Business travel behavior analysis method based on user behaviors

Abstract

The invention discloses a user behavior-based travel behavior analysis method, which comprises the following steps of: acquiring multi-dimensional multi-spatio-temporal data of a travel, and extracting dynamic characteristics to obtain a preliminary behavior sequence; if the stroke adjustment frequency exceeds the threshold value, clustering the similar adjustment mode to obtain an adjustment clustering set; integrating the associated vehicle data, supplementing missing values, and generating a complete track; extracting travel rule features from user preferences through the complete behavior track, classifying the travel rule features, and determining a preference classification group; obtaining resource configuration related data according to the preference classification group, and adjusting configuration parameters based on the configuration matching degree to obtain an optimized resource scheme; and by optimizing a resource scheme, fusing personalized service requirements, predicting a subsequent travel rule, and determining final travel service recommendation. According to the invention, through multi-level data fusion and sequence model prediction, individuation and resource configuration efficiency of travel service are significantly improved.

Description

Technical Field

[0001] This invention belongs to the field of intelligent transportation and travel service technology, and in particular relates to a method for analyzing travel behavior based on user behavior. Background Technology

[0002] Business travel analytics, as a crucial field for studying user travel patterns and preferences, is key to improving the accuracy of business travel services, optimizing resource allocation, and enhancing user experience. With the development of mobile internet and big data technologies, businesses and organizations are increasingly relying on user behavior data to understand travel needs and subsequently provide personalized services.

[0003] However, current research and applications still face numerous challenges, urgently requiring breakthroughs in technological bottlenecks to achieve more efficient behavioral analysis. Existing methods, when analyzing user travel behavior, often rely too heavily on single-dimensional information, such as focusing solely on a user's destination or travel frequency, while neglecting the comprehensive correlation of multi-dimensional data within the trip. This limitation makes it difficult for the analysis results to fully reflect the user's true behavioral patterns. For example, existing systems may be able to identify cities frequently visited by users, but they cannot delve into changes in user travel preferences across different scenarios, or accurately determine the dynamic decision-making process behind a user's choice of transportation mode. This makes it difficult for service providers to offer accurate recommendations or optimization solutions when facing complex travel scenarios.

[0004] To address the aforementioned problems in existing technologies, there is an urgent need to propose a travel behavior analysis method based on user behavior. Summary of the Invention

[0005] To address the aforementioned technical problems, this invention provides a method for analyzing travel behavior based on user behavior.

[0006] This invention proposes a method for analyzing travel behavior based on user behavior, the method comprising the following steps: Collect multidimensional spatiotemporal data of users during their business trips, extract dynamic change features, and obtain preliminary behavioral sequences; Based on the preliminary behavior sequence, the itinerary adjustment records are obtained. If the adjustment frequency is higher than the preset threshold, a clustering algorithm is applied to group similar adjustment patterns and determine the adjustment cluster set. By adjusting the cluster set, we can obtain related transportation selection data, integrate the data, and obtain an enhanced behavior sequence. Missing values ​​are filled into the enhanced behavior sequence to obtain the complete behavior trajectory; By analyzing complete behavioral trajectories, we can extract travel pattern features from user preferences, classify them, and determine preference classification groups. Based on preference classification groups, obtain resource configuration-related data, and adjust configuration parameters based on configuration matching degree to obtain an optimized resource solution; By optimizing resource allocation, integrating personalized service needs, and predicting subsequent travel patterns, the final travel service recommendations can be determined.

[0007] Optionally, multidimensional spatiotemporal data of users during their business trips can be collected, dynamic change features can be extracted, and preliminary behavioral sequences can be obtained, including: Multidimensional spatiotemporal data of users during their business trips are acquired through sensors and positioning devices and stored as structured datasets; Spatiotemporal data fusion is performed on structured datasets to generate unified feature representations; If the integrity of the unified feature representation is higher than a preset threshold, a long short-term memory network is used to perform dynamic feature extraction on the unified feature representation to obtain dynamic behavioral features. Hidden Markov Models are used to perform sequence model analysis on dynamic behavioral features to generate preliminary behavioral sequences.

[0008] Optionally, based on the preliminary behavior sequence, travel adjustment records are obtained. If the adjustment frequency is higher than a preset threshold, a clustering algorithm is applied to group similar adjustment patterns to determine the adjustment cluster set, including: Obtain itinerary adjustment records, process them using a uniform format, and obtain a standardized adjustment dataset; By standardizing and adjusting the dataset, the adjustment frequency of each record is calculated to obtain the frequency distribution results; If the frequency distribution result is higher than the preset threshold, the K-means clustering algorithm is applied to the high-frequency adjustment records to group similar adjustment patterns and obtain a preliminary cluster set. For the initial cluster set, the silhouette coefficient is used to verify the cluster quality, resulting in an optimized adjusted cluster set.

[0009] Optionally, by adjusting the cluster set, related transportation selection data can be obtained, and the data can be integrated to obtain enhanced behavior sequences, including: Obtain transportation selection data from adjusted cluster sets; The K-means algorithm was used to cluster the transportation choice data to obtain classified behavior clusters. Spatiotemporal information is extracted from the classified behavioral clusters to generate feature vectors containing time and location. If the spatiotemporal information of the feature vectors is complete, then the behavior clusters and spatiotemporal information are integrated through a weighted fusion mechanism to obtain a fused dataset; If spatiotemporal information is missing, the missing information is supplemented from the preset database to obtain a fused dataset; Based on the fused dataset, a Hidden Markov Model is used to analyze the transition probability of the behavior sequence and generate an enhanced behavior sequence.

[0010] Optionally, missing values ​​are imputed in the enhanced behavior sequence to obtain the complete behavior trajectory, including: The system detects whether there are disjoint segments in the enhanced behavior sequence and obtains the segment detection results. If discontinuous segments are detected, a sequence model is used to infer the missing information and obtain the inferred spatiotemporal information. Based on the inferred spatiotemporal information and combined with time-related attributes, continuous behavioral sequence data is generated to obtain a preliminary complete trajectory; By verifying the preliminary complete trajectory through the continuous attributes of the data, we can determine whether there are inconsistencies in time or space and obtain the verification results. If the verification results show inconsistency, the spatiotemporal information of the inconsistent segments is re-inferred using a sequence model to obtain the corrected spatiotemporal information; Complete trajectory data is generated by combining the corrected spatiotemporal information with sequence analysis.

[0011] Optionally, by analyzing complete behavioral trajectories, travel pattern features are extracted from user preferences and categorized to determine preference classification groups, including: Complete behavioral trajectories are obtained, time series features and spatial location features are extracted, and the time window segmentation method is used to calculate travel frequency and regularity indicators to obtain a user travel pattern feature set; If the time series features of the travel pattern feature set meet the preset periodic threshold, then the K-means clustering algorithm is used to classify the feature set and determine the preliminary user preference classification group. A preliminary user preference classification group is obtained, and the feature set is reduced in dimensionality using principal component analysis to obtain a low-dimensional feature representation. Based on the low-dimensional feature representation, the DBSCAN clustering algorithm is used to optimize the classification groups, determine the boundary points of the classification groups, and obtain the optimized user preference classification groups. Classification labels are extracted from the optimized user preference classification groups, and behavioral pattern recognition is performed in combination with spatial location information to obtain a description of user travel behavior patterns. Based on the description of user travel behavior patterns, the classification accuracy is calculated using preset evaluation indicators to obtain the final user preference classification results.

[0012] Optionally, based on preference classification groups, resource configuration-related data is obtained, and configuration parameters are adjusted based on configuration matching degree to obtain an optimized resource solution, including: Obtain resource configuration data for preference classification groups, extract user preference features and resource allocation records from the database to obtain the initial resource configuration dataset; For the initial resource configuration dataset, a preference feature extraction method is used to analyze the feature vectors of user preference classification groups to obtain the preference feature matrix; Based on the preference feature matrix, the matching algorithm is used to calculate the matching degree between the resource allocation data and the preference classification group, and the matching degree score is obtained. If the matching score is lower than the preset threshold, the resource configuration items that need to be adjusted are determined by the threshold comparison logic, and a list of configurations to be optimized is obtained. For the list of configurations to be optimized, the resource allocation strategy is updated using parameter adjustment rules to generate a temporary optimized resource plan; Based on the temporary optimized resource plan, the matching score is recalculated using a resource allocation strategy to determine whether the optimization goal has been achieved, and the final optimized resource plan is obtained.

[0013] Optionally, by optimizing resource allocation, integrating personalized service needs, and predicting subsequent travel patterns, the final travel service recommendation can be determined, including: Obtain users' historical travel records and preference tags, extract relevant data from the database, and obtain the initial service demand dataset; The initial service demand dataset is processed using time series analysis methods to extract time series features of user travel behavior, resulting in a travel pattern feature vector. A long short-term memory network model is used to predict subsequent travel behavior sequences based on travel pattern feature vectors, thus obtaining the predicted travel sequence. If the matching degree between the predicted travel sequence and the preference label is lower than the preset threshold, the recommendation parameters are adjusted by comparison logic to obtain the service sequence to be optimized. For the service sequence to be optimized, the service recommendation strategy is updated using preset optimization rules to generate a temporary service recommendation sequence; By using the temporary service recommendation sequence, the matching degree with the preference label is recalculated to determine whether the recommendation goal is met, and the final service recommendation sequence is obtained. Based on the final service recommendation sequence, update the service recommendation records in the database to complete the travel recommendation process.

[0014] The present invention also proposes a computer device, including a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the method.

[0015] The present invention also proposes a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method.

[0016] Compared with the prior art, the present invention has the following advantages and technical effects: This invention discloses a method that addresses business scenarios such as dynamic changes in user behavior, frequent itinerary adjustments, and suboptimal resource allocation during business travel. It collects multi-dimensional spatiotemporal user data, uses a sequence model to generate a preliminary behavior sequence, and combines itinerary adjustment records from heterogeneous sources with a clustering algorithm to identify similar adjustment patterns, forming an adjustment cluster set. Further, it integrates transportation selection data and spatiotemporal information to generate an enhanced behavior sequence, and uses a sequence model to fill in discontinuous segments, obtaining a complete behavior trajectory. This invention extracts travel pattern features from the complete trajectory, classifies user preferences, combines resource allocation data with optimization schemes, and finally integrates personalized needs to predict subsequent travel patterns, generating a precise service recommendation sequence. This invention significantly improves the personalization of business travel services and the efficiency of resource allocation through multi-level data fusion and sequence model prediction. Attached Figure Description

[0017] The accompanying drawings, which form part of this application, are used to provide a further understanding of this application. The illustrative embodiments and descriptions of this application are used to explain this application and do not constitute an undue limitation of this application. In the drawings: Figure 1 This is a schematic diagram of the method flow according to an embodiment of the present invention. Detailed Implementation

[0018] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0019] It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases the steps shown or described may be executed in a different order than that shown here.

[0020] Example 1 This embodiment provides a method for analyzing travel behavior based on user behavior, the method comprising the following steps: Collect multidimensional spatiotemporal data of users during their business trips, extract dynamic change features, and obtain preliminary behavioral sequences; Based on the preliminary behavior sequence, the itinerary adjustment records are obtained. If the adjustment frequency is higher than the preset threshold, a clustering algorithm is applied to group similar adjustment patterns and determine the adjustment cluster set. By adjusting the cluster set, we can obtain related transportation selection data, integrate the data, and obtain an enhanced behavior sequence. Missing values ​​are filled into the enhanced behavior sequence to obtain the complete behavior trajectory; By analyzing complete behavioral trajectories, we can extract travel pattern features from user preferences, classify them, and determine preference classification groups. Based on preference classification groups, obtain resource configuration-related data, and adjust configuration parameters based on configuration matching degree to obtain an optimized resource solution; By optimizing resource allocation, integrating personalized service needs, and predicting subsequent travel patterns, the final travel service recommendations can be determined.

[0021] As a specific implementation method, such as Figure 1 As shown, it includes the following steps: S101. By collecting multidimensional spatiotemporal data of users during their business trips, and using sequence models to process dynamic changes, a preliminary behavioral sequence is obtained.

[0022] Multidimensional spatiotemporal data of users during their business trips is acquired through sensors and positioning devices and stored as a structured dataset. Data preprocessing techniques are used to fuse spatiotemporal data in the structured dataset, generating a unified feature representation. If the completeness of the unified feature representation exceeds a preset threshold, a Long Short-Term Memory (LSTM) network is used to dynamically extract features from the unified feature representation, obtaining dynamic behavioral features. Hidden Markov Models (HMMs) are used to perform sequence model analysis on the dynamic behavioral features, generating preliminary behavioral sequences. Based on the preliminary behavioral sequences, a sliding window technique is used to detect behavioral patterns and identify patterns in user business trip behavior, obtaining behavioral pattern sequences. If the similarity between the behavioral pattern sequence and historical behavioral patterns exceeds a preset threshold, a sequence alignment algorithm is used to optimize the behavioral pattern sequences, obtaining optimized behavioral sequences. Based on the optimized behavioral sequences, a probabilistic prediction model is used to generate user behavior prediction results.

[0023] S102. Based on the preliminary behavior sequence, obtain the travel adjustment records from the heterogeneous sources. If the adjustment frequency is higher than the preset threshold, apply the clustering algorithm to group similar adjustment patterns and determine the adjustment cluster set.

[0024] Trip adjustment records are obtained from heterogeneous sources and processed using a unified format to obtain a standardized adjustment dataset. The adjustment frequency of each record is calculated using this standardized dataset to obtain a frequency distribution result. If the frequency distribution result is higher than a preset threshold, K-means clustering is applied to the high-frequency adjustment records to group similar adjustment patterns, resulting in a preliminary cluster set. The silhouette coefficient is used to verify the clustering quality of the preliminary cluster set, resulting in an optimized adjustment cluster set. Based on the optimized adjustment cluster set, feature vectors are extracted from each adjustment pattern group to determine the representative features of the pattern. These representative features are then matched against a preset trip adjustment template to determine the applicable scenarios for each group of adjustment patterns. The matching results are obtained to generate an adjustment pattern classification set, determining the final trip adjustment optimization scheme.

[0025] S103. By adjusting the cluster set, obtain the associated transportation selection data, and use the fusion mechanism to integrate these data with spatiotemporal information to obtain the enhanced behavior sequence.

[0026] Transportation choice data is obtained from the adjusted cluster set to determine initial behavioral patterns. The K-means algorithm is used to cluster the transportation choice data, resulting in categorized behavioral clusters. Spatiotemporal information is extracted from these clusters to generate feature vectors containing time and location. If the spatiotemporal information of the feature vectors is complete, a weighted fusion mechanism is used to integrate the behavioral clusters and spatiotemporal information to obtain a fused dataset; if the spatiotemporal information is missing, it is supplemented from a pre-defined database to obtain a fused dataset. Based on the fused dataset, a Hidden Markov Model is used to analyze the transition probabilities of behavioral sequences, generating enhanced behavioral sequences. For the enhanced behavioral sequences, key behavioral patterns are extracted using pattern recognition to obtain the final behavioral sequence features. Association rules are extracted from the final behavioral sequence features to generate a sequence model describing the user's transportation choice patterns.

[0027] S104. Based on the enhanced behavior sequence, if there are discontinuous segments in the sequence, fill in the missing spatiotemporal information through the sequence model to obtain the complete behavior trajectory.

[0028] Behavioral sequences are obtained through sequence analysis, and the presence of discontinuous segments is detected to obtain segment detection results. If discontinuous segments are detected, a sequence model is used to infer the missing information, resulting in inferred spatiotemporal information. Based on the inferred spatiotemporal information and combined with temporal correlation attributes, continuous behavioral sequence data is generated, yielding a preliminary complete trajectory. The preliminary complete trajectory is verified using data continuity attributes to determine if there are any temporal or spatial inconsistencies, obtaining verification results. If the verification results show inconsistencies, the sequence model is used to re-infer the spatiotemporal information of the inconsistent segments, resulting in corrected spatiotemporal information. Based on the corrected spatiotemporal information and sequence analysis, a final complete behavioral trajectory is generated, yielding complete trajectory data. Based on the complete trajectory data, trajectory generation techniques are used for spatiotemporal sequence optimization to obtain a smooth behavioral trajectory.

[0029] S105. Extract travel pattern features from user preferences through complete behavioral trajectories, and use clustering algorithms to classify these features and determine preference classification groups.

[0030] By extracting time-series and spatial location features from complete behavioral trajectories, and using a time window segmentation method to calculate travel frequency and regularity indicators, a user travel pattern feature set is obtained. If the time-series features of the travel pattern feature set meet a preset periodicity threshold, the K-means clustering algorithm is used to classify the feature set and determine preliminary user preference classification groups. After obtaining the preliminary user preference classification groups, principal component analysis is used to reduce the dimensionality of the feature set, resulting in low-dimensional feature representations. Based on these low-dimensional feature representations, the DBSCAN clustering algorithm is used to optimize the classification groups, identifying the boundary points of each group to obtain optimized user preference classification groups. Classification labels are extracted from the optimized user preference classification groups, and combined with spatial location information for behavioral pattern recognition, resulting in a description of user travel behavior patterns. For the user travel behavior pattern descriptions, a preset evaluation index is used to calculate the classification accuracy, yielding the final user preference classification result.

[0031] S106. Based on the preference classification group, obtain resource configuration-related data, and determine if the configuration matching degree is lower than the preset threshold. Then adjust the configuration parameters to obtain an optimized resource solution.

[0032] The process begins by acquiring resource configuration data for preference classification groups. User preference features and resource allocation records are extracted from the database using data collection methods to obtain an initial resource configuration dataset. For this dataset, preference feature extraction methods are employed to analyze the feature vectors of user preference classification groups, resulting in a preference feature matrix. Based on this matrix, a configuration matching algorithm is used to calculate the matching degree between the resource configuration data and the preference classification groups, yielding a matching score. If the matching score is lower than a preset threshold, threshold comparison logic is used to determine the resource configuration items that need adjustment, resulting in a list of configurations to be optimized. For this list, parameter adjustment rules are used to update the resource allocation strategy, generating a temporary optimized resource plan. Based on this temporary optimized plan, the matching score is recalculated using the resource allocation strategy to determine if the optimization goal has been achieved, resulting in the final optimized resource plan. Finally, the resource configuration records in the database are updated using the final optimized resource plan, completing the optimization process.

[0033] S107. By optimizing resource plans, integrating personalized service needs, and using sequence models to predict subsequent travel patterns, the final service recommendation sequence is determined.

[0034] The process begins by acquiring users' historical travel records and preference tags, extracting relevant data from the database to obtain an initial service demand dataset. Time series analysis is then used to process this dataset, extracting time-series features of user travel behavior to obtain a travel pattern feature vector. A Long Short-Term Memory (LSTM) network model is employed to predict subsequent travel behavior sequences based on this feature vector, resulting in a predicted travel sequence. If the match between the predicted travel sequence and the preference tags is lower than a preset threshold, recommendation parameters are adjusted using comparative logic to obtain a service sequence to be optimized. For this service sequence, a service recommendation strategy is updated using preset optimization rules, generating a temporary service recommendation sequence. The match between the temporary service recommendation sequence and the preference tags is recalculated to determine if the recommendation objective is met, resulting in the final service recommendation sequence. Finally, the service recommendation records in the database are updated based on the final service recommendation sequence, completing the recommendation process.

[0035] Example 2 This embodiment also discloses a computer device, including a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the method described in Embodiment 1.

[0036] Example 3 This embodiment also discloses a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the steps of the method described in Embodiment 1.

[0037] The above are merely preferred embodiments of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method of analyzing travel behavior based on user behavior, characterized by, The method comprises the following steps: Collecting multi-dimensional spatio-temporal data of the user during the trip, extracting dynamic change characteristics, and obtaining a preliminary behavior sequence; According to the preliminary behavior sequence, obtaining a trip adjustment record, if the adjustment frequency is higher than a preset threshold, applying a clustering algorithm to group similar adjustment modes, and determining an adjustment clustering set; Through the adjustment clustering set, obtaining associated transportation tool selection data, integrating the data, and obtaining an enhanced behavior sequence; Filling in missing values of the enhanced behavior sequence to obtain a complete behavior trajectory; Through the complete behavior trajectory, extracting travel rule characteristics from the user preferences, classifying, and determining a preference classification group; According to the preference classification group, obtaining resource configuration related data, and adjusting configuration parameters based on configuration matching degree to obtain an optimized resource scheme; Through the optimized resource scheme, fusing personalized service demand, predicting subsequent travel rules, and determining a final travel service recommendation.

2. The method of claim 1, wherein collecting multi-dimensional spatio-temporal data of the user during the trip, extracting dynamic change characteristics, and obtaining a preliminary behavior sequence comprises: obtaining multi-dimensional spatio-temporal data of the user during the trip through sensors and positioning devices, and storing as a structured data set; performing spatio-temporal data fusion on the structured data set to generate a unified feature representation; if the completeness of the unified feature representation is higher than a preset threshold, using a long short-term memory network to extract dynamic characteristics from the unified feature representation to obtain dynamic behavior characteristics; performing sequence model analysis on the dynamic behavior characteristics through a hidden Markov model to generate a preliminary behavior sequence.

3. The method of claim 1, wherein according to the preliminary behavior sequence, obtaining a trip adjustment record, if the adjustment frequency is higher than a preset threshold, applying a clustering algorithm to group similar adjustment modes, and determining an adjustment clustering set comprises: obtaining a trip adjustment record, using unified formatting processing to obtain a standardized adjustment data set; calculating the adjustment frequency of each record through the standardized adjustment data set to obtain a frequency distribution result; if the frequency distribution result is higher than a preset threshold, applying a K-means clustering algorithm to the high-frequency adjustment record to group similar adjustment modes to obtain a preliminary clustering set; for the preliminary clustering set, using a silhouette coefficient to verify the clustering quality to obtain an optimized adjustment clustering set.

4. The method of claim 1, wherein through the adjustment clustering set, obtaining associated transportation tool selection data, integrating the data, and obtaining an enhanced behavior sequence comprises: obtaining transportation tool selection data from the adjustment clustering set; using a K-means algorithm to cluster the transportation tool selection data to obtain classified behavior clusters; extracting spatio-temporal information from the classified behavior clusters to generate feature vectors containing time and location; if the spatio-temporal information of the feature vector is complete, integrating the behavior clusters and the spatio-temporal information through a weighted fusion mechanism to obtain a fusion data set; if the spatio-temporal information is missing, supplementing the missing information from a preset database to obtain a fusion data set; according to the fusion data set, using a hidden Markov model to analyze the transition probability of the behavior sequence to generate an enhanced behavior sequence.

5. The method of claim 1, wherein The missing value filling is performed on the enhanced behavior sequence to obtain a complete behavior trajectory, including: detecting whether the enhanced behavior sequence has an incoherent segment to obtain a segment detection result; if the incoherent segment is detected, inferring the missing information by using a sequence model to obtain inferred spatio-temporal information; generating continuous behavior sequence data according to the inferred spatio-temporal information and in combination with a time correlation attribute to obtain a preliminary complete trajectory; verifying the preliminary complete trajectory by using a data continuity attribute to determine whether there is inconsistency in time or space to obtain a verification result; if the verification result shows inconsistency, re-infer the spatio-temporal information of the inconsistent segment by using the sequence model to obtain modified spatio-temporal information; generating complete trajectory data by using the modified spatio-temporal information and in combination with sequence analysis.

6. The method of claim 1, wherein: from the complete behavior trajectory, travel regularity features are extracted from user preferences and classified to determine a preference classification group, including: obtaining the complete behavior trajectory, extracting time series features and spatial position features, calculating travel frequency and regularity indexes by using a time window segmentation method to obtain a user travel regularity feature set; if the time series features of the travel regularity feature set satisfy a preset periodicity threshold, classifying the feature set by using a K-means clustering algorithm to determine a preliminary user preference classification group; obtaining the preliminary user preference classification group, performing dimension reduction processing on the feature set by using a principal component analysis method to obtain a low-dimensional feature representation; according to the low-dimensional feature representation, optimizing the classification group by using a DBSCAN clustering algorithm to determine boundary points of the classification group to obtain an optimized user preference classification group; extracting classification labels from the optimized user preference classification group, and performing behavior pattern recognition in combination with spatial position information to obtain a user travel behavior pattern description; for the user travel behavior pattern description, calculating a classification accuracy rate by using a preset evaluation index to obtain a final user preference classification result.

7. The method of claim 1, wherein: according to the preference classification group, obtaining resource configuration related data, and adjusting configuration parameters based on a configuration matching degree to obtain an optimized resource scheme, including: obtaining resource configuration data of the preference classification group, extracting user preference features and resource allocation records from a database to obtain an initial resource configuration data set; for the initial resource configuration data set, analyzing feature vectors of the user preference classification group by using a preference feature extraction method to obtain a preference feature matrix; according to the preference feature matrix, calculating a matching degree of the resource configuration data and the preference classification group by using a configuration matching algorithm to obtain a matching degree score; if the matching degree score is lower than a preset threshold, determining resource configuration items that need to be adjusted by using threshold comparison logic to obtain a to-be-optimized configuration list; for the to-be-optimized configuration list, updating a resource allocation strategy by using a parameter adjustment rule to generate a temporary optimized resource scheme; according to the temporary optimized resource scheme, recalculating the matching degree score by using the resource allocation strategy to determine whether a scheme optimization target is reached to obtain a final optimized resource scheme.

8. The method of claim 1, wherein: By optimizing the resource scheme, combining personalized service needs, predicting subsequent travel rules, and determining the final travel service recommendation, including: Obtaining user historical travel records and preference labels, extracting relevant data from the database, and obtaining an initial service demand data set; Processing the initial service demand data set through a time series analysis method to extract the time series characteristics of user travel behavior and obtain a travel rule feature vector; Using a long short-term memory network model, predicting subsequent travel behavior sequences based on the travel rule feature vector, and obtaining a predicted travel sequence; If the matching degree of the predicted travel sequence and the preference label is lower than a preset threshold, adjusting the recommendation parameters through comparison logic to obtain a service sequence to be optimized; For the service sequence to be optimized, update the service recommendation strategy using a preset optimization rule to generate a temporary service recommendation sequence; Re-calculate the matching degree with the preference label through the temporary service recommendation sequence to determine whether the recommendation target is met to obtain a final service recommendation sequence; Update the service recommendation records in the database according to the final service recommendation sequence to complete the travel recommendation process.

9. A computer apparatus comprising a memory, a processor, and a computer program stored on the memory, wherein the computer program, when executed by the processor, causes the processor to perform the method of any one of claims 1 to 8. The processor executes the computer program to implement the steps of the method of any one of claims 1-8.

10. A computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to implement the steps of the method of any one of claims 1-8.

Images