A tourist travel route recommendation method and device based on association rule mining

Abstract

The application discloses a tourist travel route recommendation method and device based on association rule mining, which comprises the following steps: firstly, mobile signaling data is sorted and analyzed, tourist related information is extracted, and tourist characteristics are analyzed; in combination with a map information system, popular scenic spots are obtained; then, clustering analysis is performed on the behavior mode of tourists, and on this basis, the characteristics of each type of tourists are analyzed; thereafter, tourist travel routes are extracted, association rule mining is used, three indexes of support, confidence and lift are calculated to process the routes, and popular routes and new routes worth developing are obtained; finally, scenic spot development suggestions are provided to scenic spot managers, and tourist travel routes are recommended to tourists.

Description

Technical Field

[0001] This invention belongs to the field of tourism transportation planning and management, and particularly relates to a method for recommending tourist routes based on association rule mining. Background Technology

[0002] Against the backdrop of continuous socio-economic development, tourism, as a way for people to relax and enjoy leisure, can significantly improve their quality of life. However, the continuous development of the tourism industry has also given rise to issues related to scenic spot selection and route recommendations. On the one hand, how to categorize tourists and recommend popular scenic spots to tourists with different travel durations is crucial. On the other hand, tourists have limited travel time, especially international tourists, who may not be able to visit all the scenic spots in a city or region in a single trip, and must make choices. Therefore, how to plan tourist routes based on travel duration is also worthy of attention. At the same time, existing tourist routes and new tourist routes with potential demand also deserve discussion.

[0003] This invention attempts to process and analyze collected mobile phone signaling data to extract the main scenic spots visited by tourists with different travel durations, and to explore tourist behavior patterns. Based on these patterns, popular scenic spots are introduced to tourists with different needs, their travel routes are planned, the quality of travel and enjoyment is improved, and existing data is used to identify new tourist routes worth developing. Summary of the Invention

[0004] To address the aforementioned problems, the purpose of this invention is to provide a tourist route recommendation method based on association rule mining. This method can fully utilize mobile phone signaling data generated from tourists' past travel experiences to recommend travel routes to tourists with different needs according to their travel behavior patterns; at the same time, it can also discover new travel routes with potential demand.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: a tourist route recommendation method based on association rule mining, characterized by the following steps: (1) obtaining mobile phone signaling data of tourists from a tourism app and establishing a database. (2) organizing and analyzing the mobile phone signaling data, extracting relevant tourist information, and analyzing tourist characteristics. (3) performing cluster analysis on tourist behavior patterns based on the obtained tourist characteristics. (4) conducting specific analysis on tourists with different behavior patterns, and summarizing popular tourist routes and developing new routes through association rule mining.

[0006] Furthermore, in step (1), the method of obtaining mobile phone signaling data of tourists from the tourism APP and establishing a database includes the following steps: (1.1) Obtaining mobile phone signaling data of tourists from the tourism APP, wherein the signaling data includes user ID, time, latitude and longitude, nationality, mobile phone system version, etc. (1.2) Organizing the obtained mobile phone signaling data to form a tourism information database.

[0007] Furthermore, in step (2), the method for organizing and analyzing mobile phone signaling data, extracting tourist-related information, and analyzing tourist characteristics includes the following steps: (2.1) Identifying tourists in scenic areas using the coordinates and scenic area coordinates provided by the tourist mobile phone signaling data. (2.2) Classifying popular scenic areas based on the number of identified tourists in scenic areas. (2.3) Statistically analyzing tourists' travel time, nationality, and scenic areas visited to obtain tourist characteristic analysis.

[0008] Furthermore, in step (3), the method for clustering tourist behavior patterns based on the obtained tourist characteristics includes the following steps: (3.1) Using the data in the tourism information database established in step (1) as the basic dataset, clustering popular scenic spots, tourist areas, and tourist time as variables. The clustering method adopts the K-means model, and each initial cluster center is selected according to the probability distribution to avoid adverse results caused by random selection. (3.2) Using the silhouette coefficient method as the evaluation index, a suitable number of cluster groups is selected for clustering, and the clustering results are sorted according to the tourist area division. (3.3) The different tourist areas obtained by clustering are displayed on the map, and the distribution of different behavioral patterns and the duration of different tourist travel patterns are visualized and analyzed to obtain specific tourist behavior patterns.

[0009] Furthermore, in step (4), a specific analysis of tourists with different behavioral patterns is conducted, and a method for summarizing popular tourist routes and developing new routes is implemented through association rule mining. This method includes the following steps: (4.1) For a specific behavioral pattern, historical paths existing in the data are extracted. (4.2) According to association rule mining, support, confidence, and lift are calculated respectively, and a support threshold is selected to exclude paths that do not meet the requirements. (4.3) Routes with support higher than the threshold are marked and summarized as popular tourist routes. (4.4) Routes with confidence higher than the threshold are marked and summarized as routes that urgently need to be promoted. (4.5) Routes with lift higher than the threshold are marked and summarized as potential development routes.

[0010] Furthermore, in step (5), the method of providing scenic area development suggestions to scenic area managers and recommending tourist routes to tourists includes the following steps: (5.1) Sending the popular routes and two routes to be developed obtained in step (4) to the scenic area managers and pointing out the corresponding development suggestions. (5.2) Recommending the popular routes obtained in step (4) to tourists, wherein the route sorting should take into account the tourist's positioning and the tourist's expected number of days of visit.

[0011] Furthermore, this invention proposes a tourist route recommendation device based on association rule mining, characterized in that it includes:

[0012] Input module: Used to obtain tourist information from the tourist terminal. Tourist information should include tourist's geographical location, expected number of days of stay, nationality and other characteristic information.

[0013] Judgment module: used to match the tourist information obtained by the input module with the tourist behavior pattern obtained in step (3).

[0014] Calculation module: Used to calculate the recommended route and the corresponding recommendation index according to the method in step (4) by utilizing the tourist behavior pattern obtained in the judgment module.

[0015] Recommendation module: Used to select the N routes with the highest recommendation index as the final recommended routes.

[0016] Display module: Used to display recommended routes and related suggestions from the recommendation module on the tourist terminal interface. Attached Figure Description

[0017] Figure 1 This is a flowchart of the travel route recommendation method based on association rule mining according to the present invention.

[0018] Figure 2 This is a schematic diagram illustrating the database for establishing mobile phone signaling data according to the present invention.

[0019] Figure 3 This is a schematic diagram of the recommended tourist route of this invention.

[0020] Figure 4 This is a schematic diagram of the tourist route recommendation device recommended by the present invention. Detailed Implementation

[0021] To better explain and facilitate understanding of the present invention, the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0022] refer to Figure 1This application proposes a tourist route recommendation method based on association rule mining. By acquiring tourist mobile phone signaling data and performing a series of mining and calculations, it can recommend popular tourist routes and provide route development suggestions.

[0023] In step (1.2), the method for organizing the obtained mobile phone signaling data is as follows: First, remove information that is missing user name, nationality, coordinates, and time; second, mark the number of days of play as short, medium, or long (e.g., 1-3 days, 4-5 days, 6-7 days, 8-14 days), and remove data with excessively long play days (e.g., more than 14 days).

[0024] refer to Figure 2 , Figure 2 This serves as a reference template for establishing a tourism database after organizing the mobile phone signaling data in step (1.2).

[0025] In step (2.1), the method for identifying tourists in the scenic area is to combine the map data of the scenic area and, according to the three categories of large, medium and small scenic areas, expand the center coordinates of the scenic area into circles with radii of 1500 meters, 1000 meters and 500 meters respectively based on the coordinates. If the user's mobile phone signaling data coordinates fall within this circle, it is considered that the tourist has passed through the scenic area and is a tourist of the scenic area.

[0026] In step (2.2), the method for classifying popular scenic spots is to select scenic spots with a number of visitors greater than the threshold (e.g., 150 people) obtained from the method in step (2.1) as popular scenic spots in subsequent steps.

[0027] In step (2.3), the method for obtaining tourist characteristics is to summarize the data of each user and analyze the nationality distribution ratio, the tourism time distribution ratio, and the popularity of scenic spots to obtain tourist characteristics.

[0028] In step (3), the method for obtaining tourist behavior patterns using the K-means clustering model is as follows: First, the model is built using SPSS 22.0 software, and the obtained popular scenic spots, tourist areas, and tourist times are placed into the variable box. Second, in the analysis interface, an appropriate number of clusters (evaluated using the silhouette coefficient method as an evaluation index) and the number of iterations are selected, that is, to find the optimal solution to the greatest extent. Third, key data such as cluster members, distance from cluster centers, initial cluster centers, ANOVA table, and cluster information for each case are exported. Finally, the clustering results are sorted according to the tourist area division.

[0029] In step (3.2), the method for selecting a suitable number of groups using the contour coefficient method is as follows:

[0030] The silhouette coefficient is a method for evaluating the quality of clustering. It combines two factors: cohesion and separation, and can be used to evaluate the impact of different algorithms or different ways of running algorithms on the clustering results, based on the same original data.

[0031] For one of the points c:

[0032] Calculate a(c) = average(distance from vector c to all other points in the cluster it belongs to).

[0033] Calculate b(c) = min(the average distance from vector c to all points in a cluster that does not contain it).

[0034] Therefore, the contour coefficient of vector c is:

[0035] S = ba / max{a,b]

[0036] The profile coefficient is between [-1, 1]

[48] . The closer the absolute value of the profile coefficient is to 1, the better the cohesion and separation are.

[0037] In step (4.1), the method for extracting historical paths from the data is to compare the clustering results obtained in step (3.2) by sorting according to the division of tourist areas with the user's mobile phone signaling data. The itinerary that meets the time interval and tourist area requirements of the clustering results is the historical path corresponding to the behavior pattern.

[0038] In step (4.2), the method for calculating support, confidence, and lift according to association rule mining, selecting a support threshold, and excluding paths that do not meet the requirements is as follows:

[0039] Association rules are expressions that describe the events X→Y, where X and Y are called the antecedent and consequent. In the study of travel patterns, X is the previous stop in the travel sequence, and Y is the next travel destination.

[0040] As shown in formula (1), support represents the proportion of a given event among all events. In the tourist tour model, the proportion of the X→Y tourist route among all tourist routes can be obtained, that is, the popularity of the route.

[0041]

[0042] As shown in formula (2), the confidence level represents the number of events where X is the antecedent and the event X→Y is satisfied. In the tour mode, it describes the probability of going to scenic spot Y after going to scenic spot X, which can be used to explore whether the X→Y route is worth promoting.

[0043]

[0044] As shown in formula (3), the lift represents the ratio of the confidence level of the rule to the event with consequent Y. In the tour model, it is used to explore the degree of correlation of the X→Y tour route and whether it is worth developing.

[0045]

[0046] Due to the complexity of subsequent analysis and the fact that patterns with a certain frequency are identified as targets, paths with support below a suitable value (e.g., 0.005) are excluded.

[0047] In step (4.2), the method for summarizing popular tourist routes is to organize the route data, list the tourist routes with a support rate greater than a suitable threshold (e.g., 6%), and mark them as popular routes.

[0048] In step (4.3), the method for summarizing routes that urgently need to be promoted is to organize the route data, list the tourist routes with a confidence level greater than a suitable threshold (e.g., 25%), and mark them as routes that urgently need to be promoted.

[0049] In step (4.4), the method for summarizing potential development routes is to organize the route data, list the tourist routes with an elevation greater than a suitable threshold (e.g., 3), and mark them as potential development routes.

[0050] refer to Figure 3 , Figure 3 This is a schematic diagram of recommended routes after path data is mined according to association rules.

[0051] For step (5.1), the popular routes and two routes to be developed obtained in step (4) are sent to the scenic area manager, and corresponding development suggestions are provided.

[0052] For step (5.2), the popular routes obtained in step (4) are recommended to tourists, wherein the route sorting should take into account the tourist's location and the tourist's expected number of days of travel.

[0053] This application presents a tourist route recommendation device based on association rule mining. (See attached document.) Figure 4 ,include:

[0054] Input module: Used to obtain tourist information from the tourist terminal. Tourist information should include tourist's geographical location, expected number of days of stay, nationality and other characteristic information.

[0055] Judgment module: used to match the tourist information obtained by the input module with the tourist behavior pattern obtained in step (3).

[0056] Calculation module: Used to calculate the recommended route and the corresponding recommendation index according to the method in step (4) by utilizing the tourist behavior pattern obtained in the judgment module.

[0057] Recommendation module: Used to select the N routes with the highest recommendation index as the final recommended routes.

[0058] Display module: Used to display recommended routes and related suggestions from the recommendation module on the tourist terminal interface.

[0059] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A tourist travel route recommendation method based on association rule mining, characterized by, The method comprises the following steps: Step (1), obtaining mobile signaling data of tourists from a tourism APP and establishing a database; Step (2), sorting and analyzing the mobile signaling data in the database, extracting tourist-related information, and analyzing tourist characteristics; Step (3), based on the obtained tourist characteristics, performing cluster analysis on the behavior patterns of tourists; Step (4), specifically analyzing tourists with different behavior patterns, and inducing popular tourist routes and developing new routes through association rule mining; Step (5), providing scenic spot development suggestions to scenic spot managers and recommending tourist routes to tourists; In step (3), the cluster analysis method based on the obtained tourist characteristics includes the following steps: Step (3.1), taking the data in the tourism information database established in step (1) as the basic data set, taking popular scenic spots, tourist intervals and tourist times as variables for clustering, using the K-means model for clustering, and selecting each initial clustering center according to the probability distribution; Step (3.2), using the silhouette coefficient method as an evaluation index to select the number of clustering groups for clustering, and sorting the clustering results according to the tourist interval division; Step (3.3), displaying different tourist intervals obtained by clustering on a map, visualizing the distribution of different mode behaviors and the duration of different mode tours, and obtaining specific tourist behavior patterns; In step (4), the method for specifically analyzing tourists with different behavior patterns and inducing popular tourist routes and developing new routes through association rule mining includes the following steps: Step (4.1), for the behavior pattern, extracting the historical path existing in the data; Step (4.2), calculating the support, confidence and lift according to the association rule mining, selecting a support threshold, and excluding paths that do not meet the requirements; Step (4.3), marking routes with a support higher than the threshold as popular tourist routes; Step (4.4), marking routes with a confidence higher than the threshold as routes that need to be promoted; Step (4.5), marking routes with a lift higher than the threshold as potential development routes.

2. The tourist travel route recommendation method based on association rule mining according to claim 1, wherein, In step (1), the method for obtaining mobile signaling data of tourists from a tourism APP and establishing a database includes the following steps: Step (1.1), obtaining mobile signaling data of tourists from a tourism APP, wherein the signaling data includes user ID, time, latitude and longitude, nationality, and mobile system version; Step (1.2), sorting the obtained mobile signaling data to form a tourism information database.

3. The tourist travel route recommendation method based on association rule mining of claim 1, wherein, In step (2), the method for sorting and analyzing mobile signaling data, extracting tourist-related information, and analyzing tourist characteristics includes the following steps: Step (2.1), identifying tourists in a scenic spot by using the coordinates provided by the tourist mobile signaling data and the range of scenic spot coordinates; Step (2.2), dividing popular scenic spots based on the number of identified tourists in the scenic spots; Step (2.3), statistics of the tourist travel time, nationality and the scenic spot, to obtain the characteristics of the tourists.

4. The tourist travel route recommendation method based on association rule mining of claim 1, wherein, In the step (5), the method for providing the scenic spot developers with the scenic spot development suggestions and the tourists with the travel route recommendations comprises the following steps: Step (5.1), the popular routes and the two to be developed routes obtained in the step (4) are sent to the scenic spot managers, and the corresponding development suggestions are pointed out; Step (5.2), the popular routes obtained in the step (4) are recommended to the tourists, wherein the route sorting should consider the tourist positioning and the expected play days of the tourists.

5. A tourist travel route recommendation device based on association rule mining for implementing the method according to any one of claims 1 to 4, characterized in that, Comprise: The input module is used for obtaining the tourist information from the tourist terminal, and the tourist information should include the geographical position of the tourists, the expected play days, the nationality characteristic information; The judgment module is used for matching the tourist information obtained by the input module with the tourist behavior mode obtained in the step (3); The calculation module is used for calculating the to-be-recommended routes and the corresponding recommendation indexes according to the method of the step (4) by using the tourist behavior mode obtained in the judgment module; The recommendation module is used for taking the N routes with the highest recommendation indexes as the final recommended routes; The display module is used for displaying the recommended routes and the related suggestions in the recommendation module on the interface of the tourist terminal.

Images