Marketing method based on intelligent commodity background matching

Abstract

The application discloses a marketing method based on intelligent commodity background matching, and relates to the technical field of commodity marketing, and specifically comprises the following steps: real-time acquisition of new commodity browsing information generated by a user when browsing commodities that have not been paid attention to before during a backflow behavior, and analysis after acquisition, generation of a backflow behavior interest index and a backflow purchase intention coefficient; construction of an effective evaluation model for the generated backflow behavior interest index and backflow purchase intention coefficient, generation of an effective evaluation coefficient; analysis of the generated effective evaluation coefficient, evaluation of whether the backflow behavior of the user is effective, and generation of an effective signal and a non-effective signal according to the evaluation result. The application solves the problem that the existing recommendation system cannot identify and evaluate the effectiveness of the backflow behavior of the user, realizes accurate recommendation based on the backflow demand of the user by dynamically adjusting the recommendation strategy, and improves the recommendation accuracy and the platform conversion rate.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of commodity marketing, specifically to a marketing method based on intelligent commodity background matching. BACKGROUND

[0002] The marketing method based on intelligent commodity background matching is a marketing approach that uses intelligent algorithms and big data technology to achieve precise matching between commodities and target users by analyzing the background information of commodities (such as usage scenarios, user behavior data, geographic location, time, etc.) and personalized user needs. This method can push the most suitable commodities or services to users in specific situations, greatly improving marketing efficiency and user experience. The reason for using intelligent commodity background matching for marketing is that traditional marketing models usually rely on extensive coverage and inefficient promotion methods, while background matching technology can deeply explore the core value of commodities and the implicit needs of users, avoiding information overload and resource waste, and improving user recognition and purchase rate of products through precise pushing, thereby creating higher business value for enterprises.

[0003] The existing marketing technology based on intelligent commodity background matching conducts marketing in the following way: first, the system collects commodity background data, including the attributes of the commodity, the usage scenarios, the target user portrait, etc., and obtains user behavior data such as browsing history, purchase records, geographic location, real-time environment (such as weather, time), etc. Then, using big data analysis and machine learning algorithms, the commodity background data and user data are intelligently matched to generate personalized recommendation content. The system analyzes the matching results to determine the potential needs of users and pushes related commodities or services at the most suitable time, place, and scene (such as mobile applications, social media ads, emails, etc.). This process usually integrates natural language processing, image recognition, and deep learning technologies to better understand user preferences and commodity characteristics, thereby providing personalized marketing solutions with high relevance, improving user purchase willingness and experience satisfaction.

[0004] The existing technology has the following shortcomings:

[0005] In an e-commerce platform, when a user does not perform any purchase or browsing behavior within a period of time, the system usually considers that the user is no longer interested in the relevant goods. But in some cases, such as holiday promotion, advertisement push or guidance of specific events, the user's interest may suddenly flow back and start browsing goods or related accessories that have not been paid attention to before. For example, a user bought a smart sound a few months ago, and after that, he did not browse similar goods until the promotion period, when the user suddenly started browsing smart sound and its accessories. The existing recommendation technology based on smart goods background matching often only relies on the user's historical purchase and browsing behavior to generate recommendations, ignoring the dynamic changes of the user's "backflow behavior" of new demand. Therefore, the system will still recommend outdated goods based on the user's past purchase records, failing to identify the user's new demand generated during the promotion period in time. Since the existing recommendation technology fails to distinguish and evaluate the effectiveness of such backflow behavior, it cannot adjust the recommendation logic in time, so that the recommended content is still based on outdated data, failing to push goods that meet the user's current demand. This will cause the recommended goods to not match the user's actual interest, thereby reducing the accuracy of the recommendation system, leading to user dissatisfaction with the recommendation system, and even possible loss, thereby affecting the platform's conversion rate and overall sales benefit.

[0006] The above information disclosed in the background section is only intended to strengthen the understanding of the background of the present disclosure, and therefore it can include information that does not constitute prior art known to those of ordinary skill in the art. SUMMARY

[0007] The purpose of the present application is to provide a marketing method based on smart goods background matching to solve the problems in the background.

[0008] In order to achieve the above-mentioned purpose, the present application provides the following technical solution: a marketing method based on smart goods background matching, specifically comprising the following steps:

[0009] By monitoring the user's browsing and purchase behavior on the e-commerce platform in real time, the backflow behavior of the user is identified, and when the backflow behavior of the user is identified, it is detected whether the user browses goods that have not been paid attention to before during the backflow behavior, and in the case that the user browses goods that have not been paid attention to before during the backflow behavior, the following steps are performed;

[0010] Real-time acquisition of backflow behavior new goods browsing information generated when the user browses goods that have not been paid attention to before during the backflow behavior, and analysis after acquisition, respectively generating backflow behavior interest index and backflow purchase intention coefficient;

[0011] An effective evaluation model is constructed for the generated backflow behavior interest index and backflow purchase intention coefficient to generate an effective evaluation coefficient;

[0012] analyzing the generated effective evaluation coefficients to evaluate whether the backflow behavior of the user is effective, and generating an effective signal and a non-effective signal according to the evaluation result;

[0013] In the case of generating an effective signal, a plurality of effective evaluation coefficients generated subsequently are continuously acquired and analyzed after generation to generate signals of different levels, and corresponding recommended measures are taken according to the generated signals of different levels;

[0014] The backflow behavior and backflow behavior new commodity browsing information of the user are continuously monitored, and the parameters of the backflow behavior effective evaluation model and the recommended strategy are dynamically adjusted according to the real-time monitoring result, and historical data are stored and analyzed to optimize the model parameters and the recommended strategy.

[0015] Preferably, the backflow behavior new commodity browsing information generated by the user when browsing commodities that have not been previously paid attention to during the backflow behavior is acquired in real time, and analyzed after acquisition to generate a backflow behavior interest index and a backflow purchase intention coefficient, specifically including the following steps:

[0016] The backflow behavior new commodity browsing information generated by the user when browsing commodities that have not been previously paid attention to during the backflow behavior is acquired in real time, and preprocessed after acquisition;

[0017] The backflow behavior interest information and the backflow purchase intention information in the preprocessed backflow behavior new commodity browsing information are extracted;

[0018] The extracted backflow behavior interest information and the backflow purchase intention information are analyzed to generate a backflow behavior interest index and a backflow purchase intention coefficient, respectively.

[0019] Preferably, the acquisition logic of the backflow behavior interest index is as follows:

[0020] The backflow behavior interest information in the preprocessed backflow behavior new commodity browsing information is extracted, specifically including the browsing times of new commodities by the user in different time periods during the backflow period, the dwell time of new commodities, and the ratio of the click times to the display times of new commodities, and are respectively marked as , and , represents the browsing times of new commodities by the user in the time period during the backflow period, represents the dwell time of new commodities by the user in the time period during the backflow period, represents the ratio of the click times to the display times of new commodities by the user in the time period during the backflow period, , is a positive integer;

[0021] construct a set of the dwell time of the user on the new commodity in different time periods during the reflux period, and mark the maximum value in the set as ;

[0022] calculate the reflux behavior interest index, and the specific calculation formula is as follows: wherein, is the reflux behavior interest index.

[0023] Preferably, the acquisition logic of the reflux purchase intention coefficient is as follows:

[0024] extract the reflux purchase intention information in the preprocessed reflux behavior new commodity browsing information, specifically including the number of times of adding the new commodity to the shopping cart by the user in different time periods during the reflux period, the conversion rate of purchasing the new commodity, and the number of times of collecting the new commodity, and mark them as , and , respectively, wherein represents the number of times of adding the new commodity to the shopping cart by the user in the time period during the reflux period, represents the conversion rate of purchasing the new commodity by the user in the time period during the reflux period, represents the number of times of collecting the new commodity by the user in the time period during the reflux period, , is a positive integer;

[0025] construct a set of the conversion rate of purchasing the new commodity by the user in different time periods during the reflux period, and mark the maximum value in the set as ;

[0026] calculate the reflux purchase intention coefficient, and the specific calculation formula is as follows: wherein, is the reflux purchase intention coefficient.

[0027] Preferably, the effective evaluation model is constructed for the generated reflux behavior interest index and the reflux purchase intention coefficient , and the effective evaluation coefficient is generated by weighted summation, and the specific calculation formula is as follows: wherein, is the effective evaluation coefficient, and are non-zero weight coefficients of the reflux behavior interest index and the reflux purchase intention coefficient , respectively, and .

[0028] Preferably, a preset effective evaluation coefficient threshold value is determined , and after being determined, the generated effective evaluation coefficient is compared, and according to the comparison result, it is evaluated whether the reflux behavior of the user is effective, and the specific comparison analysis is as follows:

[0029] If , the reflux behavior of the user is not effective;

[0030] If , the reflux behavior of the user is effective.

[0031] Preferably, according to the evaluation result, an effective signal and an ineffective signal are generated respectively, and specifically:

[0032] If the evaluation result is that the reflux behavior of the user is not effective, an ineffective signal is generated;

[0033] If the evaluation result is that the reflux behavior of the user is effective, an effective signal is generated.

[0034] Preferably, in the case of generating an effective signal, a plurality of subsequent generated effective evaluation coefficients are continuously acquired, and are re-labeled as , is the number of the plurality of subsequent generated effective evaluation coefficients in the case of generating an effective signal, , is a positive integer;

[0035] The average value of the plurality of subsequent generated effective evaluation coefficients is calculated, according to the formula: ;

[0036] The standard deviation of the plurality of subsequent generated effective evaluation coefficients is calculated, according to the formula: .

[0037] Preferably, a preset average value threshold value and a preset standard deviation threshold value of the plurality of subsequent generated effective evaluation coefficients in the case of generating an effective signal are determined, and after being determined, the average value and the standard deviation of the plurality of effective evaluation coefficients are compared respectively, and according to the comparison result, signals of different levels are generated, and the specific comparison analysis is as follows:

[0038] If and , a first level signal is generated;

[0039] If and , a second level signal is generated;

[0040] If and , a third level signal is generated;

[0041] If and , a fourth level signal is generated.

[0042] Preferably, corresponding recommended measures are taken according to the generated different levels of signals, specifically:

[0043] If the first level signal is generated, the recommended measures taken are: increasing the recommendation frequency, improving the exposure rate of the new commodity in the user recommendation list, and pushing it to the home page and popular recommendation of the platform, etc.

[0044] If the second level signal is generated, the recommended measures taken are: reducing the recommendation frequency, maintaining the recommendation of the new commodity, but reducing its display priority and frequency;

[0045] If the third level signal is generated, the recommended measures taken are: maintaining the existing recommendation frequency, providing new commodity recommendation according to the current interest intensity of the user;

[0046] If the fourth level signal is generated, the recommended measures taken are: stopping the recommendation, and removing the new product from the user's recommendation list.

[0047] In the above technical solution, the technical effects and advantages provided by the present application are:

[0048] 1、The present application effectively solves the blind spot problem of the existing e-commerce platform recommendation system when the user's interest changes by monitoring the user's backflow behavior in real time and combining the new commodity browsing information of the backflow behavior. The traditional recommendation system usually recommends commodities based on historical data, assuming that the user's interest is fixed. This way ignores the user's backflow behavior, that is, after a period of time without purchase or browsing, the user's interest may recover or change due to promotion, advertisement push or other events. By capturing and analyzing the interest fluctuations during the user's backflow period in real time, the technical solution can identify and handle the user's new needs, thereby ensuring that the recommendation system is more flexible and accurate in responding to the user's dynamic needs.

[0049] 2、The application realizes multi-dimensional evaluation of user's return flow behavior by generating return flow behavior interest index and return flow purchase intention coefficient, and combining the two to build an effective evaluation model. These evaluation coefficients can accurately measure the user's interest intensity and purchase potential for new goods, avoiding static analysis relying only on historical behavior data. By weighted summation to generate an effective evaluation coefficient and comparison with the set threshold, the scheme can effectively distinguish between effective and ineffective return flow behavior, ensure real-time adjustment of the recommendation logic, reduce resource waste, and improve the accuracy of recommendations and user satisfaction.

[0050] 3、The application can realize personalized recommendation strategy adjustment by generating different levels of signals and taking corresponding recommendation measures according to these signals. The recommendation measures corresponding to different signal levels include increasing the recommendation frequency, adjusting the recommendation priority, maintaining the existing recommendation frequency, and stopping the recommendation, which enables the system to adjust the recommendation intensity in time according to the changes in user return flow behavior, thereby effectively improving the conversion rate and sales benefit of the platform. By continuously monitoring and analyzing historical data, the system can also optimize the parameters of the return flow behavior evaluation model and dynamically adjust the recommendation strategy, so that the platform can adapt to the long-term changes in user demand, improving the adaptability and flexibility of the system. Overall, the technical solution not only improves the accuracy of the recommendation system, but also ensures efficient use of resources, improves user experience and enhances the competitiveness of the platform. BRIEF DESCRIPTION OF DRAWINGS

[0051] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed in the embodiments will be briefly introduced as follows. Obviously, the drawings in the following description are only some embodiments described in the present application, and other drawings can also be obtained by those skilled in the art based on these drawings.

[0052] Figure 1 The flowchart of the marketing method based on intelligent commodity background matching of the present application. DETAILED DESCRIPTION

[0053] Example implementations will now be described more fully with reference to the accompanying drawings. Example implementations may, however, be implemented in many different forms and should not be construed as limited to the examples set forth herein; rather, these example implementations are provided so that this disclosure will be thorough and complete, and will fully convey the scope of example implementations to those skilled in the art.

[0054] The present application provides a marketing method based on intelligent commodity background matching as shown in Figure 1 The marketing method based on intelligent commodity background matching of the present application specifically comprises the following steps:

[0055] By monitoring the user's browsing and purchasing behavior on the e-commerce platform in real time, the user's backflow behavior is identified. When the user's backflow behavior is identified, it is detected in real time whether the user browses the previously unattended goods during the backflow behavior. If it is detected that the user browses the previously unattended goods during the backflow behavior, the following steps are performed.

[0056] By monitoring the user's browsing and purchasing behavior on the e-commerce platform in real time, the user's backflow behavior is identified. When the user's backflow behavior is identified, it is detected in real time whether the user browses the previously unattended goods during the backflow behavior. If it is detected that the user browses the previously unattended goods during the backflow behavior, the following steps are performed.

[0057] When the system identifies the user's backflow behavior, it needs to detect in real time whether the goods browsed by the user during the backflow are previously unattended goods. This can be achieved by establishing a user preference model based on the user's past browsing and purchasing data to identify the user's preferred product categories, brands and types. When backflow occurs, the system will compare the currently browsed goods of the user with the user's historical browsing or purchasing records to see if there is any overlap. If the currently browsed goods are not in the historical records, it is considered that the user has browsed previously unattended goods. This process can be quickly compared through a recommendation engine combined with collaborative filtering or content recommendation algorithm to ensure that each backflow user can accurately match their current interest points.

[0058] This implementation process can solve the problem that the existing recommendation system relies too much on historical data and fails to dynamically respond to changes in user demand. During holiday promotions, ad pushes or specific event guides, users' interests often change, and traditional recommendation systems will continue to push goods based on outdated historical behavior, failing to meet users' new demands. By detecting the user's browsing behavior of unattended goods during backflow in real time, the system can capture these new interest points in time and make corresponding adjustments in the recommendation logic, thereby pushing goods that meet the user's current needs. This not only improves the accuracy of recommendations, but also enhances user experience, avoiding user churn and low conversion rates caused by the lack of dynamic response of the recommendation system.

[0059] Real-time acquisition of new commodity browsing information generated by the user when browsing previously unattended commodities during the backflow behavior, and analysis after acquisition, respectively generating backflow behavior interest index and backflow purchase intention coefficient;

[0060] In this embodiment, real-time acquisition of new commodity browsing information generated by the user when browsing previously unattended commodities during the backflow behavior, and analysis after acquisition, respectively generating backflow behavior interest index and backflow purchase intention coefficient, specifically including the following steps:

[0061] Real-time acquisition of new commodity browsing information generated by the user when browsing previously unattended commodities during the backflow behavior, and preprocessing after acquisition;

[0062] To real-time acquire new commodity browsing information generated by the user when browsing previously unattended commodities during the backflow behavior, a real-time event tracking system can be integrated into the e-commerce platform. This system can automatically record user behavior data such as clicks, browsing, and dwell time every time the user accesses the platform. Specifically, when the user backflows to the platform, the system will capture and store the user's browsing behavior in real time, especially the first time browsing previously unattended commodities. This can be achieved through user ID-based tracking, where the system compares the user's historical browsing records with current behavior to identify whether the currently browsed commodity is one that the user has not previously browsed. In addition, the system will transmit this information to the background database in real time to ensure the immediacy and accuracy of the data, avoiding delays that affect data processing results.

[0063] Preprocessing is an important step in data analysis, aiming to convert raw behavior data into a form suitable for analysis and calculation. In the real-time acquired new commodity browsing information of backflow behavior, there may be noise data (such as abnormal clicks, invalid browsing, etc.) or inconsistent data formats, so preprocessing operations such as denoising, normalization, and standardization are needed. Denoising operations can be completed by setting filtering rules (such as excluding records with invalid browsing time below a certain threshold) to ensure data validity. Normalization is to convert different range data (such as browsing time and click frequency) to a unified scale to ensure comparability between different data types. Standardization processing is to process data so that different user behavior data is calculated under the same standard, avoiding data bias due to large differences in user behavior. Through these preprocessing steps, the system can ensure the accuracy of subsequent data analysis and evaluation, and provide clean and standardized input data for generating backflow behavior interest index and backflow purchase intention coefficient.

[0064] Extract the backflow behavior interest information and backflow purchase intention information in the preprocessed new commodity browsing information of backflow behavior;

[0065] The reflow behavior interest information and the reflow purchase intention information in the preprocessed reflow behavior new commodity browsing information can be extracted by data mining and feature extraction methods. First, the reflow behavior interest information can be extracted by analyzing the user's browsing behavior during the reflow period. This includes screening the user's browsing frequency, dwell time, page switching frequency and other features from the preprocessed data to measure the user's interest in new commodities during the reflow period. Specifically, the system can calculate the interest intensity index according to the user's browsing time and access frequency of each commodity, and integrate these data into an interest index through a set algorithm (such as weighted average method). For reflow purchase intention information, the extraction process focuses on the user's shopping cart behavior, purchase conversion rate and collection behavior. The system analyzes whether the user adds new commodities to the shopping cart, whether the purchase is completed, and whether the commodity is added to the collection during the reflow period to judge the purchase potential. In order to extract this information, the system will real-time grab user behavior logs from the database, and mark specific actions related to purchase intention, such as the number of times of adding to the shopping cart, browsing to purchase conversion rate, and the frequency of collection behavior. Through these steps, the system can quantify the user's purchase potential during the reflow period, providing necessary basis for subsequent reflow behavior analysis. This process relies entirely on data mining algorithms and real-time data analysis techniques to ensure that the extracted features accurately reflect the user's current interest and purchase intention.

[0066] The extracted reflow behavior interest information and reflow purchase intention information are analyzed to generate a reflow behavior interest index and a reflow purchase intention coefficient, respectively.

[0067] In this embodiment, the logic for obtaining the reflow behavior interest index is as follows:

[0068] The reflow behavior interest information in the preprocessed reflow behavior new commodity browsing information is extracted, including the number of times the user browses new commodities in different time periods during the reflow period, the dwell time of new commodities, and the ratio of the number of clicks to the number of displays of new commodities, and are respectively labeled as , and , represents the number of times the user browses new commodities in the time period during the reflow period, represents the dwell time of new commodities by the user in the time period during the reflow period, represents the ratio of the number of clicks to the number of displays of new commodities by the user in the time period during the reflow period, , is a positive integer;

[0069] To obtain the "number of views of new products by users at different time periods during the backflow period, the dwell time on new products, and the ratio of click-through rate to display frequency", a user behavior data collection and analysis system can be integrated in the e-commerce platform. This system needs to monitor every page access and interaction of the user in real time and record relevant data in the background. Specifically, the number of views can be obtained by recording each user's access to a product page event. Each time the user loads the product detail page, the system will increase the view counter of the product and associate the user's unique identifier (such as user ID). The dwell time can be obtained by calculating the time period of the user's stay on a product page. This is usually done by starting the timer when the user enters the product detail page and recording the length of stay on the page until the user leaves the page or jumps to another page. The ratio of click-through rate to display frequency can be obtained by recording the number of times a product is displayed when the product page is loaded, increasing the click count each time the user clicks on the product, and finally calculating the ratio of click-through rate to display frequency. All these data will be captured in real time by the event tracking system (such as JavaScript event listener) when user behavior occurs, and transmitted to the backend database through the data collection system (such as Google Analytics, Mixpanel, etc.) for storage and management. Then, the system processes and analyzes these raw data according to time periods (such as every hour, every half hour, etc.) to generate backflow behavior interest information. Finally, these processed data will be used to generate backflow behavior interest information by calculating the view frequency, dwell time and click-through rate to evaluate the user's interest in new products.

[0070] The dwell time of users on new products during the backflow period is constructed into a set, and the maximum value in the set is marked as ;

[0071] The backflow behavior interest index is calculated, and the specific formula is as follows: wherein, is the backflow behavior interest index.

[0072] The backflow behavior interest index is calculated by combining logarithmic and exponential operations, which aims to comprehensively and accurately measure the user's interest in new products during the backflow period. First, the dwell time is processed using exponential operation to emphasize the impact of long dwell time on user interest. Dwell time reflects the user's in-depth attention to a product, and long dwell time usually means higher interest. Using exponential operation can non-linearly amplify this interest intensity and highlight the user's deep interaction. Second, the view frequency The logarithmic operation is used to balance the relationship between the number of user browsing the goods and the interest intensity. The logarithmic function can suppress the excessive influence of high-frequency browsing on the interest index, while ensuring that frequently browsed goods are properly represented in the index. Furthermore, the way of calculating the click rate aims to amplify the impact of the click rate on interest while avoiding division by zero errors. The click rate itself is a direct manifestation of user interest in the goods, and a high click rate means that the user has a high interest in the goods, so the contribution of goods with high click rates to the final interest index is enhanced through logarithmic operation. In addition, the average value calculation in the formula can average the user interest in different stages in multiple time periods, further ensuring the comprehensive evaluation of user behavior during the return flow. In summary, through these operation methods, RII can better reflect the user's overall interest in new goods during the return flow, ensuring that the recommendation system can accurately capture changes in user interest and make corresponding adjustments.

[0073] The size of the return behavior interest index (RII) directly reflects the user's interest intensity in new goods during the return flow, therefore, there is a close relationship between the value of RII and "evaluating whether the user's return behavior is effective". When the RII value is high, it means that the user shows strong interest in new goods during the return flow, which usually means that the user has high purchase potential and strong engagement during the return flow. Therefore, the system considers that the return behavior is effective and adjusts the recommendation strategy accordingly, pushing more goods that meet the user's interest and improving the recommendation accuracy. Conversely, when the RII value is low, it indicates that the user has weak interest in new goods, which may be accidental browsing or unintentional interaction, and the system will judge that the return behavior is ineffective or has low interest, and the recommendation system will reduce the recommendation of the goods or maintain the existing recommendation strategy. Therefore, the high and low of RII provides a quantitative basis for the system, by evaluating the value, the system can effectively distinguish which return behaviors truly reflect the user's current needs, and then optimize the recommended content, improve the conversion rate of the platform and user experience.

[0074] In this embodiment, the acquisition logic of the return purchase intention coefficient is as follows:

[0075] Extract the return purchase intention information in the preprocessed return behavior new goods browsing information, specifically including the number of times the user adds new goods to the shopping cart in different time periods during the return period, the conversion rate of purchasing new goods, and the number of times the user collects new goods, and respectively marked as , and , represents the number of times the user adds new goods to the shopping cart in the time period during the return period, represents the conversion rate of purchasing new goods in the time period during the return period,​ represents the number of times a user adds a new product to their cart during the reflow period the number of times a user adds a new product to their cart during the reflow period , is a positive integer.

[0076] To obtain the number of times a user adds a new product to their cart, the conversion rate of purchasing a new product, and the number of times a user collects a new product during different time periods during the reflow period, a user behavior data collection and event tracking system can be integrated into the e-commerce platform. First, the number of times a product is added to the cart can be recorded in real time by the system in the background when the user clicks on the product and adds it to the cart. Each time the user clicks the "Add to Cart" button on the product page, the system records the event and marks each product with a user ID and timestamp, and groups the behavior according to the time period (e.g. hours, days, etc.) to calculate the number of additions in each time period. Second, the purchase conversion rate is obtained by purchasing data, the system records the number of times a product is viewed and purchased, and then calculates the ratio of the number of purchases to the number of views of the product during the reflow period to obtain the conversion rate. Products with high conversion rates usually indicate strong purchase intentions of users during the reflow period. Finally, the number of collections can be recorded by the user clicking on the "Add to Favorites" or "Add to Wish List" button for the product. Each time the user adds a product to the favorites, the system updates the number of collections for that product in real time, and the number of collections is counted by time period to obtain the collection behavior data of new products by users during the reflow period. These data can be captured in real time by an event tracking system (such as a JavaScript event listener or custom tracking code), and stored and analyzed by a data analysis platform (such as Google Analytics, Mixpanel, or a self-built data warehouse). The system associates these behavior data with user ID, time period, and product ID to analyze and evaluate the behavior of each user during the reflow period in detail, and ultimately helps generate the reflow purchase intention coefficient. In this way, the platform can accurately capture and record user shopping behavior in real time without interfering with the user experience.

[0077] The maximum value in the set is designated as ;

[0078] The reflow purchase intention coefficient is calculated as follows: wherein is the reflow purchase intention coefficient.

[0079] The reflow purchase intention coefficient The calculation formula design combines exponential operation and logarithmic operation, aiming to more accurately measure the purchase intention of users during the reflux period, and comprehensively consider the influence of shopping cart behavior, purchase conversion rate and collection behavior on purchase potential. First, By performing exponential operation on the purchase conversion rate ( ), the user's purchase intention is emphasized when the conversion rate is high. This operation amplifies the influence of high conversion rate on reflux purchase intention and avoids the interference of low conversion rate on intention evaluation. Second, The standardized method is adopted to process the collection times ( ), so that the collection behavior has a nonlinear amplification effect in the calculation process, but avoids over-emphasizing the influence of frequently collected goods on purchase intention. This makes the collection behavior play a reasonable auxiliary role in intention evaluation, and will not be distorted due to extreme values of collection times. Finally, The logarithmic operation is used to measure the influence of the number of shopping cart additions ( ), to prevent high-frequency shopping cart addition behavior from excessively affecting the calculation result, while ensuring that frequently added shopping cart goods are given higher weight in evaluating purchase intention. Overall, the formula can accurately capture the purchase potential of users during reflux by weighting, standardizing and nonlinearly amplifying these behavior data, and avoid the deviation of single indicator in purchase intention evaluation, thereby improving the accuracy and effectiveness of the recommendation system.

[0080] Reflux purchase intention coefficient The size of the reflux purchase intention coefficient is directly related to "whether the user's reflux behavior is effective". When the value of is high, it means that the user has shown strong purchase intention during the reflux period, which is usually because the user frequently adds goods to the shopping cart, has high conversion rate and more collection behavior, indicating that the user has strong purchase potential, so the system considers that the reflux behavior is effective and pushes more goods that meet the user's current interest and purchase intention. On the contrary, if the value of is low, it means that the user's purchase intention during the reflux period is weak, which may be accidental browsing or no clear purchase plan, indicating that the reflux behavior is invalid or the interest is low, so the system will reduce the recommendation of the goods or maintain the original recommendation strategy. Through the value of , the system can quantify the purchase potential of the user, so as to judge whether the reflux behavior is effective, help to accurately adjust the recommendation strategy, and ensure that the user sees the goods that meet his actual needs, thereby improving the conversion rate and user satisfaction of the platform.

[0081] An effective evaluation model is constructed for the generated reflux behavior interest index and reflux purchase intention coefficient to generate an effective evaluation coefficient.

[0082] In this embodiment, the generated reflux behavior interest index and return purchase intention coefficient An effective evaluation model is constructed, and effective evaluation coefficients are generated through weighted summation. The specific calculation formula is as follows: In the formula, To effectively evaluate the coefficients, and The return visit behavior interest index is as follows: and return purchase intention coefficient The non-zero weight coefficients, and .

[0083] The Effective Evaluation Coefficient (EEC) is calculated by weighted summation, combining two important indicators: the Return Behavior Interest Index (RII) and the Return Purchase Intent Coefficient (RPIC). Its purpose is to comprehensively evaluate the effectiveness of user return behavior. The Return Behavior Interest Index (RII) measures the intensity of a user's interest in new products, reflecting the level of attention a user pays upon returning; while the Return Purchase Intent Coefficient (RPIC) assesses a user's purchasing potential, indicating their actual intention to purchase products during the return period. These two indicators complement each other, providing a more accurate assessment of user behavior.

[0084] Two non-zero weighting coefficients were used in calculating EEC ( and ( ) to adjust the influence of RII and RPIC in the effective evaluation coefficient. The weighting of the return flow as an interest index controls the contribution of RII to EEC; The weight representing the return purchase intention coefficient controls the contribution of RPIC to EEC. The sum of the two weight coefficients must equal 1, i.e. This means that the influence of RII and RPIC in the evaluation is relative, but their sum is always 100%. By adjusting these two weighting coefficients, the calculation of EEC can be optimized based on business needs or historical data, flexibly adjusting the importance of user interest and purchase intent in the evaluation of return behavior. For example, if the system places more emphasis on the user's purchase potential, the weighting coefficients can be appropriately increased. The value; if the intensity of the user's interest is more critical, it can be increased. The value of EEC. In this way, EEC can flexibly adapt to different business scenarios, helping the system to more accurately evaluate the effectiveness of user return behavior, and thus optimize the recommendation strategy.

[0085] The generated valid evaluation coefficients are analyzed to assess whether the user's feedback behavior is valid, and valid signals and invalid signals are generated respectively based on the evaluation results;

[0086] In this embodiment, a pre-set effective evaluation coefficient threshold is determined. and after determining the effective evaluation coefficient The comparison is made, and according to the comparison result, the user's backflow behavior is evaluated whether effective, and the specific comparison analysis is as follows:

[0087] If , the user's backflow behavior is not effective;

[0088] This case means that the backflow behavior fails to achieve the expected user interest or purchase potential, indicating that the user's interest in the goods or goods category is weak, or the purchase intention cannot be effectively stimulated. The system will stop or reduce the recommendation of the goods in this case, avoiding wasting the recommendation resources, and avoiding pushing the goods that do not match the actual needs of the user. This helps to improve the efficiency and accuracy of the recommendation, ensures that the system resources are concentrated on the goods that are more likely to promote conversion, optimizes the overall recommendation effect, and improves the user experience and platform conversion rate.

[0089] If , the user's backflow behavior is effective.

[0090] This case indicates that the user has shown strong interest and purchase intention during the backflow period, and the system will determine that the user's backflow behavior is effective, and push the goods or recommended content that meet the user's interest and purchase potential. At this time, the system will strengthen the recommendation of related goods, increase the exposure frequency or push intensity of these goods, so as to improve the conversion rate. By effectively identifying the user's interest and purchase intention, the system can provide more personalized and accurate recommendations, improve the user's purchase experience, increase the user's interaction with the platform, and enhance the platform's sales efficiency and customer satisfaction.

[0091] The pre-set effective evaluation coefficient threshold can be determined by a data-driven method, usually optimized based on historical data analysis and business objectives. First, the user behavior data collected by the platform (such as browsing, purchasing, collecting, adding to the shopping cart, etc.) and related information of the backflow behavior can be used to make a preliminary setting by statistical analysis and machine learning models. Specifically, the system can find the dividing point between effective and ineffective backflow behaviors by analyzing the effective evaluation coefficients of these two types of behaviors based on the distribution of historical backflow data. Usually, this threshold is based on the statistical distribution of the backflow behavior interest index (RII) and the backflow purchase intention coefficient (RPIC), for example, by calculating the mean and standard deviation of these indicators, setting a threshold above the distribution so that the evaluation coefficients of most effective backflow behaviors are above the threshold. In addition, the threshold can be verified and optimized by A / B testing. In actual operation, the system can experiment with different threshold settings to evaluate their impact on user behavior and recommendation effectiveness, gradually adjusting and optimizing the threshold to achieve the best recommendation effectiveness and platform conversion rate. During the experiment, the platform can monitor key indicators such as backflow behavior conversion rate, click rate, and purchase rate in real time, and continuously adjust the threshold based on actual results to ensure that it can effectively distinguish between effective and ineffective backflow behaviors of users. Through such data analysis and experimental optimization, the system can automatically determine the most appropriate effective evaluation coefficient threshold to a certain extent, ensuring the accuracy of the recommendation system and user satisfaction.

[0092] In this embodiment, effective signals and ineffective signals are generated according to the evaluation results, specifically:

[0093] If the evaluation result is that the backflow behavior of the user is not effective, an ineffective signal is generated;

[0094] To achieve the generation of ineffective signals and their subsequent impact, the recommendation engine and user behavior analysis module in the system can be used to dynamically adjust the recommendation strategy. Specifically, when the system determines that the backflow behavior is not effective, the generated effective evaluation coefficient is compared with the pre-set threshold to trigger the ineffective signal. This signal can adjust the recommendation intensity through an algorithm, i.e., reduce the recommendation frequency of the product or remove it from the user's recommendation list. This can be achieved by modifying the recommendation weight of the product or reducing the relevance of the product in the recommendation model. For example, if the backflow behavior does not show enough interest or purchase intention, the system will reduce the display of the product and focus on recommending other products that better meet the user's current needs, thereby improving the overall accuracy and efficiency of the recommendation system.

[0095] If the evaluation result is that the backflow behavior of the user is effective, an effective signal is generated.

[0096] The generation of the effective signal is achieved by increasing the recommendation strength of the backflow behavior. Specifically, the interest degree and purchase potential of the backflow behavior are scored in the system, and after the effective signal is generated, the recommendation strength of the related goods is increased based on the signal. For example, the system can increase the weight of the goods in the recommendation algorithm, make it appear more frequently in the user's recommendation list, or increase the display frequency of the goods in the prominent position of the home page, category page, etc. In addition, the system can also analyze the user's backflow behavior mode through algorithm, and push more goods that meet the user's interest and purchase potential in real time when the user backflows, which can improve the accuracy of the recommendation, enhance the user's satisfaction, and promote the conversion rate and user stickiness of the platform.

[0097] In the case of generating an effective signal, a plurality of effective evaluation coefficients generated subsequently are continuously obtained, and are analyzed after being generated to generate signals of different levels, and corresponding recommendation measures are taken according to the generated signals of different levels;

[0098] In the case of generating an effective signal, a plurality of effective evaluation coefficients generated subsequently are continuously obtained, and are analyzed after being generated to generate signals of different levels, and corresponding recommendation measures are taken according to the generated signals of different levels; The number of the plurality of effective evaluation coefficients generated subsequently in the case of generating an effective signal is is a positive integer;

[0099] ​​To achieve continuous acquisition of several effective evaluation coefficients generated subsequently, real-time data stream processing and event-driven architecture can be used. This process can be achieved by integrating a real-time data monitoring and collection module in the platform's recommendation system, which continuously captures and records the effective evaluation coefficients in each reflux cycle when the user's reflux behavior occurs. Specifically, the system automatically calculates and generates the corresponding effective evaluation coefficients each time the user performs a reflux operation (such as browsing goods, adding to the shopping cart, clicking to purchase, etc.), and stores these coefficients in the background database. To ensure real-time data, the system can use message queues or stream data processing platforms (such as Apache Kafka, Apache Flink, etc.) to transmit the effective evaluation coefficients generated each time to the data processing layer in real time. In this way, the system can continuously and quickly obtain new effective evaluation coefficients and use them for subsequent analysis and recommendation strategy adjustment. In addition, the system can set a sliding window mechanism to continuously track and update these effective evaluation coefficients within a certain time range (such as every hour, every day). Each time new reflux behavior data is generated, the system automatically processes and analyzes the newly generated evaluation coefficients together with the previous coefficients to ensure the dynamic and timeliness of the evaluation process. In this way, the system can ensure that in the case of generating effective signals, it can obtain effective evaluation coefficients in real time and continuously, providing accurate data support for subsequent analysis and signal generation.

[0100] The average value of the several effective evaluation coefficients generated subsequently is calculated according to the formula: ;

[0101] The standard deviation of the several effective evaluation coefficients generated subsequently is calculated according to the formula: .

[0102] In this embodiment, the preset average value threshold and the preset standard deviation threshold of the several effective evaluation coefficients generated subsequently in the case of generating effective signals are determined, and the average value and the standard deviation of the several effective evaluation coefficients are compared respectively, and different levels of signals are generated according to the comparison results. The specific comparison and analysis are as follows:

[0103] If and , a first-level signal is generated.

[0104] ​​The first level signal (when the average value is greater than or equal to the preset average value threshold, and the standard deviation is greater than or equal to the preset standard deviation threshold) means that the user shows strong and stable interest in the commodity during the backflow period. The system will recognize this signal and regard it as a high purchase intention and strong demand of the user. This situation usually means that the backflow behavior is very effective, the user's interest is concentrated and continuous, and the system should strengthen the recommendation intensity, increase the exposure frequency of related commodities or increase the weight of the commodity in the recommendation algorithm, to ensure that the user can receive more commodity recommendations related to his interest, thereby improving the conversion rate and sales benefit.

[0105] If and , a second level signal is generated;

[0106] The second level signal (when the average value is less than the preset average value threshold, and the standard deviation is greater than or equal to the preset standard deviation threshold) means that the user's backflow behavior shows weak interest but large fluctuations. This situation indicates that the user's interest in the commodity during the backflow period may not be stable, and may be affected by external factors (such as promotion, advertisement push, etc.). The system should reduce the recommendation frequency and adjust the recommendation strategy moderately in this case, to avoid excessive push of commodities that do not match the user's fluctuating interest. By reducing the recommendation intensity, it can avoid pushing too frequent commodities and causing user dissatisfaction, while ensuring that platform resources are not wasted on uncertain interest.

[0107] If and , a third level signal is generated;

[0108] The third level signal (when the average value is greater than or equal to the preset average value threshold, and the standard deviation is less than the preset standard deviation threshold) means that the user shows strong but relatively stable interest during the backflow period. This situation indicates that the user's interest during the backflow period is concentrated and stable, although the interest intensity is high, but the fluctuation is small. For this case, the system can moderately push commodities related to the user's interest, instead of significantly increasing the recommendation intensity. Through appropriate recommendation strategy, the system can ensure to continuously provide commodities that meet the user's stable interest, avoid excessive push, and improve the user's purchase experience and platform conversion rate.

[0109] If and , a fourth level signal is generated.

[0110] The fourth level signal (when the average value is less than the preset average value threshold, and the standard deviation is less than the preset standard deviation threshold) means that the user's backflow behavior interest is weak and relatively stable. In this case, the user's backflow behavior does not show obvious interest changes, and the intensity of his interest is low, so the system should judge that the backflow behavior is invalid backflow. Based on this signal, the system will reduce or stop the recommendation of related goods, avoiding wasting the recommendation resources. Through this adjustment, the platform can ensure the efficiency of the recommendation system, while ensuring that the user will not receive too many irrelevant goods recommendations, thereby improving the accuracy of the recommendation system and the satisfaction of the user.

[0111] The preset average value threshold and the preset standard deviation threshold of the several effective evaluation coefficients generated subsequently in the case of generating an effective signal can be dynamically set through data analysis and machine learning methods. First, the system can analyze the distribution of the effective evaluation coefficient (EEC) according to the historical user backflow behavior data set. By calculating the average value and the standard deviation of the backflow behavior in the historical data, the system can obtain the baseline values, which can reflect the overall trend and fluctuation degree of the backflow behavior under normal circumstances. On this basis, the system can adjust these baseline values to ensure the rationality and applicability of the thresholds. Specifically, the system first performs statistical analysis on the effective evaluation coefficients in the historical data to obtain the mean and standard deviation of the EEC values in each time period. Then, the preset standard deviation threshold is set according to the fluctuation range of the historical data, which is usually set to 1-2 times the baseline value, ensuring that it can cover most of the backflow behaviors within the normal range. The preset average value threshold can be set by the median or quantile of the historical data, or optimized through A / B testing, that is, by experimentally adjusting the influence of different thresholds on the recommendation effect in actual operation to adjust the most suitable value. In addition, the system can use a dynamic adjustment mechanism to fine-tune the thresholds according to the real-time generated effective evaluation coefficients. As the platform user behavior and backflow patterns continue to change, the system can periodically update the data model and threshold adjustment strategy to adapt to changes in user behavior, ensuring that the thresholds can always accurately reflect the user's current interest and purchase potential. In this way, the system can optimize the setting of the thresholds based on historical data and real-time feedback, ensuring the accuracy and personalization of the recommendations.

[0112] In this embodiment, corresponding recommendation measures are taken according to the generated different levels of signals, specifically:

[0113] If the first level signal is generated, the recommendation measures taken are: increasing the recommendation frequency, improving the exposure rate of new goods in the user's recommendation list, and pushing them to the home page and hot recommendation positions of the platform, etc., to ensure that the user can see more goods matching his interest;

[0114] When the first-level signal is generated, the system will increase the exposure of the commodity in the user recommendation list by increasing the recommendation frequency of the commodity. The specific implementation is to adjust the weight of the commodity in the recommendation algorithm, so that the commodity obtains a higher ranking in the recommendation engine. The system can push the commodity to the prominent position such as the home page recommendation, the hot recommendation, the related commodity, etc. by adjusting the weight parameter in the recommendation algorithm in real time, and increase the display probability of the commodity in the user page. In addition, the recommendation system can also use the user interest model and the personalized recommendation algorithm to ensure that the commodities are highly matched with the current interests of the user through the analysis of the historical behavior and interests, so as to improve the click rate and the purchase conversion rate of the commodity. The purpose of doing so is to ensure that the user can see more commodities meeting the interests and needs of the user when the user returns, and to improve the participation of the user and the conversion rate of the platform.

[0115] If the second-level signal is generated, the recommendation measures taken are to reduce the recommendation frequency, keep the recommendation of the new commodity, but reduce the display priority and frequency of the new commodity, and the new commodity can be recommended to the user group with uncertain interest in the new commodity, so as to avoid excessive recommendation of invalid commodities.

[0116] When the second-level signal is generated, the system will reduce the recommendation frequency, keep the recommendation of the new commodity, but reduce the display priority and frequency of the commodity. The specific implementation is to adjust the weight in the recommendation algorithm, so that the recommendation frequency of the commodity is reduced, and the display opportunity of the commodity in the home page recommendation and the related commodity is reduced. The system can recommend the commodity to the user group with uncertain interest in the commodity, or push the commodity to the lower priority recommendation position, for example, through the personalized algorithm analysis that the interest relationship between the commodity and some users is weak, so as to present a lower weight in the recommendation list of these users. This can avoid excessive recommendation of invalid commodities to the user, ensure that the resources of the recommendation system are used efficiently, and avoid disturbing the user with irrelevant commodity recommendation, so as to improve the accuracy of the recommendation.

[0117] If the third-level signal is generated, the recommendation measures taken are to maintain the existing recommendation frequency, provide new commodity recommendation according to the current interest intensity of the user, and display the new commodity to the user in a relatively stable recommendation mode, so as to ensure that the user can continuously obtain commodities meeting the interests of the user.

[0118] When the third-level signal is generated, the system will maintain the existing recommendation frequency, ensuring that the product is displayed to the user at a moderate frequency in the recommendation list. This means that the recommendation intensity of the product remains unchanged, neither increasing nor decreasing, ensuring that the product is in a stable recommendation state. The specific implementation is to analyze the user's interest and purchase intention in real time, and the system calculates the user's interest intensity, and then decides the recommendation frequency of the product according to the intensity. If the user's interest in the product remains stable during the backflow period, the system will maintain the frequency of the product in the recommendation list without excessive intervention, ensuring that the user can continue to see the product within the normal recommendation range. This approach ensures the balance of the recommendation system, neither over-pushing products nor ignoring the user's long-term interest, thereby improving user satisfaction and maintaining a relatively stable conversion rate.

[0119] If the fourth-level signal is generated, the recommendation measures taken are to stop recommending and remove the new product from the user's recommendation list to avoid ineffective recommendations, focusing resources on recommending more suitable products to the user, and improving the efficiency of the system's recommendations.

[0120] When the fourth-level signal is generated, the system will stop recommending the product and remove it from the user's recommendation list. The specific implementation is to adjust the weight in the recommendation algorithm, reduce the recommendation weight of the product to zero, or completely exclude it from the relevant goods and recommended positions. In order to avoid the interference of ineffective product recommendations to users, the system can determine the ineffective recommendations of the product through backflow behavior analysis and stop pushing, thereby saving recommendation resources and focusing on goods that better match the user's interest. The core purpose of this approach is to ensure that system resources are concentrated on high-potential goods, improve the efficiency of the recommendation system, avoid users receiving goods that do not match their interests, and improve the user experience and accuracy of the platform's recommendations.

[0121] The user's backflow behavior and backflow behavior new product browsing information are continuously monitored, and the parameters of the backflow behavior effective evaluation model and the recommendation strategy are dynamically adjusted according to the real-time monitoring results, while the historical data is stored and analyzed to optimize the model parameters and the recommendation strategy.

[0122] To continuously monitor the user's backflow behavior and backflow behavior new product browsing information, a real-time data collection and analysis system can be integrated into the e-commerce platform, and a dynamically adjusted recommendation algorithm can be used. Specifically, the system will continuously capture relevant data such as browsing time, click count, and dwell time in each backflow behavior of the user through the user behavior tracking module in the background, and transmit it to the data processing platform in real time. The system will use event-driven architecture and stream data processing platforms such as Apache Kafka or Apache Flink for real-time processing, ensuring that the data of each user backflow and browsing behavior can be captured and analyzed in a timely manner.

[0123] On the basis of real-time monitoring, the system dynamically adjusts the parameters of the backflow behavior effective evaluation model, such as optimizing the model by adjusting the weights of interest index and purchase intention coefficient, to ensure that the model can respond to changes in user interest and purchase behavior in real time. The recommendation system updates the parameters in the algorithm based on real-time data, improves the recommendation logic, and enhances the accuracy of the recommendations. At the same time, the system also stores these real-time data in the database for historical data analysis, analyzes the trend of changes in user behavior patterns, optimizes parameter settings and recommendation strategies. Through the storage and analysis of historical data, the system can identify long-term effective user behavior characteristics, further refine the backflow behavior evaluation model, and improve the prediction ability of the model and the accuracy of the recommendation strategy. The purpose of this is to ensure that the recommendation system can continuously adapt to changes in user demand, improve recommendation efficiency, and improve user experience and platform conversion rate.

[0124] The above formulas are dimensionless numerical calculations, and the formulas are obtained by software simulation of a large amount of data to obtain a formula of the latest real situation. The preset parameters in the formula are set by a person skilled in the art according to the actual situation.

[0125] The above embodiments can be realized wholly or partially by software, hardware, firmware, or any other combination. When realized by software, the above embodiments can be realized wholly or partially in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the processes or functions described in the embodiments of the present application are wholly or partially generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium or transferred from one computer-readable storage medium to another, for example, the computer instructions can be transferred from one website, computer, server, or data center to another through wired or wireless (such as infrared, wireless, microwave, etc.) methods. The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server, data center, etc. containing one or more available medium collections. The available medium can be a magnetic medium (such as a floppy disk, a hard disk, a magnetic tape), an optical medium (such as a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state disk.

[0126] It should be understood that in various embodiments of the present application, the size of the sequence number of each process described above does not mean the order of execution, and the execution order of each process should be determined by its function and inherent logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0127] Those skilled in the art can understand that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be realized in electronic hardware or a combination of computer software and electronic hardware. Whether the functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0128] In several embodiments provided in the present application, it should be understood that the disclosed system and method can be implemented in other ways. For example, the above-described embodiments are merely illustrative, for example, the division of the units is only a logical function division, and actual implementation can have another division manner, for example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the units or components shown or discussed can be indirect coupling or communication connection through some interfaces, devices or units, and can be electrical, mechanical or other forms.

[0129] The units described as separate components can or can not be physically separated, and the components shown as units can or can not be physical units, that is, they can be located in one place or distributed on multiple network units. Part or all of the units can be selected according to actual needs to achieve the purpose of the embodiment.

[0130] In addition, each functional unit in each embodiment of the present application can be integrated into a processing unit, or each unit can exist physically independently, or two or more units can be integrated into one unit.

[0131] The above is merely a specific implementation of the present application, but the protection scope of the present application is not limited thereto, and any person skilled in the art can easily think of changes or replacements within the technical scope disclosed in the present application, which should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A marketing method based on intelligent commodity background matching, characterized in that, Specifically comprising the following steps: By monitoring the user's browsing and purchasing behavior on the e-commerce platform in real time, the user's backflow behavior is identified, and when the user's backflow behavior is identified, it is detected whether the user browses the previously unattended goods during the backflow behavior, and in the case of detecting that the user browses the previously unattended goods during the backflow behavior, the following steps are performed; Real-time acquisition of backflow behavior new product browsing information generated when the user browses previously unattended goods during the backflow behavior, and analysis after acquisition, respectively generating backflow behavior interest index and backflow purchase intention coefficient; Specifically comprising the following steps: Real-time acquisition of backflow behavior new product browsing information generated when the user browses previously unattended goods during the backflow behavior, and preprocessing after acquisition; Extract backflow behavior interest information and backflow purchase intention information from the preprocessed backflow behavior new product browsing information; Analyze the extracted backflow behavior interest information and backflow purchase intention information to generate backflow behavior interest index and backflow purchase intention coefficient respectively; The acquisition logic of the backflow behavior interest index is as follows: Extracting the reflow behavior interest information in the preprocessed reflow behavior new commodity browsing information, specifically including the browsing times of the new commodity by the user in different time periods during the reflow period, the stay time of the new commodity, and the ratio of the click times to the display times of the new commodity, and are respectively marked as , and , represents the browsing times of the new commodity by the user in the time period during the reflow period, represents the stay time of the new commodity by the user in the time period during the reflow period, represents the ratio of the click times to the display times of the new commodity by the user in the time period during the reflow period, , is a positive integer;​​​ constructing a set of dwell times of the user on the new item at different time periods within the reflow period, and designating a maximum value within the set as ; Calculate the backflow behavior interest index, and the specific calculation formula is as follows: In the formula, Rf is the reflux behavior interest index; The acquisition logic of the backflow purchase intention coefficient is as follows: Extract the new commodity browsing information in the preprocessed backflow behavior, specifically including the number of times the user adds the new commodity to the shopping cart in different time periods during the backflow period, the conversion rate of purchasing the new commodity, and the number of times of collecting the new commodity, and are respectively marked as , and , represents the number of times the user adds the new commodity to the shopping cart in the time period during the backflow period , represents the conversion rate of the user purchasing the new commodity in the time period during the backflow period , represents the number of times the user collects the new commodity in the time period during the backflow period , , is a positive integer; construct a set of conversion rates of the user to purchase a new item at different time periods within the reflow period, and designate a maximum value within the set as ; Calculate the backflow purchase intention coefficient, and the specific calculation formula is as follows: In the formula, is the coefficient of purchase intention for backflow An effective evaluation model is constructed for the generated backflow behavior interest index and backflow purchase intention coefficient to generate an effective evaluation coefficient; Analyze the generated effective evaluation coefficient to evaluate whether the user's backflow behavior is effective, and generate effective signals and ineffective signals respectively according to the evaluation results; In the case of generating effective signals, continuously acquire a plurality of effective evaluation coefficients generated subsequently, and analyze after generation to generate signals of different levels, and take corresponding recommendation measures according to the generated signals of different levels respectively; Continuously monitor the user's backflow behavior and backflow behavior new product browsing information, and dynamically adjust the parameters of the backflow behavior effective evaluation model and the recommendation strategy according to the real-time monitoring results, while storing and analyzing historical data to optimize the model parameters and the recommendation strategy. 2.The marketing method based on smart goods context matching according to claim 1, characterized in that, Reflux behavior interest index generated And reflux purchase intention coefficient An effective evaluation model is constructed, and an effective evaluation coefficient is generated by weighted summation. The specific calculation formula is as follows: wherein is an effective evaluation coefficient, and are non-zero weight coefficients of a backflow behavior interest index and a backflow purchase intention coefficient respectively, and . 3.The marketing method based on smart goods context matching according to claim 2, characterized in that, determining a preset effective evaluation coefficient threshold and comparing the generated effective evaluation coefficient with the preset effective evaluation coefficient threshold after the determination, and evaluating whether the reflux behavior of the user is effective according to the comparison result. If , the user's backflow behavior is not valid; If , the user's backflow behavior is valid. 4.The marketing method based on smart commodity context matching according to claim 3, characterized in that, According to the evaluation results, effective signals and ineffective signals are generated respectively, specifically: If the evaluation result is that the user's backflow behavior is not effective, an ineffective signal is generated; If the evaluation result is that the user's backflow behavior is effective, an effective signal is generated. 5.The marketing method based on smart commodity context matching according to claim 4, characterized in that, In case of a valid signal being generated, a number of subsequently generated valid evaluation coefficients are continuously acquired and newly indexed as , a number for a number of subsequently generated valid evaluation coefficients in case of a valid signal being generated, , is a positive integer; The average of several valid evaluation coefficients generated subsequently is calculated according to the formula:​ ； a number of valid evaluation coefficients are calculated for the subsequent generation the standard deviation according to the formula: 。 6.The marketing method based on smart commodity context matching according to claim 5, characterized in that, determining a preset average value threshold of a plurality of effective evaluation coefficients generated subsequently in the case of generating an effective signal and a preset standard deviation threshold , and comparing the average value and the standard deviation of the plurality of effective evaluation coefficients respectively after determination, and generating signals of different levels according to the comparison results, and the specific comparison analysis is as follows: If and a first level signal is generated; If and a second level signal is generated; If and a third level signal is generated; If and a fourth level signal is generated. 7.The marketing method based on smart product context matching according to claim 6, wherein, According to the generated signals of different levels, corresponding recommendation measures are taken respectively, specifically: If the first level signal is generated, the recommendation measure taken is to increase the recommendation frequency, improve the exposure rate of new products in the user's recommendation list, and push it to the home page and hot recommendation of the platform, etc. If the second level signal is generated, the recommendation measure taken is to reduce the recommendation frequency, maintain the recommendation of new products, but reduce the display priority and frequency; If the third level signal is generated, the recommendation measure taken is to maintain the existing recommendation frequency, and provide new product recommendations according to the user's current interest intensity; If the fourth level signal is generated, the recommendation measure taken is to stop the recommendation and remove the new product from the user's recommendation list.

Images