Enterprise business travel customization planning method and system based on big data

Abstract

The invention discloses an enterprise business travel customization planning method and system based on big data, and relates to the technical field of big data. The method comprises the following steps: collecting customer business travel data, extracting behavior intention, historical reservation and browsing data, and processing behavior intention information to obtain a key category; performing data feature extraction and clustering analysis, determining an optimal clustering number, and analyzing group business travel preferences; establishing a dynamic law change diagram by taking a timestamp as a benchmark, displaying a consumption mode by using a visual tool, and performing comparative analysis according to a period; calculating business travel item relevance, and setting a threshold value to judge strong and weak relevance; the real-time data and the historical data are combined, the weight is adjusted according to the value score, and the historical data are updated regularly; based on historical and real-time data of users, personalized business travel products are recommended through similarity calculation, recommendation results are optimized, data-driven customized planning and personalized recommendation schemes are provided for business travel services of enterprises, and multi-dimensional analysis and algorithm application are fused.

Description

Technical Field

[0001] The present invention belongs to the technical field of big data, and particularly relates to a method and system for enterprise business travel customization planning based on big data. Background Art

[0002] In today's rapidly changing business environment, enterprise business travel planning and management are becoming increasingly complex and crucial. Traditional methods rely on manual experience and static data, making it difficult to capture the subtle changes in market dynamics and customer needs.

[0003] With the development of big data technology, business travel customization planning methods based on big data have emerged, aiming to provide personalized, efficient and accurate solutions. The diversity and complexity of enterprise business travel activities require a higher level of planning. There are significant differences in business travel needs among different enterprises, departments and employees, and traditional methods are difficult to cope with. Big data technology can analyze a large amount of customer data, reveal patterns of needs and preferences, and provide a scientific basis for planning. The changing market environment also requires business travel planning to be more flexible and adaptable. Business travel costs fluctuate due to various factors, and customer needs change with market trends. Big data technology can monitor market dynamics in real time, help enterprises adjust strategies and respond to changes. In addition, personalized service has become an important trend in modern business travel services. Traditional standardized services are difficult to meet customer needs, and big data technology can mine customers' historical behaviors and preferences to provide customized solutions, improving customer experience and satisfaction.

[0004] In summary, a method for enterprise business travel customization planning based on big data is introduced to address the diversification, complexity of business travel activities and market changes, and meet the needs of personalized services. By deeply mining customer data, it provides accurate, efficient and personalized business travel solutions for enterprises, improving the level of business travel management and market competitiveness. Summary of the Invention

[0005] To overcome the above-mentioned drawbacks and deficiencies of the prior art, the first object of the present invention is to provide a method for enterprise business travel customization planning based on big data; the second object of the present invention is to provide a system for enterprise business travel customization planning based on big data.

[0006] The first object of the present invention adopts the following technical solution:

[0007] A method for enterprise business travel customization planning based on big data has the following process:

[0008] Step 1: Collect customer data related to business travel, preprocess the data, and extract the behavioral intention information, historical reservation information and historical browsing big data of the target user within a preset time period;

[0009] Step 2: Extract features and perform clustering analysis on the preprocessed data. Meanwhile, perform frequent behavior sequence mining on the processed behavior intention information, and output the frequent behavior intention sequences and their support degrees;

[0010] Step 3: Establish a dynamic pattern change graph, and rearrange and analyze the customer's consumption records and operation data based on the time stamp;

[0011] Step 4: Calculate the correlation between business travel items, and identify business travel item combinations with strong correlations;

[0012] Step 5: Merge the real-time standard data set with the customer's historical data to form new historical data, and adjust the weight ratio according to the value score of the data to conduct business travel customization planning;

[0013] Step 6: Use the collaborative filtering recommendation algorithm, combined with time series analysis, to recommend personalized business travel plans for customers.

[0014] Preferably, data processing also includes preprocessing the behavior intention information, eliminating noise and outliers to obtain behavior intention data, and removing data with itinerary time conflicts and price anomalies;

[0015] Process the behavior intention information to obtain key categories, and the key categories include business meeting categories, training categories, and inspection categories. The segmentation process is as follows: Sort the behavior intention data in chronological order to obtain a primary behavior intention sequence. Each behavior intention node of the user has its corresponding position in the sequence, and the behavior intention sequence represents the user's behavior intention within a certain time stamp; Sort the behavior intention data in chronological order to obtain a primary behavior intention sequence Y. Each behavior intention node of the user has its corresponding position in the sequence, and the behavior intention sequence represents the user's behavior intention within a certain time stamp;

[0016] The set of primary behavior intention sequences Y = (y1, y2, y3, y4,..., ym), where m is a positive integer. Create a corresponding point for each behavior intention ym in the sequence, and add the corresponding behavior intention identifier, time stamp, and page ID to retain the context information of each point. Sort the created points using the time stamp to obtain the set of intermediate behavior intention sequences X = (x1, x2, x3, x4,..., xm), where m is a positive integer;

[0017] Take any point xm in the space as the center of a circle, set the radius as r to form a circular area, and mark the set of all points within this circular area as the neighborhood B r (xm), B rN(xm) = {xn ∈ D │ dist(xm, xn) ≤ r}; where dist(xm, xn) represents the distance between xm and xn;

[0018] Mark the minimum value of the number of samples in the neighborhood as MinPts;

[0019] Randomly select a point xm from the set X, and determine whether |N r (xm)| is greater than or equal to MinPts. When |N r (xm)| ≥ MinPts, then determine that xm is a seed point and add it to the seed set Z;

[0020] Randomly select a seed point xn from the seed set Z, and add all the points that are density-reachable from it to a new set C1 to form the first key category. The definition of density-reachability is as follows: If xn is in the neighborhood of xm and xm is a seed point, then xn is directly density-reachable from xm. If there exist a1, a2,..., a n , where a1 = xm, a n = xn, and a i+1 is directly density-reachable from a i , then xn is density-reachable from xm;

[0021] Continue to visit the next point in the set X, repeat the above steps until all points in the dataset are processed, obtain the key categories, mark the points not included in the key categories as noise and delete them; The obtained key categories include business meeting categories, training categories, and inspection categories.

[0022] Preferably, the data feature extraction and clustering analysis steps include: performing clustering analysis using the K-means algorithm, calculating the sum of squared errors SSE under different K values and plotting the curve of SSE changing with the K value, and determining the optimal number of clusters K according to the elbow method. The calculation formula of SSE is where K is the number of clusters; C i is the i-th cluster; x is the data point of the cluster center; μ i is the centroid of the i-th cluster; Then, plot the curve of SSE changing with the K value, find the inflection point on the curve, that is, the point where the sum of squared errors begins to decrease slowly, as the optimal K value; Describe the business travel preferences of each group according to the clustering results, analyze the attribute values of the cluster center points, and identify the destinations frequently visited by different groups and the preferred accommodation standards.

[0023] Preferably, the steps for establishing and analyzing the dynamic pattern change graph include: rearranging the consumption records and operation data of customers based on timestamps, and establishing a dynamic pattern change graph over a time span; presenting the graph using a visualization tool to display the changes in the consumption patterns of customers at different time periods, analyzing the business travel characteristics and preference changes of customers at different time periods, and marking the peak and trough periods; setting a time period T, conducting a comparative analysis by intervals, comparing the differences in peak periods, trough periods, and business travel characteristics in different time intervals, and analyzing their impacts on customer behavior and market trends.

[0024] Preferably, the steps for calculating the relevance of business travel items include: for the real-time standard data set and historical data of each customer, calculating the relevance of the previous and subsequent business travel items after sorting by time; applying the FP-Growth algorithm to mine frequent item sets and generate association rules, and analyzing the item relevance between adjacent two business travel records; setting a relevance threshold, and judging the association strength according to support, confidence, and lift, where the calculation formula for support is where σ(X∪Y) is the number of transactions containing item sets X and Y; N is the total number of transactions, and the calculation formula for confidence is where σ(X) is the number of transactions containing item set X, and the calculation formula for lift is It is specified that support greater than 0.5 and confidence greater than 0.7 are used as the criteria for judging strong associations.

[0025] Preferably, the steps for combining real-time data and historical data include: obtaining the real-time standard data set and the historical data of customers, and merging the two to form new historical data; calculating a value score according to the time value, commodity association value, and continuity of the data, and the calculation formula is Value = a1×Time + a2×Association + a3×Continuity; where Time is the time value of the data; Association is the commodity association value contained in the data, Continuity is the continuity, and the weights of the time value, commodity association value, and continuity of the data in the value score are a1, a2, and a3 respectively, and the weight of valueless data in the original historical data is 0; adjusting the weight ratio according to the value score, regularly updating the historical data set, adding the latest transaction records, and recalculating the value score for business travel customization planning.

[0026] Preferably, the steps of the collaborative filtering recommendation algorithm include: based on the historical business travel records and real-time behavior data of users, using collaborative filtering technology to predict the business travel products that users may be interested in; finding similar user groups by calculating the similarity between users, and recommending business travel products for target users according to the historical behaviors of similar users, where the calculation formula for cosine similarity is where A and B are the behavior vectors of two users; n is the vector dimension.

[0027] Preferably, the combined time series analysis step includes: performing combined time series analysis, using the ARIMA model to analyze the user's historical consumption data; understanding the changing trend of user behavior over time, and further optimizing the recommendation results based on this trend.

[0028] The second object of the present invention adopts the following technical solutions:

[0029] A corporate business travel customization planning system based on big data, used to implement a corporate business travel customization planning method based on big data. The system includes:

[0030] A data collection and processing module: responsible for collecting business travel-related data from multiple sources and preprocessing the data;

[0031] A customer group division module: based on the preprocessed data, using a clustering algorithm to divide customers into different groups and conduct a detailed feature description for each group;

[0032] A dynamic law analysis module: taking the timestamp as a reference, rearranging the consumption records and operation data of customers, establishing a dynamic law change graph, and performing data analysis;

[0033] A business travel project correlation calculation module: calculating the correlation between the business travel projects of customers and identifying strongly correlated business travel project combinations;

[0034] A data merging and business travel customization planning module: merging the real-time standard data set with the customer's historical data to form new historical data, and adjusting the weight ratio according to the value score of the data to perform business travel customization planning;

[0035] A personalized recommendation module: applying a collaborative filtering recommendation algorithm and time series analysis technology to recommend personalized business travel solutions for customers.

[0036] In summary, due to the adoption of the above technical solutions, the beneficial effects of the present invention are:

[0037] 1. By performing clustering analysis on users, the present invention identifies the business travel preferences of different groups (such as high-star hotels and frequent business trips, or economy accommodation and low-frequency business trips), and combines a collaborative filtering recommendation algorithm (such as cosine similarity) to find similar user groups, providing customized services for users. In addition, the ARIMA model is used to predict future consumption trends, making the recommendation more in line with the changing needs of users.

[0038] 2. By using the dynamically changing pattern graph and interval comparison analysis in the present invention, enterprises can clearly observe the changes in customers' consumption patterns, peak periods, and off-peak periods, so as to flexibly adjust their business travel strategies. By calculating the relevance of business travel projects and setting clear thresholds and metrics, resource allocation is optimized. The combination of real-time data and historical data and the weight adjustment mechanism ensure that decisions are based on the latest and most valuable data, enhancing service competitiveness.

[0039] 3. By regularly updating the historical data set and recalculating the value scores in the present invention, the customer preference model is continuously updated. The processing of behavioral intention information includes preprocessing, region segmentation, and frequent behavior sequence mining, which helps enterprises deeply understand users' behavior patterns. Continuously improve the service according to these patterns and user feedback. For example, timely provide special ticket recommendations to customers who prefer specific regions, realizing the dynamic adjustment and optimization of the service. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0041] Figure 1 Shows the flowchart of the enterprise business travel customization planning method based on big data of the present invention;

[0042] Figure 2 Shows the module diagram of the enterprise business travel customization planning system based on big data of the present invention;

[0043] Figure 3 Shows the flowchart of the second step of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0044] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, rather than all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts fall within the protection scope of the present invention.

[0045] In addition, the described features, structures, or characteristics may be combined in any suitable manner in one or more example embodiments. In the following description, numerous specific details are provided to give a thorough understanding of the example embodiments of the present disclosure. However, those skilled in the art will realize that one or more of the specific details may be omitted in practicing the technical solutions of the present disclosure, or other methods, components, steps, etc. may be adopted. In other cases, well-known structures, methods, implementations, or operations are not shown or described in detail to avoid obscuring various aspects of the present disclosure.

[0046] Example 1:

[0047] Refer to Figure 1 As shown, the method for enterprise business travel customized planning based on big data in this embodiment is as follows:

[0048] Step 1: Collect customer data related to business travel, preprocess the collected data, and at the same time extract the behavioral intention information, historical booking information, and historical browsing big data of the candidate enterprise business travel behavior data generated by the target user within a preset time period.

[0049] Collect business travel-related data from sources such as online booking platforms, customer feedback forms, and social media, including but not limited to consumption amounts (such as air ticket costs, hotel costs), consumption frequencies (number of trips), and commodity types (destinations, accommodation standards, transportation methods).

[0050] For example, obtain air ticket costs from the airline reservation system, obtain accommodation costs from the hotel management system, and obtain feedback scores from customer satisfaction surveys.

[0051] Use web analysis tools (such as Google Analytics) to track the online behavior of customers on the online booking platform, record the clickstream data of customers, and the operation data includes the data clicked and browsed by customers, the monitored accessed pages, and the browsing duration. For example, the time a user stays on the booking platform, the number of pages viewed, etc.

[0052] At the same time, extract the behavioral intention information, including itinerary arrangement information (such as travel dates, itinerary durations), service selection information (such as transportation method selection, hotel class selection), and travel scenario information (such as business travel, tourism travel); the historical booking information includes booking service information (such as booked hotels, flights, car rentals, etc.), booking frequency information, and booking service detail information (such as room types, flight cabins). The timestamp is the time information corresponding to the customer's business travel activities (such as booking time, travel time), ensuring that all collected data is attached with accurate timestamps.

[0053] For example, each booking operation generates a timestamp marking the exact time when the operation occurs.

[0054] Ensure that all data processing activities comply with relevant privacy regulations (such as GDPR), and take necessary encryption measures to protect personal information. Clearly inform users of the purpose of data use and obtain their consent.

[0055] Step 2: Data feature extraction and clustering analysis. At the same time, after processing the behavioral intention information, perform frequent behavior sequence mining to output the frequent behavioral intention sequences and their support degrees.

[0056] Refer to Figure 3 As shown below, the specific process is as follows:

[0057] S21. Use the K-means algorithm for clustering analysis, and determine the optimal number of clusters K using the elbow method.

[0058] First, calculate the sum of squared errors (SSE) of clustering under different K values, and then plot the curve of the change of SSE with the K value. Find the inflection point on the curve, that is, the point where the sum of squared errors begins to decrease slowly, as the optimal K value.

[0059] The calculation formula of SSE is as follows: where K is the number of clusters; C i is the i-th cluster; x is the data point of the cluster center; μ i is the centroid of the i-th cluster; then, plot the curve of SSE changing with the K value, and find the inflection point on the curve, that is, the point where the sum of squared errors begins to decrease slowly, as the optimal K value.

[0060] Based on the clustering results, describe each group in detail and identify the business travel preferences of different groups. (Such as frequently visited destinations, preferred accommodation standards, etc.). Analyze the attribute values of the center point of each cluster. For example, some groups may be more inclined to high-star hotels, while others prefer budget accommodation.

[0061] S22. Process the behavioral intention information.

[0062] S201. Preprocess the behavioral intention information, clean and preprocess the data, eliminate noise and outliers, and obtain behavioral intention data through the above processing. Clean the data of various behavioral intention information in the business travel behavior of users, remove unreasonable data, such as data with conflicting travel times, abnormal service selections (such as data with extremely low or high prices), and obtain behavioral intention data.

[0063] S202. Process the behavioral intention information to obtain key categories including business meeting category, training category, and inspection category;

[0064] Specifically, the behavioral intention data is sorted in chronological order to obtain a primary behavioral intention sequence. Each behavioral intention node of the user has its corresponding position in the sequence, and the behavioral intention sequence represents the user's behavioral intention within a certain time stamp; the behavioral intention data is sorted in chronological order to obtain a primary behavioral intention sequence Y. Each behavioral intention node of the user has its corresponding position in the sequence, and the behavioral intention sequence represents the user's behavioral intention within a certain time stamp;

[0065] The set of primary behavioral intention sequences Y = (y1, y2, y3, y4,..., ym), where m is a positive integer. For each behavioral intention ym in the sequence, a corresponding point is created, and the corresponding behavioral intention identifier, time stamp, and page ID are added to retain the context information of each point. The created points are sorted using the time stamp to obtain the set of intermediate behavioral intention sequences X = (x1, x2, x3, x4,..., xm), where m is a positive integer;

[0066] Taking any point xm in the space as the center, a radius of r is set to form a circular area. The set of all points within this circular area is marked as the neighborhood B r (xm),

[0067] B r (xm) = {xn ∈ D │ dist(xm, xn) ≤ r};

[0068] where dist(xm, xn) represents the distance between xm and xn;

[0069] The minimum value of the number of samples in the neighborhood is marked as MinPts;

[0070] Randomly select a point xm from the set X and determine whether |B r (xm)| is greater than or equal to MinPts. When |B r (xm)| ≥ MinPts, then xm is determined to be a seed point and added to the seed set Z;

[0071] Randomly select a seed point xn from the seed set Z, and add all the points that are density-reachable from it to a new set C1 to form the first key category. The definition of density-reachability is as follows: if xn is within the neighborhood of xm and xm is a seed point, then xn is directly density-reachable from xm. If there exist a1, a2,..., a n , where a1 = xm, a n = xn, and a i+1 is directly density-reachable from a i , then xn is density-reachable from xm;

[0072] Continue to access the next point in set X, repeat the above steps until all points in the dataset are processed to obtain the key categories. Mark the points not included in the key categories as noise and delete them;

[0073] The key categories include business meeting category, training category, and inspection category.

[0074] S203. Use the Apriori algorithm to perform frequent behavior intention sequence mining on the key categories, output the frequent behavior intention sequences and their support degrees. Set the support degree threshold, eliminate the behavior intention sequences with support degrees lower than the threshold, retain the behavior intention sequences with support degrees higher than the threshold and their support degrees, and mark them as frequent behavior intention sequences. Here, the support degree represents the frequency of a certain user behavior intention sequence appearing in the dataset.

[0075] Step Three: Establishment and analysis of the dynamic law change graph.

[0076] First, rearrange the consumption records and operation data of customers based on the time stamp and establish a dynamic law change graph. For each business travel reservation consumption record and operation data on the reservation platform, arrange them in the order of time stamp, showing the continuity and variability of customers' business travel behaviors in the time dimension.

[0077] Next, use visualization tools to present the established dynamic law change graph. In the form of a chart, display the changes in the consumption patterns of customers in different time periods, draw a line chart of the total business travel consumption amount per month or per quarter, and determine the peak and trough periods of consumption according to the fluctuations in the consumption amount in the graph. At the same time, through in-depth analysis of the data in different time periods, summarize the characteristics and preference changes of customers in business travel, including that customers are more inclined to choose high-end hotels in a certain time period, while in other time periods, they pay more attention to transportation convenience and cost-effectiveness.

[0078] Finally, set the time period T and conduct interval comparison analysis. Divide the time period into different intervals by quarter or month, and compare the data in different intervals. The comparison content includes the differences in the occurrence time, duration, and intensity of the peak and trough periods, and also pay attention to the changes in business travel characteristics such as destination preferences and transportation mode selection preferences in different intervals. Analyze the impact of these differences on customer behaviors, including price fluctuations during peak periods causing customers to book in advance or postpone, or change the type of products booked. At the same time, study its impact on market trends, including the changes in the popularity of certain destinations in specific time periods affecting the supply and price strategies of related business travel products. Through this comprehensive analysis, it can provide an important basis for enterprises to adjust business travel customization planning strategies, enabling them to better adapt to market changes and meet customer needs. Step Four: Calculation of business travel project relevance.

[0079] S41. Association calculation.

[0080] For the real-time standard data set and historical data of each customer, after sorting by time, calculate the correlation between business travel items before and after. Analyze the item correlation between adjacent two business travel records.

[0081] Implementation details: Apply the FP-Growth algorithm to mine frequent item sets and generate association rules. For example, if it is found that a customer usually books the same chain hotel after booking a flight ticket, it can be considered that there is a strong association between these two items.

[0082] S42. Association judgment.

[0083] Set an association threshold. If the calculated association exceeds this threshold, it is considered that there is a strong association between two business travel items, otherwise it is a weak association. The association indicators include support, confidence, and lift. The formula for support is where σ(X∪Y) is the number of transactions containing item sets X and Y; N is the total number of transactions; the formula for confidence is where σ(X) is the number of transactions containing item set X; the formula for lift is

[0084] Define clear threshold criteria, such as support > 0.5, confidence > 0.7, etc., to filter out association rules with practical significance.

[0085] Example: After analysis, it is found that when a customer books a flight ticket to Shanghai, there is an 80% probability that they will choose to book a hotel of a specific brand. Therefore, corresponding business travel customization planning strategies can be set.

[0086] Step Five. Combine real-time data and historical data.

[0087] S51. Data merging and weight adjustment.

[0088] Obtain the real-time standard data set and customer historical data, merge the real-time standard data set with the original historical data to form new historical data, and at the same time adjust the weight ratio according to the value scores of the real-time standard data set and customer historical data.

[0089] The value score is comprehensively evaluated based on the time value of the data, the commodity association value contained in the data, and the continuity. The calculation formula is Value = 0.3×Time + 0.2×Association + 0.5×Continuity; where Time is the time value of the data; Association is the commodity association value contained in the data, and Continuity is the continuity. Adjust the weight ratio according to the value score. The proportion of the time value of the data in the value score is 0.3, the proportion of the commodity association value in the value score is 0.2, and the proportion of the continuity in the value score is 0.5. The weight of valueless data in the original historical data is 0. Among them, the time value of the data: newer data is usually considered to be more valuable because it can better reflect the current preferences of users. This indicator is quantified by calculating the time difference from the date when the data is generated to the present;

[0090] The commodity association value contained in the data: measures the association strength between different commodities or services involved in the data; for example, if a user often stays at the same chain hotel after booking a flight, it indicates a strong association between the flight and the hotel.

[0091] Continuity: refers to the consistency and repeatability of user behavior. High - continuity behavior means that users tend to repeat certain specific behavior patterns, which is very important for predicting future behavior.

[0092] For example: Develop a business travel plan for a customer who is based in Beijing but occasionally needs to go to Shanghai for meetings:

[0093] If it is found that this customer has chosen a certain airline and a specific brand of hotel several times when going to Shanghai recently, and this choice has lasted for some time (high continuity), then the value score of this piece of data will be very high.

[0094] If further analysis shows that after booking a ticket of this airline, this customer almost always books the same chain hotel (high commodity association value), then this piece of data will also get a relatively high value score.

[0095] Reservation behaviors that occurred in the recent few months are more important than those that occurred several years ago (time value).

[0096] S52. Business travel customization plan.

[0097] Generate a dynamic regular change graph based on the new historical data and calculate the business travel project relevance. Regularly update the historical data set, add the latest transaction records, and recalculate the value score for business travel customization planning.

[0098] Implementation details: Regularly update the historical dataset, add the latest transaction records, recalculate the value score, and conduct business travel customization planning. The value score is comprehensively evaluated and defined based on the time value of the data, the commodity association value contained in the data, and continuity. The value score takes into account factors such as the time freshness, commodity relevance, and continuity of the data to comprehensively evaluate the importance of the data.

[0099] Example: As new data is continuously added, the customer's preference model can be updated in a timely manner, and the business travel customization planning strategy can be adjusted according to the latest trends. For example, if it is found that the flight prices in a certain region have recently decreased, special ticket recommendations can be provided to customers who prefer that region.

[0100] The proportion of the time value of the data in the value score is 0.3.

[0101] The proportion of the commodity association value in the value score is 0.2.

[0102] The proportion of continuity in the value score is 0.5.

[0103] The weight of the worthless data in the original historical data is 0.

[0104] Step 6: Use the collaborative filtering recommendation algorithm to recommend personalized business travel solutions for customers.

[0105] S61. Collaborative filtering recommendation.

[0106] Based on the user's historical business travel records and real-time behavior data, use collaborative filtering technology to predict the business travel products that the user may be interested in. For example, by calculating the similarity between users (such as cosine similarity), find the user group similar to the target user, and recommend business travel products for the target user based on the historical behavior of the similar users.

[0107] The calculation formula of cosine similarity is

[0108] where A and B are the behavior vectors of two users; n is the vector dimension.

[0109] Example: For a customer who is based in Beijing but occasionally needs to go to Shanghai for meetings, the system can recommend cost-effective round-trip tickets and suitable hotel packages according to their historical records and the current market situation.

[0110] S62. Combine time series analysis.

[0111] Combine time series analysis to understand the changing trend of user behavior over time and further optimize the recommendation results. For example, use the ARIMA model to conduct time series analysis on the user's historical consumption data to predict future consumption trends, so as to more accurately recommend business travel products for users.

[0112] The beneficial effects of this embodiment are as follows: Through big data analysis, a profound understanding of customers' business travel behaviors is achieved, which not only improves the quality of personalized services but also accurately predicts customers' future needs. Cluster analysis is used to identify the preferences of different user groups, a dynamic law change graph is used to capture changes in consumption patterns, and the relevance of product recommendations is enhanced through association rule mining.

[0113] Embodiment 2:

[0114] Refer to Figure 2 As shown, the big data-based enterprise business travel customization planning system of this embodiment includes a data collection and processing module, a customer group division module, a dynamic law analysis module, a business travel project relevance calculation module, a data merging and business travel customization planning module, and a personalized recommendation module.

[0115] Data collection and processing module: Responsible for collecting business travel-related data from multiple sources and preprocessing the data to ensure the quality and usability of the data.

[0116] Customer group division module: Based on the preprocessed data, use clustering algorithms to divide customers into different groups and conduct detailed feature descriptions for each group.

[0117] Dynamic law analysis module: Based on timestamps, rearrange customers' consumption records and operation data, establish a dynamic law change graph, and conduct data analysis.

[0118] Business travel project relevance calculation module: Calculate the relevance between customers' business travel projects and identify strongly relevant business travel project combinations.

[0119] Data merging and business travel customization planning module: Merge the real-time standard data set with customers' historical data to form new historical data, adjust the weight ratio according to the value of the data, and conduct business travel customization planning.

[0120] Personalized recommendation module: Use collaborative filtering recommendation algorithms and time series analysis techniques to recommend personalized business travel solutions for customers.

[0121] The beneficial effects of this embodiment are as follows: The full process from data collection to personalized recommendation is automated, effectively improving the efficiency and accuracy of business travel planning. The system can deeply explore customers' needs, provide customized services, enhance the customer experience, and bring higher business travel management benefits to the enterprise.

[0122] As described above, only the preferred specific embodiments of the present invention are provided, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention, according to the technical solution and inventive concept of the present invention, making equivalent substitutions or changes, shall be covered by the protection scope of the present invention.

[0123] The preferred embodiments of the present invention disclosed above are only used to help explain the present invention. The preferred embodiments do not describe all the details in detail, nor do they limit the present invention to only the specific embodiments. Obviously, many modifications and variations can be made according to the content of this specification. These embodiments are selected and specifically described in this specification to better explain the principle and practical application of the present invention, so that those skilled in the art can well understand and utilize the present invention. The present invention is only limited by the claims and their full scope and equivalents.

Claims

1. A method for customizing business travel planning for enterprises based on big data, characterized in that, The method process is as follows: Step 1: Collect customer data related to business travel, preprocess the data, and extract the behavioral intention information, historical reservation information, and historical browsing big data of the target users within a preset time period; Step 2: Perform feature extraction and clustering analysis on the preprocessed data, and at the same time perform frequent behavior sequence mining processing on the behavioral intention information after processing, and output the frequent behavior intention sequences and their support degrees; Step 3: Establish a dynamic rule change graph, and rearrange and analyze the consumption records and operation data of customers based on the time stamp; Step 4: Calculate the correlation between business travel items, and identify the business travel item combinations with strong correlations; Step 5: Merge the real-time standard data set with the customer historical data to form new historical data, and adjust the weight ratio according to the value score of the data to perform business travel customization planning; Step 6: Use the collaborative filtering recommendation algorithm, combined with time series analysis, to recommend personalized business travel solutions for customers.

2. The method for customizing and planning enterprise business travel based on big data according to claim 1, wherein In Step 1, the data processing also includes preprocessing the behavioral intention information, eliminating noise and outliers to obtain behavioral intention data, and removing data with itinerary time conflicts and price anomalies; Process the behavioral intention information to obtain key categories, and the key categories include business meeting categories, training categories, and inspection categories. The segmentation process is as follows: Sort the behavioral intention data in chronological order to obtain the primary behavioral intention sequence. Each behavioral intention node of the user has its corresponding position in the sequence, and the behavioral intention sequence represents the behavioral intention of the user at a certain time stamp; Sort the behavioral intention data in chronological order to obtain the primary behavioral intention sequence Y. Each behavioral intention node of the user has its corresponding position in the sequence, and the behavioral intention sequence represents the behavioral intention of the user at a certain time stamp; The set of primary behavior intention sequences \(Y=(y_1,y_2,y_3,y_4,\cdots,y_m)\), where \(m\) is a positive integer. For each behavior intention \(y_m\) in the sequence, a corresponding point is created, and the corresponding behavior intention identifier, timestamp, and page ID are added to retain the context information of each point. The created points are sorted using the timestamp to obtain the set of intermediate behavior intention sequences \(X=(x_1,x_2,x_3,x_4,\cdots,x_m)\), where \(m\) is a positive integer; Taking any point \(x_m\) in the space as the center and setting the radius as \(r\) to form a circular region, the set of all points within this circular region is marked as the neighborhood \(B\). r (x_m), B r (x_m)=\{x_n\in D|\text{dist}(x_m,x_n)\leq r\}; where \(\text{dist}(x_m,x_n)\) represents the distance between \(x_m\) and \(x_n\); Mark the minimum value of the number of samples in the neighborhood as MinPts; Randomly select a point \(x_m\) from the set \(X\) and judge whether \(|B r (x_m)|\) is greater than or equal to MinPts. When \(|B r (x_m)|\geq\text{MinPts}\), then determine that \(x_m\) is a seed point and add it to the seed set \(Z\). Randomly select a seed point xn from the seed set Z, and add all the points that are density-reachable from it to a new set C1 to form the first key category. Among them, the definition of density-reachability is as follows: If xn is within the neighborhood of xm and xm is a seed point, then xn is directly density-reachable from xm. If there exist a1, a2,..., a n , where a1 = xm, a n = xn, and a i+1 is directly density-reachable from a i , then xn is density-reachable from xm; Continue to access the next point in set X, and repeat the above steps until all points in the data set are processed to obtain the key categories. Mark the points not included in the key categories as noise and delete them; The key categories obtained include business meeting categories, training categories, and inspection categories.

3. The method for customizing and planning enterprise business travel based on big data according to claim 1, characterized in that The data feature extraction and clustering analysis steps include: performing clustering analysis using the K-means algorithm, calculating the sum of squared errors (SSE) of clustering under different K values and plotting the curve of SSE varying with the K value, and determining the optimal number of clusters K according to the elbow method. The calculation formula of SSE is where K is the number of clusters; C i is the i-th cluster; x is the data point of the cluster center; μ i is the centroid of the i-th cluster. Then, plot the curve of SSE varying with the K value, find the inflection point on the curve, i.e., the point where the sum of squared errors begins to decrease slowly, as the optimal K value. Describe the business travel preferences of each group according to the clustering results, analyze the attribute values of the cluster centers, and identify the destinations frequently visited by different groups and the preferred accommodation standards.

4. The method for customizing and planning enterprise business travel based on big data according to claim 1, wherein The steps for establishing and analyzing the dynamic pattern change graph include: First, rearrange the consumption records and operation data of customers based on timestamps and establish a dynamic pattern change graph. For each business travel reservation consumption record and operation data on the reservation platform, arrange them in the order of timestamps, presenting the continuity and variability of customers' business travel behaviors in the time dimension. Then, use visualization tools to present the established dynamic pattern change graph. In the form of charts, show the changes in customers' consumption patterns in different time periods, draw a line graph of the total business travel consumption amount per month or per quarter, and determine the peak and trough periods of consumption according to the fluctuations in the consumption amount in the graph. At the same time, through in-depth analysis of data in different time periods, summarize the characteristics and preference changes of customers in business travel, including that customers are more inclined to choose high-end hotels in a certain time period, while in other time periods, they pay more attention to transportation convenience and cost-effectiveness. Finally, set a time period T and conduct interval comparison analysis. By dividing the time period into different intervals by quarter or month, compare the data in different intervals. The comparison content includes the occurrence time, duration, and intensity differences of the peak and trough periods, and also pay attention to the changes in business travel characteristics such as destination preferences and transportation mode selection preferences in different intervals, and analyze the impact of these differences on customers' behaviors, including price fluctuations during the peak period causing customers to book in advance or postpone, or change the type of products booked.

5. The method for customized business travel planning of enterprises based on big data according to claim 1, characterized in that The steps for calculating the relevance of business travel items include: For the real-time standard data set and historical data of each customer, calculate the relevance of the previous and subsequent business travel items after sorting by time; Apply the FP-Growth algorithm to mine frequent item sets and generate association rules, and analyze the item relevance between adjacent two business travel records; Set a relevance threshold, and judge the association strength according to support, confidence, and lift; It is clear that the support greater than 0.5 and the confidence greater than 0.7 are used as the criteria for judging strong associations.

6. The method for customizing and planning enterprise business travel based on big data according to claim 1, wherein The steps for combining real-time data and historical data include: Obtain the real-time standard data set and customer historical data, and merge the two to form new historical data; Calculate the value score according to the time value, commodity association value, and continuity of the data. The calculation formula is Value = a1×Time + a2×Association + a3×Continuity; where, Time is the time value of the data; Association is the commodity association value contained in the data, Continuity is the continuity, and the weights of the time value, commodity association value, and continuity of the data in the value score are a1, a2, and a3 respectively, and the weight of the valueless data in the original historical data is 0; Adjust the weight ratio according to the value score, regularly update the historical data set, add the latest transaction records, and recalculate the value score for business travel customization planning; Among them, the time value of the data: This indicator is quantified by calculating the time difference between the data generation date and the present. Updated data is considered to be of greater reference value and can better reflect the current preferences of users; Time = e -λt ; where t is the time difference between the data generation time and the current time; λ is the decay rate parameter; The commodity association value contained in the data: Measures the association strength between different commodities or services involved in the data; When a user often stays in the same chain hotel after booking a flight, it indicates a strong association between the flight and the hotel; Continuity: It refers to the repeatability of user behavior. High continuity behavior means that users tend to repeat certain specific behavior patterns; Among them, F i represents the repetition frequency of the i-th specific behavior pattern, that is, the number of times this behavior occurs; R i represents the relative importance weight of this behavior pattern among all behaviors, T i represents the total number of user behaviors during the observation period; W i is the time decay factor, used to measure whether the importance of a specific behavior will decrease over time, where t i is the time difference from the current time to the time when the behavior occurred, and λ is the decay rate; the relative importance weight R i is obtained as follows: Determine each behavior value index, including consumption amount, contribution to enterprise profit, and promotion of brand image, and perform normalization processing on different behavior value indexes; for index j, its normalized value where x ij is the original value of index j of behavior pattern i, min(x j ) and max(x j ) are the minimum and maximum values of index j among all behavior patterns respectively; calculate the comprehensive value score of each behavior pattern Among them, a j is the weight of index j. Finally, normalize the comprehensive value scores of all behavior patterns to obtain the relative importance weight where m is the total number of behavior patterns.

7. The method for customized business travel planning of enterprises based on big data according to claim 1, wherein, The steps of the collaborative filtering recommendation algorithm include: based on the historical business travel records and real-time behavior data of users, using collaborative filtering technology to predict business travel products that users may be interested in; finding similar user groups by calculating the similarity between users, and recommending business travel products for target users according to the historical behaviors of similar users. The calculation formula of cosine similarity is where A and B are the behavior vectors of two users; n is the vector dimension.

8. The method for customizing and planning enterprise business travel based on big data according to claim 1, wherein The combined time series analysis step includes: combining time series analysis, using the ARIMA model to analyze the user's historical consumption data; understanding the changing trend of user behavior over time, and further optimizing the recommendation results based on this trend.

9. An enterprise business travel customization planning system based on big data, which is used to implement the enterprise business travel customization planning method based on big data as described in claim 1, is characterized in that, The system includes: Data acquisition and processing module: responsible for collecting business travel-related data from multiple sources and preprocessing the data; Customer group division module: based on the preprocessed data, using clustering algorithms to divide customers into different groups and describing the detailed characteristics of each group; Dynamic law analysis module: taking the timestamp as the benchmark, rearranging the customer's consumption records and operation data, establishing a dynamic law change graph, and performing data analysis; Business travel project relevance calculation module: calculating the relevance between the customer's business travel projects and identifying strong relevant business travel project combinations; Data merging and business travel customization planning module: merging the real-time standard data set with the customer's historical data to form new historical data, adjusting the weight ratio according to the value score of the data, and performing business travel customization planning; Personalized recommendation module: using collaborative filtering recommendation algorithms and time series analysis techniques to recommend personalized business travel solutions for customers.

Images