Air ticket data refreshing method and device, medium and electronic equipment

Abstract

The invention relates to the technical field of online transaction, and provides an air ticket data refreshing method and device, a medium and electronic equipment. According to the method, factor characteristic values of respective types are obtained based on multiple types of key factor data; obtaining a plurality of predicted values of respective types and weight values corresponding to the predicted values based on the factor characteristic values of the various types; obtaining target factor values of the corresponding types based on the respective multiple predicted values of the multiple types and the weight values of the corresponding predicted values; obtaining trigger factor values of corresponding types based on the multiple types of trigger factor data; and when any type of trigger factor value meets the preset refreshing condition of the corresponding type, obtaining the refreshing frequency of the air ticket data based on the multiple types of target factor values and the multiple types of trigger factor values. The frequency of refreshing the air ticket is dynamically adjusted, the frequency of data updating is self-adapted, and it is ensured that the system can obtain key data in time.

Description

Technical Field

[0001] The present application relates to the field of online transaction technology, and in particular to a method, device, medium and electronic device for refreshing air ticket data. Background Art

[0002] Over-The-Air (OTA) is a wireless network that allows airlines to upgrade and manage aircraft avionics, sensors, and software without having to park the aircraft on the ground. It can reduce the workload and downtime of ground staff and improve the efficiency and safety of the fleet.

[0003] With the popularity of global air travel and the intensification of competition, we are facing the challenge of dynamically adjusting fares and cabins. At present, the way to update air ticket data depends on price verification, booking failure or fixed expiration time. OTA air tickets are passively refreshed or refreshed too frequently, resulting in poor user experience.

[0004] Therefore, the present application provides a method for refreshing air ticket data to solve the above technical problems. Summary of the invention

[0005] The purpose of this application is to provide a method, device, medium and electronic device for refreshing ticket data, which can solve at least one of the technical problems mentioned above. The specific solution is as follows: According to a specific implementation of the present application, in a first aspect, the present application provides a method for refreshing air ticket data, comprising: Acquire multiple types of key factor data that affect the refreshing of air ticket data, wherein the multiple types of key factor data at least include trigger factor data of corresponding types; Obtaining factor characteristic values ​​of respective types based on the multiple types of key factor data; Based on the multiple types of factor feature values, obtain multiple prediction values ​​of respective types and weight values ​​of corresponding prediction values; Obtaining a target factor value of a corresponding type based on a plurality of prediction values ​​of each of the plurality of types and weight values ​​of the corresponding prediction values; Obtaining trigger factor values ​​of corresponding types based on the multiple types of trigger factor data; When any type of trigger factor value satisfies a corresponding type of preset refresh condition, the refresh frequency of the air ticket data is obtained based on the multiple types of target factor values ​​and the multiple types of trigger factor values.

[0006] Optionally, obtaining a plurality of prediction values ​​of respective types and weight values ​​of corresponding prediction values ​​based on the plurality of types of factor feature values ​​includes: Applying the multiple types of factor characteristic values ​​to multiple prediction models respectively to obtain multiple prediction values ​​of corresponding types; Based on the multiple types of factor feature values ​​and the multiple predicted values ​​of the corresponding types, weight values ​​of the multiple predicted values ​​of the corresponding types are obtained.

[0007] Optionally, the multiple prediction models include: a time series model, a deep learning model and a Bayesian-integral model.

[0008] Optionally, obtaining the weight values ​​of the multiple predicted values ​​of the corresponding types based on the multiple types of factor feature values ​​and the multiple predicted values ​​of the corresponding types includes: Calculate the difference between each type of factor characteristic value and each predicted value of the corresponding type to obtain the residual value of the corresponding predicted value; Calculate the standard deviation of the residual value of each predicted value to obtain the standard deviation of the corresponding predicted value; The weight values ​​of the respective predicted values ​​of each type are obtained based on the respective standard deviation values ​​of the respective predicted values ​​of each type.

[0009] Optionally, obtaining the weight values ​​of the corresponding types of the plurality of prediction values ​​based on the respective standard deviation values ​​of the plurality of prediction values ​​of each type comprises the following formula: ;

[0010] in, oh ij Indicates i Type j The weight of the predicted value, s ij Indicates i Type j The standard deviation of the predicted values, s ik Indicates i Type k The standard deviation of the predicted values, N Indicates i the number of the plurality of predicted values ​​of the type, j Less than or equal to N .

[0011] Optionally, obtaining the target factor value of the corresponding type based on the multiple predicted values ​​of each of the multiple types and the weight values ​​of the corresponding predicted values ​​includes the following formula:

[0012] in, α i Indicates iTypes of target factor values, oh ik Indicates i Type k The weight of the predicted value, α ik Indicates i Type k Prediction values, N Indicates i The number of the multiple predicted values ​​of the type.

[0013] Optionally, the refresh frequency of the air ticket data is obtained based on the multiple types of target factor values ​​and the multiple types of trigger factor values, including the following formula:

[0014] in, F update Indicates the refresh frequency of ticket data. α i Indicates i Types of target factor values, P i Indicates i types of trigger factor values, and M represents the number of the multiple types.

[0015] Optionally, the multiple types include: ticket type, cabin type and external type; The key factor data of the ticket type includes a time series value and a ticket price, and the ticket price of the ticket type represents the trigger factor data of the ticket type; The key factor data of the class type includes: time series value, total number of classes and number of seats sold, and the total number of classes and number of seats sold of the class type represent the trigger factor data of the class type; The key factor data of the external type include: timing value, keyword heat value, key temperature value, key rainfall value, key wind speed value and key news sentiment value. The keyword heat value, key temperature value, key rainfall value, key wind speed value and key news sentiment value of the external type represent the trigger factor data of the external type.

[0016] Optionally, the factor characteristic values ​​of the ticket type include: time series characteristic value, ticket price mean and increase / decrease value; The factor characteristic values ​​of the class type include: time series characteristic value, number of remaining class spaces and ratio of sold seats; The external factor characteristic values ​​include: time series characteristic values, topic popularity characteristic values, weather impact characteristic values ​​and news sentiment characteristic values.

[0017] Optionally, the trigger factor value of the air ticket type includes price volatility; The trigger factor value of the space type includes the space change rate; The external type trigger factor value includes the external factor change rate.

[0018] According to a specific implementation of the present application, in a second aspect, the present application provides a device for refreshing ticket data, comprising: An acquisition unit, used for acquiring multiple types of key factor data affecting the refreshing of air ticket data, wherein the multiple types of key factor data at least include trigger factor data of corresponding types; A first obtaining unit, configured to obtain factor characteristic values ​​of respective types based on the multiple types of key factor data; A second obtaining unit, configured to obtain a plurality of prediction values ​​of respective types and weight values ​​of corresponding prediction values ​​based on the plurality of types of factor characteristic values; A third obtaining unit, configured to obtain a target factor value of a corresponding type based on the respective multiple predicted values ​​of the multiple types and the weight values ​​of the corresponding predicted values; a fourth obtaining unit, configured to obtain trigger factor values ​​of corresponding types based on the multiple types of trigger factor data; The fifth obtaining unit is used to obtain the refresh frequency of the air ticket data based on the multiple types of target factor values ​​and the multiple types of trigger factor values ​​when any type of trigger factor value meets the corresponding type of preset refresh condition.

[0019] Optionally, obtaining a plurality of prediction values ​​of respective types and weight values ​​of corresponding prediction values ​​based on the plurality of types of factor feature values ​​includes: Applying the multiple types of factor characteristic values ​​to multiple prediction models respectively to obtain multiple prediction values ​​of corresponding types; Based on the multiple types of factor feature values ​​and the multiple predicted values ​​of the corresponding types, weight values ​​of the multiple predicted values ​​of the corresponding types are obtained.

[0020] Optionally, the multiple prediction models include: a time series model, a deep learning model and a Bayesian-integral model.

[0021] Optionally, obtaining the weight values ​​of the multiple predicted values ​​of the corresponding types based on the multiple types of factor feature values ​​and the multiple predicted values ​​of the corresponding types includes: Calculate the difference between each type of factor characteristic value and each predicted value of the corresponding type to obtain the residual value of the corresponding predicted value; Calculate the standard deviation of the residual value of each predicted value to obtain the standard deviation of the corresponding predicted value; The weight values ​​of the respective predicted values ​​of each type are obtained based on the respective standard deviation values ​​of the respective predicted values ​​of each type.

[0022] Optionally, obtaining the weight values ​​of the corresponding types of the plurality of prediction values ​​based on the respective standard deviation values ​​of the plurality of prediction values ​​of each type comprises the following formula: ;

[0023] in, oh ij Indicates i Type j The weight of the predicted value, s ij Indicates i Type j The standard deviation of the predicted values, s ik Indicates i Type k The standard deviation of the predicted values, N Indicates i the number of the plurality of predicted values ​​of the type, j Less than or equal to N .

[0024] Optionally, obtaining the target factor value of the corresponding type based on the multiple predicted values ​​of each of the multiple types and the weight values ​​of the corresponding predicted values ​​includes the following formula:

[0025] in, α i Indicates i Types of target factor values, oh ik Indicates i Type k The weight of the predicted value, α ik Indicates i Type k Prediction values, N Indicates i The number of the multiple predicted values ​​of the type.

[0026] Optionally, the refresh frequency of the air ticket data is obtained based on the multiple types of target factor values ​​and the multiple types of trigger factor values, including the following formula:

[0027] in, F update Indicates the refresh frequency of ticket data.α i Indicates i Types of target factor values, P i Indicates i types of trigger factor values, and M represents the number of the multiple types.

[0028] Optionally, the multiple types include: ticket type, cabin type and external type; The key factor data of the ticket type includes a time series value and a ticket price, and the ticket price of the ticket type represents the trigger factor data of the ticket type; The key factor data of the class type includes: time series value, total number of classes and number of seats sold, and the total number of classes and number of seats sold of the class type represent the trigger factor data of the class type; The key factor data of the external type include: timing value, keyword heat value, key temperature value, key rainfall value, key wind speed value and key news sentiment value. The keyword heat value, key temperature value, key rainfall value, key wind speed value and key news sentiment value of the external type represent the trigger factor data of the external type.

[0029] Optionally, the factor characteristic values ​​of the ticket type include: time series characteristic value, ticket price mean and increase / decrease value; The factor characteristic values ​​of the class type include: time series characteristic value, number of remaining class spaces and ratio of sold seats; The external factor characteristic values ​​include: time series characteristic values, topic popularity characteristic values, weather impact characteristic values ​​and news sentiment characteristic values.

[0030] Optionally, the trigger factor value of the air ticket type includes price volatility; The trigger factor value of the space type includes the space change rate; The external type trigger factor value includes the external factor change rate.

[0031] According to a specific implementation of the present application, in a third aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, and when the program is executed by a processor, the method for refreshing air ticket data as described in any of the above items is implemented.

[0032] According to the specific implementation of the present application, in a fourth aspect, the present application provides an electronic device, comprising: one or more processors; a storage device for storing one or more programs, when the one or more programs are executed by the one or more processors, the one or more processors implement the method for refreshing the air ticket data as described in any of the above items.

[0033] Compared with the prior art, the above solution of the embodiment of the present application has at least the following beneficial effects: The present application provides a method, device, medium and electronic device for refreshing air ticket data. The present application obtains factor characteristic values ​​of various types based on various types of key factor data; obtains multiple predicted values ​​of various types and weight values ​​of corresponding predicted values ​​based on the factor characteristic values ​​of various types; obtains target factor values ​​of corresponding types based on the multiple predicted values ​​of various types and weight values ​​of corresponding predicted values; obtains trigger factor values ​​of corresponding types based on the trigger factor data of various types; when any type of trigger factor value meets the preset refresh condition of the corresponding type, obtains the refresh frequency of air ticket data based on the target factor values ​​of various types and the trigger factor values ​​of various types. The frequency of refreshing air tickets is dynamically adjusted, and the frequency of data update is adaptively adjusted to ensure that the system can obtain key data in a timely manner. It not only maintains the accuracy and adaptability of the prediction, but also improves the refresh efficiency, interpretability and flexibility, so that the refresh is more adaptable to changes in air ticket data. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 A flow chart showing a method for refreshing air ticket data according to an embodiment of the present application is shown; Figure 2 A unit block diagram of a device for refreshing air ticket data according to an embodiment of the present application is shown. DETAILED DESCRIPTION

[0035] In order to make the purpose, technical solutions and advantages of the present application clearer, the present application will be further described in detail below in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present application.

[0036] The terms used in the embodiments of the present application are only for the purpose of describing specific embodiments, and are not intended to limit the present application. The singular forms "a", "said" and "the" used in the embodiments of the present application and the appended claims are also intended to include plural forms, unless the context clearly indicates other meanings, and "multiple" generally includes at least two.

[0037] It should be understood that the term "and / or" used in this article is only a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship.

[0038] It should be understood that although the terms first, second, third, etc. may be used to describe in the embodiments of the present application, these descriptions should not be limited to these terms. These terms are only used to distinguish the descriptions. For example, without departing from the scope of the embodiments of the present application, the first may also be referred to as the second, and similarly, the second may also be referred to as the first.

[0039] As used herein, the words "if" and "if" may be interpreted as "at the time of" or "when" or "in response to determining" or "in response to detecting", depending on the context. Similarly, the phrases "if it is determined" or "if (stated condition or event) is detected" may be interpreted as "when it is determined" or "in response to determining" or "when detecting (stated condition or event)" or "in response to detecting (stated condition or event)", depending on the context.

[0040] It should also be noted that the term "includes", "comprising" or any other variation thereof is intended to cover non-exclusive inclusion, so that a commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprising a ..." do not exclude the existence of other identical elements in the commodity or device including the elements.

[0041] It should be particularly noted that any symbols and / or numbers in the specification that are not marked in the accompanying drawings are not drawing marks.

[0042] The optional embodiments of the present application are described in detail below with reference to the accompanying drawings.

[0043] The embodiment provided in this application is an embodiment of a method for refreshing air ticket data.

[0044] Combine the following Figure 1 The embodiments of the present application are described in detail.

[0045] Step S101, obtaining various types of key factor data that affect the refreshing of air ticket data.

[0046] Among them, the multiple types of key factor data all include at least corresponding types of trigger factor data.

[0047] The trigger factor data is used to obtain the trigger factor value, and the trigger factor value is used to determine whether the ticket data is refreshed.

[0048] In some specific embodiments, the multiple types include: ticket type, cabin type and external type.

[0049] The embodiment of the present application comprehensively determines the refresh frequency of air ticket data through multiple types, thereby improving the real-time response efficiency of air ticket data and reducing unnecessary waste of resources.

[0050] Key factor data refers to historical data that affects air ticket data.

[0051] The timing value is the time point when the data is collected.

[0052] The key factor data of the ticket type includes a time series value and a ticket price, and the ticket price of the ticket type represents the trigger factor data of the ticket type.

[0053] The key factor data of the class type include: time series value, total number of classes and number of seats sold, and the total number of classes and number of seats sold of the class type represent the trigger factor data of the class type.

[0054] The key factor data of the external type include: timing value, keyword heat value, key temperature value, key rainfall value, key wind speed value and key news sentiment value. The keyword heat value, key temperature value, key rainfall value, key wind speed value and key news sentiment value of the external type represent the trigger factor data of the external type.

[0055] The external key factor data is derived from social media data, news hotspot data and weather data.

[0056] Social media data: Analyze trending topics, popular destinations, and popular events. Obtained through keyword monitoring and sentiment analysis models. For example, keyword popularity values ​​on social media.

[0057] News hotspot data: Extract information related to destinations or time points, such as events, festivals, news sentiment, etc. For example, the keyword heat value and key news sentiment value of news hotspots.

[0058] Weather data: Collect weather changes at the departure and destination points to assess the impact on travel demand, such as key temperature values, key rainfall values, and key wind speed values.

[0059] This specific embodiment flexibly accesses external media data (social media, news hot spots, etc.) and real-time change information to make the prediction more comprehensive and agile, and can cope with the impact of emergencies (such as weather and news).

[0060] This specific embodiment classifies the external type of key factor data and formulates a priority strategy.

[0061] Classification by region: Tourist attractions and popular places are classified into priority attention areas and general attention areas according to historical popularity and climatic conditions.

[0062] Priority Areas: Updated monthly based on historical data, areas with higher temperatures and better climates in that month will be listed as priority areas (such as summer resorts in the north, warm areas in the south in winter, etc.).

[0063] Exceptional activities: Concerts, press conferences and other entertainment activities are not subject to regional classification restrictions. Such activities should be paid attention to one month in advance, starting from the official announcement of the specific date of the event.

[0064] Regarding the frequency of collecting social media data, if it is a priority area, the interval is 1 day / time; if it is a general area, the interval is 3 days / time; if the activity intensity changes, new entertainment activities or sudden hot events appear in the priority area or general area, the collection frequency of the corresponding area will be automatically increased to ensure the timeliness of the data.

[0065] Regarding the frequency of data collection during an event, if the event occurs in a priority time period, the period from one week before the event to one week after the event will be the focus period. One week before the event, maintain the normal frequency (one day / time interval in priority areas, three days / time interval in general areas); three days before the event, increase the frequency to 12 hours / time; one day before the event, increase the frequency of data collection to one hour / time; during the event, maintain an interval of one hour / time to ensure real-time data collection; one week after the event, the interval is 12 hours / time, and the original frequency will be restored after the event. During the focus period, the air ticket data of the relevant event areas will be updated synchronously; one week after the event, the update frequency of general areas will be restored.

[0066] Regarding the frequency of weather data collection. For the update of regular weather data, the default frequency of weather data collection for destinations and scenic spots is 1 day / time. When an event is approaching, the frequency of weather data update will be adjusted accordingly to enable timely analysis of the impact of weather on the event; for 3 days before the event and 2 days after the event, the frequency of weather data collection will be 1 hour / time; for other time periods, the frequency of weather data update will be restored to 1 day / time.

[0067] Regarding the attention mode of special events. For events with sustained popularity (for example, "Guizhou Village Super"), pay attention to them in the week before and after, follow the above-mentioned non-special events, and collect data frequently one week before and one week after the event. For multiple close events, if the interval between events does not exceed 20 days, continue to pay attention during this period, shorten the prediction frequency, and ensure that the data meets the needs of hot spots. For multiple events with long intervals, if the interval between events exceeds 20 days, return to the attention frequency of the above-mentioned non-special events, and the collection frequency will also return to the frequency of the above-mentioned non-special events.

[0068] Step S102: obtaining factor characteristic values ​​of respective types based on the multiple types of key factor data.

[0069] In some specific embodiments, the factor characteristic values ​​of the ticket type include: time series characteristic value, ticket price mean and increase / decrease value.

[0070] Time series feature values, time features extracted from time series values, for example, the day of the week the time series value is, whether it is a holiday, or whether it is a peak tourist season.

[0071] For example,

[0072] in, Price mean represents the average fare of the ticket type, N represents the total number of time series eigenvalues, Price t Indicates t The air ticket price of the time series feature value.

[0073] For example,

[0074] in, Price diff Indicates the increase or decrease value of the ticket type. Price t Indicates t The air ticket price of the time series feature value, Price t-1 Indicates t- 1 time series feature value of air ticket price.

[0075] The factor characteristic values ​​of the class type include: time series characteristic value, number of remaining classes and ratio of sold seats.

[0076] For example,

[0077] in, Seats availabel Indicates the number of remaining spaces of the space type. Total Seats Indicates the total number of spaces of the space type. Seats sold Indicates the number of seats sold for the cabin class.

[0078] For example,

[0079] in, sold_ratio Indicates the percentage of seats sold for a particular class of service. Total Seats Indicates the total number of spaces of the space type. Seats sold Indicates the number of seats sold for the cabin class.

[0080] The external factor characteristic values ​​include: time series characteristic values, topic popularity characteristic values, weather impact characteristic values ​​and news sentiment characteristic values.

[0081] For example,

[0082] in, Social trend Indicates the topic heat feature value of the external type, weight k The external type k The preset weight value of keywords, trend k The external type k The keyword heat value of keywords, K Indicates the number of keywords.

[0083] For example,

[0084] in, Weather impact The weather impact characteristic value representing the external type, F represents the mapping function, Temperature Indicates the critical temperature value of the external type, Rainfall represents the critical rainfall value of the external type, WindSpeed Indicates the critical wind speed value for the exterior type.

[0085] For example,

[0086] in, News sentiment The news sentiment feature value representing the external type, sentiment m The external type m Key news sentiment value of keywords, weight m The external type m The preset weight value of a keyword, M represents the number of external type keywords.

[0087] Before prediction, multiple types of factor feature values ​​are processed to remove duplicate data, fill in missing values, process regional data differences, enhance data, and standardize data.

[0088] The difference processing of regional data can perform hierarchical analysis on demand differences in different regions, which can improve the prediction accuracy. The demand characteristics of different destinations and departure points vary greatly, and regional refinement of the model can reduce the prediction error.

[0089] Data augmentation processing can generate diverse time series and increase the richness of training data to enhance the generalization ability of the model.

[0090] Step S103, obtaining a plurality of prediction values ​​of respective types and weight values ​​of corresponding prediction values ​​based on the plurality of types of factor characteristic values.

[0091] In some specific embodiments, obtaining a plurality of prediction values ​​of respective types and weight values ​​of corresponding prediction values ​​based on the plurality of types of factor feature values ​​includes: Step S103 - 1 , applying the multiple types of factor feature values ​​to multiple prediction models respectively to obtain multiple prediction values ​​of corresponding types.

[0092] In some specific embodiments, the multiple prediction models include: a time series model, a deep learning model, and a Bayesian-integral model.

[0093] The time series model is used to capture the linear trend and time series characteristics of price and space changes, including the following formula:

[0094] in, α SARIMAX_i The output of the time series model is i Types of predicted values, f SARIMAX represents a time series model, X it Indicated in t The time series characteristic value i Types of factor eigenvalues.

[0095] Deep learning models are used to capture nonlinear relationships in input data, including the following formula:

[0096] in, α SLTM_i Represents the output of the deep learning model i types of prediction values, SLTM represents the deep learning model, X it Indicated in t The time series characteristic value i Types of factor eigenvalues, h it-1 Indicates that the deep learning model t-1 The time series characteristic value i The hidden state carries the t-1The context information on the time series feature value mainly assists in the prediction of the current time series feature value.

[0097] The Bayesian-integral model includes the Bayesian model and the integral model. The Bayesian model uses Bayesian inference to perform uncertainty analysis, risk control, and credibility analysis. The Bayesian model specifically includes the following formulas:

[0098] in, α i Indicates i types of predicted target variable values, X it Indicated in t The time series characteristic value i Types of factor eigenvalues, P（α i |X it ） represents the credibility value, P(X it |α i ） Likelihood function, expressed in α i Observed under the conditions X it The probability of P（α i ） Indicates i The prior probability values ​​of the types, P(X it ） Indicates the preset marginal probability value.

[0099] The integral model specifically includes the following formula:

[0100] in, α Bayes_i The output of the integral model is i The integral model is used to calculate the expected value of the price forecast, which is a weighted average of all credibility values.

[0101] This specific embodiment integrates the time series model and the deep learning model, combining the linear trend and periodic characteristics that the time series model is good at with the nonlinear modeling ability of the deep learning model. It can not only cope with the long-term trend changes in prices and cabin space, but also deal with short-term nonlinear fluctuations.

[0102] By using the confidence evaluation of the Bayesian-integral model, dynamic weights are assigned to different models to ensure that the fusion results are more reliable. When the air ticket data changes significantly, the Bayesian-integral model can adaptively adjust the weights to effectively reduce the prediction error caused by model deviation.

[0103] Bayesian model inference helps the model better quantify the uncertainty of the prediction results, thereby avoiding more frequent real-time ticket data updates due to huge prediction errors. In this way, the prediction model is more robust in the face of high volatility and complex market changes, and is suitable for scenarios with dynamic pricing and real-time adjustments.

[0104] Step S103 - 2 , obtaining weight values ​​of the corresponding types of the multiple predicted values ​​based on the multiple types of factor feature values ​​and the corresponding types of the multiple predicted values.

[0105] In some specific embodiments, obtaining the weight values ​​of the multiple predicted values ​​of the corresponding types based on the multiple types of factor feature values ​​and the multiple predicted values ​​of the corresponding types includes: Step S103-21, calculating the difference between each type of factor characteristic value and each predicted value of the corresponding type, and obtaining the residual value of the corresponding predicted value.

[0106] Step S103-22, calculating the standard deviation of the residual value of each predicted value to obtain the standard deviation value of the corresponding predicted value.

[0107] Step S103-23, obtaining the weight values ​​of the corresponding types of the plurality of prediction values ​​based on the standard deviation values ​​of the plurality of prediction values ​​of each type.

[0108] In some specific embodiments, the method of obtaining the weight values ​​of the corresponding types of the plurality of prediction values ​​based on the respective standard deviation values ​​of the plurality of prediction values ​​of each type comprises the following formula: ;

[0109] in, oh ij Indicates i Type j The weight of the predicted value, s ij Indicates i Type j The standard deviation of the predicted values, s ik Indicates i Type k The standard deviation of the predicted values, N Indicates i the number of the plurality of predicted values ​​of the type, jLess than or equal to N .

[0110] Step S104: obtaining a target factor value of a corresponding type based on the respective multiple prediction values ​​of the multiple types and the weight values ​​of the corresponding prediction values.

[0111] In some specific embodiments, the obtaining of the target factor value of the corresponding type based on the multiple predicted values ​​of each of the multiple types and the weight values ​​of the corresponding predicted values ​​comprises the following formula:

[0112] in, α i Indicates i Types of target factor values, oh ik Indicates i Type k The weight of the predicted value, α ik Indicates i Type k Prediction values, N Indicates i The number of the multiple predicted values ​​of the type.

[0113] Step S105 : obtaining trigger factor values ​​of corresponding types based on the multiple types of trigger factor data.

[0114] In some specific embodiments, the trigger factor value of the ticket type includes the price fluctuation rate; the trigger factor value of the cabin type includes the cabin change rate; and the trigger factor value of the external type includes the external factor change rate.

[0115] Price volatility, specifically including the following formula:

[0116] in, represents the average fare of the ticket type, Price t Indicates the timing value t The fare of the fare type mentioned, Price t-1 Indicates the timing value t-1 The fare for the stated fare type.

[0117] The cabin space change rate includes the following formula:

[0118] in, V seat Indicates the space change rate of the space type. sold_ratio Indicates the percentage of seats sold for a particular class of service. Seats sold Indicates the number of seats sold for the cabin type. Total Seats Indicates the total number of spaces of the space type.

[0119] The rate of change of external factors includes the following formula:

[0120] in, Events impact Indicates the rate of change of external factors of external type, Social trend Indicates the topic heat feature value of the external type, Weather impact The weather impact characteristic value representing the external type, News sentiment The news sentiment feature value representing the external type, f、 and ψ are the weights of the influence of each part, and their sum is equal to 1.

[0121] Step S106, when any type of trigger factor value satisfies the preset refresh condition of the corresponding type, the refresh frequency of the air ticket data is obtained based on the multiple types of target factor values ​​and the multiple types of trigger factor values.

[0122] For example, if the price volatility of an air ticket type is greater than a price volatility threshold, the refresh frequency of the air ticket data is obtained based on the target factor values ​​of the multiple types and the trigger factor values ​​of the multiple types; the price standard deviation and the historical price change average are calculated based on the historical air ticket prices, and then the price volatility threshold is obtained based on the product of a preset multiple value (such as 1.5 times or 2 times) and the price standard deviation, and then the sum of the historical price change average.

[0123] In some specific embodiments, the refresh frequency of the air ticket data is obtained based on the multiple types of target factor values ​​and the multiple types of trigger factor values, including the following formula:

[0124] in, F update Indicates the refresh frequency of ticket data. α i Indicates i Types of target factor values, P i Indicates i types of trigger factor values, and M represents the number of the multiple types.

[0125] For example, continuing the above example, specifically, the formula is as follows:

[0126] in, F update Indicates the refresh frequency of ticket data. represents the average fare of the ticket type, α represents the target factor value of the ticket type, V seat Indicates the space change rate of the space type. β represents the target factor value of the space type, Events impact Indicates the rate of change of external factors of external type, c Represents the target factor value of the external type.

[0127] The embodiment of the present application obtains factor characteristic values ​​of respective types based on multiple types of key factor data; obtains multiple predicted values ​​of respective types and weight values ​​of corresponding predicted values ​​based on the multiple types of factor characteristic values; obtains target factor values ​​of corresponding types based on the multiple predicted values ​​of respective types and weight values ​​of corresponding predicted values; obtains trigger factor values ​​of corresponding types based on the multiple types of trigger factor data; and obtains refresh frequencies of air ticket data based on the multiple types of target factor values ​​and the multiple types of trigger factor values ​​when any type of trigger factor value meets the preset refresh conditions of the corresponding type. Dynamic adjustment of the frequency of refreshing air tickets and the frequency of adaptive data updates are adopted to ensure that the system can obtain key data in a timely manner. It not only maintains the accuracy and adaptability of the prediction, but also improves the refresh efficiency, interpretability and flexibility, making the refresh more adaptable to changes in air ticket data.

[0128] The present application also provides a device embodiment that is based on the above embodiment, which is used to implement the method steps described in the above embodiment. The explanation based on the same name meaning is the same as the above embodiment, and has the same technical effect as the above embodiment, which will not be repeated here.

[0129] like Figure 2 As shown, the present application provides a device 200 for refreshing air ticket data, comprising: An acquisition unit 201 is used to acquire multiple types of key factor data that affect the refreshing of air ticket data, wherein the multiple types of key factor data at least include trigger factor data of corresponding types; A first obtaining unit 202, configured to obtain factor characteristic values ​​of respective types based on the multiple types of key factor data; A second obtaining unit 203 is used to obtain a plurality of prediction values ​​of respective types and weight values ​​of corresponding prediction values ​​based on the plurality of types of factor characteristic values; A third obtaining unit 204 is used to obtain a target factor value of a corresponding type based on the respective multiple prediction values ​​of the multiple types and the weight values ​​of the corresponding prediction values; A fourth obtaining unit 205, configured to obtain a trigger factor value of a corresponding type based on the multiple types of trigger factor data; The fifth obtaining unit 206 is configured to obtain a refresh frequency of the air ticket data based on the multiple types of target factor values ​​and the multiple types of trigger factor values ​​when any type of trigger factor value satisfies a corresponding type of preset refresh condition.

[0130] Optionally, obtaining a plurality of prediction values ​​of respective types and weight values ​​of corresponding prediction values ​​based on the plurality of types of factor feature values ​​includes: Applying the multiple types of factor characteristic values ​​to multiple prediction models respectively to obtain multiple prediction values ​​of corresponding types; Based on the multiple types of factor feature values ​​and the multiple predicted values ​​of the corresponding types, weight values ​​of the multiple predicted values ​​of the corresponding types are obtained.

[0131] Optionally, the multiple prediction models include: a time series model, a deep learning model and a Bayesian-integral model.

[0132] Optionally, obtaining the weight values ​​of the multiple predicted values ​​of the corresponding types based on the multiple types of factor feature values ​​and the multiple predicted values ​​of the corresponding types includes: Calculate the difference between each type of factor characteristic value and each predicted value of the corresponding type to obtain the residual value of the corresponding predicted value; Calculate the standard deviation of the residual value of each predicted value to obtain the standard deviation of the corresponding predicted value; The weight values ​​of the respective predicted values ​​of each type are obtained based on the respective standard deviation values ​​of the respective predicted values ​​of each type.

[0133] Optionally, obtaining the weight values ​​of the corresponding types of the plurality of prediction values ​​based on the respective standard deviation values ​​of the plurality of prediction values ​​of each type comprises the following formula: ;

[0134] in, oh ij Indicates i Type j The weight of the predicted value, s ij Indicates i Type j The standard deviation of the predicted values, s ik Indicates i Typek The standard deviation of the predicted values, N Indicates i the number of the plurality of predicted values ​​of the type, j Less than or equal to N .

[0135] Optionally, obtaining the target factor value of the corresponding type based on the multiple predicted values ​​of each of the multiple types and the weight values ​​of the corresponding predicted values ​​includes the following formula:

[0136] in, α i Indicates i Types of target factor values, oh ik Indicates i Type k The weight of the predicted value, α ik Indicates i Type k Prediction values, N Indicates i The number of the multiple predicted values ​​of the type.

[0137] Optionally, the refresh frequency of the air ticket data is obtained based on the multiple types of target factor values ​​and the multiple types of trigger factor values, including the following formula:

[0138] in, F update Indicates the refresh frequency of ticket data. α i Indicates i Types of target factor values, P i Indicates i types of trigger factor values, and M represents the number of the multiple types.

[0139] Optionally, the multiple types include: ticket type, cabin type and external type; The key factor data of the ticket type includes a time series value and a ticket price, and the ticket price of the ticket type represents the trigger factor data of the ticket type; The key factor data of the class type includes: time series value, total number of classes and number of seats sold, and the total number of classes and number of seats sold of the class type represent the trigger factor data of the class type; The key factor data of the external type include: timing value, keyword heat value, key temperature value, key rainfall value, key wind speed value and key news sentiment value. The keyword heat value, key temperature value, key rainfall value, key wind speed value and key news sentiment value of the external type represent the trigger factor data of the external type.

[0140] Optionally, the factor characteristic values ​​of the ticket type include: time series characteristic value, ticket price mean and increase / decrease value; The factor characteristic values ​​of the class type include: time series characteristic value, number of remaining class spaces and ratio of sold seats; The external factor characteristic values ​​include: time series characteristic values, topic popularity characteristic values, weather impact characteristic values ​​and news sentiment characteristic values.

[0141] Optionally, the trigger factor value of the air ticket type includes price volatility; The trigger factor value of the space type includes the space change rate; The external type trigger factor value includes the external factor change rate.

[0142] The embodiment of the present application obtains factor characteristic values ​​of respective types based on multiple types of key factor data; obtains multiple predicted values ​​of respective types and weight values ​​of corresponding predicted values ​​based on the multiple types of factor characteristic values; obtains target factor values ​​of corresponding types based on the multiple predicted values ​​of respective types and weight values ​​of corresponding predicted values; obtains trigger factor values ​​of corresponding types based on the multiple types of trigger factor data; and obtains refresh frequencies of air ticket data based on the multiple types of target factor values ​​and the multiple types of trigger factor values ​​when any type of trigger factor value meets the preset refresh conditions of the corresponding type. Dynamic adjustment of the frequency of refreshing air tickets and the frequency of adaptive data updates are adopted to ensure that the system can obtain key data in a timely manner. It not only maintains the accuracy and adaptability of the prediction, but also improves the refresh efficiency, interpretability and flexibility, making the refresh more adaptable to changes in air ticket data.

[0143] Example 3 This embodiment provides an electronic device, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the method steps described in the above embodiment.

[0144] Example 4 An embodiment of the present application provides a non-volatile computer storage medium, wherein the computer storage medium stores computer executable instructions, and the computer executable instructions can execute the method steps described in the above embodiment.

[0145] Finally, it should be noted that the various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other. For the system or device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part description.

[0146] The above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for refreshing air ticket data, characterized in that: include: Acquire multiple types of key factor data that affect the refreshing of air ticket data, wherein the multiple types of key factor data at least include trigger factor data of corresponding types; Obtaining factor characteristic values ​​of respective types based on the multiple types of key factor data; Based on the multiple types of factor feature values, obtain multiple prediction values ​​of respective types and weight values ​​of corresponding prediction values; Obtaining a target factor value of a corresponding type based on a plurality of prediction values ​​of each of the plurality of types and weight values ​​of the corresponding prediction values; Obtaining trigger factor values ​​of corresponding types based on the multiple types of trigger factor data; When any type of trigger factor value satisfies a corresponding type of preset refresh condition, the refresh frequency of the air ticket data is obtained based on the multiple types of target factor values ​​and the multiple types of trigger factor values.

2. The method according to claim 1, characterized in that The obtaining of a plurality of prediction values ​​of respective types and weight values ​​of corresponding prediction values ​​based on the plurality of types of factor characteristic values ​​comprises: Applying the multiple types of factor characteristic values ​​to multiple prediction models respectively to obtain multiple prediction values ​​of corresponding types; Based on the multiple types of factor feature values ​​and the multiple predicted values ​​of the corresponding types, weight values ​​of the multiple predicted values ​​of the corresponding types are obtained.

3. The method according to claim 2, characterized in that The multiple prediction models include: time series model, deep learning model and Bayesian-integral model.

4. The method according to claim 2, characterized in that: The obtaining the weight values ​​of the corresponding types of the plurality of predicted values ​​based on the plurality of types of factor feature values ​​and the corresponding types of the plurality of predicted values ​​comprises: Calculate the difference between each type of factor characteristic value and each predicted value of the corresponding type to obtain the residual value of the corresponding predicted value; Calculate the standard deviation of the residual value of each predicted value to obtain the standard deviation of the corresponding predicted value; The weight values ​​of the respective predicted values ​​of each type are obtained based on the respective standard deviation values ​​of the respective predicted values ​​of each type.

5. The method according to claim 4, characterized in that The method of obtaining the weight values ​​of the corresponding types of the plurality of predicted values ​​based on the respective standard deviation values ​​of the plurality of predicted values ​​of each type comprises the following formula: in, ω ij Indicates i Type j The weight of the predicted value, σ ij Indicates i Type j The standard deviation of the predicted values, σ ik Indicates i Type k The standard deviation of the predicted values, N Indicates i the number of the plurality of predicted values ​​of the type, j Less than or equal to N .

6. The method according to claim 1, characterized in that The target factor value of the corresponding type is obtained based on the respective multiple predicted values ​​of the multiple types and the weight values ​​of the corresponding predicted values, Includes the following formulas: in, α i Indicates i Types of target factor values, ω ik Indicates i Type k The weight of the predicted value, α ik Indicates i Type k Prediction values, N Indicates i The number of the multiple predicted values ​​of the type.

7. The method according to claim 1, characterized in that The method of obtaining the refresh frequency of the air ticket data based on the multiple types of target factor values ​​and the multiple types of trigger factor values ​​includes the following formula: in, F update Indicates the refresh frequency of ticket data. α i Indicates i Types of target factor values, P i Indicates i types of trigger factor values, and M represents the number of the multiple types.

8. The method according to claim 1, characterized in that The multiple types include: ticket type, cabin type and external type; The key factor data of the ticket type includes a time series value and a ticket price, and the ticket price of the ticket type represents the trigger factor data of the ticket type; The key factor data of the class type includes: time series value, total number of classes and number of seats sold, and the total number of classes and number of seats sold of the class type represent the trigger factor data of the class type; The key factor data of the external type include: timing value, keyword heat value, key temperature value, key rainfall value, key wind speed value and key news sentiment value. The keyword heat value, key temperature value, key rainfall value, key wind speed value and key news sentiment value of the external type represent the trigger factor data of the external type.

9. The method according to claim 8, characterized in that The factor characteristic values ​​of the ticket type include: time series characteristic value, ticket price mean and increase / decrease value; The factor characteristic values ​​of the class type include: time series characteristic value, number of remaining class spaces and ratio of sold seats; The external factor characteristic values ​​include: time series characteristic values, topic popularity characteristic values, weather impact characteristic values ​​and news sentiment characteristic values.

10. The method according to claim 8, characterized in that The trigger factor value of the ticket type includes price volatility; The trigger factor value of the space type includes the space change rate; The external type trigger factor value includes the external factor change rate.

11. A device for refreshing air ticket data, characterized in that: include: An acquisition unit, used for acquiring multiple types of key factor data affecting the refreshing of air ticket data, wherein the multiple types of key factor data at least include trigger factor data of corresponding types; A first obtaining unit, configured to obtain factor characteristic values ​​of respective types based on the multiple types of key factor data; A second obtaining unit, configured to obtain a plurality of prediction values ​​of respective types and weight values ​​of corresponding prediction values ​​based on the plurality of types of factor characteristic values; A third obtaining unit, configured to obtain a target factor value of a corresponding type based on the respective multiple predicted values ​​of the multiple types and the weight values ​​of the corresponding predicted values; a fourth obtaining unit, configured to obtain trigger factor values ​​of corresponding types based on the multiple types of trigger factor data; The fifth obtaining unit is used to obtain the refresh frequency of the air ticket data based on the multiple types of target factor values ​​and the multiple types of trigger factor values ​​when any type of trigger factor value meets the corresponding type of preset refresh condition.

12. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the method according to any one of claims 1 to 10 is implemented.

13. An electronic device, characterized in that: include: one or more processors; a storage device for storing one or more programs, When the one or more programs are executed by the one or more processors, the one or more processors implement the method as claimed in any one of claims 1 to 10.

Images