Regional document travel intelligent recommendation method, system and device and storage medium

Abstract

The invention discloses a regional document travel intelligent recommendation method, which comprises the following steps of: S1, collecting a large amount of data, and preprocessing the data; s2, constructing a user portrait by using a machine learning algorithm based on the behavior data of the user; s3, performing feature extraction on the data, processing text description related to the cultural resources by using a text mining technology, and extracting specified information; s4, designing a recommendation model, matching the user portrait with the text travel resource features, and calculating the preference degree of the user to the specific text travel resource; s5, generating a personalized regional travel recommendation result for the user by using the recommendation model, and selecting the most suitable recommendation result from the database of the travel resources according to the portrait and the historical behavior of the user; and S6, updating the user portrait and the recommendation model in real time, and providing a dynamic personalized recommendation result. According to the invention, the prediction score of the target user can be improved, and more accurate tourism place recommendation can be carried out on the target user.

Description

Technical Field

[0001] The present invention relates to the field of intelligent cultural and tourism recommendations, and in particular to a method, system, device and storage medium for intelligent cultural and tourism recommendations for a region. Background Art

[0002] With the development of tourism, tourists' demand for personalized and accurate recommendations of tourist attractions is increasing. However, traditional recommendation methods often cannot meet these needs. Online travel platforms, tourist consumption, and the contentization of user habits are in a fierce development trend.

[0003] Tourists planning a trip often use user-generated information on online travel websites, such as ratings, online reviews, pictures, etc. as an important reference for their decision-making. However, for tourists, screening through massive amounts of information is a daunting challenge. Therefore, scholars have proposed tourist attraction recommendation methods to solve the "information overload" problem. With the help of machine learning, deep learning, data mining and other technologies, the characteristics of users and attractions are mined, and the recommendation quality of attraction recommendation methods is continuously improved.

[0004] The purpose of scenic spot recommendation is to solve the "information overload" problem of tourists and provide them with scenic spot recommendations that meet their preferences. Scenic spot recommendations based on ratings and online reviews are two widely used approaches in recommendation algorithms, but there are still the following problems:

[0005] (1) Lack of comprehensive recommendations: Currently, most cultural and tourism recommendation systems mainly make recommendations based on user behavior data or content features. They lack comprehensive analysis of users’ multi-dimensional information and cannot fully consider users’ different preferences and needs.

[0006] (2) Differences among user groups: Different user groups have different preferences and values ​​for cultural and tourism resources. Existing recommendation algorithms are difficult to fully consider the differences between different user groups.

[0007] To this end, this application specifically proposes a regional cultural and tourism intelligent recommendation method and system to solve the above-mentioned technical problems. Summary of the invention

[0008] The main purpose of the present invention is to provide a regional cultural tourism intelligent recommendation method and system, by building a big data platform, through the analysis and processing of massive data collection, analysis of basic static attribute characteristics and other multi-dimensional dynamic attribute characteristics of tourist user information, according to the attribute characteristics to establish a multi-level and multi-dimensional user portrait model, improve the problems existing in the traditional algorithm, group users together, improve the predicted score of the target user, at the same time according to the tourist destination information in different time periods, the tourist destination feature extraction, by matching the user portrait with the characteristic information of the tourist destination, to achieve a more accurate recommendation of the tourist destination for the target user, thereby improving the tourist experience, so as to solve the technical problems raised in the background technology.

[0009] The present invention adopts the following technical solutions to solve the above technical problems:

[0010] A regional cultural tourism intelligent recommendation method includes the following specific execution steps:

[0011] S1. Data collection and preprocessing: Collect a large amount of cultural and tourism related data through public channels and preprocess the data;

[0012] S2. User portrait construction: Based on user behavior data, use machine learning algorithms to build user portraits;

[0013] S3. Cultural and tourism resource feature extraction: feature extraction of regional cultural and tourism resources, and use text mining technology to process the text description of cultural resources and extract specified information;

[0014] S4. Recommendation model design: Design a recommendation model to match user profiles with cultural and tourism resource characteristics and calculate the user's preference for specific cultural and tourism resources;

[0015] S5. Recommendation result generation: Generate personalized regional cultural and tourism recommendation results for users using the recommendation model, and select the most suitable recommendation results from the database of cultural and tourism resources based on the user's profile and historical behavior;

[0016] S6. Real-time recommendation and feedback: Based on real-time user behavior data and feedback information, the user portrait and recommendation model are updated in real time, dynamic personalized recommendation results are provided, and real-time recommendation is realized by using real-time data processing technology.

[0017] Preferably, the specific operation steps of data collection and preprocessing in step S1 include:

[0018] S11. Data information collection: Collect the required data from different data sources as the initial data for characterizing user portraits;

[0019] S12. Data cleaning and integration: Screen, clean, process and integrate the data from multiple data sources, denoise the data, unify the data from multiple data sources and in multiple formats according to the distribution of data space, integrate them into the big data platform, and realize the sharing of data resources;

[0020] S13. Data selection and transformation: Extract the integrated data and then convert it into the data required by the system through format.

[0021] Preferably, the specific operation steps of constructing the user portrait in step S2 include:

[0022] S21. Characterize the user portrait label level through the collected user data:

[0023] S22. Implement user portrait model clustering based on the data in the user portrait tag level, wherein the algorithm steps of the model clustering include:

[0024] The preset user portrait set is U = {u1, u1, u1, ..., u m}, there are n attributes in each portrait, and the corresponding attribute matrix is:

[0025]

[0026] Each attribute is affected by sample data v. Suppose the set of sample data affected by an attribute has h sample data, sample data v i With v j The Euclidean distance between ij Indicates that n ij =||u i -u j ||, where 1≤i, j≤h, and is the density range of sample data, then user profile u i The density function centered at can be expressed as:

[0027]

[0028] Among them, h i Must meet Inner i The number of sample data around, sample data v i When there are more points around, it is similar to the sample data v i The closer the relationship is, the i The larger the value of , the greater the degree of membership to the portrait;

[0029] Through y i For analysis, introduce the density function value ρi :

[0030]

[0031] ρ i Indicates v i The surrounding sample data pair v i The influence of classification, when ρ i The larger the value, the i The greater the degree of belonging to a certain category, the greater the sample influence value ρ i The traditional algorithms introduced are:

[0032]

[0033] U represents the user portrait, C represents the cluster center, and v ij Expressed as a fuzzy classification matrix, c i Expressed as the cluster center matrix, u i It is represented as user portrait matrix, and h is the number of samples;

[0034] For the fuzzy classification matrix v ij and the cluster center matrix c i The derivatives are:

[0035]

[0036] where ρ i The size of is mainly determined by the density of the distribution around its sample space. i After that, the sample cluster center is adjusted, and the membership degree in the sample data is also adjusted because of the changes in the cluster center during the iteration process.

[0037] Preferably, the user portrait tag level in step S21 includes:

[0038] Raw data layer: used to integrate, process, and process multiple data sources into the raw data collected on the big data platform for data mining, in order to obtain a large amount of user static attribute information and dynamic attribute information;

[0039] Fact layer: contains descriptions of user-specific information, used to verify the original data layer and transmit accurate data to the feature preference layer for predictive modeling;

[0040] Feature preference layer: used to transmit data based on the fact layer, extract feature preferences by establishing an artificial intelligence algorithm model, obtain user behavior travel data information, and establish different label systems for users based on the model, analyze and process the data, and discover the user's potential knowledge;

[0041] Model prediction layer: Label user features and user behaviors by semanticizing the feature preference layer and using short text methods, and then establish a model prediction layer. The customer habit preferences of group users are predicted based on user portrait labels to explore users' potential needs.

[0042] Preferably, the specific operation steps of extracting the characteristics of Chinese tourism resources in step S3 include:

[0043] S31. Introduce time series as the change value of feature extraction. For a given time series z(i), i = 1, 2, 3, ..., N, let the impact of time series on feature extraction y z(i) It consists of two parts: linear features and nonlinear features. Then:

[0044] y z(i) = l t +k t

[0045] l t is a linear feature, k t It is a nonlinear feature;

[0046] S32. Calculate the linear features in the neural network model to obtain f(l t ), and the calculated residual part is used as the nonlinear feature to obtain:

[0047] e t =y z(i) -f(l t )-k t

[0048] where e t is the residual part;

[0049] S33. In the neural network model, the prediction results for linear features and nonlinear features are set as and Then the impact of changes in time series on feature extraction is:

[0050]

[0051] Assume that the feature extraction value is H, and its change in time series is:

[0052] Preferably, the specific operation steps of the recommended model design in step S4 include:

[0053] S41. Establishing a user-tourism destination rating matrix:

[0054] There are a*b data in total, where a represents the tourist destination, b represents the user, and D sbis the rating of the ath user for the bth tourist destination. The larger the value, the higher the user's preference for the tourist destination. Conversely, the lower the data, the lower the preference. Transformed into a mathematical model, the matrix rating is expressed as:

[0055]

[0056] S42. Calculate the similarity between the tourist destination and other tourist destinations based on the attribute characteristics of the tourist destination itself, obtain tourist destinations similar to the tourist destination in the entire big data platform based on the similarity, and make personalized recommendations.

[0057] Preferably, the specific operation steps of the recommendation model design in step S4 include: the personalized recommendation in step S42 can be made by predicting and scoring the tourist destinations, and the tourist destination with the highest predicted score is recommended, and the prediction score expression formula is:

[0058] Assume that target user C in tourist destination A predicts the rating of another tourist destination E:

[0059]

[0060] Among them, sim(A, E) is the similarity between tourist destination A and tourist destination E, D CA is the rating of user C for tourist destination A.

[0061] Preferably, a regional cultural tourism intelligent recommendation system is used to execute any of the above-mentioned regional cultural tourism intelligent recommendation methods, comprising the following modules:

[0062] Data collection and preprocessing module: used to collect a large amount of data including regional culture, tourism resources, and user evaluations, and to perform preprocessing work such as data cleaning, deduplication, and normalization;

[0063] User portrait building module: Based on the user's historical clicks, comments, ratings and other behavioral data, a user portrait is built using machine learning algorithms, including the user's interest preferences, geographic location information, and cultural background;

[0064] Cultural and tourism resource feature extraction module: extracts features of regional cultural and tourism resources, including geographic location, historical ratings, and cultural feature information, and uses text mining technology to process text descriptions of cultural resources to extract keywords and theme information;

[0065] Recommendation model design module: Design a recommendation model to calculate the user's preference for specific cultural and tourism resources by matching user portraits with cultural and tourism resource characteristics;

[0066] Recommendation result generation module: Use the recommendation model to generate personalized regional cultural and tourism recommendation results for users, and select the most suitable recommendation results from the database of cultural and tourism resources based on the user's profile and historical behavior;

[0067] Real-time recommendation and feedback module: Based on real-time user behavior data and feedback information, it updates user portraits and recommendation models in real time, provides dynamic personalized recommendation results, and uses real-time data processing technology to achieve real-time recommendations.

[0068] On the other hand, the present invention further discloses a computer-readable storage medium storing a computer program, wherein when the computer program is executed by a processor, the processor executes the steps of the above method.

[0069] On the other hand, the present invention further discloses a computer device, including a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the above method.

[0070] It can be seen from the above technical solution that the present invention provides a method for intelligently recommending regional cultural tourism. Compared with the prior art, the present invention has the following advantages:

[0071] 1. The method of the present invention can capture and adapt to the dynamic changes of user behaviors and preferences, rather than relying solely on static historical data, and further build a user portrait for continuous updating. Therefore, it can dynamically adjust the recommended content in real time according to the user's new behavior data, give priority to pushing attractions that match the user's current interests, and improve the real-time and effectiveness of the recommendation.

[0072] 2. Through time series analysis, the present invention can identify the characteristics of tourist destinations in different time periods, so that the recommendation system can provide more accurate personalized recommendations based on the needs of users at different time points. At the same time, after introducing the time series, the features can be decomposed into linear and nonlinear parts. This distinction makes feature extraction more detailed, and can better understand and predict the attractiveness of tourist destinations, and optimize recommendation strategies for this purpose, thereby improving user experience.

[0073] 3. By integrating multiple data sources such as social media, user feedback, and cultural and tourism resource information, the present invention can understand users and resources from multiple dimensions, providing a more comprehensive and in-depth perspective for the recommendation algorithm, and enhancing the insight and prediction accuracy of the recommendation system.

[0074] It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present invention, nor is it intended to limit the scope of the present invention. Other features of the present invention will become easy to understand through the following description. Of course, it is not necessary to achieve all of the advantages described above simultaneously for any product implementing the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0075] The drawings constituting a part of the present application are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0076] Figure 1 It is a schematic diagram of the overall process of the method of the present invention;

[0077] Figure 2 A schematic diagram of the user portrait label hierarchy of the present invention. DETAILED DESCRIPTION

[0078] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments. In the absence of conflict, the embodiments in this application and the features in the embodiments can be combined with each other. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0079] In the first embodiment, see Figure 1 to Figure 2 .

[0080] Traditional recommendation systems usually make recommendations based on simple user behavior data, which may ignore the personalized needs and diverse interests of users; while user portraits can comprehensively and systematically characterize the various characteristics of users, including age, gender, income level, interests and hobbies, historical behavior, etc.

[0081] Therefore, Figure 1 As shown. A regional cultural tourism intelligent recommendation method proposed in an embodiment of the present invention is based on a regional cultural tourism intelligent recommendation system, by building a big data platform, through the analysis and processing of massive data collection, the basic static attribute characteristics and other multi-dimensional dynamic attribute characteristics of tourist user information are analyzed, and a multi-level and multi-dimensional user portrait model is established according to the attribute characteristics, the problems existing in the traditional algorithm are improved, the users are grouped and clustered, and the predicted scores of the target users are improved. At the same time, according to the tourist destination information in different time periods, the tourist destination is feature extracted, and the user portrait is matched with the feature information of the tourist destination to achieve more accurate tourist destination recommendation for the target user, thereby improving the tourist experience.

[0082] This regional cultural tourism intelligent recommendation method performs corresponding operations in the middle of the system. Based on the user portrait, the system can capture the dynamic changes of user behavior and preferences instead of relying solely on static historical data. This is in sharp contrast to the traditional static recommendation system, which may not be able to adapt to changes in user preferences.

[0083] The method includes the following specific execution steps:

[0084] S1. Data collection and preprocessing: First, a large amount of data including regional culture, tourism resources, user reviews, etc. needs to be collected, and preprocessed by cleaning, deduplication, and normalization. These data can come from public data sets, social media platforms, tourism websites and other channels. The specific steps include:

[0085] S11. Data information collection: Collect the required data from different data sources as the initial data for portraying user portraits. The data source can come from the user's own data and external data, and make targeted selections based on the required data during collection, such as: users have selected their own interest preferences in their accounts, or users have liked or browsed some travel recommendation scenes on social media for a long time;

[0086] S12. Data cleaning and integration: The collected data sources are screened, cleaned, processed and integrated, and the data are denoised. According to the distribution of data space, the data from multiple data sources and in multiple formats are unified and integrated into the big data platform to realize the sharing of data resources.

[0087] S13. Data selection and transformation: extract the integrated data and then convert it into the data required by the system through the format;

[0088] In summary, by integrating multiple data sources such as social media, user feedback, and cultural and tourism resource information, the system can understand users and resources from multiple dimensions. This multi-dimensional data fusion provides a more comprehensive and in-depth perspective for the recommendation algorithm, enhancing the insight and prediction accuracy of the recommendation system.

[0089] S2. User portrait construction: Based on the user's historical click, comment, rating and other behavioral data, a user portrait is constructed, including the user's interest preferences, geographic location information, cultural background, etc. Machine learning algorithms such as cluster analysis and classification algorithms can be used to achieve the construction of user portraits. The specific steps include:

[0090] S21. Describe the user portrait label level through the collected user data. The user portrait label level is as follows: Figure 2 As shown, including:

[0091] Raw data layer: used to integrate, process, and process multiple data sources into the raw data collected on the big data platform for data mining, in order to obtain a large amount of user static attribute information and dynamic attribute information;

[0092] Fact layer: contains descriptions of user-specific information, such as user demographic attributes (age, identity, etc.), hobby characteristics, and other specific information. It is used to verify the original data layer and transmit accurate data to the feature preference layer for predictive modeling.

[0093] Feature preference layer: used to transmit data according to the fact layer, establish an artificial intelligence algorithm model through data mining, machine learning and other related algorithms to extract feature preferences, obtain user behavior travel data information, and establish different label systems for users according to the model, analyze and process the data, and discover the user's potential knowledge;

[0094] Model prediction layer: Through data mining and machine learning methods, the feature preference layer is semanticized and the user features and user behaviors are labeled using short text methods, and then the model prediction layer is established. The customer habit preferences of group users are predicted based on user portrait labels, and the user's potential needs are explored:

[0095] S22. Implement user portrait model clustering based on the data in the user portrait tag level, wherein the algorithm steps of the model clustering include:

[0096] The preset user portrait set is U = {u1, u1, u1, ..., u m}, there are n attributes in each portrait, and the corresponding attribute matrix is:

[0097]

[0098] Each attribute is affected by sample data v. Suppose the set of sample data affected by an attribute has h sample data, sample data v i With v j The Euclidean distance between ij Indicates that n ij =||u i -u j ||, where 1≤i, j≤h, and is the density range of sample data, then user profile u i The density function centered at can be expressed as:

[0099]

[0100] Among them, h i Must meet Inner i The number of sample data around, sample data v i The more points around, the closer the sample data v is to i The closer the relationship is, the i The larger the value of , the greater the degree of membership to the portrait;

[0101] Through y i For analysis, introduce the density function value ρ i :

[0102]

[0103] ρ i Indicates v i The surrounding sample data pair v i The influence of classification, when ρ i The larger the value, the i The greater the degree of belonging to a certain category, the greater the sample influence value ρ i The traditional algorithms introduced are:

[0104]

[0105] U represents the user portrait, C represents the cluster center, and v ij Expressed as a fuzzy classification matrix, c i Expressed as the cluster center matrix, u i It is represented as user portrait matrix, and h is the number of samples;

[0106] For the fuzzy classification matrix v ij and the cluster center matrix c i The derivatives are:

[0107]

[0108] where ρ i The size of is mainly determined by the density of the distribution around its sample space. i After that, the cluster center of the sample is adjusted, and the membership degree in the sample data is also adjusted because of the changes in the iterative process of the cluster center;

[0109] S3. Cultural and tourism resource feature extraction: feature extraction of regional cultural and tourism resources, which may include geographical location, historical ratings, cultural characteristics and other information, and use text mining technology to process the text description of cultural resources to extract keywords, themes and other specified information. The specific steps include:

[0110] S31. Introduce time series as the change value of feature extraction. For a given time series z(i), i = 1, 2, 3, ..., N, let the impact of time series on feature extraction y z(i) It consists of two parts: linear features and nonlinear features. Then:

[0111] y z(i) = l t +k t

[0112] lt is a linear feature, k t It is a nonlinear feature;

[0113] S32. Calculate the linear features in the neural network model to obtain f(l t ), and the calculated residual part is used as the nonlinear feature to obtain:

[0114] e t =y z(i) -f(l t )-k t

[0115] where e t is the residual part;

[0116] S33. In the neural network model, the prediction results for linear features and nonlinear features are set as and Then the impact of changes in time series on feature extraction is:

[0117]

[0118] Assume that the feature extraction value is H, and its change in time series is:

[0119]

[0120] At this time, through time series analysis, it is possible to identify the characteristics of tourist destinations in different time periods, so that the recommendation system can provide more accurate personalized recommendations based on the needs of users at different time points. For example, the system can identify the difference in scenery of a certain attraction in summer and winter, and recommend it accordingly to tourists who like different seasons;

[0121] After the introduction of time series, the features can be decomposed into linear and nonlinear parts. Linear features reflect the stable trend over time, while nonlinear features capture the impact of seasonal changes and special events on tourist destinations. Traditional methods usually cannot distinguish between the two, or can only capture simple trends. This distinction makes feature extraction more detailed and can better understand and predict the attractiveness of tourist destinations. For example, by identifying nonlinear features, the system can predict the peak of tourists during holidays or special events, and optimize the recommendation strategy for this purpose to improve user experience.

[0122] S4. Recommendation model design: Design a recommendation model to match user portraits with cultural and tourism resource features and calculate the user's preference for specific cultural and tourism resources. Commonly used recommendation algorithms include collaborative filtering algorithms, content-based recommendation algorithms, matrix decomposition algorithms, etc. The specific steps include:

[0123] S41. Establishing a user-tourism destination rating matrix:

[0124] There are a*b data in total, where a represents the tourist destination, b represents the user, and D ab is the rating of the ath user for the bth tourist destination. The larger the value, the higher the user's preference for the tourist destination. Conversely, the lower the data, the lower the preference. Transformed into a mathematical model, the matrix rating is expressed as:

[0125]

[0126] S42. Calculate the similarity between the tourist destination and other tourist destinations based on the attribute characteristics of the tourist destination itself, obtain similar tourist destinations to the tourist destination in the entire big data platform based on the similarity, and make personalized recommendations. The personalized recommendation can be made by predicting and scoring the tourist destinations, and recommending the tourist destination with the highest predicted score. The prediction score expression formula is:

[0127] Assume that target user C in tourist destination A predicts the rating of another tourist destination E:

[0128]

[0129] Among them, sim(A, E) is the similarity between tourist destination A and tourist destination E, D CA is the rating of user C on tourist destination A;

[0130] S5. Recommendation result generation: Generate personalized regional cultural and tourism recommendation results for users using the recommendation model, and select the most suitable recommendation results from the database of cultural and tourism resources based on the user's profile and historical behavior;

[0131] S6. Real-time recommendation and feedback: Based on real-time user behavior data and feedback information, the user portrait and recommendation model are updated in real time, dynamic personalized recommendation results are provided, and real-time recommendation is realized by using real-time data processing technology.

[0132] During the application of the method in this aspect, the user profile can be continuously updated, and the recommended content can be dynamically adjusted according to the user's new behavior data. For example, if the user has recently frequently visited a certain type of tourist destination, the system can instantly adjust the recommendation and give priority to pushing attractions that match the user's current interests, thereby improving the real-time and effectiveness of the recommendation.

[0133] Based on the above first embodiment, the present invention further provides a second embodiment. In the second embodiment, the present invention takes the need to promote Wugong Mountain as an example to describe specific operation steps:

[0134] Data collection: Use crawler technology to collect cultural and tourism related data from various data sources such as social media and travel websites. At the same time, collect information from user behavior data (such as in-app behavior, location data, etc.);

[0135] Preprocessing: Clean the collected data (remove irrelevant data and erroneous data), remove duplicates (ensure data uniqueness), and normalize (unify data formats). Use Python's Pandas library to clean and preprocess data to ensure data quality.

[0136] Data analysis: Use machine learning libraries in Python (such as scikit-learn) to perform cluster analysis and classification of user behavior data to identify user groups and individual preferences

[0137] Portrait tags: Build user portraits based on user behaviors (such as historical clicks, comments, and ratings), including tags such as interest preferences, geographic location information, and cultural background.

[0138] Feature extraction: Extract features from the collected cultural and tourism resource data, including geographic location, ratings, cultural characteristics, etc. NLP techniques (such as TF-IDF and LDA topic models) can be used to process text data and extract key information.

[0139] Model selection: Choose the appropriate recommendation algorithm based on specific needs. Collaborative filtering can make recommendations based on similarities between users or items; content-based recommendations focus on feature matching of items; matrix decomposition techniques (such as SVD) can discover potential associations between users and items.

[0140] Personalized recommendations: Combining user portraits and cultural and tourism resource characteristics, the recommendation model is used to calculate the user's preference for different cultural and tourism resources and generate a personalized recommendation list.

[0141] Real-time processing: Use real-time data processing tools such as Apache Kafka and Apache Flink to realize real-time collection and processing of user behavior data.

[0142] Model update: Dynamically adjust user portraits and recommendation models based on real-time user feedback (such as clicks, favorites, ratings, etc.) to reflect the latest changes in user preferences.

[0143] In the specific implementation process, in the steps of building user portraits:

[0144] Characterize the user portrait label level through the collected user data:

[0145] Raw data layer: refers to the raw data that are integrated, processed, and collected into the big data platform through data mining and other related technologies, which can obtain a large amount of user static attribute information and dynamic attribute information. The raw data layer can reveal the user's static attribute information (such as gender, age and other information that is not easy to change) and dynamic attribute information (such as consumption habits, page browsing behavior and other information that may change over time) through the application of data mining and other technologies.

[0146] Fact layer: Description of user specific information, such as user demographic attributes (age, identity, etc.), hobby characteristics, and other specific information. This layer is mainly used to verify the original data layer, and finally transmit the accurate data to the feature preference layer for predictive modeling. It is used to verify and filter the information of the original data layer, ensure that the verified data is accurate, and provide reliable basic data for subsequent modeling.

[0147] Feature preference layer: Based on the data transmitted from the fact layer, models are established through data mining, machine learning and other related algorithms to extract feature preferences, obtain user behavior travel data information, establish different label systems for users based on the models, analyze and process the data, and discover the user's potential knowledge. The key to this layer is to extract valuable features and user behavior preferences for travel from the fact layer data by building models, and create a representative label system to enable a deeper understanding and analysis of users.

[0148] Model prediction layer: Through data mining and machine learning methods, the feature preference layer is semanticized and the user features and user behaviors are labeled using short text methods, and then a model prediction layer is established to predict the customer habits and preferences of group users based on user portrait labels and explore users' potential needs.

[0149] Implement user portrait model clustering based on data in the user portrait tag level:

[0150] Assume that the user portrait set is U = {u1, u1, u1, ..., u m}, there are n attributes in each portrait, and the corresponding attribute matrix is:

[0151]

[0152] Each attribute is affected by sample data v. Suppose the set of sample data affected by an attribute has h sample data, sample data v i With v j The Euclidean distance between ij Indicates that n ij =||u i -u j ||, where 1≤i, j≤h, and is the density range of the sample data. In this process, the Euclidean distance between each attribute and the sample data is calculated to determine the influence range and density of the attribute. This analysis helps quantify the relationship between user attributes and their behavior. i The density function centered at can be expressed as:

[0153]

[0154] Among them, h i Must meet Inner i The number of sample data around, sample data v i The more points around, the closer the sample data v is to i The closer the relationship is, the i The larger the value of , the greater the degree of membership to the portrait; by calculating the density function, the concentration of the user portrait on a specific attribute can be evaluated, thereby determining the possibility that the user belongs to a certain type of preference; the higher the density function, the stronger the user's preference for a certain type of travel preference;

[0155] Through y i For analysis, introduce the density function value ρ i :

[0156]

[0157] ρ i Indicates v i The surrounding sample data pair v i The influence of classification, when ρ i The larger the value, the i The greater the degree of belonging to a certain category, the greater the sample influence value ρ i The traditional algorithms introduced are:

[0158]

[0159] U represents the user portrait, C represents the cluster center, and v ij Expressed as a fuzzy classification matrix, c i Expressed as the cluster center matrix, u i It is represented as user portrait matrix, and h is the number of samples;

[0160] For the fuzzy classification matrix v ij and the cluster center matrix c i The derivatives are:

[0161]

[0162] where ρ iThe size of is mainly determined by the density of the distribution around its sample space. i The cluster center of the sample is adjusted, and the membership in the sample data is also adjusted because of the changes in the iterative process of the cluster center; the cluster center is adjusted using the density function value, and the user's classification results are gradually optimized. Through repeated iterations, a stable cluster center and membership matrix are finally obtained. These results can be used to predict the user's travel preferences.

[0163] In some embodiments, the user portrait includes: the user's age, gender, income level, geographical location, etc.; in the process of making travel decisions, people are not only affected by common characteristics, that is, the difference in perceived environment, but also by individual characteristics, that is, national conditions, folk customs, and even personal interests, hobbies and personality differences. We call this difference in personality characteristics travel preference. The impact of travel preferences on decision makers can be studied from the aspects of living environment, age, occupation, education, etc. The attributes of user portraits are different travel preference choices, such as: different types of attractions (museums, cultural landscapes, etc.), different types of travel projects (hiking, climbing, rafting, etc.), different travel season preferences (spring, summer, autumn, winter), etc.; The sample data is used to further perform cluster analysis on the attributes to obtain their precise travel preferences. For example, if the tourist's personal information is, age 23, male, and it is found in his sample data that his historical travel data includes five mountain climbings, one museum, and one hiking. According to the above formula, cluster analysis is performed on these sample data, and it can be obtained that his travel preference is mountain climbing. If the final result is: prefer holiday travel, and the travel preference is mountain climbing; personalized recommendations are made based on his preferences. Of course, this embodiment only explains one aspect, and the specific circumstances of cluster analysis based on different attributes are also different. For example, based on the seasonal attribute, the cluster analysis shows that the tourist's demand for tourist destinations or tourist projects in different seasons.

[0164] Through the above method, user groups can be effectively classified in a more detailed and targeted manner, and more accurate personalized recommendations and services can be achieved. This data-driven analysis method helps to explore the deep-seated needs of users and provide more targeted strategies and decision-making support for cultural and tourism promotion.

[0165] The steps of extracting cultural and tourism resource features include:

[0166] In the process of feature extraction, the time series is introduced as the variation value of feature extraction. For a given time series z(i), i = 1, 2, 3, ..., N, let the influence of the time series on feature extraction y z(i) It consists of two parts: linear features and nonlinear features. Then:

[0167] y z(i) = l t+k t

[0168] l t is a linear feature, k t It is a nonlinear feature. In the neural network model, the linear feature is calculated to obtain f(l t ), and the calculated residual part is used as the nonlinear feature to obtain:

[0169] e t =y z(i) -f(l t )-k t

[0170] where e t is the residual part. In the neural network model, the prediction results for linear features and nonlinear features are set as and Then the impact of changes in time series on feature extraction is:

[0171]

[0172] Assume that the feature extraction value is H, and its change in time series is:

[0173]

[0174] Time series data has the characteristic of changing dynamically over time, which is a key dimension, especially for analyzing cultural and tourism resources (such as the number of visits to tourist attractions, the popularity of seasonal events, etc.). In the process of feature extraction, the introduction of time series can help capture this dynamic change. For example, the scenery of tourist destinations will change in different seasons. Common ones include enjoying flowers in spring and watching fallen leaves in autumn. These changes in scenery are also important factors in determining whether target users will visit these tourist destinations. In the feature extraction of cultural and tourism resources, the introduction of time series data makes it possible to capture features that change over time. By analyzing linear features (stable trends over time) and nonlinear features (such as seasonal changes, the impact of special events, etc.), and using neural network models to process these data, more accurate features can be obtained. To better understand and predict the attractiveness of cultural and tourism resources, linear features and nonlinear features: linear features are those that show a stable trend over time, while nonlinear features include complex and irregular patterns that change over time, neural network model processing: the predicted value of the linear feature is calculated through the neural network model, and the residual part after calculation is regarded as information containing nonlinear features, so as to achieve comprehensive extraction of features, dynamic changes in feature extraction values: linear and nonlinear features are combined to form a comprehensive feature extraction value, which changes dynamically with the time series, reflecting the changing trend of cultural and tourism resource characteristics over time; through this method of combining linear and nonlinear features, the complexity of cultural and tourism resources changing over time can be more comprehensively captured, providing data support for the management and promotion of cultural and tourism resources.

[0175] Based on the above first embodiment, the present invention further provides a third embodiment. In the third embodiment, the present invention takes the recommendation of a tourist attraction, "Glacier Lake", to tourists as an example. The characteristics of this attraction are very different in winter and summer:

[0176] Winter: The glacier lake freezes over, making it suitable for ice activities (such as skating and ice sculpture viewing).

[0177] Summer: The lake melts and the surrounding trees are green, making it suitable for hiking and boating.

[0178] Therefore, there are the following specific steps:

[0179] 1. Time Series Data Collection

[0180] We collected data on the number of visitors to Glacier Lake, participation in activities, user reviews, etc. over the past few years and compiled them into time series data by month. For example:

[0181] Visitors: The total number of visitors per month.

[0182] Activity participation: The number of people who participate in activities such as skating and rowing each month.

[0183] User evaluation: sentiment analysis is performed based on the ratings and comments of users in different months and quantified into scores.

[0184] 2. Linear feature extraction

[0185] By analyzing these time series data, we can identify stable trends over time, i.e. linear features:

[0186] Linear trend of visitor volume: Hypothetically, the analysis found that there are stable peaks in visitor volume in summer (June to August) and winter (December to February) each year, while the number of visitors is lower in spring and autumn.

[0187] Linear trend of user evaluation: Users’ evaluation scores for Glacier Lake in winter and summer are relatively stable and high.

[0188] These stable trends reflect the long-term appeal of glacial lakes during certain seasons, which can still attract large numbers of visitors even if annual variations are modest.

[0189] 3. Nonlinear feature extraction

[0190] Through residual analysis, the part of the time series that cannot be explained by the linear trend, that is, the nonlinear characteristics, are extracted:

[0191] Seasonal variations: In certain months, such as the festive season in mid-December, there is a sudden surge in visitor numbers due to the holiday effect.

[0192] Impact of special events: In the summer of a certain year, a large outdoor music festival was held, which led to a significant increase in the number of tourists in the short term. This is also a nonlinear feature.

[0193] These nonlinear characteristics reflect the abrupt changes of glacial lakes at specific points in time, which cannot be predicted by simple linear trends.

[0194] 4. Feature synthesis and calculation

[0195] Combining linear and nonlinear features to form a comprehensive feature value for glacial lakes:

[0196] Linear feature prediction value: Predict the trend of summer and winter tourist volume in the next year through the neural network model.

[0197] Non-linear feature prediction: predicting sudden changes in visitor volume during festivals or special events.

[0198] Final comprehensive characteristic value: This comprehensive characteristic value represents the overall attractiveness of the glacial lake in each month in the future.

[0199] 5. Recommended Applications

[0200] Winter recommendations: If a user prefers ice activities and his historical travel records show that he has traveled more in winter, the system will recommend that the user visit the glacial lake, participate in ice skating and watch ice sculptures between December and February based on the user's travel preferences.

[0201] Summer recommendation: Another user prefers summer outdoor activities, and historical data shows that he has a strong interest in hiking and boating. Based on his travel preferences, the system will recommend that the user go to the glacial lake from June to August to enjoy the fun of hiking and boating on the lake.

[0202] This specific example shows how to use time series analysis to extract linear and nonlinear features to calculate the comprehensive feature values ​​of tourist attractions in different time periods. This method helps the recommendation system to more accurately capture the temporal changes of tourist attractions, and thus provide users with personalized travel recommendations.

[0203] Of course, it is not only the time series that affects the feature extraction of tourist destinations, but also a variety of other features such as climate / temperature. For example, people who prefer snowy scenery have a stronger demand for travel on snowy days, and vice versa, their demand for travel is reduced.

[0204] On the other hand, based on the above first embodiment, the present invention further provides a fourth embodiment. In the fourth embodiment, the present invention proposes a regional cultural tourism intelligent recommendation system, which is used to execute any regional cultural tourism intelligent recommendation method in the above embodiments, and includes the following modules:

[0205] (1) Data collection and preprocessing module: used to collect a large amount of data including regional culture, tourism resources, user reviews, etc., and perform preprocessing such as data cleaning, deduplication, and normalization. These data can come from public data sets, social media platforms, tourism websites, and other channels;

[0206] (2) User portrait construction module: Build user portraits based on the user's historical click, comment, rating and other behavioral data, including the user's interest preferences, geographic location information, cultural background, etc. Machine learning algorithms such as cluster analysis and classification algorithms can be used to build user portraits;

[0207] Multi-level labeling system: Based on the above user portrait labeling hierarchy, from the raw data layer to the model prediction layer, a multi-dimensional feature labeling system for users is gradually constructed, including but not limited to basic demographic information, interests and hobbies, travel behavior preferences, consumption habits, etc.;

[0208] User classification: Apply machine learning methods such as fuzzy clustering algorithms to segment users and classify them into precise market segments based on their behavior and preference characteristics;

[0209] (3) Cultural and tourism resource feature extraction module: extract features of regional cultural and tourism resources, which may include information such as geographic location, historical ratings, and cultural characteristics. Use text mining technology to process the text description of cultural resources and extract keywords, themes, and other information.

[0210] Time series analysis: Use time series data to analyze the dynamic trends of cultural and tourism resources, distinguish linear and nonlinear characteristics, and capture the impact of seasonal changes and special events;

[0211] Resource labeling: Label cultural and tourism resources accordingly based on the extracted features, such as "best choice for family travel", "spring flower viewing location", etc., to optimize the matching degree of the recommendation algorithm

[0212] (4) Intelligent recommendation algorithm model design module: Design a recommendation model to match user profiles with cultural and tourism resource features and calculate the user's preference for specific cultural and tourism resources. Commonly used recommendation algorithms include collaborative filtering algorithms, content-based recommendation algorithms, matrix decomposition algorithms, etc.

[0213] Personalized recommendation model: Combining user portraits and cultural and tourism resource characteristics, design content-based recommendation systems, collaborative filtering recommendation systems, or hybrid recommendation systems to provide personalized recommendations for tourist attractions and activities;

[0214] Real-time recommendation and feedback learning: Implement a real-time recommendation mechanism and introduce a feedback learning mechanism to continuously optimize the recommendation model based on users' actual click, browse and consumption data, thereby improving the accuracy of recommendations and user satisfaction.

[0215] (5) Recommendation result generation module: Use the recommendation model to generate personalized regional cultural and tourism recommendation results for users, and select the most suitable recommendation results from the database of cultural and tourism resources based on the user's profile and historical behavior;

[0216] Front-end display: build a user-friendly application interface, display recommended content, provide search, screening, evaluation and other functions, and enhance user experience

[0217] (6) Real-time recommendation and feedback module: Based on the user's real-time behavior data and feedback information, the user profile and recommendation model are updated in real time to provide dynamic personalized recommendation results. Real-time recommendations can be achieved by using real-time data processing technologies such as streaming computing and real-time data mining;

[0218] Evaluate system performance by setting indicators (such as click-through rate, conversion rate, and user satisfaction) and perform regular iterative updates to ensure the long-term effectiveness of the recommendation system and continuous improvement of user experience;

[0219] (7) User Privacy and Data Security Module:

[0220] Compliance: Ensure that data collection, storage and processing comply with relevant laws and regulations and user agreements, and protect user privacy;

[0221] Security measures: Take security measures such as data encryption and access control to prevent data leakage and abuse.

[0222] On the other hand, the present invention further discloses a computer-readable storage medium storing a computer program, wherein when the computer program is executed by a processor, the processor executes the steps of the above method.

[0223] On the other hand, the present invention further discloses a computer device, including a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the above method.

[0224] In another embodiment provided in the present application, a computer program product including instructions is also provided, which, when executed on a computer, enables the computer to execute any of the regional cultural and tourism intelligent recommendation methods in the above embodiments.

[0225] It is understandable that the system provided by the embodiment of the present invention corresponds to the method provided by the embodiment of the present invention, and the explanation, examples and beneficial effects of the relevant contents can refer to the corresponding parts in the above method.

[0226] The embodiment of the present application also provides an electronic device, including a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other through the communication bus.

[0227] Memory, used to store computer programs;

[0228] The processor is used to implement the above-mentioned regional cultural tourism intelligent recommendation method when executing the program stored in the memory.

[0229] The communication bus mentioned in the above electronic device can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. The communication bus can be divided into an address bus, a data bus, a control bus, etc.

[0230] The communication interface is used for communication between the above electronic device and other devices.

[0231] The memory may include a random access memory (RAM) or a non-volatile memory (NVM), such as at least one disk memory. Optionally, the memory may also be at least one storage device located away from the aforementioned processor.

[0232] The above-mentioned processor can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, and discrete hardware components.

[0233] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium, for example, the computer instructions may be transmitted from a website site, computer, server or data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) mode to another website site, computer, server or data center. The computer-readable storage medium may be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more available media integrated. The available medium may be a magnetic medium, (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium (e.g., a solid-state drive Solid State Disk (SSD)), etc.

[0234] 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 principle of the present invention should be included in the protection scope of the present invention.

[0235] In addition, it should be noted that if the embodiments of the present invention involve directional indications (such as up, down, left, right, front, back, etc.), the directional indications are only used to explain the relative position relationship, movement status, etc. between the components in a certain specific posture. If the specific posture changes, the directional indication will also change accordingly.

[0236] In addition, if there are descriptions involving "first", "second", etc. in the embodiments of the present invention, the descriptions of "first", "second", etc. are only used for descriptive purposes and cannot be understood as indicating or implying their relative importance or implicitly indicating the number of technical features indicated. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In addition, the meaning of "and / or" appearing in the full text includes three parallel schemes. Taking "A and / or B" as an example, it includes scheme A, or scheme B, or schemes that A and B meet at the same time. In addition, in the embodiments of the present invention, "multiple" refers to more than two. In addition, the technical solutions between the various embodiments can be combined with each other, but it must be based on the ability of ordinary technicians in the field to implement. When the combination of technical solutions is contradictory or cannot be implemented, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection required by the present invention.

Claims

1. A regional cultural tourism intelligent recommendation method, characterized in that: The specific implementation steps include the following: S1. Data collection and preprocessing: Collect a large amount of cultural and tourism related data through public channels and preprocess the data; S2. User portrait construction: Based on user behavior data, use machine learning algorithms to build user portraits; S3. Cultural and tourism resource feature extraction: feature extraction of regional cultural and tourism resources, and use text mining technology to process the text description of cultural resources and extract specified information; S4. Recommendation model design: Design a recommendation model to match user profiles with cultural and tourism resource characteristics and calculate the user's preference for specific cultural and tourism resources; S5. Recommendation result generation: Generate personalized regional cultural and tourism recommendation results for users using the recommendation model, and select the most suitable recommendation results from the database of cultural and tourism resources based on the user's profile and historical behavior; S6. Real-time recommendation and feedback: Based on real-time user behavior data and feedback information, the user portrait and recommendation model are updated in real time, dynamic personalized recommendation results are provided, and real-time recommendation is realized by using real-time data processing technology.

2. The regional cultural tourism intelligent recommendation method according to claim 1, characterized in that: The specific operation steps of data collection and preprocessing in step S1 include: S11. Data information collection: Collect the required data from different data sources as the initial data for characterizing user portraits; S12. Data cleaning and integration: Screen, clean, process and integrate the data from multiple data sources, denoise the data, unify the data from multiple data sources and in multiple formats according to the distribution of data space, integrate them into the big data platform, and realize the sharing of data resources; S13. Data selection and transformation: Extract the integrated data and then convert it into the data required by the system through format.

3. The regional cultural tourism intelligent recommendation method according to claim 1, characterized in that: The specific operation steps of constructing the user portrait in step S2 include: S21. Characterize the user portrait label level through the collected user data: S22. Implement user portrait model clustering based on the data in the user portrait tag level, wherein the algorithm steps of the model clustering include: The preset user portrait set is U = {u1, u1, u1, ..., u m }, there are n attributes in each portrait, and the corresponding attribute matrix is: Each attribute is affected by sample data v. Suppose the set of sample data affected by an attribute has h sample data, sample data v i With v j The Euclidean distance between ij Indicates that n ij =||u i -u j ||, where 1≤i, j≤h, and is the density range of sample data, then user profile u i The density function centered at can be expressed as: Among them, h i Must meet Inner i The number of sample data around, sample data v i The more points around, the closer the sample data v is to i The closer the relationship is, the i The larger the value of , the greater the degree of membership to the portrait; Through y i For analysis, introduce the density function value ρ i : ρ i Indicates v i The surrounding sample data pair v i The influence of classification, when ρ i The larger the value, the i The greater the degree of belonging to a certain category, the greater the sample influence value ρ i The traditional algorithms introduced are: U represents the user portrait, C represents the cluster center, and v ij Expressed as a fuzzy classification matrix, c i Expressed as the cluster center matrix, u i It is represented as user portrait matrix, and h is the number of samples; For the fuzzy classification matrix v ij and the cluster center matrix c i The derivatives are: where ρ i The size of is mainly determined by the density of the distribution around its sample space. i After that, the sample cluster center is adjusted, and the membership degree in the sample data is also adjusted because of the changes in the cluster center during the iteration process.

4. The regional cultural tourism intelligent recommendation method according to claim 3, characterized in that: The user portrait tag levels in step S21 include: Raw data layer: used to integrate, process, and process multiple data sources into the raw data collected on the big data platform for data mining, in order to obtain a large amount of user static attribute information and dynamic attribute information; Fact layer: contains descriptions of user-specific information, used to verify the original data layer and transmit accurate data to the feature preference layer for predictive modeling; Feature preference layer: used to transmit data based on the fact layer, extract feature preferences by establishing an artificial intelligence algorithm model, obtain user behavior travel data information, and establish different label systems for users based on the model, analyze and process the data, and discover the user's potential knowledge; Model prediction layer: Label user features and user behaviors by semanticizing the feature preference layer and using short text methods, and then establish a model prediction layer. The customer habit preferences of group users are predicted based on user portrait labels to explore users' potential needs.

5. The regional cultural tourism intelligent recommendation method according to claim 1, characterized in that: The specific operation steps of extracting the characteristics of Chinese travel resources in step S3 include: S31. Introduce time series as the change value of feature extraction. For a given time series z(i), i = 1, 2, 3, ..., N, let the impact of time series on feature extraction y z(i) It consists of two parts: linear features and nonlinear features. Then: and z(i) =l t +k t l t is a linear feature, k t It is a nonlinear feature; S32. Calculate the linear features in the neural network model to obtain f(l t ), and the calculated residual part is used as the nonlinear feature to obtain: e t =y z(i) -f(l t )-k t where e t is the residual part; S33. In the neural network model, the prediction results for linear features and nonlinear features are set as and Then the impact of changes in time series on feature extraction is: Assume that the feature extraction value is H, and its change in time series is:

6. The regional cultural tourism intelligent recommendation method according to claim 1, characterized in that: The specific operation steps of the recommended model design in step S4 include: S41. Establishing a user-tourism destination rating matrix: There are a*b data in total, where a represents the tourist destination, b represents the user, and D ab is the rating of the ath user for the bth tourist destination. The larger the value, the higher the user's preference for the tourist destination. Conversely, the lower the data, the lower the preference. Transformed into a mathematical model, the matrix rating is expressed as: S42. Calculate the similarity between the tourist destination and other tourist destinations based on the attribute characteristics of the tourist destination itself, obtain tourist destinations similar to the tourist destination in the entire big data platform based on the similarity, and make personalized recommendations.

7. The regional cultural tourism intelligent recommendation method according to claim 6, characterized in that: The specific operation steps of the recommendation model design in step S4 include: in step S42, the personalized recommendation can be made by predicting and scoring the tourist destinations, and the tourist destination with the highest predicted score is recommended. The prediction score expression formula is: Assume that target user C in tourist destination A predicts the rating of another tourist destination E: Among them, sim(A, E) is the similarity between tourist destination A and tourist destination E, D CA is the rating value given by user C to tourist destination A.

8. A regional cultural tourism intelligent recommendation system, characterized in that: The method for executing the regional cultural tourism intelligent recommendation method according to any one of claims 1 to 7 above comprises the following modules: Data collection and preprocessing module: used to collect a large amount of data including regional culture, tourism resources, and user evaluations, and to perform preprocessing work such as data cleaning, deduplication, and normalization; User portrait building module: Based on the user's historical clicks, comments, ratings and other behavioral data, a user portrait is built using machine learning algorithms, including the user's interest preferences, geographic location information, and cultural background; Cultural and tourism resource feature extraction module: extracts features of regional cultural and tourism resources, including geographic location, historical ratings, and cultural feature information, and uses text mining technology to process text descriptions of cultural resources to extract keywords and theme information; Recommendation model design module: Design a recommendation model to calculate the user's preference for specific cultural and tourism resources by matching user portraits with cultural and tourism resource characteristics; Recommendation result generation module: Use the recommendation model to generate personalized regional cultural and tourism recommendation results for users, and select the most suitable recommendation results from the database of cultural and tourism resources based on the user's profile and historical behavior; Real-time recommendation and feedback module: Based on real-time user behavior data and feedback information, it updates user portraits and recommendation models in real time, provides dynamic personalized recommendation results, and uses real-time data processing technology to achieve real-time recommendations.

9. A computer-readable storage medium, characterized in that: A computer program is stored, and when the computer program is executed by a processor, the processor is caused to perform the steps of the method according to any one of claims 1 to 7.

10. A computer device, characterized in that: The method comprises a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the method according to any one of claims 1 to 7.

Images