Personalized scenic spot travel line recommendation method

Abstract

The invention relates to a personalized scenic spot travel line recommendation method, which comprises the following steps of: constructing a scenic spot database which comprises a scenic spot information classification and grading database and a scenic spot dimension crowd portrait adaptability classification database; updating the scenic spot database in real time according to the hotspot database; constructing a travel crowd portrait from three dimensions of time, space and behavior; the scenic spot attributes of the scenic spot database are deeply coupled with the travel crowd portraits, scenic spots conforming to coupling labels are extracted from the scenic spot dimension crowd portrait adaptability classification database, then the scenic spots are accessed to the scenic spot information classification and grading database, and according to the scenic spot grading result, the scenic spots with different classifications and high recommendation levels are selected from each coupling type of scenic spots; and providing a basic scenic spot list for personalized game line customization. By adopting a scenic spot information scientific classification and grading method and based on a scenic spot attribute and crowd portrait deep coupling technology, a database is updated in real time according to scenic spot popularity, the information cocoon house problem is avoided, and personalized tourism service is provided.

Description

Technical Field

[0001] The present invention relates to the technical field of tourism services, and in particular to a method for recommending personalized scenic spot itineraries. Background Art

[0002] Traditional cultural tourism suffers from fragmented information, homogenized content, and a lack of system integration, making it difficult to accurately match cultural tourism services to tourist needs. Furthermore, there's a lack of systematic categorization of tourist preferences, leading to insufficiently personalized experiences and irrational allocation of cultural tourism resources. With the continuous development of the tourism market and the increasing demands of travelers, providing users with personalized, diverse, and high-quality travel recommendation services has become a pressing issue.

[0003] The existing technology proposes a solution: based on the user information of the Mafengwo travel platform, a multi-dimensional user portrait model is constructed, and personalized recommendations are then made to users through recommendation technology. The main contents include: (1) constructing a multi-dimensional user portrait model, extracting specific scenario tags from the user's travel notes information, constructing the user's scenario dimension portrait, extracting the user's interests and theme keywords from the user's travel notes data, constructing the user's interest portrait, extracting emotional words from the user's comments on the scenic spots, scoring and judging their emotions, and constructing the user's emotional dimension portrait; (2) making personalized recommendations based on the user portrait model, and making association rule recommendations based on the portraits of the three dimensions (scenario dimension, interest dimension, and emotional dimension); (3) verifying the personalized recommendation effect based on the user portrait model.

[0004] It has the following technical problems:

[0005] This solution only implements personalized recommendations based on the integration of users' existing data. Its recommendation method is prone to information cocoons and cannot be updated as user preferences change. Summary of the Invention

[0006] In response to the problems existing in the prior art, the purpose of the present invention is to provide a personalized scenic spot tour route recommendation method that can avoid the emergence of information cocoons and can combine user-defined needs and recommendations to generate personalized travel routes that are both efficient and flexible.

[0007] In order to achieve the above object, the present invention adopts the following technical solutions:

[0008] A personalized scenic spot tour route recommendation method includes the following steps:

[0009] Construct a scenic spot database, including a scenic spot information classification and grading database and a scenic spot dimension crowd portrait adaptive classification database;

[0010] Update the scenic spot database in real time based on the hotspot database;

[0011] Construct a portrait of the cultural tourism population from three dimensions: time, space, and behavior;

[0012] The scenic spot attributes of the scenic spot database are deeply coupled with the portrait of the cultural and tourism population. The scenic spots that meet the coupling labels are extracted from the adaptive classification database of the scenic spot dimension population portrait, and then connected to the scenic spot information classification and grading database. Based on the scenic spot grading results, scenic spots with different classifications and high recommendation levels are selected from each coupled type of scenic spots to provide a basic scenic spot list for personalized tour line customization.

[0013] Furthermore, the scenic spot information classification and grading database includes cultural and tourism space resource data, cultural and tourism time resource data, cultural and tourism scenic spot popularity value data and cultural and tourism scenic spot attribute data, classifies scenic spot information according to different interest types, and grades scenic spot information according to cultural and tourism scenic spot popularity value.

[0014] Furthermore, the adaptive classification database of crowd portraits in the scenic spot dimension includes attribute data of cultural and tourism scenic spots and data on suitable crowds for cultural and tourism scenic spots. Based on crowd preferences, coupling labels of different tourist crowds are assigned to each scenic spot.

[0015] Furthermore, the attraction attributes in the attraction database are deeply coupled with the portrait of the cultural tourism population, including determining the basic social attributes of the tourism population based on the composition and age of the cultural tourism population, determining the travel rhythm preference based on the number of days of play and daily itinerary, determining the travel behavior preference based on the travel purpose and must-see attractions, and determining the travel space preference based on the travel scope;

[0016] Among the four dimensions of basic social attributes, travel rhythm preferences, travel behavior preferences and travel space preferences of the determined tourist population, only one dimension is set as a one-dimensional portrait, two dimensions are set as a two-dimensional portrait, three dimensions are set as a three-dimensional portrait, and four dimensions are set as a four-dimensional portrait, so as to identify the portrait of the cultural and tourism population.

[0017] Furthermore, providing a basic attraction list for personalized tour route customization includes first screening out some attractions based on the basic social attributes of the tourist population when making attraction recommendations, then further screening or supplementing attractions based on travel rhythm preferences, then further screening or supplementing attractions based on travel behavior preferences, and finally screening attractions based on travel space preferences.

[0018] Furthermore, we filter out scenic spots from the list of scenic spots corresponding to the one-dimensional portraits, and define the set of scenic spots corresponding to each dimension portrait as S d :

[0019] S d ={s d1 ,sd2 ,…,s dn}

[0020] Various types of attractions involved in the portraits in various dimensions di ∈S d In the , select 3-5 recommended attractions according to the recommendation level to obtain a list of recommended attractions:

[0021]

[0022] in:

[0023] Attraction type s under the corresponding dimension di A sublist of recommended attractions;

[0024] Attraction type s under the corresponding dimension di The corresponding candidate scenic spot set;

[0025] Select the top recommendation in descending order attractions,

[0026] L d : A list of recommended attractions in the corresponding dimension.

[0027] Furthermore, the multi-dimensional portrait lists (including two-dimensional portraits, three-dimensional portraits, and four-dimensional portraits) are merged, and the basic recommended scenic spot list is obtained after integrating all dimensions:

[0028]

[0029] L base A basic recommendation list.

[0030] Furthermore, providing a basic list of attractions for personalized tour line customization also includes dynamic adjustment of the itinerary. The total number of attractions in the itinerary is obtained by multiplying the travel pace and number of days selected by the tourists.

[0031] N total =D×R;

[0032] Where: D is the number of travel days (D≥1), R is the rhythm intensity;

[0033] The first optimization of the basic recommended attractions list is performed with 125% of the total number of attractions in the itinerary. The target total number of attractions in the itinerary is:

[0034] N target =1.25×N total

[0035] The first optimization method adopts the principle of deleting if there are too many and adding if there are too few. When the number of attractions in the basic recommended attractions list is less than 125% of the total number of attractions in the itinerary, the difference between the two is divided by the number of portrait dimensions to obtain the number of attractions that need to be added in each dimension. Then, according to this number, non-duplicate attractions are added to the attractions involved in the portraits of each dimension according to the recommendation level. When the number of attractions in the basic attractions list is greater than 125% of the total number of attractions in the itinerary, it is deleted according to this method to obtain the optimized attraction list for the corresponding group.

[0036] The difference is defined as: Δ=|L base |-N target

[0037] If Δ<0, additional attractions are required:

[0038] Then optimize the attractions list:

[0039]

[0040] If Δ>0, attractions need to be deleted:

[0041] Then optimize the attractions list:

[0042]

[0043] in:

[0044] TopRank(C d \L d ,Δ add ): From the set C d Take the first Δ add Highly popular non-repetitive attractions;

[0045] BottomRank(L d ,Δ remove ): From the set L d After taking the middle Δ remove low-population attractions;

[0046] Finally generate the optimized scenic spot list L opt1 , and satisfy:

[0047] |L opt1 |=N target .

[0048] Furthermore, after the first optimization, a second optimization is provided.

[0049] L opt2 =L opt1 ∪U

[0050] in:

[0051] U is a user-defined attraction collection.

[0052] Furthermore, based on the second optimized version list, according to the recommendation level of scenic spots, the first-level recommended scenic spots are classified as must-see scenic spots, and the second-level and below scenic spots are classified as recommended scenic spots, and finally a personalized scenic spot list that can be checked is formed.

[0053] L 必去 ={s∈L opt2 Level(s)=1

[0054] L 推荐 ={s∈L opt2 |Level(s)≥2}

[0055] Among them: Level(s) is the recommended level of the attraction;

[0056] The number of attractions checked by tourists should be within 25% of the total number of attractions in the itinerary.

[0057] N final =|L 必去 |+N 勾选

[0058] Need to meet: N 勾选 ∈[0.75×N total ,1.25×N total ]

[0059] Then, based on the tourists’ selections and the number of days and pace of their travel, the itinerary is arranged in different days.

[0060] In general, the present invention has the following advantages:

[0061] By employing a scientific classification and grading approach for attraction information, and leveraging deep coupling technology based on attraction attributes and demographic profiles, we provide a more integrated, systematic, personalized, and dynamic cultural tourism service solution. By updating the database in real time based on attraction popularity, we avoid the information cocooning problem that can exist in existing technologies, effectively integrating cultural tourism resources, and precisely matching them with visitor needs, providing personalized travel services. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] Figure 1 Schematic diagram of the scenic spot database construction method of the present invention.

[0063] Figure 2 This is an example of a hierarchical classification database for cultural and tourism attractions.

[0064] Figure 3 Technical model diagram for profiling the “4W” population.

[0065] Figure 4 Create a portrait identification process for the cultural and tourism groups.

[0066] Figure 5 This is the framework diagram of the personalized scenic spot tour route recommendation module.

[0067] Figure 6 Flowchart of the method for recommending personalized tourist attraction itineraries. DETAILED DESCRIPTION

[0068] The present invention will be described in further detail below.

[0069] like Figure 5 、 Figure 6 As shown, a personalized scenic spot tour route recommendation method is characterized by comprising the following steps:

[0070] Construct a scenic spot database, including a scenic spot information classification and grading database and a scenic spot dimension crowd portrait adaptive classification database;

[0071] Update the scenic spot database in real time based on the hotspot database;

[0072] Construct a portrait of the cultural tourism population from three dimensions: time, space, and behavior;

[0073] The scenic spot attributes of the scenic spot database are deeply coupled with the portrait of the cultural and tourism population. The scenic spots that meet the coupling labels are extracted from the adaptive classification database of the scenic spot dimension population portrait, and then connected to the scenic spot information classification and grading database. Based on the scenic spot grading results, scenic spots with different classifications and high recommendation levels are selected from each coupled type of scenic spots to provide a basic scenic spot list for personalized tour line customization.

[0074] The present invention realizes the systematic data classification and processing of urban cultural and tourism resources, creates a classification and grading database of scenic spot information and a classification database of crowd portrait adaptability, uses the classification and grading and crowd portrait adaptability classification methods to fully divide the attributes of scenic spots, and updates the hotspot database in real time through the popularity value of scenic spots. The tourist portrait results are integrated and matched according to the attributes of scenic spots. For group and content classification portraits, the portrait results are matched according to the attributes of scenic spots to form exclusive routes; for personalized tourist group portraits, according to the scenic spot grading results, the portrait types involved are considered separately, and the scenic spots with higher recommendation levels are extracted for comprehensive recommendation, so as to realize personalized travel route recommendations and accurately match tourist needs. The recommendation system of the present invention can clarify the attributes of scenic spots and their corresponding relationship with crowd portraits based on real data and crowd needs, recommend more extensive scenic spot information, and can also update and adapt to changes in user preferences in real time to provide more dynamic and personalized services.

[0075] The specific steps are as follows:

[0076] 1. Construction of scenic spot database: Build a spatiotemporal resource database of urban cultural and tourism resources, including a classification and grading database of scenic spot information and an adaptive classification database of crowd portraits in scenic spot dimensions.

[0077] (1) Scenic spot information classification and grading database: including cultural tourism space resource data, cultural tourism time resource data, cultural tourism scenic spot heat value data, cultural tourism scenic spot attribute data, etc., based on big data, the urban cultural tourism resource spatiotemporal resource database is scientifically graded and classified at the scenic spot level and portrait level, to obtain a more systematic and accurate scenic spot information classification and grading database and scenic spot dimension crowd portrait adaptive classification database, to improve the correlation between scenic spot data and crowd portrait, thereby improving the accuracy and applicability of the portrait. Figure 1 As shown, where:

[0078] ① Classification method - 11 interest tags: Based on the spatiotemporal resource data and attribute data of cultural and tourism attractions, cultural and tourism interest types are classified from the perspectives of history, culture, landscape, crowd activities, etc. Here, taking the cultural and tourism information of Guangzhou as an example, it is mainly divided into 11 types of attractions, including history, culture and art, nature and super landscapes, immersive experience scenes, performances, festivals, regional IP, leisure and entertainment, life experience, food and shopping, ritual atmosphere, and popular check-ins, to realize the classification of attraction information. The types can be locally adjusted and increased or decreased according to specific regions.

[0079] ② Grading method - 5-level popularity grading: According to the popularity value of cultural and tourism attractions, from 0 to 10 points, from high to low, with each 2 points as a level, determine the recommended level of the attractions and form grading data, such as Figure 2 shown.

[0080] (2) Database of adaptive classification of crowd portraits in the dimension of scenic spots: including attribute data of cultural and tourism scenic spots and data on suitable crowds for cultural and tourism scenic spots. Based on crowd preferences, each scenic spot is given a coupling label of different tourist crowds.

[0081] (3) Real-time update of popular scenic spot database and scenic spots: The popular scenic spot database is composed of cultural and tourism hotspots. The scenic spot database and the popular scenic spot database are updated in real time based on the popularity value of the scenic spots, realizing dynamic scenic spot database update.

[0082] 2.4W technology application: Combined with the "4W" model (who, when, what, where), such as Figure 3 As shown in the figure, a cultural tourism group portrait model is constructed from the three dimensions of time, space, and behavior, avoiding the limitations of single-level portraits. Through the progressively superimposed portrait model of the four Ws, different tourist groups are identified and targeted itinerary recommendations are provided to meet the needs of different cultural tourism preference groups.

[0083] The cultural tourism group survey questionnaire is designed based on the "4W" technology. The questionnaire obtains the overall characteristics of the cultural tourism group portrait based on the basic social attributes and tourism preference attributes of the tourism group, and then identifies the group. The basic social attributes are understood through "who" to understand the composition and age of the tourism group. The tourism preference attributes first understand the travel time arrangement and travel rhythm preferences through "when", then understand the travel hobbies and travel purposes through "what", and finally understand the regional preferences through "where". Through the collection of tourist questionnaire information, the tourism group is identified, such as Figure 4 As shown, the scenic spots are screened step by step, and ultimately personalized travel route recommendations are made to tourists.

[0084] (1) First, determine the basic social attributes of tourists:

[0085] The "who" factor is used to determine the nature of a travel group at the group level. The "who" factor identifies the nature of the group based on age and demographic composition. The group type is denoted by A, and the five categories are individuals (A1), families (A2), seniors (A3), couples (A4), and team building (A5).

[0086] (2) Then, determine the tourism preference attributes of the tourists, which are divided into time preference, behavior preference and space preference:

[0087] By dividing the time dimension by "when," the total number of recommended attractions and the travel rhythm are determined by the number of days of play and the daily itinerary. Time preference is marked with B, and the classification of time preference is divided into two categories: special forces type (B1) and roaming and lying type (B2);

[0088] The "what" dimension is used to divide travelers into different categories based on their travel preferences and goals, and their travel objectives and must-see attractions. Behavioral preferences are denoted by C, and are categorized into seven categories: historical, cultural, and artistic (C1), natural and super-landscape (C2), immersive experience (C3), performing arts (C4), urban IP (C5), leisure and entertainment (C6), and ritual atmosphere (C7).

[0089] The spatial dimension "where" is used to divide the area, and the location attribute is determined by the travel range. Spatial preference is marked with D, and the classification of spatial preference is divided into core urban areas (D1) and remote suburbs (D2).

[0090] (3) Finally, after determining the basic social attributes and tourism preference attributes of the tourist population, the population portrait can be identified. For those who only select one of the four dimensions ABCD, it is a one-dimensional portrait; for those who include two of the four dimensions ABCD, it is a two-dimensional portrait; for those who include three of the four dimensions ABCD, it is a three-dimensional portrait; and for those who include all four dimensions, it is a four-dimensional portrait, such as A3B2C7D1.

[0091] When recommending scenic spots, we first select some scenic spots based on the basic social attributes of the crowd, group characteristics, and big data. For example, for A3 elderly tourism, we recommend healing scenery, Lotus Mountain, Lantern Festival, and other related attractions. Then, based on the relevant tourism preference attributes, we further select or supplement scenic spots based on time preference. For example, for B2 roaming and lying style, we select one-stop eating, drinking, and entertainment, rural homestays, and other related attractions. Then, based on behavioral preferences, we further select or supplement scenic spots. For example, for C7, we select scenic spots related to Zen. Finally, we select scenic spots based on spatial preferences. For example, for D1, we select scenic spots in the core urban area among the recommended scenic spots.

[0092] 3. Deep coupling of scenic spot attributes and crowd portraits: Integrate and match tourist portrait results based on scenic spot attributes.

[0093] Based on the one-dimensional portrait results, all attractions related to the coupling tags of this dimension are extracted from the crowd portrait adaptive classification database. Then, the database of attraction information classification and grading is accessed. Based on the attraction classification, 3-5 recommended attractions are selected in each category according to the recommendation level to provide a basic attraction list for personalized tour line customization.

[0094] For the results of two-dimensional portraits, three-dimensional portraits, and four-dimensional portraits, the portraits of each dimension are matched according to the attributes of the scenic spot, and the scenic spots that meet all the coupling labels are extracted from the crowd portrait adaptive classification database. Then, they are connected to the scenic spot information classification and grading database. According to the scenic spot grading results, scenic spots with different classifications and high recommendation levels are selected from each coupling type of scenic spots to provide a basic scenic spot list for personalized tour line customization.

[0095] 4. Personalized travel service recommendations: Establish a phased method for dynamically optimizing the list of scenic spots. Starting from the basic recommendation list, through itinerary adjustments, personalized additions, constraint selection, and daily planning, combined with user-defined needs and recommendations, a personalized travel route that takes into account both efficiency and flexibility is generated.

[0096] Phase 1: Basic recommendation list generation

[0097] For one-dimensional portraits, in addition to recommending attractions under the dimension selected by tourists, in order to avoid the limitations of recommendations, on the one hand, we will supplement selected routes for tourists, and on the other hand, we will focus on listing attractions with higher recommendation levels under other unselected dimensions for tourists to make secondary choices and improve personalized customized tour routes.

[0098] For multi-dimensional portraits (including two-dimensional portraits, three-dimensional portraits, and four-dimensional portraits), based on the deep coupling results of scenic spot attributes and crowd portraits in each dimension, for each dimension d∈{Who, When, What, Where}, the set of various types of scenic spots corresponding to each dimensional portrait is defined as S d :

[0099] S d ={s d1 ,s d2 ,…,s dn}

[0100] That is: when d=Who, S d ={Individual, Family, Elderly, Couple, Team Building}

[0101] (1) Filtering the one-dimensional image attraction list

[0102] Various types of attractions involved in one-dimensional portraits di ∈S d In the , select 3-5 recommended attractions according to the recommendation level to obtain the recommended attraction list under each dimension:

[0103]

[0104] in:

[0105] Attraction type s under the corresponding dimension di Sublist of recommended attractions

[0106] Attraction type s under the corresponding dimension di Corresponding candidate attraction set

[0107] Select the top recommendation in descending order attractions,

[0108] L d : Re-recommend the list of scenic spots in the corresponding dimension;

[0109] (2) Merge the list of multi-dimensional portraits (including two-dimensional portraits, three-dimensional portraits, and four-dimensional portraits) of scenic spots,

[0110] After integrating all dimensions, we get a basic list of recommended attractions:

[0111]

[0112] L base :Basic recommendation list (initial unoptimized)

[0113] Phase 2: Dynamic adjustment of travel (first optimization)

[0114] (1) Calculation of the total number of attractions

[0115] According to the travel rhythm and number of days selected by tourists, the total number of attractions in the itinerary is obtained by multiplying the two.

[0116] N total =D×R

[0117] in:

[0118] D: Travel days (D≥1)

[0119] R: Rhythm intensity (e.g., fast rhythm R=4, medium rhythm R=3, slow rhythm R=2)

[0120] (2) Target quantity threshold

[0121] The first optimization of the basic recommended attractions list is performed with 125% of the total number of attractions in the itinerary. The target total number of attractions in the itinerary is:

[0122] N target =1.25×N total

[0123] (3) List adjustment rules

[0124] The first optimization method follows the principle of removing more attractions and adding more attractions. When the number of attractions in the basic recommended attraction list is less than 125% of the total number of attractions in the itinerary, the difference between the two is divided by the number of portrait dimensions to determine the number of attractions to be added in each dimension. Based on this number, non-duplicate attractions are then added to the attractions in each dimension according to their recommendation level. Similarly, when the number of attractions in the basic attraction list is greater than 125% of the total number of attractions in the itinerary, the same method is used to remove them. This results in an optimized attraction list for the corresponding group.

[0125] The difference is defined as: Δ=|L base |-N target

[0126] If Δ<0 (additional attractions are required): (Average across four dimensions, rounded up)

[0127] Then optimize the attractions list:

[0128]

[0129] If Δ>0 (attractions need to be deleted): (Average across four dimensions, rounded down)

[0130] Then optimize the attractions list:

[0131]

[0132] in:

[0133] TopRank(C d \L d ,Δ add ): From the set C d Take the first Δ add Highly popular non-repetitive attractions

[0134] BottomRank(L d ,Δ remove ): From the set L d After taking the middle Δ remove Low-popularity attractions

[0135] Finally generate the optimized scenic spot list L opt1 , and satisfy:

[0136] |L opt1 |=N target

[0137] Phase 3: Personalized Addition (Second Optimization)

[0138] After that, the second round of list optimization is conducted. This optimization focuses on personalized additions by tourists. This involves adding attractions that tourists want to visit but are not included in the recommended itineraries due to low recommendation levels. Tourists can add attractions to their itineraries on their own. This completes the second round of list optimization.

[0139] L opt2 =L opt1 ∪U

[0140] in:

[0141] U: Add a collection of attractions for users to customize.

[0142] Stage 4: Final checklist generation

[0143] (1) Must-see and recommended categories

[0144] According to the second optimized version list, according to the recommendation level of attractions, the first-level recommended attractions are classified as must-see attractions, and the second-level and below attractions are classified as recommended attractions, and finally a personalized attraction list that can be checked is formed.

[0145] L 必去 ={s∈L opt2 Level(s)=1

[0146] L 推荐 ={s∈L opt2 |Level(s)≥2}

[0147] Among them: Level(s) is the recommended level of the attraction.

[0148] (2) Check the quantity constraint

[0149] Tourists will be reminded here that the number of attractions they check should be within 25% of the total number of attractions in the itinerary.

[0150] N final =|L 必去 |+N 勾选

[0151] Need to meet: N 勾选 ∈[0.75×N total ,1.25×N tota1 ]

[0152] Phase 5: Arrangement by days

[0153] Then, based on the results of the tourists' selections and combined with the number of days and pace of the tourists' travel, a multi-day itinerary is formed, including three recommended tour routes and an integrated arrangement of food, accommodation and transportation. The arrangement of attractions on the same day follows the principle of proximity, and the integrated arrangement of food, accommodation and transportation includes surrounding hotels and restaurants. Hotels are recommended based on star rating, restaurants based on ratings and prices, and reverse travel recommendations.

[0154] This invention utilizes a scientific classification and grading method for scenic spot information, a "4W" crowd profiling technique based on spatiotemporal causality, and a deep coupling of scenic spot attributes and crowd profiling. These techniques enable the construction of a more systematic scenic spot database, the development of a more personalized crowd profiling technique, and the deep coupling of scenic spot attributes and crowd profiling. Compared to existing technologies, this invention provides a more integrated, systematic, personalized, and dynamic cultural tourism service solution.

[0155] First, the present invention constructs a more systematic and multidimensional crowd profiling technology model. This model utilizes "4W" technology, based on spatiotemporal causal textures, to deeply analyze user preferences from four perspectives: who (travel group type), when (travel pace), what (content), and where (space). These preferences are then layered and layered to derive cultural tourism crowd profile types. Second, the present invention provides a more integrated method for scientifically classifying and grading scenic spot databases. First, through classification and grading, and adaptive crowd profile classification methods, a database for the classification and grading of scenic spot information and an adaptive crowd profile classification database are constructed. Second, the database is updated in real time based on scenic spot popularity, dynamically matching it with user preferences. This avoids the information cocoon problem that can exist in existing technologies and effectively integrates cultural tourism resources. Finally, the present invention achieves a deep coupling between scenic spot attributes and crowd profiles, enhancing visitor experience and engagement. Building on the "4W" crowd profiling technology and scenic spot information database, the model systematically analyzes the types and characteristics of cultural tourism group profiles and matches them to scenic spot attributes, achieving a precise match between cultural tourism resources and visitor needs, providing personalized tourism services.

[0156] The above embodiments are preferred implementation modes of the present invention, but the implementation modes of the present invention are not limited to the above embodiments. Any other changes, modifications, substitutions, combinations, and simplifications that do not deviate from the spirit and principles of the present invention should be considered as equivalent replacement methods and are included in the scope of protection of the present invention.

Claims

1. A personalized scenic spot itinerary recommendation method, characterized by: The following steps are included: Construct a scenic spot database, including a scenic spot information classification and grading database and a scenic spot dimension crowd portrait adaptive classification database; Update the scenic spot database in real time based on the hotspot database; Construct a portrait of the cultural tourism population from three dimensions: time, space, and behavior; The scenic spot attributes of the scenic spot database are deeply coupled with the portrait of the cultural and tourism population. The scenic spots that meet the coupling labels are extracted from the adaptive classification database of the scenic spot dimension population portrait, and then connected to the scenic spot information classification and grading database. Based on the scenic spot grading results, scenic spots with different classifications and high recommendation levels are selected from each coupled type of scenic spots to provide a basic scenic spot list for personalized tour line customization.

2. The personalized scenic spot tour route recommendation method according to claim 1, characterized in that: The scenic spot information classification and grading database includes cultural and tourism space resource data, cultural and tourism time resource data, cultural and tourism scenic spot popularity value data and cultural and tourism scenic spot attribute data. It classifies scenic spot information according to different interest types and grades scenic spot information according to the popularity value of cultural and tourism scenic spots.

3. The personalized scenic spot tour route recommendation method according to claim 1, characterized in that: The adaptive classification database of crowd portraits in the scenic spot dimension includes attribute data of cultural and tourism scenic spots and data on suitable crowds for cultural and tourism scenic spots. Based on the preferences of the crowd, each scenic spot is given a coupling label of different tourist groups.

4. The personalized scenic spot tour route recommendation method according to claim 1, characterized in that: Deeply coupling the scenic spot attributes of the scenic spot database with the portrait of the cultural tourism population includes determining the basic social attributes of the tourism population based on their composition and age, determining the travel rhythm preference based on the number of days of play and daily itinerary, determining the travel behavior preference based on the travel purpose and must-see attractions, and determining the travel space preference based on the travel scope; Among the four dimensions of basic social attributes, travel rhythm preferences, travel behavior preferences and travel space preferences of the determined tourist population, only one dimension is set as a one-dimensional portrait, two dimensions are set as a two-dimensional portrait, three dimensions are set as a three-dimensional portrait, and four dimensions are set as a four-dimensional portrait, so as to identify the portrait of the cultural and tourism population.

5. The personalized scenic spot tour route recommendation method according to claim 4, characterized in that: Providing a basic list of attractions for personalized tour route customization includes first screening out some attractions based on the basic social attributes of the tourist population when making attraction recommendations, then further screening or supplementing attractions based on travel rhythm preferences, then further screening or supplementing attractions based on travel behavior preferences, and finally screening attractions based on travel space preferences.

6. The personalized scenic spot tour route recommendation method according to claim 5, characterized in that: Filter out scenic spots from the list of scenic spots corresponding to the one-dimensional portrait, and define the set of various types of scenic spots corresponding to each dimensional portrait as S d : S d ={s d1 ,s d2 ,…,s dn } Various types of attractions involved in the portraits in various dimensions di ∈S d In the , select 3-5 recommended attractions according to the recommendation level to obtain a list of recommended attractions: in: Attraction type s under the corresponding dimension di A sublist of recommended attractions; Attraction type s under the corresponding dimension di The corresponding candidate scenic spot set; Select the top recommendation in descending order attractions, L d : A list of recommended attractions in the corresponding dimension.

7. The personalized scenic spot tour route recommendation method according to claim 6, characterized in that: The two-dimensional, three-dimensional, and four-dimensional portrait lists are merged, and after integrating all dimensions, a basic list of recommended attractions is obtained: L base A basic recommendation list.

8. The personalized scenic spot tour route recommendation method according to claim 7, characterized in that: Providing a basic list of attractions for personalized tour line customization also includes dynamic adjustment of itinerary, According to the travel rhythm and number of days selected by tourists, the total number of attractions in the itinerary is obtained by multiplying the two. N total =D×R; Where: D is the number of travel days (D≥1), R is the rhythm intensity; The first optimization of the basic recommended attractions list is performed with 125% of the total number of attractions in the itinerary. The total number of attractions in the target itinerary is: N target =1.25×N total ; The first optimization method adopts the principle of deleting if there are too many and adding if there are too few. When the number of attractions in the basic recommended attractions list is less than 125% of the total number of attractions in the itinerary, the difference between the two is divided by the number of portrait dimensions to obtain the number of attractions that need to be added in each dimension. Then, according to this number, non-duplicate attractions are added to the attractions involved in the portraits of each dimension according to the recommendation level. When the number of attractions in the basic attractions list is greater than 125% of the total number of attractions in the itinerary, it is deleted according to this method to obtain the optimized attraction list for the corresponding group. The difference is defined as: Δ=|L base |-N target If Δ<0, additional attractions are required: Then optimize the attractions list: If Δ>0, attractions need to be deleted: Then optimize the attractions list: in: TopRank(C d \L d ,Δ add ): From the set C d Take the first Δ add Highly popular non-repetitive attractions; BottomRank(L d ,Δ remove ): From the set L d After taking the middle Δ remove low-population attractions; Finally generate the optimized scenic spot list L opt1 , and satisfy: |L opt1 |=N target 。 9. The personalized scenic spot tour route recommendation method according to claim 8, characterized in that: After the first optimization, there is a second optimization. L opt2 =L opt1 ∪U Among them: U is a collection of scenic spots added by users.

10. The personalized scenic spot tour route recommendation method according to claim 9, characterized in that: According to the second optimized version list, according to the recommended level of attractions, the first-level recommended attractions are classified as must-see attractions, and the second-level and below attractions are classified as recommended attractions, and finally a personalized attraction list that can be checked is formed. L 必去 ={s∈L opt2 |Level(s)=1] L 推荐 ={s∈L opt2 |Level(s)≥2} Among them: Level(s) is the recommended level of the attraction; The number of attractions checked by tourists should be within 25% of the total number of attractions in the itinerary. N final =|L 必去 |+N 勾选 Need to meet: N 勾选 ∈[0.75×N total ,1.25×N total ] Then, based on the tourists’ selections and the number of days and pace of their travel, the itinerary is arranged in different days.

Images