Multimedia resource providing method, device, electronic device and storage medium
By dynamically adjusting the number of multimedia resource candidate queues based on the traffic distribution at the received content request moment, the problem of waste of computing resources during the trough period is solved, and efficient resource utilization and better advertising recommendation results are achieved.
Patent Information
- Application Number
- CN202210036543.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-01-12
- Publication Date
- 2024-12-06
- Estimated Expiration
- 2042-01-12
AI Technical Summary
When the number of requests for processing is at a low point, the server will cause a certain amount of computing resources to be wasted.
By determining the estimated number of traffic requests at the time interval at which the target time of receiving the content request sent by the client is located, the number of multimedia resource candidate queues is dynamically adjusted, and filtering and sorting processing is performed based on this to provide multimedia resources to the client.
Avoid the oversupply of computing resources during non-traffic peak periods, make full use of computing resources, and provide clients with more matching multimedia resources.
Smart Images

Figure CN114418208B_ABST
Abstract
Description
Technical Field
[0001] The present disclosure relates to the field of computer technology, and more specifically, to a method, device, electronic device and storage medium for providing multimedia resources. Background Art
[0002] At present, online traffic at every time of the day and every day has a certain distribution pattern, and is not evenly distributed. For example, at 5 or 6 p.m., most users have already gone off work, and those who have already gone off work have free time to shop online, watch videos, or browse the web, etc., so the time period between 5 and 6 p.m. is a peak traffic period; and at around 2 or 3 a.m., most users have already gone to bed, so the time period between 2 or 3 a.m. is a low traffic period.
[0003] In related technologies, in order to ensure the availability of services, the server generally uses the number of machines actually deployed online as the number of machines required during the peak traffic period to process corresponding requests. However, since the number of machines actually deployed online is fixed, when the number of requests processed is at a low point, a certain amount of computing resources will be wasted. Summary of the invention
[0004] The present disclosure provides a multimedia resource providing method, device, electronic device and storage medium to at least solve the technical problem in the above-mentioned related technologies that when the processing request volume is at a low point, a certain amount of computing resources will be wasted.
[0005] According to a first aspect of an embodiment of the present disclosure, a method for providing multimedia resources is provided, comprising: determining an estimated number of traffic requests for a time interval at a target moment at which a content request sent by a client is received, wherein the estimated number of traffic requests is the total number of content requests received by a server within the time interval; determining a first number of multimedia resource candidate queues based on the estimated number of traffic requests and a first preset number corresponding to multimedia resource candidate queues, wherein the first preset number is the number corresponding to the multimedia resource candidate queues determined by the server according to a maximum number of traffic requests; screening out the first number of candidate multimedia resources, and sorting the screened candidate multimedia resources; and providing multimedia resources to the client according to the sorting result.
[0006] Optionally, determining the first number corresponding to the multimedia resource candidate queue based on the estimated number of traffic requests and the first preset number corresponding to the multimedia resource candidate queue includes: determining the ratio of the maximum number of traffic requests to the estimated number of traffic requests in the time interval of the target moment of receiving the content request sent by the client; determining the first number based on the ratio and the first preset number corresponding to the multimedia resource candidate queue determined by the server according to the maximum number of traffic requests.
[0007] Optionally, the sorting process of the screened candidate multimedia resources includes: performing a first granularity sorting process on the screened candidate multimedia resources to obtain candidate multimedia resources that have been sorted by the first granularity; determining a second number of the candidate multimedia resources that need to be sorted by a second granularity based on the target time; selecting the second number of candidate multimedia resources from the candidate multimedia resources that have been sorted by the first granularity; and performing a second granularity sorting process on the second number of candidate multimedia resources to obtain the sorting result.
[0008] Optionally, determining the second number of candidate multimedia resources that require second granularity sorting includes: determining a ratio of the maximum number of traffic requests to the estimated number of traffic requests in the time interval of the target moment of receiving the content request sent by the client; determining the second number based on the ratio and a second preset number; wherein the second preset number is: the number corresponding to the candidate queues of multimedia resources that require second granularity sorting determined by the server according to the maximum number of traffic requests.
[0009] Optionally, the first number is negatively correlated with the estimated number of traffic requests, and the second number is negatively correlated with the estimated number of traffic requests.
[0010] Optionally, the first granularity sorting process is performed on the screened candidate multimedia resources to obtain the candidate multimedia resources that have been sorted with the first granularity, including: obtaining first recommended evaluation index information for each candidate multimedia resource in the screened candidate multimedia resources; based on the first recommended evaluation index information, the first granularity sorting process is performed on the screened candidate multimedia resources to obtain the candidate multimedia resources that have been sorted with the first granularity.
[0011] Optionally, performing a second granularity sorting process on the second number of candidate multimedia resources to obtain the sorting result includes: obtaining second recommended evaluation index information for each candidate multimedia resource in the second number of candidate multimedia resources; and performing a second granularity sorting process on the second number of candidate multimedia resources based on the second recommended evaluation index information to obtain the sorting result.
[0012] Optionally, the estimated number of traffic requests is obtained by inputting features related to the time interval into a traffic request prediction model.
[0013] According to a second aspect of an embodiment of the present disclosure, a multimedia resource providing device is provided, comprising: a determination module, configured to determine an estimated number of traffic requests for a time interval at a target moment of receiving a content request sent by a client, wherein the estimated number of traffic requests is the total number of content requests received by a server within the time interval; the determination module, configured to determine a first number of multimedia resource candidate queues based on the estimated number of traffic requests and a first preset number of multimedia resource candidate queues, wherein the first preset number is the number of multimedia resource candidate queues determined by the server according to a maximum number of traffic requests; a sorting module, configured to screen out the first number of candidate multimedia resources and sort the screened candidate multimedia resources; a providing module, configured to provide multimedia resources to the client according to the sorting result.
[0014] Optionally, the determination module is configured to: determine the ratio of the maximum number of traffic requests to the estimated number of traffic requests in the time interval at the target moment of receiving the content request sent by the client; determine the first number based on the ratio and the first preset number corresponding to the multimedia resource candidate queue determined by the server according to the maximum number of traffic requests.
[0015] Optionally, the sorting module is configured to: perform a first granularity sorting process on the screened candidate multimedia resources to obtain candidate multimedia resources that have been sorted with the first granularity; determine a second number of candidate multimedia resources that need to be sorted with a second granularity based on the target time; select the second number of candidate multimedia resources from the candidate multimedia resources that have been sorted with the first granularity; perform a second granularity sorting process on the second number of candidate multimedia resources to obtain the sorting result.
[0016] Optionally, the sorting module is configured to: determine the ratio of the maximum number of traffic requests to the estimated number of traffic requests in the time interval of the target moment of receiving the content request sent by the client; determine the second number based on the ratio and a second preset number; wherein the second preset number is: the number corresponding to the multimedia resource candidate queues that need to be sorted with the second granularity determined by the server according to the maximum number of traffic requests.
[0017] Optionally, the first number is negatively correlated with the estimated number of traffic requests, and the second number is negatively correlated with the estimated number of traffic requests.
[0018] Optionally, the sorting module is configured to: obtain first recommendation evaluation index information of each candidate multimedia resource among the screened candidate multimedia resources; based on the first recommendation evaluation index information, perform first granularity sorting processing on the screened candidate multimedia resources to obtain candidate multimedia resources that have undergone first granularity sorting processing.
[0019] Optionally, the sorting module is configured to: obtain second recommendation evaluation index information for each candidate multimedia resource in the second number of candidate multimedia resources; and perform second granularity sorting processing on the second number of candidate multimedia resources based on the second recommendation evaluation index information to obtain the sorting result.
[0020] Optionally, the estimated number of traffic requests is obtained by inputting features related to the time interval into a traffic request prediction model.
[0021] According to a third aspect of an embodiment of the present disclosure, there is provided an electronic device, comprising: a processor; and a memory for storing instructions executable by the processor; wherein the processor is configured to execute the instructions to implement the multimedia resource providing method according to the present disclosure.
[0022] According to a fourth aspect of an embodiment of the present disclosure, a computer-readable storage medium is provided. When instructions in the computer-readable storage medium are executed by a processor of an electronic device, the electronic device is enabled to execute the multimedia resource providing method according to the present disclosure.
[0023] According to a fifth aspect of an embodiment of the present disclosure, a computer program product is provided, including a computer program, and when the computer program is executed by a processor, the multimedia resource providing method according to the present disclosure is implemented.
[0024] The technical solution provided by the embodiments of the present disclosure brings at least the following beneficial effects: since the online traffic distribution at different times is different, the present disclosure can dynamically determine the corresponding number of multimedia resource candidate queues according to the traffic distribution corresponding to the time when the client's content request is received, and perform subsequent screening and sorting operations based on this. On the one hand, it avoids the situation of excess computing resources during non-traffic peak periods and can make full use of computing resources; on the other hand, it can provide the client with more matching multimedia resources.
[0025] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present disclosure. BRIEF DESCRIPTION OF THE DRAWINGS
[0026] The drawings herein are incorporated into and constitute a part of the specification, illustrate embodiments consistent with the present disclosure, and together with the description are used to explain the principles of the present disclosure, and do not constitute improper limitations on the present disclosure.
[0027] Figure 1 is a flowchart illustrating a multimedia resource providing method according to an exemplary embodiment of the present disclosure;
[0028] Figure 2 is a block diagram showing a multimedia resource providing apparatus according to an exemplary embodiment of the present disclosure;
[0029] Figure 3 is a block diagram illustrating an electronic device according to an exemplary embodiment of the present disclosure. DETAILED DESCRIPTION
[0030] In order to enable ordinary persons in the art to better understand the technical solutions of the present disclosure, the technical solutions in the embodiments of the present disclosure will be clearly and completely described below in conjunction with the accompanying drawings.
[0031] It should be noted that the terms "first", "second", etc. in the specification and claims of the present disclosure and the above-mentioned 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 data used in this way can be interchanged where appropriate, so that the embodiments of the present disclosure described herein can be implemented in an order other than those illustrated or described herein. The implementation methods described in the following examples do not represent all implementation methods consistent with the present disclosure. Instead, they are merely examples of devices and methods consistent with some aspects of the present disclosure as detailed in the attached claims.
[0032] It should be noted that the phrase "at least one of the items" in the present disclosure includes three types of parallel situations: "any one of the items", "a combination of any number of the items", and "all of the items". For example, "including at least one of A and B" includes the following three parallel situations: (1) including A; (2) including B; (3) including A and B. Another example is "executing at least one of step 1 and step 2" which means the following three parallel situations: (1) executing step 1; (2) executing step 2; (3) executing step 1 and step 2.
[0033] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data for display, data for analysis, etc.) involved in this disclosure are all information and data authorized by the user or fully authorized by all parties.
[0034] Figure 1 is a flowchart illustrating a multimedia resource providing method according to an exemplary embodiment of the present disclosure.
[0035] Reference Figure 1In step 101, the estimated number of traffic requests in the time interval at which the target time of receiving the content request sent by the client is located can be determined. The estimated number of traffic requests can be the total number of content requests received by the server in the time interval.
[0036] It should be noted that the content request sent by the client may be a request for a natural work, and the natural work may be a video; the candidate multimedia resource may be an advertisement. For example, when a user uses a video application on the client to watch a video, the client may send a content request to the server to update the video every time the user finishes watching a fixed number of videos, for example, every time the user finishes watching 5 videos; or the client may also send a content request to the server to update the video every preset time. The server may return a natural work to the client based on the received content request, for example, the latest recommended video, and may also return several advertisements to the client at the same time.
[0037] When the server is screening advertisements to be pushed to the client, it can generally be divided into a targeting stage, a recall stage and a sorting stage, wherein the sorting stage can also include a rough sorting stage and a fine sorting stage.
[0038] It should be noted that before receiving the content request sent by the client, offline data mining can be performed for different users. For example, it is possible to analyze which products the user has recently purchased, which videos he has watched, and which web pages he has browsed, and then determine the type of people the user belongs to, whether he has a car, whether he has a house, whether he has children, and the region he belongs to, etc. Among them, the type of people can be people who love sports, people who love watching movies, people who like shopping, etc. For example, if a user has recently watched parenting videos and often bought maternal and child products, it can be determined that the user has children. In addition, when advertisers place advertisements, they will circle the user groups targeted by the advertisements they place. For example, the advertisements placed by advertiser A are targeted at people who love sports; the advertisements placed by advertiser B are targeted at people with children; the advertisements placed by advertiser C are targeted at people with cars; and the advertisements placed by advertiser D are targeted at people who love watching movies. The content request sent by the client to the server may include a user ID. The server can then find the user's features previously mined offline based on the user ID, such as the type of person the user belongs to, whether the user has a car, a house, children, and the region to which the user belongs.
[0039] After the server receives the content request from the client, it needs to go through the targeting stage to filter the advertisements. For example, the server can determine that the user who sent the content request is A based on the user ID in the received content request, and the server can determine that A loves sports, has no children, and has a car based on the features of A previously mined offline. At this time, the advertisements that can be recommended to A are the advertisements placed by advertiser A and the advertisements placed by advertiser C. That is, the targeting stage uses some filtering conditions to hard filter a large number of, for example, hundreds of thousands of advertisements, and then screen out a part of the advertisements, such as tens of thousands of advertisements, for further screening in the subsequent recall stage.
[0040] In the recall phase, we can obtain user behavior information such as what products the user who sent the content request has recently bought and what videos he has watched. Then, we can input the obtained user behavior information and tens of thousands of advertisements screened out in the targeting phase into the neural network, and the neural network will screen out thousands of advertisements that are more matched with the user behavior information. Then, the thousands of advertisements screened out by the neural network, that is, a certain number of advertisements screened out in the recall phase, can be supplied to the subsequent rough sorting phase for further screening. It should be noted that the recall phase is a relatively coarse-grained screening phase, the purpose of which is to exclude most of the advertisements that are irrelevant to the user.
[0041] According to an exemplary embodiment of the present disclosure, the first recommendation evaluation index information of each candidate multimedia resource in the screened candidate multimedia resources can be obtained, and then the screened candidate multimedia resources can be sorted at a first granularity based on the first recommendation evaluation index information to obtain the candidate multimedia resources that have been sorted at the first granularity. The first granularity sorting process is a coarse sorting process.
[0042] In the rough ranking stage, for each of the thousands of advertisements provided in the recall stage, the click-through rate (CTR) model can be used to predict the click-through rate of each advertisement; the conversion rate (CVR) model can be used to predict the conversion rate of each advertisement. The click-through rate (CTR) refers to the percentage of the number of clicks on an advertisement to the number of times the advertisement is displayed; the conversion rate (CVR) refers to the percentage of the number of direct revenue conversions of an advertisement to the number of clicks on the advertisement, that is, the conversion rate from a user clicking on an advertisement to becoming an effectively activated or registered or even a paying user. Next, the product of the click-through rate of each advertisement, the conversion rate of each advertisement, and the bid of the advertiser of the advertisement can be calculated, and the product is the first recommendation evaluation indicator information of the advertisement. Moreover, the larger the above product corresponding to the advertisement, the more revenue the advertisement brings to the advertising publishing platform, and the more inclined to recommend the advertisement. The bid of the above advertiser can be the price of one conversion of the advertisement. In this way, the rough ranking stage can sort the advertisements according to the estimated value of the CTR model and the estimated value of the CVR model, according to the revenue brought by the advertisement to the advertising publishing platform, and then select the first certain number of advertisements in the rough ranking results to provide to the subsequent fine ranking stage for further fine ranking.
[0043] According to an exemplary embodiment of the present disclosure, second recommendation evaluation index information of each candidate multimedia resource in the second number of candidate multimedia resources can also be obtained, and then based on the second recommendation evaluation index information, a second granularity sorting process can be performed on the second number of candidate multimedia resources to obtain a sorting result. The second granularity sorting process is a refined sorting process.
[0044] In the fine ranking stage, for the advertisements selected in the rough ranking stage, the click-through rate model with higher prediction accuracy than that in the rough ranking stage can be used to predict the click-through rate of each advertisement, and the conversion rate model with higher prediction accuracy than that in the rough ranking stage can be used to predict the conversion rate of each advertisement. Next, the product of the click-through rate of each advertisement, the conversion rate of each advertisement and the bid of the advertiser of the advertisement can be calculated, and the product is the second recommendation evaluation index information of the advertisement. Moreover, the larger the above product corresponding to the advertisement, the more revenue the advertisement brings to the advertising publishing platform, and the more inclined to recommend the advertisement. Compared with the rough ranking stage, the CTR model and the CVR model in the fine ranking stage have higher accuracy, and the predicted click-through rate and conversion rate are more accurate. In this way, the fine ranking stage can sort the advertisements according to the estimated value of the CTR model and the estimated value of the CVR model, according to the revenue brought by the advertisement to the advertising publishing platform, and then select the advertisements that are finally delivered and can bring greater revenue to the advertising publishing platform.
[0045] It can be seen that the more advertisements that need to be screened out in the recall stage, the more advertisements that need to be calculated in the subsequent rough ranking stage; the more advertisements that need to be screened out in the rough ranking stage, the more advertisements that need to be calculated in the subsequent fine ranking stage.
[0046] Each advertisement screening stage can correspond to its own service cluster. The number of machines contained in each service cluster is generally fixed. Therefore, the computing power of the service cluster corresponding to each stage is also fixed. The actual computing power requirement for each advertisement screening stage can be measured by multiplying the number of content requests received by the server at the current moment by the queue length of the candidate advertisements in that stage. Since the computing power of each stage is fixed, in the related art, for each sorting stage, the queue length of the candidate advertisements to be sorted in the sorting stage is usually determined uniformly according to the maximum number of content requests that the server can bear (i.e., the peak traffic situation) to ensure that the actual computing power of the sorting stage will not exceed the bearing capacity. However, this often results in that when the number of content requests received does not reach the maximum number of content requests (i.e., the non-peak traffic situation), the sorting calculation is still performed according to the fixed number of candidate advertisements set for the peak traffic, which does not fully utilize the computing power and wastes computing resources. Therefore, the present disclosure takes into account that the queue length of candidate ads at each sorting stage may be inversely proportional to the number of content requests received by the server at the current moment, that is, the queue length of candidate ads at each sorting stage may change with changes in traffic distribution, that is, the supply quantity of candidate ads at each sorting stage may change with changes in traffic distribution. In this way, when the number of content requests received by the server is small, more candidate ads can be sorted and calculated for each content request. On the one hand, computing resources can be fully utilized. On the other hand, the more candidate ads there are, the more room for sorting options there are, which can also improve the advertising recommendation effect.
[0047] As mentioned above, the server can determine the estimated number of traffic requests for the time interval at the target time when the content request sent by the client is received. For example, the time interval can be set according to actual needs and actual conditions. For example, the time interval can be every hour or every 10 minutes, and the present disclosure does not limit this. For example, assuming that the time when the server receives the content request from the client is 12:05, the number of traffic requests for the time interval from 12:00 to 12:10 can be estimated. Next, the first number can be determined based on the estimated number of traffic requests. The first number can be the queue length of the candidate advertisements in the rough sorting stage, that is, the supply number of candidate advertisements in the rough sorting stage; it can also be the queue length of the candidate advertisements in the fine sorting stage, that is, the supply number of candidate advertisements in the fine sorting stage. Among them, the first number is negatively correlated with the estimated number of traffic requests. For example, the queue length of the candidate advertisements in the rough sorting stage is inversely proportional to the estimated number of traffic requests in the time interval where the current moment is located; or, the queue length of the candidate advertisements in the fine sorting stage is inversely proportional to the estimated number of traffic requests in the time interval where the current moment is located.
[0048] It should be noted that the number of candidate ads provided to the fine ranking stage in the rough ranking stage will directly affect the size of the candidate pool of ads in the fine ranking stage. If the recall stage and the rough ranking stage can supply more candidate ads, then the final fine ranking stage can also choose from a larger pool of candidate ads, thereby obtaining a better ad recommendation effect. Therefore, the present disclosure can dynamically adjust the queue length of candidate ads in the rough ranking stage from the recall stage, and the queue length of candidate ads in the fine ranking stage from the rough ranking stage, according to the traffic distribution corresponding to the time interval at the time when the content request from the client is received, so as to make full use of computing resources during non-peak traffic periods, and obtain high-quality ads that are more matched with users and can bring more revenue to the advertising publishing platform by screening from more candidate ads.
[0049] In step 102, a first number of multimedia resource candidate queues may be determined based on the estimated number of traffic requests and a first preset number of multimedia resource candidate queues, wherein the first preset number is the number of multimedia resource candidate queues determined by the server according to the maximum number of traffic requests.
[0050] According to an exemplary embodiment of the present disclosure, the ratio of the maximum number of traffic requests to the estimated number of traffic requests in the time interval at the target time of receiving the content request sent by the client can be determined. Then, the first number can be determined based on the ratio and the first preset number corresponding to the multimedia resource candidate queue determined by the server according to the maximum number of traffic requests. For example, the first preset number can be the number of candidate advertisements supplied in the rough sorting stage when the server receives the content request with the maximum number of traffic requests.
[0051] According to an exemplary embodiment of the present disclosure, as mentioned above, the first number may be negatively correlated with the estimated number of traffic requests.
[0052] Furthermore, the following formula may be used to calculate the first number corresponding to the multimedia resource candidate queue, that is, the queue length of candidate advertisements in the rough sorting stage at the time interval of the target time at which the content request of the client is received:
[0053]
[0054] Among them, L 粗排 is the first number corresponding to the multimedia resource candidate queue, that is, the queue length of candidate advertisements in the rough sorting stage at the target time of receiving the content request from the client, L 原始粗排 is the first preset number, that is, the number of candidate advertisements provided in the rough sorting stage when the server receives the content request with the maximum traffic request number, T max is the maximum number of traffic requests that the server can carry in each time interval, T predict is the estimated number of traffic requests in the time interval of the target time when the content request from the client is received, R1 is the weight ratio, and the value range of R1 is 0 to 1. In fact, in the related art, regardless of the actual situation, the queue length of candidate advertisements in the rough sorting stage is always fixed to L 原始粗排 .
[0055] For example, assuming that the value of the trade-off ratio R is 1, during the peak traffic period, the number of traffic requests may reach a peak of 1,000 in a time interval. At this time, the queue length of candidate advertisements in the rough sorting stage may be 2,000. After receiving the content request from the client, if the estimated number of traffic requests in the time interval at the target moment of receiving the content request is only half of the number of traffic requests during the peak traffic period, that is, 500, then the queue length of candidate advertisements in the rough sorting stage can be increased from 2,000 during the peak traffic period to 4,000 during the trough traffic period. Therefore, the present disclosure can dynamically adjust the queue length of candidate advertisements in the rough sorting stage in the recall stage according to the traffic distribution corresponding to the time interval at the target moment of receiving the content request from the client, so that the rough sorting stage can screen more candidate advertisements, that is, there is more room for selection, and it is possible to screen out high-quality advertisements that are more matched with users and can bring more revenue to the advertising publishing platform.
[0056] According to an exemplary embodiment of the present disclosure, features related to the time interval at which the target moment of receiving the content request from the client is located can be input into a traffic request prediction model to obtain an estimated number of traffic requests. Among them, the features related to the time interval at which the target moment of receiving the content request from the client is located can be whether it is a weekday and the specific moment of receiving the content request from the client, etc. It should be noted that the traffic request prediction model can estimate the number of traffic requests for each time period of the day before the start of the day, and can also estimate the number of traffic requests for the time interval at which the target moment of receiving the content request is located in real time after receiving the content request from the client.
[0057] In step 103, a first number of candidate multimedia resources may be screened out, and the screened candidate multimedia resources may be sorted.
[0058] According to an exemplary embodiment of the present disclosure, the selected candidate multimedia resources may be sorted at a first granularity to obtain the candidate multimedia resources that have been sorted at the first granularity. Then, according to the target time, a second number of candidate multimedia resources that need to be sorted at a second granularity may be determined. Next, a second number of candidate multimedia resources may be selected from the candidate multimedia resources that have been sorted at the first granularity. Then, a second number of candidate multimedia resources may be sorted at a second granularity to obtain a sorting result.
[0059] As mentioned above, the sorting stage may include a rough sorting stage and a fine sorting stage. In the rough sorting stage, the first number of advertisements screened out in the recall stage may be received, and the first number of advertisements screened out in the recall stage may be sorted at a first granularity. As mentioned above, the first number of advertisements may be sorted at a first granularity according to the revenues brought to the advertisement publishing platform by the advertisements based on the estimated values of the CTR model and the estimated values of the CVR model.
[0060] Next, the second number of candidate multimedia resources that need to be sorted at the second granularity can be determined according to the target time when the content request is received, that is, the queue length of candidate advertisements that need to be input into the fine sorting stage can be determined. Then, the first second number of candidate multimedia resources in the sorting queue obtained after the first granularity sorting process can be sorted to obtain a sorting result, that is, the first second number of candidate advertisements in the sorting queue obtained after the sorting process in the rough sorting stage can be sorted to obtain a final sorting result.
[0061] According to an exemplary embodiment of the present disclosure, the ratio of the maximum number of traffic requests to the estimated number of traffic requests in the time interval of the target moment of receiving the content request sent by the client can be determined. Then, based on the ratio and the second preset number, the second number can be determined, that is, the queue length of the candidate advertisements in the fine ranking stage can be determined. Among them, the second preset number is: the corresponding number of the candidate queues of multimedia resources that need to be sorted with the second granularity determined by the server according to the maximum number of traffic requests. That is, the second preset number can be the supply number of candidate advertisements in the fine ranking stage when the server receives the content request with the maximum number of traffic requests.
[0062] According to an exemplary embodiment of the present disclosure, the second number may be negatively correlated with the estimated number of traffic requests.
[0063] Furthermore, the second number of candidate advertisements that need to be sorted at the second granularity may be calculated using the following formula, that is, the queue length of candidate advertisements in the fine sorting phase at the time interval of the target time when the content request of the client is received:
[0064]
[0065] Among them, L 精排 is the second number of candidate advertisements that need to be sorted at the second granularity, that is, the queue length of candidate advertisements in the fine sorting stage at the time interval of the target time when the content request of the client is received, L 原始精排 is the second preset number, that is, the number of candidate advertisements provided in the fine ranking stage when the server receives the content request with the maximum traffic request number, T max is the maximum number of traffic requests that the server can carry in each time interval, T predict is the estimated number of traffic requests in the time interval of the target time when the content request from the client is received, R2 is the weighted ratio, and the value range of R2 is 0 to 1. In fact, in the related art, regardless of the actual situation, the queue length of candidate advertisements in the fine ranking stage is always fixed to L. 原始精排 .
[0066] For example, assuming that the value of the trade-off ratio R2 is 1, during the peak traffic period, the number of traffic requests may reach a peak of 1,000 in a time interval. At this time, the queue length of the candidate advertisements in the fine ranking stage may be 100. After receiving the content request from the client, if the estimated number of traffic requests in the time interval at the target moment of receiving the content request is only half of the number of traffic requests in the peak traffic period, that is, 500, then the queue length of the candidate advertisements in the fine ranking stage can be increased from 100 in the peak traffic period to 200 in the trough traffic period. Therefore, the present disclosure can dynamically adjust the queue length of the candidate advertisements in the fine ranking stage from the rough ranking stage according to the traffic distribution corresponding to the time interval at the target moment of receiving the content request from the client, so that the fine ranking stage can screen from more candidate advertisements, that is, there is more room for selection, and it is possible to screen out high-quality advertisements that are more matched with users and can bring more revenue to the advertising publishing platform.
[0067] In step 104, multimedia resources may be provided to the client according to the ranking result. That is, the higher the ranking, the higher the priority provided. For example, the finally selected advertisements may be provided to the client according to the ranking result.
[0068] Figure 2 is a block diagram illustrating a multimedia resource providing apparatus according to an exemplary embodiment of the present disclosure.
[0069] Reference Figure 2 The device 200 may include a determination module 201, a sorting module 202 and a providing module 203.
[0070] The determination module 201 is configured to determine the estimated number of traffic requests in the time interval at which the target time of receiving the content request sent by the client is located, wherein the estimated number of traffic requests is the total number of content requests received by the server in the time interval;
[0071] The determination module 201 is configured to determine a first number of multimedia resource candidate queues based on the estimated number of traffic requests and a first preset number of multimedia resource candidate queues, wherein the first preset number is the number of multimedia resource candidate queues determined by the server according to the maximum number of traffic requests;
[0072] The sorting module 202 is configured to screen out the first number of candidate multimedia resources and sort the screened candidate multimedia resources;
[0073] The providing module 203 is configured to provide multimedia resources to the client according to the sorting result.
[0074] According to an exemplary embodiment of the present disclosure, the determining module 201 is configured to:
[0075] Determine a ratio of the maximum number of traffic requests to an estimated number of traffic requests in a time interval at which a target time instant at which a content request sent by a client is received;
[0076] The first number is determined based on the ratio and a first preset number corresponding to the multimedia resource candidate queue determined by the server according to the maximum traffic request number.
[0077] According to an exemplary embodiment of the present disclosure, the sorting module 202 is configured to:
[0078] Performing a first granularity sorting process on the screened candidate multimedia resources to obtain candidate multimedia resources that have undergone the first granularity sorting process;
[0079] Determining, according to the target time, a second number of the candidate multimedia resources that need to be sorted at a second granularity;
[0080] Selecting the second number of candidate multimedia resources from the candidate multimedia resources that have been sorted at the first granularity;
[0081] Perform a second granularity sorting process on the second number of candidate multimedia resources to obtain the sorting result.
[0082] According to an exemplary embodiment of the present disclosure, the sorting module 202 is configured to:
[0083] Determine a ratio of the maximum number of traffic requests to an estimated number of traffic requests in a time interval at which a target time instant at which a content request sent by a client is received;
[0084] Determining the second number based on the ratio and a second preset number;
[0085] The second preset number is: the number of multimedia resource candidate queues that need to be sorted at the second granularity determined by the server according to the maximum traffic request number.
[0086] According to an exemplary embodiment of the present disclosure, the first number is negatively correlated with the estimated number of traffic requests, and the second number is negatively correlated with the estimated number of traffic requests.
[0087] According to an exemplary embodiment of the present disclosure, the sorting module 202 is configured to:
[0088] Acquire first recommendation evaluation index information of each candidate multimedia resource among the screened candidate multimedia resources;
[0089] Based on the first recommendation evaluation index information, the screened candidate multimedia resources are sorted at a first granularity to obtain candidate multimedia resources that have undergone the sorting at the first granularity.
[0090] According to an exemplary embodiment of the present disclosure, the sorting module 202 is configured to:
[0091] Acquire second recommendation evaluation index information of each candidate multimedia resource in the second number of candidate multimedia resources;
[0092] Based on the second recommendation evaluation index information, a second granularity sorting process is performed on the second number of candidate multimedia resources to obtain the sorting result.
[0093] According to an exemplary embodiment of the present disclosure, the estimated number of traffic requests is obtained by inputting features related to the time interval into a traffic request prediction model.
[0094] Figure 3 is a block diagram illustrating an electronic device 300 according to an exemplary embodiment of the present disclosure.
[0095] Reference Figure 3 The electronic device 300 includes at least one memory 301 and at least one processor 302. The at least one memory 301 stores instructions. When the instructions are executed by the at least one processor 302, the multimedia resource providing method according to the exemplary embodiment of the present disclosure is executed.
[0096] As an example, the electronic device 300 may be a PC, a tablet device, a personal digital assistant, a smart phone, or other device capable of executing the above instructions. Here, the electronic device 300 is not necessarily a single electronic device, but may also be any device or circuit capable of executing the above instructions (or instruction sets) individually or in combination. The electronic device 300 may also be part of an integrated control system or a system manager, or may be configured as a portable electronic device interconnected with a local or remote (e.g., via wireless transmission) interface.
[0097] In the electronic device 300, the processor 302 may include a central processing unit (CPU), a graphics processing unit (GPU), a programmable logic device, a dedicated processor system, a microcontroller or a microprocessor. As an example and not limitation, the processor may also include an analog processor, a digital processor, a microprocessor, a multi-core processor, a processor array, a network processor, etc.
[0098] The processor 302 may execute instructions or codes stored in the memory 301, wherein the memory 301 may also store data. Instructions and data may also be sent and received over a network via a network interface device, wherein the network interface device may employ any known transmission protocol.
[0099] The memory 301 may be integrated with the processor 302, for example, by placing RAM or flash memory within an integrated circuit microprocessor or the like. In addition, the memory 301 may include a separate device, such as an external disk drive, a storage array, or any other storage device that can be used by a database system. The memory 301 and the processor 302 may be operatively coupled, or may communicate with each other, such as through an I / O port, a network connection, etc., so that the processor 302 can read files stored in the memory.
[0100] In addition, the electronic device 300 may also include a video display (such as a liquid crystal display) and a user interaction interface (such as a keyboard, a mouse, a touch input device, etc.) All components of the electronic device 300 may be connected to each other via a bus and / or a network.
[0101] According to an exemplary embodiment of the present disclosure, a computer-readable storage medium may also be provided, and when the instructions in the computer-readable storage medium are executed by a processor of an electronic device, the electronic device is enabled to perform the above-mentioned multimedia resource providing method. Examples of computer-readable storage media here include: read-only memory (ROM), random access programmable read-only memory (PROM), electrically erasable programmable read-only memory (EEPROM), random access memory (RAM), dynamic random access memory (DRAM), static random access memory (SRAM), flash memory, non-volatile memory, CD-ROM, CD-R, CD+R, CD-RW, CD+RW, DVD-ROM, DVD-R, DVD+R, DVD-RW, DVD+RW, DVD-RAM, BD-ROM, BD-R, BD-R LTH, BD-RE, Blu-ray or optical disk storage, hard disk drive (HDD), solid state drive (SSD), card storage (such as, multimedia card, secure digital (SD) card or extreme digital (XD) card), magnetic tape, floppy disk, magneto-optical data storage device, optical data storage device, hard disk, solid state disk and any other device, any other device is configured to store computer programs and any associated data, data files and data structures in a non-transitory manner and provide the computer programs and any associated data, data files and data structures to a processor or computer so that the processor or computer can execute the computer program. The computer program in the above-mentioned computer-readable storage medium can be run in an environment deployed in a computer device such as a client, a host, an agent device, a server, etc. In addition, in one example, the computer program and any associated data, data files and data structures are distributed on a networked computer system, so that the computer program and any associated data, data files and data structures are stored, accessed and executed in a distributed manner by one or more processors or computers.
[0102] According to an exemplary embodiment of the present disclosure, a computer program product may also be provided, including a computer program, and when the computer program is executed by a processor, the multimedia resource providing method according to the present disclosure is implemented.
[0103] According to the multimedia resource providing method, device, electronic device and storage medium disclosed in the present invention, since the online traffic distribution at different times is different, the present invention can dynamically determine the number of multimedia resource candidate queues corresponding to the traffic distribution corresponding to the target time when the content request of the client is received, and perform subsequent screening and sorting operations based on this, on the one hand, avoiding the situation of excess computing resources during non-peak traffic periods, and making full use of computing resources; on the other hand, providing clients with more matching multimedia resources. Furthermore, the present invention can dynamically adjust the queue length of candidate advertisements from the recall stage to the rough sorting stage, and from the rough sorting stage to the fine sorting stage, according to the traffic distribution corresponding to the time interval of the target time when the content request of the client is received, so as to make full use of computing resources during non-peak traffic periods, and obtain high-quality advertisements that are more matched with users and can bring more revenue to the advertising publishing platform by screening among more candidate advertisements.
[0104] Those skilled in the art will readily appreciate other embodiments of the present disclosure after considering the specification and practicing the invention disclosed herein. The present disclosure is intended to cover any variations, uses or adaptations of the present disclosure that follow the general principles of the present disclosure and include common knowledge or customary techniques in the art that are not disclosed in the present disclosure. The description and examples are to be considered exemplary only, and the true scope and spirit of the present disclosure are indicated by the following claims.
[0105] It should be understood that the present disclosure is not limited to the exact structures that have been described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present disclosure is limited only by the appended claims.
Claims
1. A method for providing multimedia resources, characterized in that: include: Determine an estimated number of traffic requests in a time interval at which a target time at which a content request sent by a client is received, wherein the estimated number of traffic requests is a total number of content requests received by the server in the time interval; Determine a first number of multimedia resource candidate queues based on the estimated number of traffic requests and a first preset number of multimedia resource candidate queues, wherein the first preset number is the number of multimedia resource candidate queues determined by the server according to the maximum number of traffic requests; Screening out the first number of candidate multimedia resources, and sorting the screened candidate multimedia resources; According to the sorting result, providing multimedia resources to the client; The determining, based on the estimated number of traffic requests and the first preset number of multimedia resource candidate queues, of the first number of multimedia resource candidate queues includes: Determine a ratio of the maximum number of traffic requests to an estimated number of traffic requests in a time interval at which a target time instant at which a content request sent by a client is received; The first number is determined based on the ratio and a first preset number corresponding to the multimedia resource candidate queue determined by the server according to the maximum traffic request number.
2. The multimedia resource providing method according to claim 1, characterized in that: The sorting process of the screened candidate multimedia resources includes: Performing a first granularity sorting process on the screened candidate multimedia resources to obtain candidate multimedia resources that have undergone the first granularity sorting process; Determining, according to the target time, a second number of the candidate multimedia resources that need to be sorted at a second granularity; Selecting the second number of candidate multimedia resources from the candidate multimedia resources that have been sorted at the first granularity; Perform a second granularity sorting process on the second number of candidate multimedia resources to obtain the sorting result.
3. The multimedia resource providing method according to claim 2, characterized in that: The determining the second number of the candidate multimedia resources that need to be sorted at the second granularity includes: Determine a ratio of the maximum number of traffic requests to an estimated number of traffic requests in a time interval at which a target time instant at which a content request sent by a client is received; Determining the second number based on the ratio and a second preset number; The second preset number is: the number of multimedia resource candidate queues that need to be sorted at the second granularity determined by the server according to the maximum traffic request number.
4. The multimedia resource providing method according to claim 3, characterized in that: The first number is negatively correlated with the estimated number of traffic requests, and the second number is negatively correlated with the estimated number of traffic requests.
5. The multimedia resource providing method according to claim 2, characterized in that: The step of performing a first granularity sorting process on the screened candidate multimedia resources to obtain the candidate multimedia resources that have undergone the first granularity sorting process includes: Acquire first recommendation evaluation index information of each candidate multimedia resource among the screened candidate multimedia resources; Based on the first recommendation evaluation index information, the screened candidate multimedia resources are sorted at a first granularity to obtain candidate multimedia resources that have undergone the sorting at the first granularity.
6. The multimedia resource providing method according to claim 2, characterized in that: The performing a second granularity sorting process on the second number of candidate multimedia resources to obtain the sorting result includes: Acquire second recommendation evaluation index information of each candidate multimedia resource in the second number of candidate multimedia resources; Based on the second recommendation evaluation index information, a second granularity sorting process is performed on the second number of candidate multimedia resources to obtain the sorting result.
7. The multimedia resource providing method according to claim 1, characterized in that: The estimated number of traffic requests is obtained by inputting features related to the time interval into a traffic request prediction model.
8. A multimedia resource providing device, characterized in that: include: A determination module is configured to determine an estimated number of traffic requests in a time interval at which a target time at which a content request sent by a client is received, wherein the estimated number of traffic requests is a total number of content requests received by the server in the time interval; The determination module is configured to determine a first number of multimedia resource candidate queues based on the estimated number of traffic requests and a first preset number of multimedia resource candidate queues, wherein the first preset number is the number of multimedia resource candidate queues determined by the server according to the maximum number of traffic requests; a sorting module, configured to screen out the first number of candidate multimedia resources, and sort the screened candidate multimedia resources; A providing module, configured to provide multimedia resources to the client according to the sorting result; Wherein, the determination module is configured as follows: Determine a ratio of the maximum number of traffic requests to an estimated number of traffic requests in a time interval at which a target time instant at which a content request sent by a client is received; The first number is determined based on the ratio and a first preset number corresponding to the multimedia resource candidate queue determined by the server according to the maximum traffic request number.
9. The multimedia resource providing device according to claim 8, characterized in that: The sorting module is configured to: Performing a first granularity sorting process on the screened candidate multimedia resources to obtain candidate multimedia resources that have undergone the first granularity sorting process; Determining, according to the target time, a second number of the candidate multimedia resources that need to be sorted at a second granularity; Selecting the second number of candidate multimedia resources from the candidate multimedia resources that have been sorted at the first granularity; Perform a second granularity sorting process on the second number of candidate multimedia resources to obtain the sorting result.
10. The multimedia resource providing device according to claim 9, characterized in that: The sorting module is configured to: Determine a ratio of the maximum number of traffic requests to an estimated number of traffic requests in a time interval at which a target time instant at which a content request sent by a client is received; Determining the second number based on the ratio and a second preset number; The second preset number is: the number of multimedia resource candidate queues that need to be sorted at the second granularity determined by the server according to the maximum traffic request number.
11. The multimedia resource providing device according to claim 10, characterized in that: The first number is negatively correlated with the estimated number of traffic requests, and the second number is negatively correlated with the estimated number of traffic requests.
12. The multimedia resource providing device according to claim 9, characterized in that: The sorting module is configured to: Acquire first recommendation evaluation index information of each candidate multimedia resource among the screened candidate multimedia resources; Based on the first recommendation evaluation index information, the screened candidate multimedia resources are sorted at a first granularity to obtain candidate multimedia resources that have undergone the sorting at the first granularity.
13. The multimedia resource providing device according to claim 9, characterized in that: The sorting module is configured to: Acquire second recommendation evaluation index information of each candidate multimedia resource in the second number of candidate multimedia resources; Based on the second recommendation evaluation index information, a second granularity sorting process is performed on the second number of candidate multimedia resources to obtain the sorting result.
14. The multimedia resource providing device according to claim 8, characterized in that: The estimated number of traffic requests is obtained by inputting features related to the time interval into a traffic request prediction model.
15. An electronic device, characterized in that: include: processor; a memory for storing instructions executable by the processor; The processor is configured to execute the instructions to implement the multimedia resource providing method as described in any one of claims 1 to 7.
16. A computer-readable storage medium, characterized in that: When the instructions in the computer-readable storage medium are executed by a processor of an electronic device, the electronic device is enabled to execute the multimedia resource providing method as claimed in any one of claims 1 to 7.
17. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the multimedia resource providing method according to any one of claims 1 to 7 is implemented.
Citation Information
Patent Citations
Advertisement sorting method and device, electronic equipment and storage medium
CN111369271A
Information processing method and device, electronic equipment and storage medium
CN112767053A
