An AI-based self-service ticket-purchasing human-computer interaction method and system

Abstract

The present invention relates to the technical field of data processing. The present invention relates to a self-service ticket-purchasing human-computer interaction method and system based on artificial intelligence, including: obtaining the ticket-purchasing data of users within the current time period, and using the local outlier factor algorithm to perform anomaly detection on it. The detection results include suspected abnormal data and suspected normal data, and the k-nearest neighbor value in the algorithm is the initial k value. Calculate the evaluation index of the suspected normal data, and this index is positively correlated with the classification effectiveness and data similarity. When the evaluation index is lower than the threshold, update the k value, and the update rule is adjusted according to the relationship between the current and historical abnormal data ratios until the index is higher than the threshold, and finally determine the optimal value. Finally, re-detect the user's ticket-purchasing data based on the optimal value to identify and prevent abnormal ticket-purchasing behaviors. The present invention solves the problem of inaccurate evaluation results of the user's ticket-purchasing risk degree.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing. More specifically, the present invention relates to a self-service ticket-purchasing human-computer interaction method and system based on artificial intelligence. Background Art

[0002] Artificial intelligence relies on technologies such as natural language processing, machine learning, computer vision, intelligent dialogue systems, and dynamic pricing in self-service ticket-purchasing human-computer interaction, enabling the system to understand users' voice and text inputs, analyze users' behaviors, provide personalized recommendations and real-time pricing, and at the same time enhance security and user experience through identity verification and sentiment analysis. These technologies together optimize the intelligence and efficiency of the ticket-purchasing process. In a self-service ticket-purchasing system, artificial intelligence algorithms such as the Local Outlier Factor (lof) algorithm and the Isolation Forest algorithm are used to analyze ticket-purchasing behaviors and identify potential fraud, violations, or other abnormal behaviors in the risk assessment of users' ticket purchases. The realization of this function not only depends on advanced algorithms and technologies, but also requires comprehensive consideration of user experience and business requirements. The system needs to minimize the false alarm rate while accurately detecting abnormal behaviors to avoid interfering with the ticket-purchasing experience of normal users. This requires the system to have powerful data processing and analysis capabilities during the design and implementation process, and be able to flexibly adapt to different scenarios and requirements.

[0003] For example, when using the Local Outlier Factor algorithm to evaluate the risk degree of users' ticket purchases, since the Local Outlier Factor algorithm is an unsupervised outlier detection method and belongs to a density-based outlier detection method, when performing outlier detection, it often needs to set the k value. Therefore, when detecting outliers in the ticket-purchasing data of each user, it is greatly affected by the k value. When the k value is selected too large, the sensitivity of the Local Outlier Factor algorithm is low, and risk data may be misjudged as normal data; when the k value is selected too small, the sensitivity of the Local Outlier Factor algorithm is high, and normal data may be falsely alarmed, resulting in inaccurate evaluation results of the risk degree of users' ticket purchases. Summary of the Invention

[0004] To solve the problem of inaccurate evaluation results of the risk degree of users' ticket purchases proposed in the above background art, the present invention provides solutions in the following aspects.

[0005] In a first aspect, the present invention provides a self-service ticket-purchasing human-computer interaction method based on artificial intelligence, including:

[0006] Obtaining the ticket-purchasing data of different users within the current time period; performing outlier detection on the ticket-purchasing data of each user using the Local Outlier Factor algorithm to obtain suspected abnormal data and suspected normal data of each ticket-purchasing data; wherein, the k-nearest neighbor value in the Local Outlier Factor algorithm is the initial k value;

[0007] Calculate the evaluation index of suspected normal data, where the evaluation index is positively correlated with the classification effectiveness and data similarity; the classification effectiveness characterizes the difference change between the current suspected abnormal data and the abnormal data in the historical time period; in response to the evaluation index being lower than the set threshold, update the initial value until the updated evaluation index is higher than the set threshold, stop the update, and obtain the updated optimal value; among them, the update rule is: when the proportion of suspected abnormal data is lower than the proportion of abnormal data in the historical time period, use the product of the previous k value and the classification effectiveness as the current updated k value; when the proportion of suspected abnormal data is higher than or equal to the proportion of abnormal data in the historical time period, use the ratio of the previous k value to the classification effectiveness as the current updated k value;

[0008] Based on the optimal value, perform anomaly detection on the ticket purchase data of each user, obtain the final abnormal data of each ticket purchase data, and promptly prevent the abnormal ticket purchase behavior of the user corresponding to the final abnormal data.

[0009] The above solution realizes accurate anomaly data detection by optimizing the k value in the local outlier factor algorithm, enabling it to adaptively adjust in different time periods, thereby improving the problem of inaccurate evaluation results of user ticket purchase risk.

[0010] Furthermore, the initial k value is:

[0011]

[0012] In the formula, A′ represents the total number of ticket purchase data in the current time period, A represents the total number of ticket purchase data in the historical time period, B represents the total number of abnormal data in the historical time period, represents the proportion of abnormal data in the historical time period.

[0013] The above solution multiplies the proportion of abnormal data in the historical time period by the total number of ticket purchase data in the current time period to obtain the initial value of the k value through historical experience, enabling the detection algorithm to adapt to data fluctuations in different time periods, ensuring the effectiveness of the anomaly detection algorithm when the data scale changes, and improving the sensitivity and accuracy of anomaly detection.

[0014] Furthermore, the data similarity is:

[0015] S n，n+1 =α×d n，n+1 +(1 - α)×y n，n+1 ;

[0016] In the formula, S n，n+1 represents the data similarity between the nth data and the (n + 1)th data in the suspected normal data, d n，n+1 represents the Manhattan distance between the nth data and the (n + 1)th data in the suspected normal data, y n，n+1It represents the cosine similarity between the nth data and the (n + 1)th data in the suspected normal data, and α represents the distance metric weight.

[0017] The above technical solution combines the Manhattan distance and the cosine similarity, dynamically adjusts the weights of the two according to the sparsity degree of the data distribution, so that when the data distribution is sparse, more reliance is placed on the cosine similarity, thereby improving the accuracy and robustness of the similarity calculation, and effectively meeting the anomaly detection requirements under different data distributions.

[0018] Furthermore, the calculation formula for the distance metric weight α is:

[0019]

[0020] In the formula, C represents the total number of suspected normal data, r represents the total number of feature parameters in the ticket purchase data, d n，n+1 represents the Manhattan distance between the nth data and the (n + 1)th data in the suspected normal data, G i，j represents the jth feature parameter of the ith data in the suspected normal data, G i+1，j represents the jth feature parameter of the (i + 1)th data in the suspected normal data.

[0021] The above technical solution dynamically calculates the distance metric weight α by comprehensively considering the overall distribution characteristics of the suspected normal data and the differences of each feature parameter, thereby accurately reflecting the actual similarity between the data. This method can effectively adapt to the complexity of the data features.

[0022] Furthermore, the classification effectiveness is:

[0023]

[0024] In the formula, H represents the classification effectiveness, A′ represents the total number of ticket purchase data in the current time period; B′ represents the total number of suspected abnormal data in the current time period, A represents the total number of ticket purchase data in the historical time period, B represents the total number of abnormal data in the historical time period, and e is the base of the natural logarithm function.

[0025] The above technical solution quantifies the difference in the proportion of abnormal data between the current time period and the historical time period through the calculation of the classification effectiveness. By controlling the influence of the difference through the exponential function, it ensures that the classification effectiveness is close to 1 when the change in the abnormal proportion is small, thereby providing stability during dynamic adjustment and sensitively reflecting the anomaly in the case of significant differences, improving the adaptability and accuracy of the detection algorithm.

[0026] Furthermore, the evaluation index is:

[0027]

[0028] In the formula, U represents the evaluation index, H represents the classification effectiveness, Sn，n+1 represents the data similarity between the nth data and the (n + 1)th data in the suspected normal data, and C represents the total number of suspected normal data.

[0029] The above method effectively combines the influence of historical abnormal data and the similarity of current data, provides a balanced evaluation criterion, and helps to ensure the classification accuracy and stability of detection during the optimization process.

[0030] Furthermore, the ticket purchase data includes multiple characteristic parameters, and the characteristic parameters are: ticket purchase time, ticket purchase frequency, total ticket purchase price, account login frequency, and average flight interval time of the ticket.

[0031] Furthermore, it also includes: preprocessing the ticket purchase data, and the preprocessing process is: data normalization, missing value filling, and data smoothing.

[0032] In a second aspect, the present invention provides an artificial intelligence-based self-service ticket purchase human-computer interaction system, including a memory and a processor. The memory stores computer program instructions, and when the computer program instructions are executed by the processor, the artificial intelligence-based self-service ticket purchase human-computer interaction method described in any one of the above is implemented.

[0033] The beneficial effects of the present invention are as follows:

[0034] By dynamically adjusting the k value in the local outlier factor algorithm, combining the proportion of historical and current abnormal data and data similarity, the present invention optimizes the abnormal detection process, improves the sensitivity and accuracy of the detection algorithm. By continuously updating and optimizing the k value, it ensures that the evaluation index reaches the optimal, thereby realizing the accurate abnormal detection of user ticket purchase behavior and real-time risk control, solving the problem of inaccurate evaluation results of user ticket purchase risk degree, and effectively preventing potential abnormal ticket purchase behaviors. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] By referring to the following detailed description with reference to the accompanying drawings, the above and other objects, features, and advantages of the exemplary embodiments of the present invention will become readily understood. In the drawings, several embodiments of the present invention are shown in an exemplary rather than restrictive manner, and the same or corresponding reference numerals represent the same or corresponding parts, wherein:

[0036] Figure 1 is a flowchart of an artificial intelligence-based self-service ticket purchase human-computer interaction method schematically showing an embodiment of the present invention;

[0037] Figure 2 is a structural block diagram of an artificial intelligence-based self-service ticket purchase human-computer interaction system schematically showing an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0038] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative efforts belong to the protection scope of the present invention.

[0039] The following will describe in detail the specific implementation manners of the present invention in conjunction with the accompanying drawings.

[0040] An embodiment of a self-service ticket-purchasing human-computer interaction method based on artificial intelligence.

[0041] As Figure 1 shown, a flowchart of a self-service ticket-purchasing human-computer interaction method based on artificial intelligence according to an embodiment of the present invention includes:

[0042] S1: Obtain the ticket-purchasing data of different users within the current time period.

[0043] In one embodiment, the ticket-purchasing data of different users within the current time period can be obtained by calling the application programming interface of the ticket-purchasing platform. Of course, it can also be recorded by using a user behavior tracking tool, such as using Google Analytics tool to obtain the ticket-purchasing data of different users; and the ticket-purchasing data includes multiple characteristic parameters, and the characteristic parameters are: ticket-purchasing time, ticket-purchasing frequency, total ticket-purchasing price, account login frequency, and average flight interval time of the flight ticket.

[0044] Among them, the ticket-purchasing time can reveal whether the user's ticket-purchasing habit conforms to its historical behavior pattern. For example, purchasing tickets late at night may indicate that the account has been misused or the user's behavior is abnormal; frequent ticket-purchasing may imply that the user is engaged in ticket scalping or trying to take advantage of fare fluctuations, and there is a risk of the account being stolen; a high total ticket-purchasing price may be related to fraud, such as purchasing tickets with a stolen credit card. In addition, if a user usually buys large-value flight tickets but suddenly starts buying small-value flight tickets, this may also indicate that there is something abnormal with the account. Too frequent account logins may indicate that the account has been stolen or there are abnormal access attempts. At the same time, booking multiple flights within a short period of time, especially when the flight departure times are too close or the flight arrival time conflicts with the departure time of the subsequent flight, resulting in a very short or even negative average flight interval time, may also be related to fraud, evading supervision or abnormal personal behavior.

[0045] Specifically, it also includes preprocessing the ticket-purchasing data, and the preprocessing process is: data normalization, missing value filling, and data smoothing. Data normalization is used to convert data in different ranges to a unified scale so that each feature is within the same range; missing value filling is used to fill in the missing values in the dataset to ensure data integrity; data smoothing is used to reduce the noise and irregular fluctuations in the data to make the data smoother and more representative.

[0046] S2: Apply the local outlier factor algorithm to detect anomalies in the ticket purchase data, and use the initial k value to distinguish between suspected abnormal data and suspected normal data.

[0047] Specifically, apply the local outlier factor algorithm to the ticket purchase data of each user to obtain the anomaly scores of the ticket purchase data of each user; summarize the calculated anomaly scores of the ticket purchase data of each user and draw a histogram of the anomaly scores, and based on the histogram of the anomaly scores, classify the ticket purchase data of each user into suspected abnormal data and suspected normal data. Data with an anomaly score higher than the threshold is marked as suspected abnormal data, while data lower than the threshold is marked as suspected normal data; among them, the k-nearest neighbor value in the local outlier factor algorithm is the initial k value.

[0048] In one embodiment, the initial k value is:

[0049]

[0050] In the formula, A' represents the total number of ticket purchase data in the current time period, A represents the total number of ticket purchase data in the historical time period, B represents the total number of abnormal data in the historical time period, represents the proportion of abnormal data in the historical time period, and the historical time period ticket purchase data and abnormal data can be directly obtained from the historical database.

[0051] S3: Calculate the evaluation index of the suspected normal data. If the evaluation index is lower than the threshold, adjust the k value according to the update rule until the evaluation index meets the requirements, and then obtain the optimal value.

[0052] In one embodiment, calculate the evaluation index of the suspected normal data. The evaluation index is positively correlated with the classification effectiveness and data similarity. The classification effectiveness characterizes the difference change between the current suspected abnormal data and the abnormal data in the historical time period; the data similarity is:

[0053] S n，n+1 =α×d n，n+1 +(1 - α)×y n，n+1 ;

[0054] In the formula, S n，n+1 represents the data similarity between the nth data and the (n + 1)th data in the suspected normal data, d n，n+1 represents the Manhattan distance between the nth data and the (n + 1)th data in the suspected normal data, y n，n+1 represents the cosine similarity between the nth data and the (n + 1)th data in the suspected normal data, and α represents the distance metric weight.

[0055] The calculation formula of the distance metric weight α is as follows:

[0056]

[0057] In the formula, C represents the total number of suspected normal data, r represents the total number of characteristic parameters in the ticket purchase data, and d n，n+1 represents the Manhattan distance between the nth data and the (n + 1)th data in the suspected normal data, and G i，j represents the jth characteristic parameter of the ith data in the suspected normal data, and G i+1，j represents the jth characteristic parameter of the (i + 1)th data in the suspected normal data.

[0058] The classification validity is as follows:

[0059]

[0060] In the formula, H represents the classification validity, A′ represents the total number of ticket purchase data in the current time period; B′ represents the total number of suspected abnormal data in the current time period, A represents the total number of ticket purchase data in the historical time period, B represents the total number of abnormal data in the historical time period, and e is the base of the natural logarithm function.

[0061] The evaluation index is as follows:

[0062]

[0063] In the formula, U represents the evaluation index, H represents the classification validity, and S n，n+1 represents the data similarity between the nth data and the (n + 1)th data in the suspected normal data, and C represents the total number of suspected normal data.

[0064] Among them, the update rule is: when the proportion of suspected abnormal data is lower than the proportion of abnormal data in the historical time period, the product of the previous k value and the classification validity is used as the current updated k value; when the proportion of suspected abnormal data is higher than or equal to the proportion of abnormal data in the historical time period, the ratio of the previous k value to the classification validity is used as the current updated k value.

[0065] Exemplarily, when performing the second update, when the proportion of suspected abnormal data is lower than the proportion of abnormal data in the historical time period, the updated k value is: k1 = H × k; when the proportion of suspected abnormal data is higher than or equal to the proportion of abnormal data in the historical time period, the updated k value is: where k1 represents the updated k value, H represents the classification validity, and k is the initial k value.

[0066] It should be noted that the proportion of abnormal data refers to the proportion of abnormal data in all ticket purchase data in this time period within this time period.

[0067] Exemplarily, the threshold can be 0.5, and of course, it can also be determined according to the actual situation.

[0068] S4: Perform final anomaly detection based on the optimal value, identify the final abnormal data, and take measures to prevent abnormal ticket-purchasing behavior.

[0069] Among them, the k-nearest neighbor value in the local outlier algorithm is the optimal value. Use the local outlier algorithm at this time to perform anomaly detection on the ticket-purchasing data of each user, classify the ticket-purchasing data into normal data and abnormal data. After determining the final abnormal data, identify signs of potential fraud or account abuse, and promptly prevent the abnormal ticket-purchasing behavior of the user corresponding to the final abnormal data.

[0070] In this embodiment, when determining the abnormal ticket-purchasing behavior of a user, a reminder can also be sent, such as in the form of an email or text message to remind the administrator to verify whether there is indeed fraud or abuse, and then temporarily freeze the accounts of suspicious users to prevent them from further abnormal operations.

[0071] The present invention optimizes the anomaly detection process by dynamically adjusting the k value in the local outlier factor algorithm, combining the historical and current abnormal data ratios and data similarities, improving the sensitivity and accuracy of the detection algorithm. By continuously updating and optimizing the k value, ensuring that the evaluation index reaches the optimal, thus realizing accurate anomaly detection and real-time risk control of users' ticket-purchasing behavior, solving the problem of inaccurate evaluation results of users' ticket-purchasing risk levels, and effectively preventing potential abnormal ticket-purchasing behavior.

[0072] An embodiment of a self-service ticket-purchasing human-computer interaction system based on artificial intelligence.

[0073] As Figure 2 shown, the structural block diagram of a self-service ticket-purchasing human-computer interaction system based on artificial intelligence of the present invention includes a processor and a memory.

[0074] The present invention also provides a self-service ticket-purchasing human-computer interaction system based on artificial intelligence. As Figure 2 shown, the system includes a processor and a memory, and the memory stores computer program instructions. When the computer program instructions are executed by the processor, it realizes a self-service ticket-purchasing human-computer interaction method based on the above-mentioned present invention.

[0075] The self-service ticket-purchasing human-computer interaction system based on artificial intelligence also includes other components well-known to those skilled in the art such as a communication interface, and its settings and functions are known in the art, so they will not be elaborated here.

[0076] In the present invention, the aforementioned memory may be any tangible medium that contains or stores a program, which can be used by or in conjunction with an instruction execution system, apparatus, or device. For example, a computer-readable storage medium may be any suitable magnetic storage medium or magneto-optical storage medium, such as a resistive random access memory (RRAM), a dynamic random access memory (DRAM), a static random access memory (SRAM), an enhanced dynamic random access memory (EDRAM), a high-bandwidth memory (HBM), a hybrid memory cube (HMC), etc., or any other medium that can be used to store the required information and can be accessed by an application, a module, or both. Any such computer storage medium may be part of the device or accessible or connectable to the device. Any application or module described in the present invention may be implemented using computer-readable / executable instructions that can be stored or otherwise maintained by such a computer-readable medium.

[0077] In the description of this specification, the meanings of "a plurality of" and "several" are at least two, for example, two, three, or more, etc., unless otherwise specifically and clearly defined.

[0078] Although this specification has shown and described multiple embodiments of the present invention, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Those skilled in the art will think of many changes, alterations, and alternative ways without departing from the spirit and idea of the present invention. It should be understood that various alternatives to the embodiments of the present invention described herein may be employed in the practice of the present invention.

Claims

1. A self-service ticket-purchasing human-computer interaction method based on artificial intelligence, characterized in that, Including: Obtain the ticket purchase data of different users within the current time period; The local outlier factor algorithm is used to perform anomaly detection on the ticket purchase data of each user, and the suspected abnormal data and suspected normal data of each ticket purchase data are obtained; among them, the nearest neighbor value in the local outlier factor algorithm is the initial value; Calculate the evaluation index of suspected normal data, and the calculation formula is: ; In the formula, represents the evaluation index, represents the classification effectiveness, represents the data similarity between the -th data and the -th data in the suspected normal data, represents the total number of suspected normal data; The data similarity is: , represents the Manhattan distance between the -th data and the -th data in the suspected normal data, represents the cosine similarity between the -th data and the -th data in the suspected normal data, represents the distance metric weight; the distance metric weight has the following calculation formula: , represents the total number of feature parameters in the ticket purchase data, represents the -th feature parameter of the -th data in the suspected normal data, represents the -th feature parameter of the -th data in the suspected normal data; The classification validity characterizes the difference change between the current suspected abnormal data and the abnormal data in the historical time period; the classification validity is: , represents the total number of ticket purchase data in the current time period, represents the total number of suspected abnormal data in the current time period, represents the total number of ticket purchase data in the historical time period, represents the total number of abnormal data in the historical time period, is the base of the natural logarithm function; When the evaluation index is lower than the set threshold, update the initial value until the updated evaluation index is higher than the set threshold, then stop the update and obtain the updated optimal value; wherein, the update rule is: when the proportion of suspected abnormal data is lower than the proportion of abnormal data in the historical time period, use the product of the previous value and the classification validity as the current updated value; when the proportion of suspected abnormal data is higher than or equal to the proportion of abnormal data in the historical time period, use the ratio of the previous value and the classification validity as the current updated value; Based on the optimal value, perform anomaly detection on the ticket purchase data of each user to obtain the final anomaly data of each ticket purchase data, and timely prevent the abnormal ticket purchase behavior of the user corresponding to the final anomaly data.

2. The self-service ticket-purchasing human-computer interaction method based on artificial intelligence according to claim 1, wherein The initial value is: ； Wherein, represents the total number of ticket purchase data in the current time period, represents the total number of ticket purchase data in the historical time period, represents the total number of abnormal data in the historical time period, represents the proportion of abnormal data in the historical time period.

3. A self-service ticket-purchasing human-computer interaction method based on artificial intelligence according to claim 1, characterized in that, The ticket purchase data includes multiple feature parameters, and the feature parameters are: ticket purchase time, ticket purchase frequency, total ticket purchase price, account login frequency, and average flight interval time of the ticket.

4. A self-service ticket purchasing human-computer interaction method based on artificial intelligence according to claim 1, characterized in that, It also includes: Preprocess the ticket purchase data, and the preprocessing process is: data normalization, missing value filling, and data smoothing.

5. A self-service ticket-purchasing human-computer interaction system based on artificial intelligence, characterized in that, It includes a memory and a processor, and computer program instructions are stored in the memory. When the computer program instructions are executed by the processor, a self-service ticket purchase human-computer interaction method based on artificial intelligence according to any one of claims 1 to 4 is implemented.

Images