CN114066565A

CN114066565A - Commodity determination method and device, electronic equipment and storage medium

Info

Publication number: CN114066565A
Application number: CN202111303389.3A
Authority: CN
Inventors: 王毅诚; 吴运浩; 查涛; 刘奎; 赵宇; 王媛琼; 徐夙龙; 龙波
Original assignee: Beijing Jingdong Century Trading Co Ltd; Beijing Wodong Tianjun Information Technology Co Ltd
Current assignee: Beijing Jingdong Century Trading Co Ltd; Beijing Wodong Tianjun Information Technology Co Ltd
Priority date: 2021-11-05
Filing date: 2021-11-05
Publication date: 2022-02-18

Abstract

The embodiment of the application provides a commodity determining method, a commodity determining device, electronic equipment and a computer storage medium, wherein the method comprises the following steps: determining a value comparison result of a category to which each commodity to be selected belongs; the value comparison result represents a comparison result of the value of the category to which each commodity to be selected belongs and the reference value; obtaining the commodity fusing proportion of the category to which each commodity to be selected belongs according to the value comparison result and the comparison result of the set threshold; sorting each commodity to be selected by using the trained fusion model to obtain a sorting result, and determining a first commodity needing to be fused from each commodity to be selected based on the sorting result and the commodity fusing proportion; the fusion model is used for reflecting the relation between the self characteristics of each to-be-selected commodity and the order quantity and/or click rate.

Description

Commodity determination method and device, electronic equipment and storage medium

Technical Field

The present application relates to the field of e-commerce operation technologies, and in particular, to a method and an apparatus for determining a commodity, an electronic device, and a computer storage medium.

Background

Due to the operation requirement of the e-commerce platform search service, extra flow needs to be distributed to specific commodities, the display positions of the commodities in the search sequencing result are improved, so that the service target of commodity exposure or click is achieved, and the original search sequencing result is disturbed by the intervention; however, for the goods with the right to be proposed and with the lifted part of the display positions, the traffic brought by the positions cannot be accepted after the ranking is lifted, the efficiency of the goods is low, the traffic distribution efficiency is lowered, and the category or the negative direction of the large plate is caused; in addition, even if some of the entitled goods cannot bear the traffic brought by ranking promotion within a period of time, the goods are always given high exposure, which causes the continuous reduction of traffic distribution efficiency, and further, the business target of promoting the goods exposure or click cannot be reached.

Disclosure of Invention

The application provides a commodity determining method, a commodity determining device, electronic equipment and a computer storage medium.

The technical scheme of the application is realized as follows:

the embodiment of the application provides a commodity determining method, which comprises the following steps:

determining a value comparison result of a category to which each commodity to be selected belongs; the value comparison result represents a comparison result of the value of the category to which each commodity to be selected belongs and the reference value;

obtaining the commodity fusing proportion of the category to which each commodity to be selected belongs according to the value comparison result and the comparison result of the set threshold;

sorting each commodity to be selected by using the trained fusion model to obtain a sorting result, and determining a first commodity needing to be fused from each commodity to be selected based on the sorting result and the commodity fusing proportion; the fusion model is used for reflecting the relation between the self characteristics of each to-be-selected commodity and the order quantity and/or click rate.

In some embodiments, the method further comprises:

obtaining a second commodity needing to be subjected to right reduction according to the comparison result of the value comparison result and a set threshold value; the second commodity comprises at least one commodity to be selected;

and performing right reduction on the second commodity needing to be subjected to the right reduction according to historical data related to the click rate of the second commodity.

In some embodiments, the setting threshold includes a first setting threshold and a second setting threshold, and obtaining the fusing ratio of the commodities belonging to the category of each commodity to be selected according to the comparison result between the value comparison result and the setting threshold includes:

when the value comparison result is determined to be smaller than the first set threshold value, setting the increment of the negative days to be 1; the first set threshold value is a value less than 0;

upon determining that the value comparison result is greater than or equal to the second set threshold, setting an increment of negative days to-1; the second set threshold is a value greater than 0;

upon determining that the value comparison result is between the first set threshold and the second set threshold, setting an increment of negative days to 0;

and determining the commodity fusing proportion of the category to which each commodity to be selected belongs based on the increment of the negative days.

In some embodiments, the determining the percentage of fusing of the commodities belonging to the category of each commodity to be selected based on the increment of the negative days includes:

determining the product of the increment of the negative days and a set numerical value as the fusing proportion of the commodities of the category to which each commodity to be selected belongs; the set value is a value between 0 and 1.

In some embodiments, the fusion model includes a first model and a second model, and the ranking the each to-be-selected commodity by using the trained fusion model to obtain a ranking result includes:

scoring each commodity to be selected by using the trained first model to obtain a first scoring result;

scoring each to-be-selected commodity by using the trained second model to obtain a second scoring result;

and calculating the average score values of the first scoring result and the second scoring result, and sorting the average score values to obtain a sorting result.

In some embodiments, the first model is used for reflecting the relation between the exposure of each commodity to be selected and the order quantity.

In some embodiments, the setting of the threshold includes a third setting of the threshold, and obtaining the second commodity needing to be subjected to weight reduction according to the comparison result between the value comparison result and the setting of the threshold includes:

determining the category needing to be subjected to weight reduction by using the value comparison result and the comparison result of the third set threshold; the third set threshold is a value less than 0;

and selecting a second commodity needing to be subjected to weight reduction from the categories needing to be subjected to weight reduction according to a preset strategy.

In some embodiments, the reducing the weight of the second product to be reduced according to the historical data related to the click rate of the second product includes:

determining the weight reduction proportion of the second commodity according to historical data related to the click rate of the second commodity;

and performing right reduction on the second commodity needing to be subjected to the right reduction by using the right reduction proportion.

The embodiment of the application also provides a commodity determining device, which comprises a determining module, an obtaining module and a processing module, wherein,

the determining module is used for determining the value comparison result of the category to which each commodity to be selected belongs; the value comparison result represents a comparison result of the value of the category to which each commodity to be selected belongs and the reference value;

the obtaining module is used for obtaining the commodity fusing proportion of the category to which each commodity to be selected belongs according to the value comparison result and the comparison result of the set threshold;

the processing module is used for sequencing each commodity to be selected by utilizing the trained fusion model to obtain a sequencing result, and determining a first commodity needing to be fused from each commodity to be selected on the basis of the sequencing result and the commodity fusing proportion; the fusion model is used for reflecting the relation between the self characteristics of each to-be-selected commodity and the order quantity and/or click rate.

The embodiment of the present application provides an electronic device, where the device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, and when the processor executes the computer program, the processor implements the method for determining a commodity provided by one or more of the foregoing technical solutions.

The embodiment of the application provides a computer storage medium, wherein a computer program is stored in the computer storage medium; the computer program can implement the commodity determination method provided by one or more of the above technical solutions after being executed.

It can be seen that the commodity fusing proportion of the related categories can be obtained by utilizing the relationship between the value comparison result of the category to which each commodity to be selected belongs and the set threshold value; further, determining the commodities needing to be fused in related categories according to the commodity fusing proportion; compared with the prior art that the fusing proportion of each commodity to be selected needs to be adjusted according to manual experience, the commodity fusing proportion can be obtained without manually spending a large amount of time for analyzing data, and a large amount of human resources can be saved.

Drawings

Fig. 1 is a schematic flow chart of a commodity determination method in an embodiment of the present application;

FIG. 2a is a schematic diagram illustrating a process of determining an article to be fused according to an embodiment of the present application;

fig. 2b is a schematic flow chart illustrating a process of performing right reduction on a to-be-selected commodity in the embodiment of the present application;

FIG. 2c is a graph illustrating the comparison between the value of an independent visitor (uv) before and after fusing of a commodity to be fused according to an embodiment of the present invention;

fig. 3 is a schematic structural diagram of a component of the commodity determining apparatus according to an embodiment of the present application;

fig. 4 is a schematic structural diagram of an electronic device according to an embodiment of the present application.

Detailed Description

The present application will be described in further detail below with reference to the accompanying drawings and examples. It should be understood that the examples provided herein are merely illustrative of the present application and are not intended to limit the present application. In addition, the following examples are provided as partial examples for implementing the present application, not all examples for implementing the present application, and the technical solutions described in the examples of the present application may be implemented in any combination without conflict.

It should be noted that in the embodiments of the present application, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a method or apparatus that comprises a list of elements does not include only the elements explicitly recited, but also includes other elements not explicitly listed or inherent to the method or apparatus. The term "comprising" is used to specify that an element, without limitation, is not intended to exclude the presence of other elements, either in the method or in the apparatus comprising the element (e.g., steps in the method or elements in the apparatus, such as a component of a circuit, a processor, a program, software, etc.).

The term "and/or" herein is merely an associative relationship that describes an associated object, meaning that three relationships may exist, e.g., I and/or J, may mean: the three cases of the single existence of I, the simultaneous existence of I and J and the single existence of J. In addition, the term "at least one" herein means any one of a plurality or any combination of at least two of a plurality, for example, including at least one of I, J, R, and may mean including any one or more elements selected from the group consisting of I, J and R.

For example, although the product specifying method provided in the embodiment of the present application includes a series of steps, the product specifying method provided in the embodiment of the present application is not limited to the described steps, and similarly, the product specifying device provided in the embodiment of the present application includes a series of modules, but the product specifying device provided in the embodiment of the present application is not limited to the modules explicitly described, and may include modules that are required to acquire the relevant time series data or perform processing based on the time series data.

Embodiments of the application are operational with numerous other general purpose or special purpose computing system environments or configurations. Here, the server may be a distributed cloud computing technology environment including a small computer system, a large computer system, and the like.

The electronic device such as the server can realize corresponding functions through the execution of the program module. Generally, program modules may include routines, programs, objects, components, logic, data structures, etc. that perform particular tasks or implement particular abstract data types. The computer system/server may be practiced in distributed cloud computing environments where tasks are performed by remote processing devices that are linked through a communications network. In a distributed cloud computing environment, program modules may be located in both local and remote computer system storage media including memory storage devices.

In the related art, for identifying low-effect commodities in commodities to be selected, commodities which are blown out in a weight-reducing mode are defined through experience, and then the commodities are deleted from a commodity candidate set. However, this identification method is not only heavy in workload, but also needs to be operated for each business activity, and especially there are hundreds of business activities in a short period, so it is difficult to realize automatic low-efficiency commodity identification, and the work efficiency is reduced.

In view of the above technical problems, the following embodiments are proposed.

In some embodiments of the present Application, the commodity determining method may be implemented by using a Processor in the commodity determining Device, where the Processor may be at least one of an Application Specific Integrated Circuit (ASIC), a Digital Signal Processor (DSP), a Digital Signal Processing Device (DSPD), a Programmable Logic Device (PLD), a Field Programmable Gate Array (FPGA), a Central Processing Unit (CPU), a controller, a microcontroller, and a microprocessor.

Fig. 1 is a schematic flowchart of a commodity determination method in an embodiment of the present application, and as shown in fig. 1, the method includes the following steps:

step 100: determining a value comparison result of a category to which each commodity to be selected belongs; the value comparison result represents a comparison result between the value of the category to which each commodity to be selected belongs and the reference value.

In the embodiment of the application, the to-be-selected commodity represents an active commodity with a display position in a search sequencing result being promoted by allocating additional unnatural traffic, for example, a weighted commodity; for example, assuming that the initial display position of a certain movable commodity in the search ranking result is ranked at the eighth position, if the right of the movable commodity is lifted, the current display position of the movable commodity may be ranked at the sixth position, and the movable commodity originally ranked at the sixth position is ranked at the seventh position. Here, the active merchandise may represent any type of merchandise that an e-commerce platform or seller is transacting over the Internet; for example, the product may be a clothing product, a food product, or the like.

Each commodity to be selected can be various commodities in the same business activity or various commodities in a plurality of different business activities; for example, the business activity may be a novelty activity, a promotional activity, and the like.

For example, each item to be selected may belong to one category or may belong to a plurality of different categories; for example, the categories may be clothing, food, and the like.

In the embodiment of the present application, the manner of determining the value comparison result of the category to which each to-be-selected commodity belongs is not limited, for example, the value comparison result of the category to which each to-be-selected commodity belongs may be determined through an ab experiment, or the value comparison result may be determined in other manners, and the ab experiment is taken as an example to be described below.

For example, in the case that the value comparison result of the category to which each to-be-selected commodity belongs is determined through an ab experiment, the value comparison result may be a uv value comparison result, may also be a UCVR comparison result, and may also be other indexes capable of determining the category to which each to-be-selected commodity belongs under the ab experiment, which is not limited in the embodiment of the present application.

Illustratively, in the case where each item to be selected belongs to a category, the uv value of the category to which each item to be selected belongs is equal to the ratio of the sales amount to the number of visitors in the category.

For example, when each to-be-selected commodity belongs to a plurality of different categories, the uv value of each category in the plurality of categories may be calculated in the above manner, and details are not described here.

Here, in the case where the value comparison result is a uv value comparison result, the reference value may be a reference uv value set in advance according to an actual business activity; the uv value of the category to which each to-be-selected commodity belongs before being subjected to the right-giving can also be the uv value, so that the uv value comparison result can represent the comparison result of the uv value of the category to which each to-be-selected commodity belongs and the uv value of the category to which each to-be-selected commodity belongs before being subjected to the right-giving; for example, uv value of a category to which each to-be-selected commodity belongs before being entitled may be calculated according to the above manner, and details are not described here.

Exemplarily, assuming that the category 1 includes goods 1 to 10, and the search ordering results of the 10 goods are goods 1 to 10 in turn, if the right of the goods 8 is lifted, the display position of the goods 8 is lifted to the display position of the goods 6, and at this time, the goods 8 is the goods to be selected. The uv value of the category to which each commodity to be selected belongs represents the uv value of the category 1 to which the commodity 8 to be selected belongs, namely, the uv value of the category 1 after the commodity 8 is subjected to weight lifting; the uv value of the category to which each candidate item belongs before being entitled represents the uv value of the category 1 before the item 8 is entitled, that is, the uv value of the category 1 before the item 8 is entitled.

Exemplarily, the uv value of the category to which each commodity to be selected belongs represents the uv value corresponding to the experimental group b in the ab experiment; the reference uv value represents the uv value corresponding to the control group a in the ab experiment; in this case, the uv value comparison result is (uv value for experimental group b-uv value for control group a)/uv value for control group a.

For example, in the case that the value comparison result is a UCVR comparison result, the reference value may be a reference UCVR preset according to an actual business activity; or the UCVR of the category to which each to-be-selected commodity belongs before being entitled, wherein the determination mode of the UCVR comparison result is similar to the uv value comparison result, and is not repeated here.

Illustratively, in the case that each article to be selected belongs to a category, the UCVR of the category to which each article to be selected belongs is equal to the ratio of the total order line to the number of visitors under the category.

Step 101: and obtaining the fusing ratio of the commodities of the category to which each commodity to be selected belongs according to the comparison result of the value comparison result and the set threshold value.

In one embodiment, the set threshold may include a first set threshold and a second set threshold; the first set threshold is a value smaller than 0, and the second set threshold is a value larger than 0; for example, the first set threshold may be opposite to the second set threshold; for example, in the case where the first set threshold value is-0.5%, the second set threshold value may be 0.5%.

For example, the implementation manner of obtaining the fusing ratio of the commodities belonging to the category to which each commodity to be selected belongs according to the comparison result of the value comparison result and the set threshold may be: when the value comparison result is determined to be smaller than a first set threshold value, setting the increment of the negative days to be 1; the first set threshold value is a value less than 0; when the value comparison result is determined to be greater than or equal to the second set threshold, setting the increment of the negative days to be-1; the second set threshold value is a value greater than 0; when the value comparison result is determined to be between the first set threshold and the second set threshold, setting the increment of the negative days to be 0; determining the commodity fusing proportion of the category to which each commodity to be selected belongs based on the increment of the negative days; the following examples of the present application will be described with the uv value comparison results as examples.

Exemplarily, assuming that the first set threshold value is-0.5%, the second set threshold value is 0.5%, and when it is determined that the uv value comparison result is less than-0.5%, the increment of the negative days is set to 1; when it is determined that the uv value comparison result is greater than or equal to 0.5%, setting the increment of the negative days to-1; when it was determined that the uv value comparison was between-0.5% and 0.5%, the increment for negative days was set to 0.

Illustratively, the negative days are represented using a character n, and the initial value of the negative days n is 0; here, negative days n are in days; if the increment of the negative days is determined to be set to 1 on the Tth day, the negative days n of the Tth day is equal to the negative days n of the T-1 th day plus the set increment of 1; if the increment of the negative days is determined to be-1 on the Tth day, the negative days n on the Tth day is equal to the negative days n on the T-1 th day plus the set increment of-1; if day T determines that the increment for negative days is set to 0, then day T negative days n is equal to day T-1 negative days n plus the set increment of 0.

In the embodiment of the application, the negative days n of each day can be correspondingly obtained according to the increment of the negative days determined each day; thus, based on the negative days n of each day, the fusing proportion of the commodities of the category to which each commodity to be selected belongs can be correspondingly determined; it can be understood that if the negative days of the T-1 th day are different from the negative days of the T-1 th day, the commodity fusing proportion of the category to which each commodity to be selected belongs determined on the T-1 th day is different from the commodity fusing proportion of the category to which each commodity to be selected belongs determined on the T-1 th day.

Therefore, the method and the device can realize the day-level update of the commodity fusing proportion through the negative days increment determined every day, so that the negative influence on the large disk caused by the business activities corresponding to the commodities to be selected is avoided, and the flow distribution efficiency is improved.

In some embodiments, determining the fusing percentage of the commodities belonging to the category of each commodity to be selected based on the increment of the negative days may include: determining the product of the increment of the negative days and a set numerical value as the fusing proportion of the commodities of the category to which each commodity to be selected belongs; the value is set to a value between 0 and 1.

Illustratively, the set value may be a value between 0 and 1, for example, the set value may take a value of 5%; here, the specific value of the setting value may be determined according to the actual service activity, and the embodiment of the present application is not limited.

In the embodiment of the application, after the increment of the negative days is obtained according to the steps, the current negative days can be determined according to the increment of the negative days; and further multiplying the current negative days by a set numerical value, and taking the product of the current negative days and the set numerical value as the fusing proportion of the commodities of the category to which each commodity to be selected belongs.

For example, when the number of negative days on the T-1 th day is 1, the increment of the number of negative days on the T th day is 1, and the set value is 5%, it may be determined that the number of negative days on the T th day is 2, and at this time, the number of negative days on the T th day 2 is multiplied by the set value of 5%, that is, 10% may be used as the fusing ratio of the target commodities to which each commodity to be selected belongs on the T th day.

Therefore, in the embodiment of the application, the commodity fusing proportion can be automatically determined every day according to the mode, and further, the subsequent commodity fusing is performed according to the commodity fusing proportion determined every day; namely, the fusing proportion of the movable commodity does not need to be adjusted according to manual experience, so that the fusing efficiency during the period of promoting activities is improved, and meanwhile, the labor is effectively saved.

In some embodiments, according to the comparison result of the value comparison result and the set threshold value, a second commodity needing to be subjected to weight reduction can be obtained, wherein the second commodity comprises at least one to-be-selected commodity; specifically, the category to be subjected to weight reduction can be determined by using the comparison result of the value comparison result and the third set threshold; and selecting a second commodity needing to be subjected to weight reduction from the category needing to be subjected to weight reduction according to a preset strategy.

Exemplarily, the set threshold further includes a third set threshold, and the third set threshold is a value smaller than 0, for example, the value of the third set threshold may be-0.5%; the third set threshold may be the same as the first set threshold.

In the embodiment of the application, after the uv value comparison result is obtained, the uv value comparison result is compared with a third set threshold value to obtain a comparison result; if the uv value comparison result is smaller than the third set threshold value according to the comparison result, indicating that the category to which each commodity to be selected belongs is the category needing to be subjected to weight reduction; otherwise, if the uv value comparison result is determined to be larger than or equal to the third set threshold according to the comparison result, it is indicated that the category to which each to-be-selected commodity belongs is not the category needing to be subjected to weight reduction.

In the embodiment of the application, after determining that the category to which each commodity to be selected belongs is the category to which the right is required to be reduced, selecting a second commodity to be reduced from the categories to which the right is required to be reduced according to a preset strategy.

Here, the preset strategy may be a commodity with exposure of 70% at the head and 30% at the tail of the category that needs to be weighted down. It can be understood that the preset policy may be adjusted according to an actual activity scenario, which is not limited in the embodiment of the present application.

Exemplarily, assuming that the category needing to be subjected to the weight reduction is category 2, and the category 2 includes 100 goods to be selected, according to the preset strategy, 70(100 × 70%) goods to be selected with the previous exposure amount can be selected from the category 2, and then 21 goods to be selected with the next click rate can be selected from the 70 goods to be selected, so that the finally determined 21 goods to be selected are the second goods needing to be subjected to the weight reduction.

Exemplarily, if the uv value comparison result is smaller than a third set threshold, it indicates that the flow distribution efficiency of the category to which the current to-be-selected commodity belongs is reduced; that is, the part of the commodities to be selected under the category can not bear the flow brought by ranking promotion; therefore, the second commodities are some commodities to be selected which can cause flow loss; in the embodiment of the application, the weights of the commodities to be selected are reduced, the exposure of the commodities are reduced, the exposure is distributed to the commodities with positive change of efficiency, and the flow efficiency is improved. That is, it is possible to solve the problem in the related art that even if the distribution efficiency of the flow rate of some of the commodities to be selected is lowered, these commodities are always given high exposure.

Step 102: and sequencing each commodity to be selected by utilizing the trained fusion model to obtain a sequencing result, and determining a first commodity to be fused from each commodity to be selected based on the sequencing result and the commodity fusing proportion.

Illustratively, the fusion model is used for reflecting the relation between the self characteristics of each to-be-selected commodity and the order quantity and/or click rate; the fusion model may include a first model and a second model; the first model is used for reflecting the relation between the self characteristic of each commodity to be selected and the order quantity, wherein the self characteristic represents the characteristic related to the commodity exposure, namely the first model is used for reflecting the relation between the exposure of each commodity to be selected and the order quantity; here, the first model may be referred to as a high-exposure low-rotation model; the second model may be a Click Through Rate (CTR) model, which is used to reflect the relationship between the characteristics of each candidate item and the Click Through Rate.

It can be seen that, according to the method and the device for fusing the commodities, the commodities to be fused and to be selected can be determined quickly based on the sequencing result and the commodity fusing proportion, and the flow efficiency in a large promotion period is ensured.

In some embodiments, the ranking each to-be-selected commodity by using the trained fusion model to obtain a ranking result may include: scoring each commodity to be selected by using the trained first model to obtain a first scoring result; scoring each commodity to be selected by using the trained second model to obtain a second scoring result; and calculating the average scoring values of the first scoring result and the second scoring result, and sorting the average scoring values to obtain a sorting result.

Illustratively, the following definition may be made for the tag data label of the high-exposure low-rotation model (first model), wherein the positive sample definition is shown in expression (1):

the negative examples are defined as shown in expression (2):

wherein, pv_testThe exposure, pv, of the test bucket_baseRepresents the exposure of the base bucket; ctr_cid3Representing the average click rate under the third category; orderlines_testIndicating the number of orders for the test bucket, orderlines_baseThe amount of orders representing the base bucket; here, the test bucket corresponds to the experimental group b in the ab experiment, and the base bucket corresponds to the experimental group a in the ab experiment.

Illustratively, when training the high-exposure low-rotation model, firstly, a corresponding training data set needs to be acquired; here, the training data set consists of a large number of pairs of feature vectors and corresponding label data label (0-1 labels); in the training data set, if the category to which each commodity to be selected belongs meets the positive sample definition, label data label is 1; and if the category of each to-be-selected commodity meets the negative sample definition, the label data label is 0.

Illustratively, the feature vectors in the training dataset are generated from the raw features; here, the original feature may be a continuous feature, and for example, the current day, the last 3 days, the last 7 days, and the last 15 days of CTR, conversion Rate (Click Value Rate, CVR), IPV (number of times of entering into the item detail page), thousands of display Revenue (revnue Per mill, RPM), total transaction amount (Gross trade Volume, GMV), and number of purchases (addcart) of each item to be selected in the third category may be used; the original features may also be discrete features, such as whether to place an advertisement, whether to have a total stop, etc. In addition, the present embodiment does not limit the original feature types of the high-exposure low-rotation model, and for example, the continuous features may further include features such as store identification and brand identification.

The original features are screened and cleaned, then code conversion (including feature crossing) is carried out, finally, feature vectors are generated, and the feature vectors and corresponding 0-1 labels are input into a high-exposure low-rotation model for training, so that the trained high-exposure low-rotation model is obtained.

Illustratively, during prediction, the feature vector related to each commodity to be selected is input into a trained high-exposure low-rotation model, a number between 0 and 1 is output by the model, and each commodity to be selected is sequenced by the number.

For example, assuming that the output number of the model is 0.8 after a feature vector related to the product 1 to be selected is input into the high-exposure low-rotation model, 0.2(1-0.8 ═ 0.2) is taken as the score of the high-exposure low-rotation model for the product 1 to be selected.

Illustratively, after each commodity to be selected is scored according to the high exposure and low rotation model, the score of each commodity to be selected can be obtained; i.e., the first scoring result.

Exemplarily, the following definitions may be made for the tag data label of the CTR model (second model), where a positive sample defines: clicks were made in the last 7 days; negative examples define: there were no clicks for the last 7 days.

Similarly, when training the CTR model, first, a corresponding training data set needs to be acquired; here, the training data set consists of a large number of pairs of feature vectors and corresponding label data label (0-1 labels); in the training data set, if the category to which each commodity to be selected belongs meets the positive sample definition, label data label is 1; and if the category of each to-be-selected commodity meets the negative sample definition, the label data label is 0.

Illustratively, the feature vectors in the training dataset are generated from the raw features; here, the original feature may be a continuous feature, and the continuous feature may include a commodity feature, and may be, for example, a normalized value such as total sales, number of hits in search, number of focus in search, CVR, CTR, and the like of each commodity to be selected on the current day, approximately 3 days, approximately 7 days, and approximately 15 days; user characteristics such as age, gender, etc. of the user may also be included; may also include brand characteristics of the goods, such as brand click-through, brand, etc.; store characteristics such as store scores, store goodness, store badness, store wind vane, etc. may also be included. The original features may also be discrete features, which may include merchandise features, e.g., whether or not it is a self-contained merchandise; user characteristics may also be included, for example, users of different age segments; merchandise brand characteristics may also be included, such as whether a set brand; store characteristics such as store tiering may also be included. It should be noted that, in the embodiment of the present application, the original feature type of the CTR model is not limited.

The original features of the CTR model are screened and cleaned, then code conversion (including feature crossing) is carried out, finally, feature vectors are generated, and the feature vectors and the corresponding 0-1 labels are input into the CTR model for training to obtain the trained CTR model.

Illustratively, during prediction, the feature vector related to each commodity to be selected is input into the trained CTR model, the model outputs a number between 0 and 1, and each commodity to be selected is sequenced by using the number.

Illustratively, assuming that after a feature vector related to the to-be-selected commodity 1 is input into the CTR model, the output number of the model is 0.4, and 0.4 is taken as the score of the CTR model on the to-be-selected commodity 1.

Illustratively, after each commodity to be selected is scored according to the CTR model, the score of each commodity to be selected can be obtained; i.e. the second scoring result.

Illustratively, the score of each commodity to be selected, namely a first scoring result and a second scoring result, can be obtained according to the high exposure and low rotation model and the CTR model; here, the average score value of each item to be selected in the first scoring result and the second scoring result may be calculated by expression (3); and sorting the average scoring values in the order of scores from high to low to obtain a final sorting result.

Here, score represents the above fusion model; score₁Representing the first model, namely a high-exposure low-rotation model; score₂Representing the second model described above, the CTR model.

Exemplarily, assuming that the category needing to be subjected to weight reduction is category 3, and category 3 includes 3 to-be-selected commodities, namely commodities 1 to 3, if the scoring results of commodities 1 to 3 through the high-exposure low-rotation model are 0.2, 0.5 and 0.3 in sequence; the scoring results of the products 1 to 3 through the CTR model are 0.4, 0.5 and 0.5 in sequence; calculating the average score values of the commodities 1 to 3 in the first scoring result and the second scoring result to obtain the average score values of the commodities 1 to 3 which are 0.3, 0.5 and 0.4 in sequence; after the average score values are sorted in the order of scores from high to low, the final sorting results of the commodities 1 to 3 in the category 3 can be determined to be 0.3, 0.4 and 0.5.

In the embodiment of the application, after the sorting result of each commodity to be selected is obtained, the first commodity needing to be fused is determined from each commodity to be selected according to the sorting result and the commodity fusing proportion.

Exemplarily, it is assumed that 10 goods to be selected are included in a category, and the final ordering results of the 10 goods to be selected are goods 1 to goods 10 in turn from large to small; if the commodity fusing ratio of the purpose is determined to be 10% according to the commodity fusing ratio of step 101, the number 1 of first commodities needing to be fused (10 × 10% — 1) can be determined; further, the end product 10 may be fused.

It can be seen that in the embodiment of the application, each commodity to be selected is scored through the first model and the second model respectively, and the first commodity to be fused is determined according to the scoring results of the first model and the second model together, so that the accuracy of fusing the commodities can be improved.

In some embodiments, the reducing the weight of the second product that needs to be reduced according to the historical data related to the click rate of the second product may specifically include: determining the weight reduction proportion of the second commodity according to historical data related to the click rate of the second commodity; and reducing the right of the second commodity needing to be subjected to the right reduction by using the right reduction proportion.

Illustratively, historical data related to the click rate of the second commodity is obtained first, and after the historical data related to the click rate of the second commodity is obtained, the real-time weight reduction parameter γ of each to-be-selected commodity in the second commodity can be determined according to expression (4)_i,t：

Wherein α and β are adjustable coefficients; ctri, t represents the click rate of the ith to-be-selected commodity at the time t, ctr_c,t-1And the average click rate of the category to which the ith to-be-selected commodity belongs at the time t-1 is represented.

Here, the real-time weight reduction parameter γ described above may be applied using expression (5)_i,tCorrecting to obtain corrected real-time weight-reducing parameters

According to the expression (5), the weight-reducing parameter gamma is reduced in real time_i,tWhen the current value is greater than or equal to 0.5, the corresponding commodity to be selected is considered not to cause flow loss, and at the moment, the right of the commodity to be selected is not reduced; otherwise, the weight parameter gamma is reduced in real time_i,tWhen the flow rate is less than 0.5, the corresponding commodity to be selected is considered to be an inefficient commodity causing flow loss; at this time, the real-time weighting reducing parameter gamma_itValue of 2

And the right reducing proportion is used for reducing the right of the to-be-selected commodity, and the right of the to-be-selected commodity is reduced by utilizing the right reducing proportion.

Illustratively, historical data related to the click rate of the second commodity can be acquired once per hour, and the weight reduction proportion of each to-be-selected commodity is determined according to the formula; therefore, the embodiment of the application can reduce the right to a certain degree for the commodities with large disks and categories and causing real-time efficiency loss by counting the hourly feedback historical data of the commodities to be selected.

The embodiment of the application provides a commodity determining method, a commodity determining device, electronic equipment and a computer storage medium, wherein the method comprises the following steps: determining a value comparison result of a category to which each commodity to be selected belongs; the value comparison result represents a comparison result of the value of the category to which each commodity to be selected belongs and the reference value; according to the comparison result of the value comparison result and the set threshold value, obtaining the fusing proportion of the commodities of the category to which each commodity to be selected belongs; sorting each commodity to be selected by using the trained fusion model to obtain a sorting result, and determining a first commodity to be fused from each commodity to be selected based on the sorting result and the commodity fusing proportion; the fusion model is used for reflecting the relation between the self characteristics of each commodity to be selected and the order quantity and/or click rate. It can be seen that the commodity fusing proportion of the related categories can be obtained by utilizing the relationship between the value comparison result of the category to which each commodity to be selected belongs and the set threshold value; further, determining the commodities needing to be fused in related categories according to the commodity fusing proportion; compared with the prior art that the fusing proportion of each commodity to be selected needs to be adjusted according to manual experience, the commodity fusing proportion can be obtained without spending a large amount of time on analyzing data manually, and a large amount of manpower resources can be saved; in addition, according to the embodiments, since the second commodities needing to be subjected to the right reducing are some candidate commodities which cause flow loss, the embodiments of the present application can improve the flow distribution efficiency by performing the right reducing processing on the candidate commodities.

In order to further embody the object of the present application, the present application will be further described with reference to the claimed product as an example in addition to the above-described embodiments. FIG. 2a is a schematic diagram illustrating a process of determining an article to be fused according to an embodiment of the present application; as shown in fig. 2a, the process includes the following steps:

step A1: and starting fusing.

For example, before determining the goods needing to be fused, the operator issues a fusing starting instruction to the e-commerce platform, and the e-commerce platform starts fusing after receiving the instruction.

Step A2: and acquiring the trained first model and the trained second model.

Illustratively, the first model and the second model can be trained respectively by using historical data of each weighted commodity in the business activity, so as to obtain the trained first model and second model.

Step A3: and performing fusion scoring sequencing on the goods to be entitled.

Illustratively, after a trained first model and a trained second model are obtained, scoring is respectively performed on each weighted commodity by using the two models, an average scoring value corresponding to each weighted commodity is calculated, and the average scoring values corresponding to each weighted commodity are sorted from large to small to obtain scoring sorting results.

Step A4: and determining the uv value comparison result of the category to which the goods to be entitled belong.

Exemplarily, determining a uv value comparison result of a category to which each commodity belongs in the business activity under an ab experiment; the uv value comparison result represents the comparison result of the uv value of the category to which each of the weighted commodities belongs and the uv value of the category to which each of the non-weighted commodities belongs.

Step A5: and judging whether the uv value comparison result is less than-0.5%.

Exemplarily, -0.5% represents the first set threshold, and if the determination result is yes, step a6 is executed; otherwise, step a7 is performed.

Step A6: the increment for negative days is set to 1.

Illustratively, the initial value of the negative days n is 0, and the increment of the negative days is set to 1 in the case where it is determined that the uv value comparison result is less than-0.5%.

Step A7: and judging whether the uv value comparison result is greater than or equal to 0.5%.

Exemplarily, 0.5% represents the second set threshold, and if the determination result is yes, step A8 is executed; otherwise, step a9 is performed.

Step A8: the increment for negative days is set to-1.

Illustratively, where it is determined that the uv value comparison result is greater than or equal to 0.5%, the increment for negative days is set to-1.

Step A9: the increment for negative days is set to 0.

Illustratively, where it is determined that the uv value comparison is less than 0.5%, the increment for the negative days is set to 1.

Step A10: and determining the fusing ratio of the commodities.

Exemplarily, after the final negative days n are obtained according to the steps, 0.5% x n is used as a commodity fusing proportion, and the tail part of the scoring and sorting result is fused according to the commodity fusing proportion, so that the commodity needing to be fused is determined; here, the commodity fusing ratio may be determined once per day from uv value data fed back.

It can be seen that the method and the device can automatically adjust the fusing proportion of the commodities in time according to the day-level uv value feedback data of the class to which each authorized commodity belongs, the goal is to finally achieve the goal that the large-disk efficiency is not negative, the efficiency of manual fusing in a large promotion period and a weekday is improved, and manpower is effectively saved.

Illustratively, for a general tuning problem, there is a definition shown in expression (6):

here, f (x)_ij) Representing a business objective; the reduction in conversion for the large plate is given by the function g (x)_ij) Indicating that the business objective is maximized in the case that the large panel drop does not exceed Δ; the flow rate obtained by the authorized commodity is S, S belongs to C, and C is a constant; x is the number of_iIndicating that the exposure amount obtained by the ith weighted commodity is more than 0; x is the number of_ijIndicating that the exposure of the ith piece of authorized goods is increased for the jth user.

Constructing a commodity fusing target, namely, a fuse target, based on the definition shown in the above expression (6); here, it may be determined that the melt target should be consistent with the ownership activity target, maximizing the number of hits under the constraint condition that the large disk is lossless, corresponding to expression (7):

here, Pctr_i*x_ijIndicating an ith product to be authorized for a jth userThe ideal number of clicks.

Constructing a hypothesis condition according to the melt object shown in the expression (7); here, assuming that the right-lift activity traffic distribution is reasonable, the commodity does not need to be fused, but in reality, the traffic distribution may not reach the optimal solution, so that there may be a difference between the real number of clicks and the ideal number of clicks, and the melt object is to optimize the difference between the two, so there are sub-objects, as shown in expression (8):

here, ,

indicating the real number of clicks of the ith piece of entitled goods for the jth user.

In the embodiment of the application, the first commodity which is determined to be fused is fused; or, the weight of the second commodity needing to be reduced can be reduced, and the real number of clicks shown in the expression (8) can be optimized

And ideal number of clicks Pctr_i*x_ijThe difference between the two is used for realizing the business target of improving the exposure or click of the commodity.

Fig. 2b is a schematic flowchart of a process of reducing the right of an authorized product in the embodiment of the present application, and as shown in fig. 2b, the flowchart includes five modules, which are a real-time data bus (JDQ) module, a Doris module, a Mysql module, a real-time right reducing module, and a monitoring billboard module; the JDQ module is used for performing data embedding on behavior data of a user for the goods to be authorized at the bottom layer; the Doris module is used for acquiring buried point data from the JDQ module, processing the data in real time and storing a data processing result; the Mysql module is used for acquiring a corresponding data processing result from the JDQ module to perform subsequent data processing, and the data can be data related to uv value; the real-time right reducing module is used for determining the right reducing proportion of each right-proposed commodity every hour; the monitoring billboard module is used for checking the performance of the second commodity after the right of the second commodity is reduced.

Illustratively, the monitoring billboard module is further used for monitoring the fusing situation of the commodity in real time, and precipitating the result data of each dimension on a data platform, so as to provide a more comprehensive explanation for the fusing of the commodity. The monitoring billboard module can monitor whether the fusing of the commodity is effective or not, and the fusing of the commodity can be determined according to the quantity distribution of ab experiment exposure lifting, fusing commodity exposure proportion and the degree of the dropped weight; whether the efficiency of the fused product meets the requirements or not can be determined through the monitoring billboard module, and the determination can be specifically carried out according to the efficiency of a large plate, the efficiency of classification items, uv value under a trigger search term (query), the efficiency of fused commodities and unblown commodities, the efficiency comparison before and after fusing of the fused commodities and the triggering flow efficiency comparison before and after fusing of the fused commodities; the fusion model can be evaluated through the monitoring billboard module, and can be determined according to the following modes: the recall rate of the fuse is predicted to be correct and needs to be fused/actually needs to be fused; the accuracy rate of the fuse is equal to the accuracy of the fuse which needs to be fused in prediction/the accuracy of the fuse which needs to be fused in prediction; the missed recall rate of the fuse is the article which needs to be fused but is not predicted/the article which actually needs to be fused; commodities with negative uv value to be fused are needed; predicting the correct goods to be fused; here, the uv value comparison results of the next day after the inefficient commodity that correctly needs to be blown is blown change from negative to positive, see FIG. 2 c.

Fig. 2c is a schematic graph of uv value comparison results before and after fusing of a commodity needing to be fused in the embodiment of the present application, and as shown in fig. 2c, if the commodity needing to be fused is not fused starting at time t, the uv value comparison result in an ab experiment is decreased, that is, the unblown authorized commodity causes flow loss to a large disk; if the commodity that needs to be fused fuses, uv value contrast result under the ab experiment rises, promptly, can not cause the flow loss to the large plate through commodity after fusing, promotes flow distribution efficiency.

Fig. 3 is a schematic structural diagram of a component of a product determining apparatus according to an embodiment of the present application, and as shown in fig. 3, the apparatus includes: a determining module 300, an obtaining module 301 and a processing module 302, wherein:

the determining module 300 is configured to determine a value comparison result of a category to which each to-be-selected commodity belongs; the value comparison result represents a comparison result of the value of the category to which each commodity to be selected belongs and the reference value;

an obtaining module 301, configured to obtain a commodity fusing ratio of a category to which each commodity to be selected belongs according to a comparison result between the value comparison result and a set threshold;

the processing module 302 is configured to sort each to-be-selected commodity by using the trained fusion model to obtain a sorting result, and determine a first commodity to be fused from each to-be-selected commodity based on the sorting result and the commodity fusing proportion; the fusion model is used for reflecting the relation between the self characteristics of each to-be-selected commodity and the order quantity and/or click rate.

In some embodiments, the processing module 302 is further configured to:

In some embodiments, the obtaining module 301 is configured to obtain, according to a comparison result between the value comparison result and the set threshold, a fusing ratio of the commodities belonging to the category to which each commodity to be selected belongs, and includes:

In some embodiments, the obtaining module 301 is configured to determine, based on the increment of the negative days, a fusing percentage of the commodities of the category to which each commodity to be selected belongs, where the fusing percentage of the commodities includes:

In some embodiments, the fusion model includes a first model and a second model, and the processing module 302 is configured to sort each to-be-selected commodity by using the trained fusion model to obtain a sorting result, including:

In some embodiments, the setting threshold includes a third setting threshold, and the obtaining module 301 is configured to obtain, according to a comparison result between the value comparison result and the setting threshold, a second commodity needing to be subjected to weight reduction, including:

In some embodiments, the processing module 302 is configured to perform a derating on the second product needing to be derated according to historical data related to the click rate of the second product, and includes:

In practical applications, the determining module 300, the obtaining module 301 and the processing module 302 may be implemented by a processor located in an electronic device, and the processor may be at least one of an ASIC, a DSP, a DSPD, a PLD, an FPGA, a CPU, a controller, a microcontroller and a microprocessor.

In addition, each functional module in this embodiment may be integrated into one processing unit, or each unit may exist alone physically, or two or more units are integrated into one unit. The integrated unit can be realized in a form of hardware or a form of a software functional module.

Based on the understanding that the technical solution of the present embodiment essentially or a part contributing to the related art, or all or part of the technical solution, may be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for enabling a computer device (which may be a personal computer, a server, or a network device, etc.) or a processor (processor) to execute all or part of the steps of the method of the present embodiment. And the aforementioned storage medium includes: a U disk, a removable hard disk, a Read Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, and so on.

Specifically, the computer program instructions corresponding to a commodity determination method in the present embodiment may be stored on a storage medium such as an optical disc, a hard disc, or a usb disk, and when the computer program instructions corresponding to a commodity determination method in the storage medium are read or executed by an electronic device, any one of the commodity determination methods of the foregoing embodiments is implemented.

Based on the same technical concept of the foregoing embodiment, referring to fig. 4, it shows an electronic device 400 provided by the present application, which may include: a memory 401 and a processor 402; wherein,

a memory 401 for storing computer programs and data;

a processor 402 for executing a computer program stored in the memory to implement any one of the article determination methods of the previous embodiments.

In practical applications, the memory 401 may be a volatile memory (RAM); or a non-volatile memory (non-volatile memory) such as a ROM, a flash memory (flash memory), a Hard Disk (HDD), or a Solid-State Drive (SSD); or a combination of the above types of memories and provides instructions and data to the processor 402.

The processor 402 may be at least one of an ASIC, a DSP, a DSPD, a PLD, an FPGA, a CPU, a controller, a microcontroller, and a microprocessor. It is to be understood that the electronic device for implementing the above-mentioned processor function may be other devices for different object property determination devices, and the embodiments of the present application are not particularly limited.

In some embodiments, functions of or modules included in the apparatus provided in the embodiments of the present application may be used to execute the method described in the above method embodiments, and specific implementation thereof may refer to the description of the above method embodiments, and for brevity, will not be described again here.

The foregoing description of the various embodiments is intended to highlight various differences between the embodiments, and the same or similar parts may be referred to each other, and for brevity, will not be described again herein.

The methods disclosed in the method embodiments provided by the present application can be combined arbitrarily without conflict to obtain new method embodiments.

Features disclosed in various product embodiments provided by the application can be combined arbitrarily to obtain new product embodiments without conflict.

The features disclosed in the various method or apparatus embodiments provided herein may be combined in any combination to arrive at new method or apparatus embodiments without conflict.

As will be appreciated by one skilled in the art, embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of a hardware embodiment, a software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, optical storage, and the like) having computer-usable program code embodied therein.

The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

The above description is only a preferred embodiment of the present application, and is not intended to limit the scope of the present application.

Claims

1. A method for merchandise identification, the method comprising:

2. The method of claim 1, further comprising:

3. The method according to claim 1, wherein the setting threshold comprises a first setting threshold and a second setting threshold, and the obtaining of the fusing ratio of the commodities belonging to the category of each commodity to be selected according to the comparison result of the value comparison result and the setting threshold comprises:

4. The method of claim 3, wherein the determining the percentage of fusing of the items of the category to which each item of merchandise belongs based on the increment of negative days comprises:

5. The method according to claim 1, wherein the fusion model includes a first model and a second model, and the ranking of each item to be selected by using the trained fusion model to obtain a ranking result includes:

6. The method of claim 5, wherein the first model is used for reflecting the relation between the exposure of each commodity to be selected and the order quantity.

7. The method of claim 2, wherein the set threshold comprises a third set threshold, and obtaining a second product to be de-weighted according to the comparison result of the value comparison result and the set threshold comprises:

8. The method of claim 2, wherein the derating the second item needing derating according to historical data related to click rate of the second item comprises:

9. An article determination device, the device comprising:

10. An electronic device, characterized in that the device comprises a memory, a processor and a computer program stored on the memory and executable on the processor, which when executing the program implements the method of any of claims 1 to 8.

11. A computer storage medium on which a computer program is stored, characterized in that the computer program realizes the method of any one of claims 1 to 8 when executed by a processor.