A method for calculating popular content based on time decay and click position factors

Abstract

This application discloses a method for calculating popular content that integrates time decay and click location factors, including: obtaining user click exposure behavior data for the past N days; calculating the number of days between the date the user performed an action on an item and the current date; obtaining a click set A and an exposure set B; internally correlating exposure set B with click set A to obtain set C; filtering exposure data near the click location where the click behavior occurred from set C to obtain exposure set D; merging click set A and exposure set D to obtain set E; performing an aggregation operation on set E to calculate the number of clicks and exposures for each item per day to obtain set F; calculating the click-through rate (CTR) for each item in set F based on the time decay rate to obtain an item CTR result set G; filtering the item CTR result set G and then sorting the CTR values ​​in descending order to obtain the final popular results. This application can achieve accurate calculation of the click-through rate (CTR) of content.

Description

Technical Field

[0001] The present invention relates to the field of network technology, and in particular to a method and device for calculating popular content by integrating time decay factors and click position factors. Background Art

[0002] In the current hot Internet field, there are extremely rich Internet products and content, and there are many forms of content, such as news information, popular games, e-commerce products, advertisements, videos, music, etc. The amount of content in most products is very large, but the content browsing time of each user is limited. How to push high-quality content within the user's limited browsing time period is the most concerned topic of Internet products.

[0003] The click-through rate (CTR) of content is the number of clicks on the product content within a period divided by the number of displays within the same period. Generally, higher-quality content can attract users' attention, and thus the click-through rate index value is higher. Therefore, the click-through rate index can reflect the popularity of product content to a certain extent.

[0004] However, conventional content CTR calculations tend to be limited in their considerations, focusing solely on clicks and impressions, failing to factor in the changing patterns of content popularity. This can hinder the timely updating of popular content. Furthermore, during actual user browsing, some content may be successfully exposed but not actually noticed, resulting in invalid exposure data. These factors can affect the actual CTR calculation of content.

[0005] Therefore, how to introduce the user attention factor and add the exposure data near the click location that the user actually noticed into the CTR calculation, while taking into account the time decay factor of the content's popularity, so as to more accurately calculate the content's click-through rate (CTR), is particularly important in the promotion and display of product content. Summary of the Invention

[0006] The purpose of the present invention is to provide a method and device for calculating popular content by integrating time decay factors and click position factors, so as to solve the problems raised in the above technical background.

[0007] To achieve the above object, the present invention adopts the following technical solutions:

[0008] The first aspect of the present application discloses a method for calculating popular content by integrating time decay factors and click position factors, including:

[0009] Obtain the user's click and exposure behavior data for the past N days, where the click and exposure behavior data includes the date when the user clicked or exposed the item;

[0010] Based on the click / exposure behavior data, calculate the number of days between the date when the user clicked / exposure behavior on the item and the current date;

[0011] Obtain click set A and exposure set B from the click-exposure behavior data, where each of click set A and exposure set B includes the user ID, item ID, the ID of the batch of recommended content obtained by the user ID, the location record of the user ID's click / exposure behavior, and the number of days between the date when the user ID clicked / exposure behavior on the item and the current date. A batch of recommended content IDs includes at least one item ID.

[0012] Using the user ID and the batch ID of the recommended content, we internally associate the exposure set B with the click set A to obtain set C. Set C includes the exposure data in the batch where the click occurred.

[0013] From set C, the exposure data near the click location where the click behavior occurred is filtered according to the preset first filtering strategy to obtain exposure set D;

[0014] Modify the field names of the filtered exposure set D to make them consistent with the click set A. Merge click set A with the filtered exposure set D, and remove the batch recommendation number field obtained by the user number and the location field where the user number clicked / exposed. This yields set E.

[0015] Aggregate set E based on preset fields to calculate the number of clicks and impressions for each item on each day, resulting in set F. The preset fields include the item ID, the number of days between the date the user ID clicked or had an impression of the item and the current date.

[0016] Calculate the CTR for each item in set F based on the time decay rate, and obtain the item CTR result set G;

[0017] The item CTR result set G is first filtered out of items whose exposure count is less than the preset exposure threshold, and then the CTR values ​​are sorted in descending order to obtain the final popular results, that is, the CTR popular content improved by the time decay factor and the click position factor.

[0018] In a preferred embodiment, the acquisition of the user's click-exposure behavior data for the past N days includes the user number user_id, the item number item_id, the batch recommended content number trace_id, the position of the user's click / exposure behavior position, the label label, label=1 indicates click, label=0 indicates exposure, and the date bhv_date when the user clicked / exposed the item; wherein, a batch recommended content number trace_id includes at least one item number item_id.

[0019] In a more preferred embodiment, the click exposure behavior data of the user in the past N days is filtered by the condition label=1 to obtain the click set A;

[0020] The user's click exposure behavior data for the past N days is filtered using the condition label = 0 to obtain exposure set B.

[0021] In a preferred embodiment, after the exposure set B is internally associated with the click set A to obtain the set C, if the user only performs exposure behavior in a certain batch of recommended items, then the exposure data of this entire batch will not be included in the subsequent item CTR calculation.

[0022] In a preferred embodiment, the obtained set C includes the user number user_id, the number of the exposed item expose_item_id, the number of the batch recommended content obtained by the user number trace_id, the exposure position expose_position, the number of the clicked item click_item_id, the click position click_position, the label label, label = 1 means click, label = 0 means exposure, and the number of days day_gap between the date when the user number clicks / exposes the item and the current date.

[0023] In a preferred embodiment, the preset first screening strategy includes: allowing the K1 exposure data closest to the bottom of the click position and the K2 exposure data closest to the top of the click position to participate in the calculation, where K1 and K2 are both integers greater than or equal to 1.

[0024] In a preferred embodiment, the calculation of the CTR for each item in the set F based on the time decay rate includes: the CTR calculation formula for item i is:

[0025] ctr_i=sum(decay_rate^day_gap*click_num) / (sum(decay_rate^day_gap*expose_num)+1);

[0026] Among them, decay_rate is the time decay rate, which controls the speed at which the popularity of content changes over time, and its value range is (0.5, 1); day_gap is the number of days between the date when the user ID clicks / exposes the item and the current date; click_num is the number of clicks per day; and expose_num is the number of exposures per day.

[0027] A second aspect of the present application discloses a device for calculating popular content by integrating time decay factors and click position factors, comprising:

[0028] An information acquisition module is used to obtain the user's click and exposure behavior data for the past N days, where the click and exposure behavior data includes the date when the user clicked or exposed the item;

[0029] New information field calculation module, used to calculate the number of days between the date when a user clicked / exposed an item and the current date based on click / exposure behavior data;

[0030] A click set A generation module is configured to obtain click set A from the click exposure behavior data, wherein click set A includes a user ID, an item ID, an ID of a batch of recommended content obtained by the user ID, a record of the location where the user ID clicked / exposed the content, and the number of days between the date when the user ID clicked / exposed the content and the current date. A batch of recommended content IDs includes at least one item ID.

[0031] An exposure set B generation module is configured to obtain exposure set B from the click exposure behavior data, wherein exposure set B includes a user ID, an item ID, the ID of the batch of recommended content obtained by the user ID, a record of the location where the user ID's click / exposure behavior occurred, and the number of days between the date when the user ID's click / exposure behavior occurred on the item and the current date. A batch of recommended content IDs includes at least one item ID.

[0032] A set C generation module is used to internally associate exposure set B with click set A using the user ID and the batch ID of the recommended content to generate set C. Set C includes exposure data in batches where clicks occurred.

[0033] An exposure set D generation module is configured to filter exposure data near a click location where a click behavior occurs from set C according to a preset first filtering strategy to obtain exposure set D;

[0034] The module for generating set E is used to modify the field names of the filtered exposure set D to make them consistent with the click set A, merge click set A with the filtered exposure set D, and remove the number field of the batch recommendation content obtained by the user number and the location field where the user number has clicked / exposed behavior, to obtain set E;

[0035] The set F generation module is used to aggregate the set E according to preset fields to calculate the number of clicks and impressions of each item on a daily basis to obtain the set F. The preset fields include the item number, the number of days between the date when the user number clicked or was exposed to the item and the current date;

[0036] The click-through rate calculation module calculates the CTR of each item in the set F based on the time decay rate and obtains the item CTR result set G;

[0037] The CTR popular content selection module is used to first filter out items with exposure counts less than a preset exposure threshold from the item CTR result set G, and then sort the items in descending order of CTR values ​​to obtain the final popular results, namely the CTR popular content improved by time decay factors and click position factors.

[0038] In a preferred embodiment, the user's click-exposure behavior data for the past N days includes the user number user_id, the item number item_id, the batch recommended content number trace_id, the position of the user's click / exposure behavior position, the label label, label=1 indicates click, label=0 indicates exposure, and the date bhv_date when the user clicked / exposed the item; wherein, a batch recommended content number trace_id includes at least one item number item_id.

[0039] In a preferred embodiment, the set C includes the user number user_id, the number of the exposed item expose_item_id, the number of the batch recommended content obtained by the user number trace_id, the exposure position expose_position, the number of the clicked item click_item_id, the click position click_position, the label label, label=1 means click, label=0 means exposure, and the number of days day_gap between the date when the user number clicked / exposed the item and the current date.

[0040] The third aspect of the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements any of the above-mentioned methods for calculating popular content based on the comprehensive time decay factor and click position factor.

[0041] A fourth aspect of the present application provides an electronic device, including:

[0042] processor; and

[0043] a memory for storing executable instructions of the processor;

[0044] The processor is configured to execute any one of the above-mentioned methods for calculating popular content based on the comprehensive time decay factor and click position factor by executing the executable instructions.

[0045] Compared with the prior art, the technical solution of the present invention has the following beneficial effects:

[0046] The present application discloses a method and device for calculating popular content by integrating time decay factors and click position factors. The method introduces the user attention factor and adds the exposure data near the click position that the user actually notices into the CTR calculation. At the same time, the time decay factor of the content popularity and the changing pattern of the product content popularity are taken into account, so as to more accurately calculate the click-through rate (CTR) of the content, and then obtain improved CTR popular content by integrating time decay factors and click position factors, which can effectively realize the promotion and display of product content. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] The accompanying drawings, which constitute part of this application, are intended to provide a further understanding of this application. The exemplary embodiments and descriptions of this application are intended to explain this application and do not constitute an improper limitation on this application. In the accompanying drawings:

[0048] Figure 1 A flowchart schematically illustrates a method for calculating popular content by integrating time decay factors and click position factors;

[0049] Figure 2 The following schematically shows the structure of a popular content calculation device that integrates time decay factors and click position factors. DETAILED DESCRIPTION

[0050] In order to make the purpose, technical solution and effect of the present invention clearer and more specific, the present invention is further described in detail below with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0051] It should be noted that the terms "first," "second," and the like in the specification and claims of the present invention and the accompanying drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the terms used in this manner are interchangeable under appropriate circumstances. In addition, the terms "including," "having," and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product, or apparatus that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units that are not explicitly listed or that are inherent to these processes, methods, products, or apparatuses.

[0052] Example:

[0053] In some embodiments, when an application on a terminal device accesses the network, the terminal device can display the page the user has visited. Examples of terminal devices include various computing devices such as desktop computers, laptops, tablets, mobile phones, or handheld game consoles. Applications on the terminal device include various software such as browsers, WeChat, QQ, Weibo, and apps. Furthermore, the visited page can typically display various additional content, such as news and advertisements. Examples of additional content include various media formats, such as static images, dynamic images, text messages, and videos.

[0054] Taking the news product app as an example, a large amount of user behavior data, such as exposure, clicks, and other behavior data, can be obtained in the news product app. The following describes the CTR popular content calculation method of this application in detail based on this application scenario.

[0055] Figure 1 The figure is a flowchart of a method for calculating popular content that integrates time decay factors and click position factors.

[0056] See Figure 1 As shown, a method for calculating popular content that integrates time decay factors and click position factors includes the following steps:

[0057] Step S1: Obtain the user's click exposure behavior data for the past N days.

[0058] The main information fields are: user_id, item_id, trace_id, position, label, bhv_date;

[0059] in,

[0060] user_id: indicates the user ID;

[0061] item_id: indicates the item number;

[0062] trace_id: indicates the ID of the current batch of recommended content. One ID can contain multiple items. For example, one ID can contain about 15 to 30 items.

[0063] Position: The position of the user behavior. For example, if the user clicks / exposures the first news item, position = 1.

[0064] label: label, where label=1 means click; label=0 means exposure;

[0065] bhv_date: indicates the date when the user clicks / exposes the item.

[0066] The following is a sample of the data obtained:

[0067] user_id item_id trace_id position label bhv_date a1 i1 t1 3 1 2022-12-28 a1 i2 t1 7 0 2022-12-28 a2 i3 t2 1 1 2022-12-31 a2 i4 t2 2 0 2022-12-31 …… …… …… …… …… ……

[0068] Step S2: A new information field can be calculated from the click / exposure behavior data: day_gap, which means the number of days between the user's click / exposure behavior time on the item and the current time.

[0069] Step S3: Obtain click set A and exposure set B from the click exposure behavior data.

[0070] Specifically, the click set A is obtained by filtering through the condition label=1; the exposure set B is obtained by filtering through the condition label=0.

[0071] The data sample obtained is as follows (assuming the current time is 2023-01-01):

[0072] Click on Set A:

[0073]

[0074]

[0075] Exposure set B:

[0076] user_id item_id trace_id position label day_gap a1 i2 t1 7 0 4 a3 i4 t2 2 0 1 …… …… …… …… …… ……

[0077] Step S4: Exposure set B is internally associated with click set A through user_id and trace_id to obtain set C, so that set C only contains exposure data in batches where click behavior occurred.

[0078] In this way, if a user only exposes recommended items in a certain batch, the exposure data of this entire batch will not be included in the subsequent item CTR calculation.

[0079] At this point, set C contains the following information:

[0080] user_id, expose_item_id, trace_id, expose_position, label, click_item_id, click_position, day_gap.

[0081] Among them, expose_item_id is the ID of the exposed item; expose_position is the exposure position; click_item_id is the ID of the clicked item, and click_position is the click position.

[0082] The data sample is as follows:

[0083] Set C:

[0084] user_id expose_item_id trace_id expose_position label click_item_id click_position day_gap a1 i2 t1 7 0 i1 3 4 a2 i4 t2 2 0 i3 1 1 …… …… …… …… …… …… ……

[0085] Step S5: Filter out the exposure data near the click location to obtain an exposure set D.

[0086] For example, the filter criteria are:

[0087] click_position–K2<=expose_position<=click_position+K1

[0088] Explanation of parameters K1 and K2: The K1 impression data points below the click position and the K2 impression data points above the click position are allowed to participate in subsequent calculations. This is because this part of the displayed content is likely to be noticed by the user and is suitable for calculating the item's corrected CTR. K1 and K2 are both integers greater than or equal to 1.

[0089] Assuming K1=K2=2, data that does not satisfy click_position-2<=expose_position<=click_position+2 will be filtered out (for example, in the above example, the first data in set C will be filtered out because 7>3+2).

[0090] Thus, the data sample of exposure set D is obtained as follows:

[0091] user_id expose_item_id trace_id expose_position label click_item_id click_position day_gap a2 i4 t2 2 0 i3 1 1 …… …… …… …… …… …… ……

[0092] Step S6: Modify the field name of the filtered exposure set D to make it consistent with the click set A, so as to facilitate the merging of the click set A and the filtered exposure set D. Then remove the trace_id and position fields to obtain the data set E.

[0093] The data sample is as follows:

[0094] Set E:

[0095] user_id item_id label day_gap a1 i1 1 4 a2 i3 1 1 a2 i4 0 1 …… …… …… ……

[0096] Step S7: Based on the item_id and day_gap fields, perform an aggregation operation on set E to calculate the number of clicks and exposures of each item on each day to obtain the data set F.

[0097] The data sample is as follows:

[0098] Item CTR result set F:

[0099] item_id day_gap click_num expose_nums i1 1 180 3600 i1 2 500 8000 i2 1 300 15000 …… …… …… ……

[0100] Step S8: Calculate the CTR for each item in the set F to obtain the item CTR result set G.

[0101] The CTR calculation formula for item i is:

[0102] ctr_i=sum(decay_rate^day_gap*click_num) / (sum(decay_rate^day_gap*expose_num)+1);

[0103] Among them, decay_rate is the time decay rate, which controls the speed at which the popularity of content changes over time, and its value range is (0.5, 1); day_gap is the number of days between the date when the user ID clicks / exposes the item and the current date; click_num is the number of clicks per day; and expose_num is the number of exposures per day.

[0104] The data sample of item CTR result set G is as follows (the decay_rate is set to 0.9 in this example):

[0105] item_id mod_click_nums mod_expose_nums CTR i1 567 9721 0.0583 i2 270 13501 0.0200 …… …… …… ……

[0106] Data calculation instructions:

[0107] Taking i1 as an example, mod_click_nums=sum(decay_rate^day_gap*click_num), that is, mod_click_nums=0.9^1*180+0.9^2*500=567;

[0108] mod_expose_nums=sum(decay_rate^day_gap*expose_num)+1, that is, mod_expose_nums=0.9^1*3600+0.9^2*8000+1=9721;

[0109] Then the CTR of i1 = 567 / 9721≈0.0583.

[0110] Step S9: Filter out items whose exposure counts are less than the preset exposure threshold from the item CTR result set G, and then sort the CTR values ​​in descending order to obtain the final popular results, i.e., the CTR popular content improved based on the time decay factor and the click position factor.

[0111] It should be noted that the above application scenario is only an example of an embodiment of the present invention. The embodiment of the present invention is not limited to the above application scenario, but can be applied to any application scenario to which the embodiment of the present invention is applicable.

[0112] On the other hand, the present application also discloses a device for calculating popular content based on a combination of time decay factors and click position factors. Referring to FIG. 2 , the device for calculating popular content based on a combination of time decay factors and click position factors includes:

[0113] The information acquisition module 101 is used to obtain the user's click exposure behavior data in the past N days, wherein the click exposure behavior data includes the date when the user clicked / exposed the item;

[0114] New information field calculation module 102, used to calculate the number of days between the date when the user clicked / exposed the item and the current date based on the click exposure behavior data;

[0115] A click set A generating module 103 is configured to obtain a click set A from the click exposure behavior data, wherein the click set A includes a user ID, an item ID, an ID of a batch of recommended content obtained by the user ID, a record of the location where the user ID clicked / exposed the content, and the number of days between the date when the user ID clicked / exposed the content and the current date. A batch of recommended content includes at least one item ID.

[0116] Exposure set B generation module 104 is configured to obtain exposure set B from the click exposure behavior data, wherein exposure set B includes a user ID, an item ID, an ID of a batch of recommended content obtained by the user ID, a record of the location where the user ID clicked / exposed the content, and the number of days between the date when the user ID clicked / exposed the content and the current date. A batch of recommended content IDs includes at least one item ID.

[0117] The set C generating module 105 is configured to internally associate the exposure set B with the click set A using the user ID and the batch ID of the recommended content to generate set C. Set C includes the exposure data in the batch where clicks occurred.

[0118] An exposure set D generating module 106 is configured to filter exposure data near a click location where a click behavior occurs from set C according to a preset first filtering strategy to obtain exposure set D;

[0119] Set E generation module 107 is used to modify the field names of the filtered exposure set D to make them consistent with the click set A, merge click set A with the filtered exposure set D, and remove the batch recommendation content number field obtained by the user number and the location field where the user number has clicked / exposed behavior, to obtain set E;

[0120] The set F generation module 108 is configured to aggregate the set E based on preset fields to calculate the number of clicks and impressions for each item on a daily basis, thereby generating set F. The preset fields include the item number, the number of days between the date when the user's click / exposure behavior occurred and the current date.

[0121] The click rate calculation module 109 calculates the CTR for each item in the set F based on the time decay rate to obtain the item CTR result set G;

[0122] The CTR popular content selection module 110 is used to first filter out items with exposure counts less than a preset exposure threshold from the item CTR result set G, and then sort the items in descending order of CTR values ​​to obtain the final popular results, i.e., the CTR popular content improved by the time decay factor and the click position factor.

[0123] In the above-mentioned device, the user's click-exposure behavior data for the past N days includes the user number user_id, the item number item_id, the batch recommended content number trace_id, the position of the user's click / exposure behavior position, the label label, label=1 indicates click, label=0 indicates exposure, and the date bhv_date when the user clicked / exposed the item; wherein, a batch recommended content number trace_id includes at least one item number item_id.

[0124] In the above device, the set C includes the user number user_id, the number of the exposed item expose_item_id, the number of the batch recommended content obtained by the user number trace_id, the exposure position expose_position, the number of the clicked item click_item_id, the click position click_position, the label label, label = 1 means click, label = 0 means exposure, and the number of days day_gap between the date when the user number clicked / exposed the item and the current date.

[0125] In addition, the examples of the present application can be realized by a data processing program executed by a data processing device such as a computer. Obviously, the data processing program constitutes the present application. In addition, the data processing program usually stored in a storage medium is executed by directly reading the program out of the storage medium or by installing or copying the program into a storage device (such as a hard disk and or memory) of the data processing device. Therefore, such a storage medium also constitutes the present invention. The storage medium can use any type of recording method, such as paper storage media (such as paper tape, etc.), magnetic storage media (such as floppy disk, hard disk, flash memory, etc.), optical storage media (such as CD-ROM, etc.), magneto-optical storage media (such as MO, etc.), etc.

[0126] Therefore, the present application also discloses a non-volatile storage medium, in which a data processing program is stored, and the data processing program is used to execute any instance of the popular content calculation method that integrates time decay factors and click position factors of the present application.

[0127] In summary, the present application discloses a method and device for calculating popular content that integrates time decay factors and click position factors. The method introduces the user attention factor and adds the exposure data near the click position that the user actually notices to the CTR calculation. At the same time, the time decay factor of the content popularity and the changing pattern of the product content popularity are taken into account, so as to more accurately calculate the click-through rate (CTR) of the content, and then obtain the improved CTR popular content that integrates the time decay factor and the click position factor, which can effectively realize the promotion and display of product content.

[0128] While the specific embodiments of the present invention have been described in detail above, these are merely exemplary and the present invention is not limited thereto. For those skilled in the art, any equivalent modifications and substitutions to the present invention are also within the scope of the present invention. Therefore, any equivalent changes and modifications made without departing from the spirit and scope of the present invention are intended to be encompassed within the scope of the present invention.

Claims

1. A method for calculating popular content that integrates time decay factors and click position factors, characterized in that: include: Obtain the user's click and exposure behavior data for the past N days, where the click and exposure behavior data includes the date when the user clicked or exposed the item; Based on the click / exposure behavior data, calculate the number of days between the date when the user clicked / exposure behavior on the item and the current date; Obtain click set A and exposure set B from the click-exposure behavior data, where each of click set A and exposure set B includes the user ID, item ID, the ID of the batch of recommended content obtained by the user ID, the location record of the user ID's click / exposure behavior, and the number of days between the date when the user ID clicked / exposure behavior on the item and the current date. A batch of recommended content IDs includes at least one item ID. Using the user ID and the batch ID of the recommended content, we internally associate the exposure set B with the click set A to obtain set C. Set C includes the exposure data in the batch where the click occurred. From set C, the exposure data near the click location where the click behavior occurred is filtered according to the preset first filtering strategy to obtain exposure set D; Modify the field names of the filtered exposure set D to make them consistent with the click set A. Merge click set A with the filtered exposure set D, and remove the batch recommendation number field obtained by the user number and the location field where the user number clicked / exposed. This yields set E. Aggregate set E based on preset fields to calculate the number of clicks and impressions for each item on each day, resulting in set F. The preset fields include the item ID, the number of days between the date the user ID clicked or had an impression of the item and the current date. Calculate the CTR for each item in set F based on the time decay rate, and obtain the item CTR result set G; The item CTR result set G is first filtered out of items whose exposure count is less than the preset exposure threshold, and then the CTR values ​​are sorted in descending order to obtain the final popular results, that is, the CTR popular content improved by the time decay factor and the click position factor.

2. The method for calculating popular content based on the time decay factor and click position factor according to claim 1, characterized in that: The method obtains the user's click and exposure behavior data for the past N days, including the user number user_id, the item number item_id, the batch recommended content number trace_id, the position of the user's click / exposure behavior position, the label label, label = 1 indicates click, label = 0 indicates exposure, and the date bhv_date when the user clicked / exposed the item; wherein, a batch recommended content number trace_id includes at least one item number item_id.

3. The method for calculating popular content based on the time decay factor and click position factor according to claim 2, characterized in that: Filter the user's click exposure behavior data for the past N days using the condition label = 1 to obtain click set A; The user's click exposure behavior data for the past N days is filtered using the condition label = 0 to obtain exposure set B.

4. The method for calculating popular content based on the time decay factor and click position factor according to claim 1, characterized in that: After the exposure set B is internally associated with the click set A to obtain the set C, if the user only has exposure behavior in a batch of recommended items, then the exposure data of this entire batch will not be included in the subsequent item CTR calculation.

5. The method for calculating popular content based on the time decay factor and click position factor according to claim 1, characterized in that: The resulting set C includes the user ID user_id, the exposed item ID expose_item_id, the batch recommendation content ID trace_id obtained by the user ID, the exposure position expose_position, the clicked item ID click_item_id, the click position click_position, the label label, where label = 1 indicates click, and label = 0 indicates exposure, and the number of days day_gap between the date when the user ID clicked / exposed the item and the current date.

6. The method for calculating popular content based on the time decay factor and click position factor according to claim 1, characterized in that: The preset first screening strategy includes: allowing K1 exposure data closest to below the click position and K2 exposure data closest to above the click position to participate in the calculation, wherein K1 and K2 are both integers greater than or equal to 1.

7. The method for calculating popular content based on the time decay factor and click position factor according to claim 1, characterized in that: The CTR calculation for each item in the set F based on the time decay rate includes: the CTR calculation formula for item i is: ctr_i=sum(decay_rate^day_gap*click_num) / (sum(decay_rate^day_gap*expose_num)+1); Among them, decay_rate is the time decay rate, which controls the speed at which the popularity of content changes over time, and its value range is (0.5, 1); day_gap is the number of days between the date when the user ID clicks / exposes the item and the current date; click_num is the number of clicks per day; and expose_num is the number of exposures per day.

8. A device for calculating popular content by integrating time decay factors and click position factors, characterized in that: include: An information acquisition module is used to obtain the user's click and exposure behavior data for the past N days, where the click and exposure behavior data includes the date when the user clicked or exposed the item; New information field calculation module, used to calculate the number of days between the date when a user clicked / exposed an item and the current date based on click / exposure behavior data; A click set A generation module is configured to obtain click set A from the click exposure behavior data, wherein click set A includes a user ID, an item ID, an ID of a batch of recommended content obtained by the user ID, a record of the location where the user ID clicked / exposed the content, and the number of days between the date when the user ID clicked / exposed the content and the current date. A batch of recommended content IDs includes at least one item ID. An exposure set B generation module is configured to obtain exposure set B from the click exposure behavior data, wherein exposure set B includes a user ID, an item ID, the ID of the batch of recommended content obtained by the user ID, a record of the location where the user ID's click / exposure behavior occurred, and the number of days between the date when the user ID's click / exposure behavior occurred on the item and the current date. A batch of recommended content IDs includes at least one item ID. A set C generation module is used to internally associate exposure set B with click set A using the user ID and the batch ID of the recommended content to generate set C. Set C includes exposure data in batches where clicks occurred. An exposure set D generation module is configured to filter exposure data near a click location where a click behavior occurs from set C according to a preset first filtering strategy to obtain exposure set D; The module for generating set E is used to modify the field names of the filtered exposure set D to make them consistent with the click set A, merge click set A with the filtered exposure set D, and remove the number field of the batch recommendation content obtained by the user number and the location field where the user number has clicked / exposed behavior, to obtain set E; The set F generation module is used to aggregate the set E according to preset fields to calculate the number of clicks and impressions of each item on a daily basis to obtain the set F. The preset fields include the item number, the number of days between the date when the user number clicked or was exposed to the item and the current date; The click-through rate calculation module calculates the CTR of each item in the set F based on the time decay rate and obtains the item CTR result set G; The CTR popular content selection module is used to first filter out items with exposure counts less than a preset exposure threshold from the item CTR result set G, and then sort the items in descending order of CTR values ​​to obtain the final popular results, namely the CTR popular content improved by time decay factors and click position factors.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method for calculating popular content based on the comprehensive time decay factor and click position factor as described in any one of claims 1 to 7 is implemented.

10. An electronic device, characterized in that: include: processor; as well as a memory for storing executable instructions of the processor; The processor is configured to execute the method for calculating hot content based on the comprehensive time decay factor and click position factor as described in any one of claims 1 to 7 by executing the executable instructions.

Images