Method and system for personalized coupon distribution based on user behavior profiling

Abstract

The application is suitable for the technical field of e-commerce, and provides a personalized coupon distribution method and system based on user behavior portrait, which comprises the following steps: collecting user behavior events to form a user behavior sequence; extracting events related to a target category from the user behavior sequence, including adjacent cross-category events, constructing an extended behavior sequence, and labeling a category continuity label for each event in the sequence; based on behavior value weight, conversion gain value and time decay information, calculating a weighted contribution value for each event in the extended behavior sequence and summing up to generate a time sequence continuity elasticity index; determining impulsive operation count and coupon fatigue, and calculating a coupon distribution suitability index; comparing the coupon distribution suitability index with a hierarchical threshold to determine the coupon distribution level to which the user belongs, and configuring the coupon according to the corresponding strategy of the level. Through dynamic hierarchical differentiation of coupon amount, threshold, validity period and applicable scope, the platform transaction ecology is protected.

Description

Technical Field

[0001] This invention relates to the field of e-commerce technology, specifically to a personalized coupon distribution method and system based on user behavior profiles. Background Technology

[0002] Existing coupon distribution methods generally rely on aggregated statistical characteristics of user behavior, such as click frequency, purchase frequency, and preferred categories, neglecting sequential patterns such as the order of behavior, type conversion relationships between adjacent behaviors, and the interval and depth of user dwell time. Current solutions cannot distinguish between loyal users with long-term stable interests and impulsive users who frequently operate due to short-term stimuli, nor can they perceive the dynamic process of user interests migrating from one category to another. Furthermore, the continuous high-frequency issuance of coupons in pursuit of short-term redemption rates easily fosters "coupon hunters." Therefore, a personalized coupon distribution method and system based on user behavior profiles is needed to address these problems. Summary of the Invention

[0003] In view of the shortcomings of the existing technology, the purpose of this invention is to provide a personalized coupon distribution method and system based on user behavior profiles, so as to solve the problems existing in the above-mentioned background technology.

[0004] This invention is implemented as follows: a personalized coupon distribution method based on user behavior profiles, the method comprising the following steps:

[0005] Collect users' atomic behavior events, generate event records containing behavior type, timestamp, dwell time and category identification, and sort them by time to form a user behavior sequence;

[0006] For the target category to be evaluated, events related to the target category are extracted from the user behavior sequence, and adjacent cross-category pre-positioned events are included to construct an extended behavior sequence. Each event in the extended behavior sequence is labeled with a category coherence tag. The category coherence tag indicates whether the event belongs to the same category as the previous event.

[0007] Based on preset behavioral value weights, conversion gain values, and differentiated time decay information conditioned on category coherence tags, the weighted contribution values ​​of the extended behavioral sequence are calculated and summed for each event to generate a time-series coherence elasticity index that reflects the coherence and migration intensity of users’ interest in the target category.

[0008] Determine the number of impulsive actions and coupon fatigue of users within a set time window, and combine the time series coherence elasticity index with the number of impulsive actions and coupon fatigue to calculate the coupon issuance suitability index;

[0009] The coupon issuance suitability index is compared with the tiered threshold to determine the coupon issuance tier to which the user belongs. The coupon amount, usage threshold, validity period and applicable scope are configured according to the strategy corresponding to the tier before the coupon is issued.

[0010] As a further aspect of the present invention, the step of constructing the extended behavior sequence specifically includes:

[0011] Iterate through each event in the user behavior sequence. If the product category associated with the event matches the target category, mark the event as the target event and retain it.

[0012] For each retained target event, obtain its adjacent preceding event in the user behavior sequence. When the product category associated with the preceding event is different from the target category, include the preceding event and mark it as a cross-category preceding event.

[0013] All target events and included cross-class current events are arranged in their respective timestamp order to form an extended behavior sequence.

[0014] As a further aspect of the present invention, the step of calculating and summing the weighted contribution values ​​for each event of the extended behavior sequence to generate a temporal coherence elasticity index specifically includes:

[0015] To obtain the behavior value weight for each event in the extended behavior sequence, the weight is determined by looking up the behavior type of the event in a pre-defined behavior value weight table.

[0016] For each event, a differential time decay factor is calculated based on the time difference between the current moment and the timestamp, as well as the category coherence tag corresponding to the event.

[0017] Obtain the conversion gain value between the event and the behavior type of the previous event, which is determined by looking up the table in the preset conversion gain table based on the two behavior types;

[0018] Obtain the dwell depth of the event, and then smooth the dwell depth by taking the logarithm to get the depth factor;

[0019] Multiply the behavioral value weight, time decay factor, conversion gain value and depth factor of each event to obtain the weighted contribution value of the event to the target category;

[0020] The temporal coherence elasticity index is obtained by summing the weighted contribution values ​​of all events in the extended behavior sequence.

[0021] As a further aspect of the present invention, the steps of determining the user's impulsive action count and coupon fatigue level within a set time window specifically include:

[0022] The cumulative number of times a user triggers a preset impulsive behavior pattern within the first time window is counted to obtain an impulsive operation count. The preset impulsive behavior pattern includes at least the following: completing the redemption within a preset time after receiving a coupon, cyclical operation of adding and removing the same product from the cart within a single day, immediately placing an order with a coupon after browsing across multiple categories, and returning the product after placing an order with a coupon.

[0023] The system counts the total number of coupons pushed to users within the second time window, the number of coupons that users actively claimed but did not use and thus expired, and the number of full-price transactions using unused coupons. Based on the relative relationship between the total number of coupons pushed and the number of full-price transactions, and combined with the number of expired and unused coupons, a coupon fatigue level is generated, which represents the decrease in users' sensitivity to coupons.

[0024] As a further aspect of the present invention, the formula for calculating the coupon issuance suitability index M is as follows:

[0025] M=

[0026] Where n is the total number of events in the extended behavior sequence; ωi is the behavior value weight of the i-th event; λ is the baseline decay coefficient; Δti is the time difference between the current time and the time when the i-th event occurs; μ is the category coherence decay suppression factor; SCi is the category coherence tag corresponding to the i-th event, which is 1 when the event belongs to the same target category as the previous event, and 0 otherwise; di is the dwell depth of the i-th event, which is the actual dwell time in seconds; To be from the previous behavior type To the current behavior type The conversion gain; α is the impulse penalty coefficient; Rm is the impulse operation count; β is the fatigue penalty coefficient; Hf is the coupon fatigue degree.

[0027] As a further aspect of the present invention, when the coupon redemption rate continuously declines for a preset number of days after the coupon is issued and pushed, the fatigue penalty coefficient β is increased; when the coupon redemption rate rises above a preset ratio, the impulse penalty coefficient α is increased.

[0028] Another object of the present invention is to provide a personalized coupon distribution system based on user behavior profiles, the system comprising:

[0029] The behavior sequence module is used to collect users' atomic behavior events, generate event records containing behavior type, timestamp, dwell depth and category identification, and sort them by time to form a user behavior sequence;

[0030] The extended sequence module is used to extract events related to the target category from the user behavior sequence for the target category to be evaluated, and to include adjacent cross-category current events to construct an extended behavior sequence. Each event in the extended behavior sequence is labeled with a category coherence tag, which indicates whether the event belongs to the same category as the previous event.

[0031] The coherence index module is used to calculate and sum the weighted contribution values ​​of the extended behavior sequence on an event-by-event basis, based on preset behavioral value weights, conversion gain values, and differentiated time decay information conditioned on category coherence tags, to generate a time-series coherence elasticity index that reflects the coherence and migration intensity of users’ interest in the target category.

[0032] The coupon issuance suitability module is used to determine the number of impulsive actions and coupon fatigue of users within a set time window. It integrates the time-series coherence elasticity index with the number of impulsive actions and coupon fatigue to calculate the coupon issuance suitability index.

[0033] The coupon configuration module compares the coupon issuance suitability index with the tiered threshold to determine the coupon issuance tier to which the user belongs. Then, it configures the coupon denomination, usage threshold, validity period, and applicable scope according to the strategy corresponding to the tier before issuing the coupon.

[0034] Compared with the prior art, the beneficial effects of the present invention are:

[0035] By constructing extended behavioral sequences and labeling them with category coherence tags, the continuity of user behavior within target categories and cross-category jump relationships can be captured. Differentiated time decay is implemented based on category coherence, integrating behavioral value weights and type conversion gains. The resulting temporal coherence elasticity index simultaneously reflects both the long-term accumulation of user loyalty to the target category and the intensity of instantaneous migration. Further integration of this index with impulsive action counts and coupon fatigue yields a coupon issuance suitability index that fundamentally distinguishes between loyal users, migrating users, impulsive users, and fatigued users. Finally, dynamic stratification allows for differentiated configuration of coupon amounts, thresholds, validity periods, and applicable scopes, protecting the platform's transaction ecosystem. Attached Figure Description

[0036] Figure 1 This is a flowchart of a personalized coupon distribution method based on user behavior profiles.

[0037] Figure 2 A flowchart for constructing an extended behavior sequence in a personalized coupon distribution method based on user behavior profiles.

[0038] Figure 3 A flowchart for generating a time-series coherent elasticity index in a personalized coupon distribution method based on user behavior profiles.

[0039] Figure 4 This is a flowchart for determining the count of impulsive actions and coupon fatigue in a personalized coupon distribution method based on user behavior profiles.

[0040] Figure 5 This is a structural diagram of a personalized coupon distribution system based on user behavior profiles. Detailed Implementation

[0041] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0042] The specific implementation of the present invention will be described in detail below with reference to specific embodiments.

[0043] like Figure 1 As shown, this embodiment of the invention provides a personalized coupon distribution method based on user behavior profiles, the method comprising the following steps:

[0044] S100 collects user atomic behavior events, generates event records containing behavior type, timestamp, dwell depth, and category identification, and sorts them by time to form a user behavior sequence.

[0045] In this embodiment of the invention, each user's operation on the terminal (such as browsing a product list, entering product details, adding to cart, removing from cart, adding to favorites, claiming coupons, placing an order with coupons, placing an order at full price, etc.) is defined as an atomic behavior event. A structured record is generated for each event, which must include: behavior type, precise timestamp, dwell time (if it is a page view, record the number of seconds the page was spent; if it is an instantaneous operation, record the number of seconds between the end of the previous behavior and the decision interval of this behavior), and the category of the product associated with the event. All events are arranged by timestamp to form the user's original behavior sequence.

[0046] S200, for the target category to be evaluated, extract events related to the target category from the user behavior sequence, and include adjacent cross-category pre-positioned events to construct an extended behavior sequence. Label each event in the extended behavior sequence with a category coherence tag. The category coherence tag indicates whether the event belongs to the same category as the previous event.

[0047] In this embodiment of the invention, the target category to be evaluated is determined as follows: if a user enters a product details page and stays there for more than a certain period of time (e.g., 10 seconds), then the category to which that product belongs is taken as the target category to be evaluated. The extended behavior sequence not only includes the target category behavior but also incorporates the preceding cross-category behavior that triggers the migration, and each event is marked to indicate whether it maintains category coherence with the previous event, providing a basis for differentiation decay. The category coherence tag is a binary identifier. When determining the category coherence tag, for each event in the extended behavior sequence, the product category target identifier it carries is extracted, and the position of the current event in the extended behavior sequence is checked: if the current event is the first event in the sequence, then there is no preceding event, and the category coherence tag value of 0 is directly assigned to it. If the current event is not the first event, then the immediately preceding event in the sequence is used as the comparison object. If the two events are in the same category, it indicates that the behavior is a continuous operation of the user within the target category, and the category continuity label is assigned a value of 1. If the two events are different, it indicates that a category jump or a cross-category pre-event occurred before the behavior, and the category continuity label is assigned a value of 0.

[0048] S300, based on preset behavioral value weights, conversion gain values, and differentiated time decay information conditioned on category coherence tags, calculates and sums the weighted contribution values ​​of the extended behavioral sequence for each event, generating a time-series coherence elasticity index.

[0049] In this embodiment of the invention, the temporal coherence elasticity index is a comprehensive quantification of a user's behavioral history regarding a target category. Different behavior types contribute different weights; transitions between adjacent behaviors (such as browsing → adding to cart) are given a gain to capture sequence patterns; decay is applied based on time difference, and if the category coherence tag indicates that the event belongs to the same target category as the previous event, the decay rate is suppressed. Finally, the summation yields the temporal coherence elasticity index, which reflects the coherence of interests and the intensity of migration.

[0050] S400 determines the number of impulsive actions and coupon fatigue of users within a set time window, and integrates the time-series coherence elasticity index with the number of impulsive actions and coupon fatigue to calculate the coupon issuance suitability index.

[0051] In this embodiment of the invention, the formula for calculating the coupon issuance suitability index M is:

[0052] M=

[0053] The numerator is the temporal coherence elasticity index, and the impulse count and fatigue level are included in the denominator. This effectively reduces the coupon suitability index for users with impulsive characteristics or who have become fatigued with coupons, thus integrating interest intensity and risk dimension into a single decision indicator. Where n is the total number of events in the extended behavioral sequence; ωi is the behavioral value weight of the i-th event; λ is the baseline decay coefficient; Δti is the time difference between the current time and the time of occurrence of the i-th event; μ is the category coherence decay suppression factor (0 < μ < 1); SCi is the category coherence tag corresponding to the i-th event, which is 1 when the event belongs to the same target category as the previous event, and 0 otherwise; di is the dwell depth of the i-th event, taken as the actual dwell time in seconds. To be from the previous behavior type To the current behavior type The formula is: α is the conversion gain; α is the impulsive penalty coefficient; Rm is the impulsive action count; β is the fatigue penalty coefficient; and Hf is the coupon fatigue level. This formula allows M to comprehensively reflect interest intensity, migration trend, impulsive risk, and fatigue level, with a single indicator driving differentiated distribution.

[0054] S500 compares the coupon issuance suitability index with the tiered threshold to determine the coupon issuance tier to which the user belongs. Then, it configures the coupon denomination, usage threshold, validity period, and applicable scope according to the strategy corresponding to the tier before issuing the coupon.

[0055] In this embodiment of the invention, the appropriateness index for issuing coupons is compared with a preset stratification threshold to divide users into, for example, a high-potential migration layer, a stable interest layer, a wait-and-see activation layer, and a suppressed layer. Different layers correspond to different combinations of coupon attributes. For example, for users in the high-potential migration layer, coupons of medium denomination with a minimum spending requirement, valid for 24 hours, and applicable only to the target category and related products they have migrated to are issued; for users in the stable interest layer, coupons of low denomination with a low minimum spending requirement, valid for 7 days, and applicable only to the category of products they have recently added to their cart but have not yet purchased are issued; for users in the wait-and-see activation layer, coupons of very small denomination with no minimum spending requirement, valid for 3 days, are issued and are valid for all stores; for users in the suppressed layer, no coupons are issued, and non-price-based content recommendations are provided instead.

[0056] like Figure 2 As shown, in a preferred embodiment of the present invention, the step of constructing the extended behavior sequence specifically includes:

[0057] S201, Traverse each event in the user behavior sequence. If the product category associated with the event is consistent with the target category, mark the event as the target event and retain it.

[0058] S202, For each retained target event, obtain its adjacent preceding event in the user behavior sequence. When the product category associated with the preceding event is different from the target category, include the preceding event and mark it as a cross-category preceding event.

[0059] S203, arrange all target events and included cross-class current events in their respective timestamp order to form an extended behavior sequence.

[0060] In this embodiment of the invention, the complete user behavior sequence is scanned sequentially. For each event, it is checked whether the corresponding product category target identifier is exactly the same as the target category to be evaluated. If they are the same, it indicates that the event is a direct interaction between the user and the target category, and it is retained and marked as a target event. For example, if a user browses the details of a product under the target category, this event is retained. In order to capture the migration source of the user jumping from other categories to the target category, the preceding event of each target event is further checked. If the category associated with the preceding event is different from the target category, it is also included in the sequence and marked as a cross-category pre-positioned event. For example, a migration chain from browsing the clothing category to directly adding items to the shoes category is completely retained. All marked target events and cross-category pre-positioned events are collected, deduplicated, and reordered according to their original timestamps to generate a new ordered sequence, namely the extended behavior sequence. This sequence contains both the coherent interaction within the target category and the cross-category context that triggers each interaction in the target category, providing a complete adjacency relationship for subsequent labeling of category coherence tags and calculation of behavior conversion gain.

[0061] like Figure 3 As shown, in a preferred embodiment of the present invention, the step of calculating and summing the weighted contribution values ​​for each event of the extended behavior sequence to generate the temporal coherence elasticity index specifically includes:

[0062] S301, to obtain the behavior value weight for each event in the extended behavior sequence, the behavior value weight is determined by looking up the table according to the behavior type of the event from the preset behavior value weight table;

[0063] S302, For each event, calculate the differential time decay factor based on the time difference between the current time and the timestamp, and the category coherence tag corresponding to the event;

[0064] S303, obtain the conversion gain value between the event and the behavior type of the previous event, which is determined by looking up the table in the preset conversion gain table based on the two behavior types;

[0065] S304, obtain the dwell depth of the event, and obtain the depth factor after smoothing the logarithm of the dwell depth;

[0066] S305, multiply the behavioral value weight, time decay factor, conversion gain value and depth factor of each event to obtain the weighted contribution value of the event to the target category;

[0067] S306, the weighted contribution values ​​of all events in the extended behavior sequence are summed to obtain the temporal coherence elasticity index.

[0068] In this embodiment of the invention, a behavior value weight table is maintained in the server memory. For example: browsing the list 0.3, entering details 0.6, functional interaction 0.8, adding to cart 1.5, favorite 1.2, placing an order with coupon 1.8, placing an order at full price 2.5, removing from cart -0.8, etc. For each event in the extended sequence, its behavior type field is extracted, and the corresponding behavior value weight ωi is obtained by looking up the table. The expression for the time decay factor is... When event i belongs to the same target category as the previous event, the exponent term becomes Since μ is between 0 and 1, the attenuation coefficient λ is suppressed, and the attenuation slows down; when SCi = 0, the exponential term becomes The decay coefficient remains at λ, decaying at a normal rate. This mechanism allows the historical value of consecutive similar behaviors to be preserved for a longer period. The server also maintains a conversion gain table, for example: Details → Add to Cart gain is 2.0, Add to Cart → Remove gain is 0.1, Browse List → Add to Cart gain is 1.6, etc. For the first event in the sequence (without a preceding event), the gain is conventionally set to 1.0. The dwell depth di is taken from the actual dwell time in the event record (or the equivalent number of seconds converted from the decision interval). To suppress the excessive influence of extremely long dwell times, ln(1+di) is calculated as a depth factor to ensure diminishing marginal contributions. For the i-th event, the weighted contribution value is... Finally, the contribution values ​​of the n events in the sequence are summed to obtain the temporal coherence elasticity index.

[0069] like Figure 4 As shown, in a preferred embodiment of the present invention, the steps of determining the user's impulsive action count and coupon fatigue level within a set time window specifically include:

[0070] S401, count the cumulative number of times a user triggers a preset impulsive behavior pattern within the first time window to obtain an impulsive operation count. The preset impulsive behavior pattern includes at least the following: completing the redemption within a preset time after receiving the coupon, cyclical operation of adding and removing the same product within a single day, immediately placing an order with the coupon after browsing across multiple categories, and returning the product after placing an order with the coupon.

[0071] S402: Calculate the total number of times a user is pushed coupons within the second time window, the number of coupons that the user actively claimed but did not use and thus expired, and the number of full-price transactions with unused coupons. Based on the relative relationship between the total number of pushes and the number of full-price transactions, and combined with the number of expired and unused coupons, generate a coupon fatigue level that represents the user's decreased sensitivity to coupons.

[0072] In this embodiment of the invention, the server scans the user's full behavioral sequence and coupon redemption records for the past 14 days (first time window), identifies the following impulse patterns, and accumulates them: redemption is completed within 8 minutes of receiving a coupon, and the user enters the details of the relevant category no more than once before redemption; the same product is added to the cart and removed from the cart twice or more in a single day; the user browses 3 different categories consecutively within 1 hour and then immediately places an order with a coupon; the user initiates a return of the order within 7 days after placing the order with a coupon; within 24 hours, the user adds to the cart more than 4 times with a decision interval of less than 5 seconds, and at least 2 of these orders are redeemed with coupons. The number of times the impulse pattern is matched is added together to obtain the impulse operation count.

[0073] In this embodiment of the invention, the server collects user data for the past 30 days (second time window): total number of coupon pushes P, number of coupons claimed but expired and unredeemed E, and number of full-price orders placed without using any coupons N. If P=0, the fatigue level is 0. If P>0, the base ratio r=(PN) / (N+1) is calculated, and then a temporary value tmp=r+0.5×E is generated. Finally, tmp is normalized to a specified interval (e.g., [0, 5]) using a preset segmented mapping rule to obtain the fatigue level. The higher the fatigue level, the lower the user's sensitivity to coupons and the stronger their resistance.

[0074] In this embodiment of the invention, when the coupon redemption rate continuously declines for a preset number of days after coupon distribution and push notification, the fatigue penalty coefficient β is increased. Specifically, the active redemption rate indicator after global or group coupon push notification is monitored. If this indicator continuously declines for a preset number of days (e.g., a continuous decline for 3 days), it indicates that the user group's response to the push notification is weakening, and the overall fatigue level may be increasing. At this time, the fatigue penalty coefficient β is increased by one step (e.g., 0.05), which strengthens the influence of the fatigue term in the denominator of the formula, thereby lowering the coupon distribution suitability index M for more users in the future, reducing the coupon distribution coverage, and avoiding further exacerbating fatigue. When the coupon redemption rate rises above a preset proportion, the impulse penalty coefficient α is increased. Specifically, if it is detected that the coupon redemption rate has abnormally increased above a preset proportion (e.g., 30%) compared to the previous period, it indicates that there may be impulsive arbitrage behavior mixed in. The server automatically increases the impulse penalty coefficient α to lower the index of this part of the users back to a reasonable level, preventing resources from being occupied by coupon hunters.

[0075] like Figure 5As shown, this embodiment of the invention also provides a personalized coupon distribution system based on user behavior profiles, the system comprising:

[0076] The behavior sequence module 100 is used to collect users' atomic behavior events, generate event records containing behavior type, timestamp, dwell depth and category identification, and sort them by time to form a user behavior sequence;

[0077] The extended sequence module 200 is used to extract events related to the target category from the user behavior sequence for the target category to be evaluated, and to include adjacent cross-category current events to construct an extended behavior sequence. Each event in the extended behavior sequence is labeled with a category coherence label, which indicates whether the event belongs to the same category as the previous event.

[0078] The coherence index module 300 is used to calculate and sum the weighted contribution value of the extended behavior sequence on an event-by-event basis based on the preset behavior value weight, conversion gain value, and differentiated time decay information with category coherence tags as conditions, to generate a time-series coherence elasticity index that reflects the user's interest coherence and migration intensity in the target category.

[0079] The coupon issuance suitability module 400 is used to determine the number of impulsive operations and coupon fatigue of users within a set time window. It integrates the time series coherence elasticity index with the number of impulsive operations and coupon fatigue to calculate the coupon issuance suitability index.

[0080] The coupon configuration module 500 is used to compare the coupon issuance suitability index with the tier threshold to determine the coupon issuance tier to which the user belongs, and then execute the issuance after configuring the coupon denomination, usage threshold, validity period and applicable scope according to the strategy corresponding to the tier.

[0081] In a preferred embodiment of the present invention, the extended sequence module 200 includes:

[0082] The target event unit is used to traverse each event in the user behavior sequence. If the product category associated with the event is consistent with the target category, the event is marked as the target event and retained.

[0083] The event inclusion unit is used to obtain the preceding adjacent event in the user behavior sequence for each retained target event. When the product category associated with the preceding event is different from the target category, the preceding event is included and marked as a cross-category preceding event.

[0084] The extended sequence unit is used to arrange all target events and the included cross-class current events in their respective timestamp order to form an extended behavior sequence.

[0085] In a preferred embodiment of the present invention, the coherence index module 300 includes:

[0086] The value weight unit is used to obtain the behavior value weight for each event in the extended behavior sequence. It is determined by looking up the behavior value weight table based on the behavior type of the event.

[0087] The time decay unit is used to calculate the differential time decay factor for each event based on the time difference between the current time and the timestamp, as well as the category coherence tag corresponding to the event.

[0088] The conversion gain unit is used to obtain the conversion gain value between the event and the previous event behavior type. It is determined by looking up the conversion gain table based on the two behavior types.

[0089] The depth factor unit is used to obtain the dwell depth of an event. The depth factor is obtained by smoothing the dwell depth by taking the logarithm.

[0090] The weighted contribution unit is used to multiply the behavioral value weight, time decay factor, conversion gain value and depth factor of each event to obtain the weighted contribution value of the event to the target category.

[0091] The elasticity index unit is used to sum the weighted contribution values ​​of all events in the extended behavior sequence to obtain the temporal coherence elasticity index.

[0092] In a preferred embodiment of the present invention, the coupon suitability module 400 includes:

[0093] The impulsive operation unit is used to count the cumulative number of times a user triggers a preset impulsive behavior pattern within the first time window, and obtain the impulsive operation count. The preset impulsive behavior pattern includes at least the following: completing the redemption within a preset time after receiving the coupon, cyclical operation of adding and removing the same product within a single day, immediately placing an order with a coupon after browsing across multiple categories, and returning the product after placing an order with a coupon.

[0094] The fatigue unit is used to count the total number of times a user is pushed coupons within the second time window, the number of coupons that are actively claimed but not used and thus expired, and the number of full-price transactions with unused coupons. Based on the relative relationship between the total number of pushes and the number of full-price transactions, and combined with the number of expired and unused coupons, a coupon fatigue level is generated that represents the user's decreased sensitivity to coupons.

[0095] The above description only details the preferred embodiments of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

[0096] It should be understood that although the steps in the flowcharts of the various embodiments of the present invention are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the various embodiments may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least a portion of the sub-steps or stages of other steps.

[0097] Other embodiments of this disclosure will readily occur to those skilled in the art upon consideration of the disclosure in the specification and embodiments. This application is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not disclosed herein. The specification and embodiments are to be considered exemplary only, and the true scope and spirit of this disclosure are indicated by the claims.

Claims

1. A personalized coupon distribution method based on user behavior profiles, characterized in that, The method includes the following steps: Collect users' atomic behavior events, generate event records containing behavior type, timestamp, dwell time and category identification, and sort them by time to form a user behavior sequence; For the target category to be evaluated, events related to the target category are extracted from the user behavior sequence, and adjacent cross-category pre-positioned events are included to construct an extended behavior sequence. Each event in the extended behavior sequence is labeled with a category coherence tag. The category coherence tag indicates whether the event belongs to the same category as the previous event. Based on preset behavioral value weights, conversion gain values, and differentiated time decay information conditioned on category coherence tags, the weighted contribution values ​​of the extended behavioral sequence are calculated and summed for each event to generate a time-series coherence elasticity index that reflects the coherence and migration intensity of users’ interest in the target category. Determine the number of impulsive actions and coupon fatigue of users within a set time window, and combine the time series coherence elasticity index with the number of impulsive actions and coupon fatigue to calculate the coupon issuance suitability index; The coupon issuance suitability index is compared with the tiered threshold to determine the coupon issuance tier to which the user belongs. The coupon amount, usage threshold, validity period and applicable scope are configured according to the strategy corresponding to the tier before the coupon is issued.

2. The personalized coupon distribution method based on user behavior profiles according to claim 1, characterized in that, The steps for constructing an extended behavior sequence include: Iterate through each event in the user behavior sequence. If the product category associated with the event matches the target category, mark the event as the target event and retain it. For each retained target event, obtain its adjacent preceding event in the user behavior sequence. When the product category associated with the preceding event is different from the target category, include the preceding event and mark it as a cross-category preceding event. All target events and included cross-class current events are arranged in their respective timestamp order to form an extended behavior sequence.

3. The personalized coupon distribution method based on user behavior profiles according to claim 1, characterized in that, The steps for calculating and summing the weighted contribution values ​​for each event in the extended behavior sequence to generate the time-series coherent elasticity index specifically include: To obtain the behavior value weight for each event in the extended behavior sequence, the weight is determined by looking up the behavior type of the event in a pre-defined behavior value weight table. For each event, a differential time decay factor is calculated based on the time difference between the current moment and the timestamp, as well as the category coherence tag corresponding to the event. Obtain the conversion gain value between the event and the behavior type of the previous event, which is determined by looking up the table in the preset conversion gain table based on the two behavior types; Obtain the dwell depth of the event, and then smooth the dwell depth by taking the logarithm to get the depth factor; Multiply the behavioral value weight, time decay factor, conversion gain value and depth factor of each event to obtain the weighted contribution value of the event to the target category; The temporal coherence elasticity index is obtained by summing the weighted contribution values ​​of all events in the extended behavior sequence.

4. The personalized coupon distribution method based on user behavior profiles according to claim 1, characterized in that, The steps for determining the number of impulsive actions and coupon fatigue of users within a set time window include: The cumulative number of times a user triggers a preset impulsive behavior pattern within the first time window is counted to obtain an impulsive operation count. The preset impulsive behavior pattern includes at least the following: completing the redemption within a preset time after receiving a coupon, cyclical operation of adding and removing the same product from the cart within a single day, immediately placing an order with a coupon after browsing across multiple categories, and returning the product after placing an order with a coupon. The system counts the total number of coupons pushed to users within the second time window, the number of coupons that users actively claimed but did not use and thus expired, and the number of full-price transactions using unused coupons. Based on the relative relationship between the total number of coupons pushed and the number of full-price transactions, and combined with the number of expired and unused coupons, a coupon fatigue level is generated, which represents the decrease in users' sensitivity to coupons.

5. The personalized coupon distribution method based on user behavior profiles according to claim 1, characterized in that, The formula for calculating the coupon issuance suitability index M is as follows: M= Where n is the total number of events in the extended behavior sequence; ωi is the behavior value weight of the i-th event; λ is the baseline decay coefficient; Δti is the time difference between the current time and the time when the i-th event occurs; μ is the category coherence decay suppression factor; SCi is the category coherence tag corresponding to the i-th event, which is 1 when the event belongs to the same target category as the previous event, and 0 otherwise; di is the dwell depth of the i-th event, which is the actual dwell time in seconds; To be from the previous behavior type To the current behavior type The conversion gain; α is the impulse penalty coefficient; Rm is the impulse operation count; β is the fatigue penalty coefficient; Hf is the coupon fatigue degree.

6. The personalized coupon distribution method based on user behavior profiles according to claim 5, characterized in that, When the coupon redemption rate declines continuously for a preset number of days after the coupon is issued, the fatigue penalty coefficient β is increased; when the coupon redemption rate rises above a preset ratio, the impulse penalty coefficient α is increased.

7. A personalized coupon distribution system based on user behavior profiles, characterized in that, The system includes: The behavior sequence module is used to collect users' atomic behavior events, generate event records containing behavior type, timestamp, dwell depth and category identification, and sort them by time to form a user behavior sequence; The extended sequence module is used to extract events related to the target category from the user behavior sequence for the target category to be evaluated, and to include adjacent cross-category current events to construct an extended behavior sequence. Each event in the extended behavior sequence is labeled with a category coherence tag, which indicates whether the event belongs to the same category as the previous event. The coherence index module is used to calculate and sum the weighted contribution values ​​of the extended behavior sequence on an event-by-event basis, based on preset behavioral value weights, conversion gain values, and differentiated time decay information conditioned on category coherence tags, to generate a time-series coherence elasticity index that reflects the coherence and migration intensity of users’ interest in the target category. The coupon issuance suitability module is used to determine the number of impulsive actions and coupon fatigue of users within a set time window. It integrates the time-series coherence elasticity index with the number of impulsive actions and coupon fatigue to calculate the coupon issuance suitability index. The coupon configuration module compares the coupon issuance suitability index with the tiered threshold to determine the coupon issuance tier to which the user belongs. Then, it configures the coupon denomination, usage threshold, validity period, and applicable scope according to the strategy corresponding to the tier before issuing the coupon.

8. The personalized coupon distribution system based on user behavior profiles according to claim 7, characterized in that, The extended sequence module includes: The target event unit is used to traverse each event in the user behavior sequence. If the product category associated with the event is consistent with the target category, the event is marked as the target event and retained. The event inclusion unit is used to obtain the preceding adjacent event in the user behavior sequence for each retained target event. When the product category associated with the preceding event is different from the target category, the preceding event is included and marked as a cross-category preceding event. The extended sequence unit is used to arrange all target events and the included cross-class current events in their respective timestamp order to form an extended behavior sequence.

9. The personalized coupon distribution system based on user behavior profiles according to claim 7, characterized in that, The coherence index module includes: The value weight unit is used to obtain the behavior value weight for each event in the extended behavior sequence. It is determined by looking up the behavior value weight table based on the behavior type of the event. The time decay unit is used to calculate the differential time decay factor for each event based on the time difference between the current time and the timestamp, as well as the category coherence tag corresponding to the event. The conversion gain unit is used to obtain the conversion gain value between the event and the previous event behavior type. It is determined by looking up the conversion gain table based on the two behavior types. The depth factor unit is used to obtain the dwell depth of an event. The depth factor is obtained by smoothing the dwell depth by taking the logarithm. The weighted contribution unit is used to multiply the behavioral value weight, time decay factor, conversion gain value and depth factor of each event to obtain the weighted contribution value of the event to the target category. The elasticity index unit is used to sum the weighted contribution values ​​of all events in the extended behavior sequence to obtain the temporal coherence elasticity index.

10. The personalized coupon distribution system based on user behavior profiles according to claim 7, characterized in that, The coupon issuance suitability module includes: The impulsive operation unit is used to count the cumulative number of times a user triggers a preset impulsive behavior pattern within the first time window, and obtain the impulsive operation count. The preset impulsive behavior pattern includes at least the following: completing the redemption within a preset time after receiving the coupon, cyclical operation of adding and removing the same product within a single day, immediately placing an order with a coupon after browsing across multiple categories, and returning the product after placing an order with a coupon. The fatigue unit is used to count the total number of times a user is pushed coupons within the second time window, the number of coupons that are actively claimed but not used and thus expired, and the number of full-price transactions with unused coupons. Based on the relative relationship between the total number of pushes and the number of full-price transactions, and combined with the number of expired and unused coupons, a coupon fatigue level is generated that represents the user's decreased sensitivity to coupons.

Images