Crowd orientation method based on big data analysis

Abstract

The invention discloses a crowd orientation method based on big data analysis, relates to the technical field of crowd orientation, and is used for solving the problems that the shopping cart conversion efficiency is reduced and the resource waste rate of a recommendation system is increased. Counting access intervals of various commodities, accessing a historical database to extract a conversion period reference of a user, calculating interaction attenuation characteristics in combination with the access intervals, analyzing a cooling ratio to judge whether the user is in an interest cooling stage, and extracting commodity retention duration and commodity source information for the interest cooling stage; and channel viscosity indexes of different commodity sources and channels are calculated, the channel viscosity indexes and the commodity retention duration are comprehensively calculated to obtain a scheduling priority score, and the display sequence of the shopping cart commodities is adjusted, so that the display weight of high potential commodities is improved, and the shopping cart conversion efficiency and recommendation accuracy are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of crowd targeting, and more particularly to a crowd targeting method based on big data analysis. BACKGROUND

[0002] With the rapid development of Internet e-commerce platforms and digital marketing systems, user behavior data on network shopping platforms presents the characteristics of high frequency, diversification and fragmentation. Merchants usually evaluate user purchase interest and potential demand through user browsing records, shopping cart operation records and order conversion behavior.

[0003] The prior art has the following disadvantages: At present, the crowd targeting method in the prior art only groups users based on static behavior tags, lacks a dynamic analysis mechanism for user access behavior timing characteristics and interest decay law, cannot accurately identify the interest cooling stage of users and adjust the display strategy, and is difficult to realize fine targeting and product priority scheduling based on real-time interest state, resulting in reduced shopping cart conversion efficiency and increased waste of recommendation system resources. Therefore, a crowd targeting method based on big data analysis is proposed.

[0004] 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

[0005] In order to overcome the above-mentioned defects of the prior art, the embodiments of the present application provide a crowd targeting method based on big data analysis, which uses user access behavior timing analysis, interaction decay modeling, channel stickiness calculation and product scheduling priority scoring mechanism to solve the problems raised in the above background technology.

[0006] To achieve the above-mentioned purpose, the present application provides the following technical scheme, a crowd targeting method based on big data analysis, comprising the following steps: Step S1: Set an analysis time, and determine whether to enter a product analysis mechanism according to the number of times the user accesses the shopping cart within the analysis time. When entering the product analysis mechanism, the access interval of various products is counted; Step S2: Obtain the conversion cycle reference of the user from the access history database, analyze the interaction decay characteristics using the access interval, and determine whether the user's conversion interest stage is in the interest cooling stage in combination with the conversion cycle reference; Step S3: When the user is in the interest cooling stage, obtain the product retention time and product source information of various products, and evaluate the channel stickiness index of the user to the product source information according to the product source information; Step S4: The channel stickiness index and the commodity retention time are combined to evaluate the scheduling priority score of each type of commodity, the commodities in the shopping cart are sorted based on the scheduling priority score, and the arrangement order of the commodities in the shopping cart is adjusted according to the sorting result.

[0007] In a preferred embodiment, in step S1, a preset analysis time is set, the number of times of accessing the shopping cart by the user is monitored in real time within the preset analysis time, and the ratio of the number of times of accessing the shopping cart by the user to the preset analysis time is taken as the access frequency of the shopping cart. If the access frequency of the shopping cart is greater than a preset access frequency threshold, it is determined that the commodity analysis mechanism is entered. Otherwise, it is determined that the commodity analysis mechanism is not entered.

[0008] In a preferred embodiment, in step S1, when the commodity analysis mechanism is entered, the access events of each type of commodity are called through the log database, and the time stamp of the access event is taken as the commodity access time. The commodity access times of the same type of commodity are sorted in time sequence and combined into an access time sequence. The access time sequence is analyzed, and the adjacent commodity access times are processed by difference to obtain an access interval.

[0009] In a preferred embodiment, in step S2, the conversion cycle benchmark of the user is obtained from the access history database. The reciprocal of the conversion cycle benchmark is taken as the interest decay rate. The access intervals of the same type of commodity are sequentially sorted in time sequence, and the change amplitude between adjacent access intervals is analyzed. The current access interval is processed by difference with the previous access interval to obtain an interval change amplitude, and the access interval growth rate is obtained by dividing the interval change amplitude by the previous access interval. The average value of the access interval growth rate is taken as the interaction decay feature of each type of commodity.

[0010] In a preferred embodiment, in step S2, after the interaction decay feature is standardized by the Max-min standardization method, the category interest deviation value is calculated by combining the interest decay rate. The number of commodity categories with a category interest deviation value greater than 1 is counted as the total number of declining categories, and the ratio of the total number of declining categories to the total number of commodity categories in the shopping cart is taken as the cooling ratio. If the cooling ratio is greater than a preset cooling ratio threshold, it is determined that the conversion interest stage of the user is in the interest cooling stage. Otherwise, it is determined that the conversion interest stage of the user is in the interest active stage.

[0011] In a preferred embodiment, in step S3, when the user is in the interest cooling stage, the residence time length of the goods in the shopping cart and the source information of the goods are obtained through the log database; The source information of the goods is the source channel identifier when the user adds the goods to the shopping cart; The residence time length of the goods of the same kind is obtained by averaging the residence time length of the goods of the same kind.

[0012] In a preferred embodiment, in step S3, the historical conversion data corresponding to each source channel identifier is retrieved through the conversion statistics database, including the number of goods purchases corresponding to different source channel identifiers; The number of goods purchases corresponding to different source channel identifiers is accumulated and summed to obtain the cumulative purchase number; The ratio of the number of goods purchases corresponding to the source channel identifier to the cumulative purchase number is the channel stickiness index of the source information of the goods.

[0013] In a preferred embodiment, in step S4, in the source information of the goods of the same kind, the channel stickiness index is averaged to obtain the average channel stickiness index of the goods of the same kind; The channel stickiness coefficient and the residence time length coefficient are obtained by standardizing the average channel stickiness index and the residence time length of the goods, respectively.

[0014] In a preferred embodiment, in step S4, the dispatch priority score of each kind of goods is evaluated by using the channel stickiness coefficient and the residence time length coefficient: , wherein, is the dispatch priority score, is the channel stickiness coefficient, is the residence time length coefficient, is a preset time decay factor, is a natural constant; The dispatch priority score of each kind of goods is sorted in descending order according to the sorting result, and the arrangement order of each kind of goods in the shopping cart is adjusted according to the sorting result.

[0015] The technical effects and advantages of the present application are: This invention monitors the frequency of user visits to the shopping cart by setting an analysis time, determining whether to enter the product analysis mechanism. When the product analysis mechanism is entered, it statistically analyzes the access interval of various products, retrieves the user's conversion cycle benchmark from the historical database, calculates the interaction decay characteristics based on the access interval, and then analyzes the cooling ratio to determine whether the user is in the interest cooling stage. For the interest cooling stage, it extracts the product dwell time and product source information, calculates the channel stickiness index of different product source channels, and calculates the scheduling priority score by comprehensively calculating the channel stickiness index and product dwell time. Based on this, it adjusts the display order of products in the shopping cart, realizing a closed-loop control from the identification of user conversion interest stages to dynamic display optimization, increasing the display weight of high-potential products, and improving the shopping cart conversion efficiency and recommendation accuracy. Attached Figure Description

[0016] Fig. 1 This is a flowchart illustrating the implementation of a big data analysis-based population targeting method according to the present invention.

[0017] Fig. 2 This is a schematic diagram illustrating the steps of a population targeting method based on big data analysis according to the present invention. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] This invention monitors the frequency of user visits to the shopping cart by setting an analysis time, determines whether to enter the product analysis mechanism, and when the product analysis mechanism is entered, it counts the access interval of various products, retrieves the user's conversion cycle benchmark from the historical database, calculates the interaction decay characteristics based on the access interval, and then analyzes the cooling ratio to determine whether the user is in the interest cooling stage. For the interest cooling stage, it extracts the product dwell time and product source information, calculates the channel stickiness index of different product source channels, and calculates the scheduling priority score by combining the channel stickiness index and product dwell time. Based on this, it adjusts the display order of products in the shopping cart, realizing a closed-loop control from the identification of the user's interest conversion stage to dynamic display optimization, and increasing the display weight of high-potential products.

[0020] Example 1, such as Figs. 1-2 As shown, a population targeting method based on big data analysis includes the following steps: Step S1: Set the analysis time, and determine whether to enter the product analysis mechanism according to the number of times the user accesses the shopping cart within the analysis time. When entering the product analysis mechanism, the access interval of each type of product is counted. Step S2: Obtain the conversion cycle reference of the user from the access history database, analyze the interaction attenuation characteristics by using the access interval, and determine whether the conversion interest stage of the user is the interest cooling stage by combining the conversion cycle reference. Step S3: When the user is in the interest cooling stage, obtain the product retention time and product source information of each type of product, and evaluate the channel stickiness index of the user to the product source information according to the product source information. Step S4: Evaluate the scheduling priority score of each type of product by combining the channel stickiness index and the product retention time, sort the products of the shopping cart based on the scheduling priority score, and adjust the arrangement order of the products of the shopping cart according to the sorting result.

[0021] The specific implementation is as follows: In step S1, a preset analysis time is set, and the number of times the user accesses the shopping cart within the preset analysis time is monitored in real time. The ratio of the number of times the user accesses the shopping cart to the preset analysis time is taken as the shopping cart access frequency. The preset access frequency threshold is compared with the shopping cart access frequency to determine whether to enter the product analysis mechanism. If the shopping cart access frequency is greater than the preset access frequency threshold, it is determined that the product analysis mechanism is entered. Conversely, it is determined that the product analysis mechanism is not entered. When the shopping cart access frequency is greater than the preset access frequency threshold, it indicates that the user has a high product attention frequency for the shopping cart within the preset analysis time. After entering the product analysis mechanism, the conversion interest stage of the user is further analyzed.

[0022] When entering the product analysis mechanism, the access events of each type of product are retrieved from the log database, and the timestamp of the access event is taken as the product access time. Among them, each type of product refers to the division of products into different categories according to the category of the product. Within a preset statistical time, for example, it includes digital accessory products, shoe and single product, and instant food products, etc.

[0023] The product access times of the same type of product are sorted in time sequence and combined into an access time sequence. The access time sequence is analyzed, and the adjacent product access times are processed by difference to obtain the access interval, which is used to reflect the continuous access trend of the user to the same type of product.

[0024] It needs to be explained that the preset analysis time refers to a time window for counting the user's access to the shopping cart behavior, which can be set according to the platform activity and the commodity cycle; the preset access frequency threshold is a behavior trigger threshold for judging whether the user enters the commodity analysis mechanism, which can be set according to the activity level of different user groups, for example, calling the user's group label, for daily active users, the average value of the shopping cart access frequency of the daily active users is used as the preset access frequency threshold; the log database is a data storage module for recording user behavior event information, recording the behavior log of the user access, adding the shopping cart and the commodity source channel in time sequence, and used for generating the time sequence and source record of the commodity in the shopping cart.

[0025] This step can analyze the user behavior characteristics from the time dimension by monitoring and counting the user's access to the shopping cart behavior, judge whether the user enters the continuous shopping decision process, and realize the preliminary layering of the user interest cycle, and provide high timeliness and comparable data basis for the subsequent conversion interest evaluation stage.

[0026] In step S2, the access history database obtains the user's conversion cycle benchmark, which refers to the average time period that the user experiences from accessing the shopping cart to completing the purchase of the commodity in the past shopping behavior.

[0027] It needs to be explained that the history database refers to the continuous archiving and statistics of the historical data of users and different categories of commodities, and in this embodiment, it is used to extract the user's conversion cycle benchmark; After sorting the access intervals of similar commodities in time sequence, the change amplitude between adjacent access intervals is analyzed, the current access interval is subtracted from the previous access interval to obtain the interval change amplitude, and the access interval growth rate is obtained by dividing the previous access interval, which reflects the change trend of the access interval; The average value of the access interval growth rate is taken as the interaction decay characteristic of each type of commodity; The greater the interaction decay characteristic, the more the access interval generally shows an upward trend, and the more obvious the interest decay of the commodity; the smaller the interaction decay characteristic, the shorter or stable the access interval, and the interest in the commodity is maintained or enhanced.

[0028] The reciprocal of the conversion cycle benchmark is taken as the interest decay rate, which reflects the average rate of completing the purchase decision per unit time of the user; The interaction decay characteristic is standardized by using the Max-min standardization method: , is the interaction decay characteristic of each type of commodity, is the minimum value of the interaction decay characteristic of each type of commodity, is the maximum value of the interaction decay characteristic of each type of commodity, The result of the normalized interaction decay feature.

[0029] The product of the interest decay rate and the preset adjustment coefficient is taken as the adjusted decay rate. The ratio of the normalized interaction decay feature and the adjusted decay rate is taken as the category interest offset value. For example, the interest decay rate is set to 0.25, the result of the Max-Min normalization processing of the interaction decay feature of the earphone category product is 0.39, and the preset adjustment coefficient is 0.8. Then the adjusted decay rate is 0.2. The ratio of the normalized interaction decay feature of the earphone category product and the adjusted decay rate is calculated to obtain the category interest offset value of the earphone category product, which is 1.95.

[0030] When the category interest offset value is greater than 1, it indicates that the user's interest in the category product decreases faster than the interest decay rate, and the user is in the interest cooling stage. When the category interest offset value is less than or equal to 1, it indicates that the user's interest in the category product is maintained or enhanced, and the user is in the interest active stage.

[0031] The category interest offset values are analyzed to determine whether the user's conversion interest stage is in the interest cooling stage. The number of product categories with a category interest offset value greater than 1 is counted as the total number of decreasing categories. The ratio of the total number of decreasing categories to the total number of product categories in the shopping cart is taken as the cooling ratio. The preset cooling ratio threshold is compared with the cooling ratio to determine whether the user's conversion interest stage is in the interest cooling stage. If the cooling ratio is greater than the preset cooling ratio threshold, it is determined that the user's conversion interest stage is in the interest cooling stage. Conversely, it is determined that the user's conversion interest stage is in the interest active stage.

[0032] The interest cooling stage refers to the weakening of the user's attention to the products in the shopping cart and the delay of the purchase decision. The interest active stage refers to the high level of interaction between the user and the products in the shopping cart.

[0033] It should be explained that the Max-min normalization method is a data normalization method that maps the original data to the [0, 1] interval through linear transformation. The preset adjustment coefficient refers to the coefficient parameter used to adjust the influence weight of the interest decay rate in the calculation of the category interest offset value, with a value range of 0 to 1. It can be set according to the historical data distribution characteristics. The preset cooling ratio threshold can be calculated based on the historical user behavior samples. For example, by analyzing the cooling ratio of the historical user in the shopping cart access stage and the actual conversion result, a linear fitting is performed to select the cooling ratio corresponding to the turning point of the actual conversion result as the preset cooling ratio threshold.

[0034] The step is used for identifying the conversion interest stage of the user, quantifying the change of attention of the user to different commodities through dynamic analysis of interaction attenuation features, taking the cooling rate as a measurement index, converting the conversion interest stage of the user into a quantifiable index, realizing the identification of the interest cooling stage, and providing a behavior basis for the subsequent priority display strategy.

[0035] In step S3, when the user is in the interest cooling stage, the conversion potential of various commodities is further analyzed, and the commodities with high conversion potential are prioritized; The retention time of the commodity in the shopping cart and the commodity source information are obtained through the log database; The retention time of the commodity in the shopping cart and the commodity source information are obtained through the log database; The retention time of the commodity is the cumulative residence time of the user from the first time the commodity is added to the shopping cart to the current analysis time; the commodity source information is the source channel identifier when the user adds the commodity to the shopping cart, for example, the commodity source information includes homepage recommendation, search result page, activity page and different source channel identifiers.

[0036] The historical conversion data corresponding to each source channel identifier is retrieved through the conversion statistics database, the historical conversion data is the statistical sample information generated based on the historical shopping behavior of the current user under different source channels, including the number of commodity purchases corresponding to different source channel identifiers; The number of commodity purchases corresponding to different source channel identifiers is accumulated and summed as the cumulative purchase number, and the ratio of the number of commodity purchases corresponding to the source channel identifier to the cumulative purchase number is taken as the channel stickiness index of the commodity source information, reflecting the attraction degree of different commodity source information to the user's purchase behavior; For example, the purchase number of the homepage recommendation channel is 30 times, the purchase number of the activity page channel is 15 times, and the purchase number of the search page channel is 5 times. The purchase numbers of the three source channels are accumulated and summed to obtain a cumulative purchase number of 50 times. The ratio of the purchase number of each source channel to the cumulative purchase number is 0.6, 0.3 and 0.1 respectively. The corresponding values are taken as the channel stickiness index of each commodity source information.

[0037] The greater the channel stickiness index, the greater the proportion of the frequency of completing the purchase by the user under the source channel, and the higher the purchase conversion rate of the user; the smaller the channel stickiness index, the smaller the proportion of the frequency of completing the purchase by the user under the source channel, and the lower the purchase conversion rate of the user.

[0038] It should be explained that the conversion statistics database refers to a database for storing and managing historical shopping behavior data of the user under different commodity source information, for retrieving historical conversion data corresponding to the commodity source information.

[0039] The step evaluates the conversion potential of different category goods when the user is in the interest cooling stage, calculates the channel stickiness index based on the product source information, evaluates the user's dependence on different channels, and can dynamically identify the category goods that still have conversion potential in the cooling stage, and preferentially display the category goods with high stickiness.

[0040] In step S4, the channel stickiness index of the same category goods is averaged to obtain the average channel stickiness index of the same category goods in the product source information of the same category goods; The channel stickiness coefficient and the retention time coefficient are obtained by standardizing the average channel stickiness index and the product retention time, respectively; The dispatch priority score of each category of goods is evaluated by using the channel stickiness coefficient and the retention time coefficient: , wherein, is the dispatch priority score, is the channel stickiness coefficient, is the retention time coefficient, is a preset time decay factor, is a natural constant.

[0041] It needs to be explained that the standardization method includes but is not limited to standard linear transformation based on interval scaling, Z-Score standardization method based on statistics, or normalization method based on nonlinear mapping function. Here, the application method of standardization is not described. The preset time decay factor is used to adjust the decay strength of the retention time coefficient to the dispatch priority score. The value range is 0 to 1 interval. The average access interval of each category of goods and the ratio of the conversion period benchmark can be mapped to the 0 to 1 interval to obtain the preset time decay factor.

[0042] For example, the channel stickiness index and the product retention time are standardized to obtain the channel stickiness coefficient and the retention time coefficient. Assuming that the standardization result of the earphone category goods is the channel stickiness coefficient 0.72 and the retention time coefficient 0.45, and the preset time decay factor is set to 0.6, the dispatch priority score of the earphone category goods is obtained by substituting into the dispatch priority score calculation model: , wherein, is the dispatch priority score; The longer the product retention time, the longer the time the user is interested in the product in the shopping cart, and the user's interest decays. The shorter the product retention time, the more recently the user is interested in the product. The larger the dispatch priority score, the larger the channel stickiness coefficient and the smaller the retention time coefficient of the category goods, and the higher the purchase potential of the user for the category goods. The smaller the dispatch priority score, the smaller the channel stickiness coefficient and the longer the retention time coefficient of the category goods, and the lower the purchase potential of the user for the category goods. The various types of commodities are sorted in descending order according to the dispatch priority scores of the various types of commodities, and the arrangement order of the various types of commodities in the shopping cart is adjusted according to the sorting result.

[0043] The arrangement order of the various types of commodities in the shopping cart is adjusted by comparing the numerical values of the dispatch priority scores, so that the arrangement order of the commodities in the shopping cart interface can be updated when the user is in the interest cooling stage, thereby improving the conversion probability of the commodities.

[0044] This step dynamically adjusts the display order of the commodities in the shopping cart according to the dispatch priority scores, and realizes real-time optimization of the shopping cart display layer by sorting the commodities in descending order and rearranging them, so that the display order of the commodities in the shopping cart interface of the user can be consistent with the change of the conversion interest, thereby improving the exposure probability and the final conversion rate of the high-potential commodities, and forming a closed-loop optimization from interest recognition to commodity order display adjustment.

[0045] Finally, it should be noted that in this document, the terms such as first and second are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations.

[0046] Moreover, the terms "include", "contain" or any other variants thereof are intended to cover non-exclusive inclusion, so that the process, method, article or device including a series of elements not only includes those elements, but also includes other elements not explicitly listed or inherent to such process, method, article or device. Without more limitations, the element defined by the statement "includes a" does not exclude the presence of another identical element in the process, method, article or device including the element.

[0047] In this document, the singular forms "a", "an" and "the" can also include the plural forms unless the context clearly indicates otherwise. It should also be understood that the terms "include", "contain" or "have" and the like specify the presence of the stated features, integers, steps, operations, components, parts or combinations thereof, but do not exclude the presence or addition of one or more other features, integers, steps, operations, components, parts or combinations thereof. The possibility, while the term "and / or" used in this specification includes any and all combinations of the related listed items.

[0048] The various embodiments in the specification are described in a progressive manner, and each embodiment focuses on the difference from other embodiments. The various embodiments can be combined as needed, and the same and similar parts refer to each other.

[0049] Those skilled in the art will appreciate that the foregoing description is by way of example only, and is not intended to limit the application solely thereto. Numerous modifications and adaptations will be apparent to those skilled in this art. Therefore, the true scope of the application is indicated by the appended claims, rather than by the foregoing description, and all modifications which come within the scope of the claims are intended to be embraced therein.

Claims

1. A crowd targeting method based on big data analysis, characterized in that: The method comprises the following steps: Step S1: setting an analysis time, and determining whether to enter a product analysis mechanism according to the number of times of accessing the shopping cart within the analysis time; when entering the product analysis mechanism, the access intervals of various products are counted; Step S2: obtaining a conversion cycle benchmark of the user from an access history database, analyzing the interaction attenuation characteristics by using the access intervals, and determining whether the conversion interest stage of the user is an interest cooling stage in combination with the conversion cycle benchmark; Step S3: when the user is in the interest cooling stage, obtaining the product retention time and product source information of various products, and evaluating the channel stickiness index of the user to the product source information according to the product source information; Step S4: evaluating the scheduling priority score of various products in combination with the channel stickiness index and the product retention time, sorting the various products in the shopping cart based on the scheduling priority score, and adjusting the arrangement order of the products in the shopping cart according to the sorting result.

2. The method according to claim 1, wherein in step S1, a preset analysis time is set, the number of times of accessing the shopping cart is monitored in real time within the preset analysis time, and the ratio of the number of times of accessing the shopping cart to the preset analysis time is taken as the shopping cart access frequency; if the shopping cart access frequency is greater than a preset access frequency threshold, it is determined that the product analysis mechanism is entered; otherwise, it is determined that the product analysis mechanism is not entered.

3. The method according to claim 2, wherein in step S1, when the product analysis mechanism is entered, the access events of various products are called from a log database, and the time stamps of the access events are taken as product access times; the product access times of the same product are sorted in time sequence and combined into an access time sequence; the access time sequence is analyzed, and the adjacent product access times are processed by difference to obtain access intervals.

4. The method according to claim 1, wherein in step S2, the conversion cycle benchmark of the user is obtained from an access history database; the reciprocal of the conversion cycle benchmark is taken as an interest attenuation rate; the access intervals of the same product are sequentially sorted in time sequence, and the change amplitudes between adjacent access intervals are analyzed; the current access interval is processed by difference with the previous access interval to obtain an interval change amplitude, and the interval change amplitude is divided by the previous access interval to obtain an access interval growth rate; the average value of the access interval growth rates is taken as the interaction attenuation characteristics of various products.

5. The method according to claim 4, wherein in step S2, after the interaction attenuation characteristics are standardized by using a Max-min standardization method, the category interest deviation value is calculated in combination with the interest attenuation rate; the number of product categories with a category interest deviation value greater than 1 is counted as a total number of descending categories, and the ratio of the total number of descending categories to the total number of product categories in the shopping cart is taken as a cooling ratio; if the cooling ratio is greater than a preset cooling ratio threshold, it is determined that the conversion interest stage of the user is an interest cooling stage. ​ ​ ​ ​ If not, the conversion interest stage of the user is determined as the interest active stage.

6. The crowd targeting method based on big data analysis according to claim 1, characterized in that: In step S3, when the user is in the interest cooling stage, the retention time length of the goods in the shopping cart and the goods source information are obtained from the log database; The goods source information is the source channel identifier when the user adds the goods into the shopping cart; The retention time length of the same kind of goods is obtained by averaging the retention time length of the same kind of goods.

7. The crowd targeting method based on big data analysis according to claim 6, characterized in that: In step S3, the historical conversion data corresponding to each source channel identifier is retrieved from the conversion statistics database, including the number of goods purchases corresponding to different source channel identifiers; The number of goods purchases corresponding to different source channel identifiers is summed up as the cumulative purchase number; The ratio of the number of goods purchases corresponding to each source channel identifier to the cumulative purchase number is the channel stickiness index of the goods source information.

8. The crowd targeting method based on big data analysis according to claim 7, characterized in that: In step S4, in the goods source information of the same kind of goods, the average of the channel stickiness index is obtained as the average channel stickiness index of the same kind of goods; The channel stickiness coefficient and the retention time length coefficient are obtained by standardizing the average channel stickiness index and the retention time length respectively.

9. The crowd targeting method based on big data analysis according to claim 8, characterized in that: In step S4, the dispatch priority score of each type of commodity is evaluated by using the channel stickiness coefficient and the retention time coefficient: wherein, is the dispatch priority score, is the channel stickiness coefficient, is the retention time coefficient, is a preset time decay factor, is a natural constant; According to the dispatch priority score of each kind of goods, each kind of goods is sorted in descending order, and the arrangement order of each kind of goods in the shopping cart is adjusted according to the sorting result.

Images