectations.

[0127] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, a system, or a computer program product. Therefore, the present application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0128] The present application is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to the embodiments of the present application. It should be understood that each flow and / or block in the flowchart and / or block diagram can be implemented by computer program instructions, and the combination of the flows and / or blocks in the flowchart and / or block diagram can also be implemented. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate a device for realizing the functions specified in Figure 1 one or more of the flows Figure 1 or blocks or the combination of blocks.

[0129] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory generate a manufactured article including an instruction device, and the instruction device realizes the functions specified in Figure 1 one or more of the flows Figure 1 or blocks or the combination of blocks.

[0130] These computer program instructions can also be loaded onto a computer or other programmable data processing device, so that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process. Therefore, the instructions executed on the computer or other programmable device provide steps for realizing the functions specified in Figure 1 one or more of the flows Figure 1 or blocks or the combination of blocks.

[0131] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them. Although the present invention has been described in detail with reference to the above embodiments, those of ordinary skill in the art should understand that: the specific implementation manners of the present invention can still be modified or equivalently replaced, and any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered by the protection scope of the claims of the present invention.

Claims

1. A tourism recommendation method based on a large model, characterized in that: include: Collecting historical travel information of the user; wherein the historical travel information includes the user's residential address, the places the user has traveled, the distance of the user's single trip, the number of attractions the user has visited in a single trip, and the season of the user's travel; Obtain multiple recommended tourist spots based on the user's residential address and the user's single travel distance, count the user's travel seasons, count the number of times the user travels in each season, and determine the recommended season of the recommended tourist spots based on the number of times the user travels in each season; Collect the number of tourists in each season at the recommended tourist spot, and collect the number of tourists in each season at the tourist spot that the user has visited, judge whether the recommended tourist spot is in the off-season or peak season in the recommended season according to the number of tourists in each season at the tourist spot that the user has visited, judge whether the tourist spot that the user has visited is in the off-season or peak season when the user travels according to the number of tourists in each season at the tourist spot that the user has visited, and judge whether to delete the recommended tourist spot according to whether the recommended season is in the off-season or peak season and whether the tourist spot that the user has visited is in the off-season or peak season; The average number of tourist attractions per visit of the user is calculated based on the number of tourist attractions visited by the user in a single visit, the number of attractions of the recommended tourist locations is collected, and a preliminary judgment is made whether to delete the recommended tourist locations based on the average number of tourist attractions per visit of the user and the number of attractions of the recommended tourist locations; if the preliminary judgment result is that deletion is required, the range of the number of attractions is calculated based on the average number of tourist attractions per visit of the user, and the recommended locations are deleted based on the range of the number of attractions and the number of attractions of the recommended tourist locations to obtain the final recommended tourist locations.

2. The tourism recommendation method based on a large model according to claim 1, characterized in that: The method of obtaining a plurality of recommended tourist spots according to the user's residential address and the user's single travel distance includes: Setting a distance fluctuation value, obtaining the maximum value of the single travel distance of the user, and calculating the upper limit value of the single travel distance of the user; wherein the upper limit value of the single travel distance of the user is the sum of the maximum value of the single travel distance of the user and the distance fluctuation value; A circular area is drawn on the map with the user's residential address as the center and the upper limit of the user's single travel distance as the radius, and the cities within the circular area are counted and a city sequence is constructed. The places where the user has traveled are removed from the city sequence, and the city sequence after removing the places where the user has traveled is used as a tourist recommended place sequence; wherein the cities within the tourist recommended place sequence are the tourist recommended places.

3. The tourism recommendation method based on a large model according to claim 2, characterized in that: The counting of the user's travel seasons, counting the number of user's travel times in each season, and determining the recommended season of the recommended tourist location according to the number of travel times in each season, includes: Count the total number of user trips and the number of user trips in each season; Calculate the proportion of users' trips in each season to their total trips; The recommended season of the recommended tourist destination is determined according to the travel proportion.

4. The tourism recommendation method based on a large model according to claim 3 is characterized in that: The determining of the recommended season of the recommended tourist location according to the travel proportion includes: If the travel ratio in a certain season is zero, it is determined that the recommended tourist location is not recommended to the user in that season; If the travel proportion in a certain season is greater than zero, it is determined that the recommended tourist location is recommended to the user in that season.

5. The tourism recommendation method based on a large model according to claim 4 is characterized in that: The method of judging whether the recommended tourist spot is in the off-season or the peak season in the recommended season according to the number of tourists in each season at the recommended tourist spot, and judging whether the tourist spot that the user has visited when the user travels is in the off-season or the peak season according to the number of tourists in each season at the tourist spot that the user has visited, comprises: Counting the number of tourists visiting the recommended tourist spots in each season, calculating the first number ratio of the number of tourists visiting the recommended tourist spots in each season to the first number ratio of tourists visiting the whole year, and arranging the first number ratios in descending order; Counting the number of tourists in each season at the tourist destination that the user has visited, calculating the second proportion of the number of tourists in each season at the tourist destination that the user has visited to the total number of tourists throughout the year, and arranging the second proportions in descending order; Among them, the two seasons with the highest proportion of the first population are set as the peak season, and the two seasons with the lowest proportion of the first population are set as the off-season; the two seasons with the highest proportion of the second population are set as the peak season, and the two seasons with the lowest proportion of the second population are set as the off-season.

6. The tourism recommendation method based on a large model according to claim 5, characterized in that: The determining whether to delete the recommended tourist spot according to whether the recommended season is in the off-season or the peak season and the tourist spot the user has visited is in the off-season or the peak season includes: If the user travels to the places that the user has traveled to during the off-season, and the recommended season is the off-season, the recommended tourist places will not be deleted; If the user is in the peak season when visiting the places that the user has visited, and the recommended season is the peak season, the recommended tourist places will not be deleted; Otherwise, delete the recommended tourist spot.

7. The tourism recommendation method based on a large model according to claim 6, characterized in that: The calculating the average number of tourist attractions per visit of the user according to the number of tourist attractions per visit of the user, collecting the number of tourist attractions of the recommended tourist spots, and preliminarily determining whether to delete the recommended tourist spots according to the average number of tourist attractions per visit of the user and the number of tourist attractions of the recommended tourist spots, includes: If the number of attractions in the recommended tourist spots is equal to the average number of attractions per trip of the user, it is preliminarily determined that the recommended tourist spots will not be deleted; If the number of attractions in the recommended tourist location is not equal to the average number of attractions visited by the user in a single trip, it is preliminarily determined that the recommended tourist location should be deleted.

8. The tourism recommendation method based on a large model according to claim 7, characterized in that: If the preliminary judgment result is that deletion is needed, the number range of scenic spots is calculated according to the average number of scenic spots per visit of the user, and the recommended spots are deleted according to the number range of scenic spots and the number of scenic spots in the recommended tourist spots, and the final recommended tourist spots are obtained, including: Calculate the range of the number of attractions according to the average number of attractions per single visit of the user, and if the number of attractions of the recommended tourist spot is within the range of the number of attractions, determine not to delete the recommended tourist spot; If the number of attractions of the recommended tourist spot is not within the range of the number of attractions, it is determined that the recommended tourist spot is to be deleted, and the tourist city recommended spot in the recommended tourist spot sequence after the recommended tourist spot is deleted is used as the final recommended tourist spot.

9. The tourism recommendation method based on a large model according to claim 8, characterized in that: The calculation of the number of attractions based on the average number of attractions per visit by users includes: Set the fluctuation value of the number of attractions and calculate the range of the number of attractions by the following formula: The range of the number of attractions = [the average number of attractions per visit for users - the fluctuation value of the number of attractions, the average number of attractions per visit for users + the fluctuation value of the number of attractions].

10. A tourism recommendation system based on a large model, used for applying the tourism recommendation method based on a large model as described in any one of claims 1 to 9, characterized in that: include: A collection module is configured to collect historical travel information of the user; wherein the historical travel information includes the user's residential address, the user's travel locations, the user's single travel distance, the number of attractions the user has visited in a single trip, and the user's travel season; A preliminary recommendation determination module is configured to obtain a plurality of recommended tourist spots according to the user's residence address and the user's single travel distance, count the user's travel seasons, count the number of times the user travels in each season, and determine the recommended season of the recommended tourist spot according to the number of times the user travels in each season; The recommendation judgment module is configured to collect the number of tourists in each season at the recommended tourist spot, and collect the number of tourists in each season at the tourist spot that the user has visited, judge whether the recommended tourist spot is in the off-season or peak season at the recommended season according to the number of tourists in each season at the tourist spot that the user has visited, judge whether the tourist spot that the user has visited is in the off-season or peak season when the user travels according to the number of tourists in each season at the tourist spot that the user has visited, and judge whether to delete the recommended tourist spot according to whether the recommended season is in the off-season or peak season and whether the tourist spot that the user has visited is in the off-season or peak season; The final recommendation determination module is configured to calculate the average number of tourist attractions per visit of the user based on the number of tourist attractions per visit of the user, collect the number of attractions of the recommended tourist locations, and preliminarily determine whether to delete the recommended tourist locations based on the average number of tourist attractions per visit of the user and the number of attractions of the recommended tourist locations; if the preliminary determination result is that deletion is required, calculate the range of the number of attractions based on the average number of tourist attractions per visit of the user, delete the recommended locations based on the range of the number of attractions and the number of attractions of the recommended tourist locations, and obtain the final recommended tourist locations.

Images