Quota adjustment method for recall results, result display method and related products
By calculating the filtering result distribution information in the recall result set, adjusting the recall result quota of the recall branch, solving the problem of increased resource utilization in the recall system and achieving resource optimization and cost savings.
Patent Information
- Application Number
- CN202310595768.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2023-05-23
- Publication Date
- 2024-11-26
AI Technical Summary
With the increase of recall strategies, the number of recall branches has increased, resulting in the recall system occupying more and more computing resources, resulting in the expansion of machine usage costs and the reduction of effective usage.
By obtaining the recall result set and filter result set of the target recall branch, calculate the filter result retention rate corresponding to different recall result quotas, adjust the target recall result quota to optimize the resource use of the recall branch and ensure that the search effect and recommendation effect are not affected.
Automatically adjust the recall result quota of the recall branch, reduce the computing resources occupied by the recall system, save machine costs, and improve the effective use rate of costs.
Smart Images

Figure CN119025744A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of data processing technology, and in particular to a quota adjustment method for recall results, a result display method, and related products. Background Art
[0002] With the rapid development of computer technology, the Internet has become an indispensable part of people's study, work and daily life. For example, users can search and obtain search results based on their immediate needs through the Internet, or quickly learn about hot topics or events of interest through Internet recommendations. However, the rapid development of the Internet has also led to an explosive growth in information resources, and the information related to users' immediate needs or hot topics may be thousands or tens of thousands of results.
[0003] At present, the recall system can be used to recall thousands or tens of thousands of results, and the final results can be obtained through layers of screening. In the related art, multiple recall branches of the recall system are usually used to filter out relevant results from a candidate result set containing a large number of results. Each recall branch is set with a corresponding recall strategy and recall result quota, so that the candidate result set can be screened based on the recall strategy to obtain the same number of recall results as the recall result quota; the recall results of these multiple recall branches are then screened layer by layer to obtain the final result to present to the user.
[0004] However, in order to show users better results, the recall system may gradually add recall strategies during the evolution and upgrade of the recall system. Since there is a one-to-one correspondence between recall strategies and recall branches, the number of recall branches will continue to grow. The problem that may exist in the above-mentioned solution is that the gradually increasing recall branches will cause the recall system to occupy more and more computing resources, which will lead to the continuous expansion of the machine usage cost. Summary of the invention
[0005] The embodiments of the present application provide a quota adjustment method for recall results, a result display method, and related products, which aim to ensure that the search effect and recommendation effect of the recall system are not affected while reducing the computing resources occupied by the recall system, thereby saving the cost of using the machine.
[0006] The first aspect of the present application provides a quota adjustment method for recall results, comprising:
[0007] Acquire a recall result set of a target recall branch in a current cycle and a first screening result set of the current cycle; the recall result set is obtained in a candidate result set based on a recall strategy of the target recall branch and a current recall result quota of the target recall branch; the first screening result set includes a plurality of screening results further screened from the recall result set;
[0008] Based on the distribution information of the multiple screening results in the recall result set, a screening result retention rate corresponding to configuring different recall result quotas for the target recall branch is obtained; the screening result retention rate is used to indicate the proportion of the multiple screening results that are still selected into the second screening result set; the second screening result set is a screening result set obtained by configuring different recall result quotas for the target recall branch;
[0009] Based on the screening result retention rates corresponding to the different recall result quotas, determining a target recall result quota from the different recall result quotas;
[0010] If the target recall result quota is inconsistent with the recall result quota corresponding to the recall result set, the current recall result quota of the target recall branch is adjusted to the target recall result quota.
[0011] The second aspect of the present application provides a result display method based on content request, comprising:
[0012] Based on the content request, the recall strategies of different recall branches and the recall result quotas of the different recall branches, the recall result sets of the different recall branches are obtained by screening from the candidate result sets; at least one recall result quota of the different recall branches is obtained according to the quota adjustment method for recall results provided in the first aspect;
[0013] Screening the recall result sets of the different recall branches to obtain a rough sorting result set, and screening the rough sorting result set to obtain a fine sorting result set;
[0014] Scattering the refined ranking results in the refined ranking result set to obtain the results corresponding to the content request;
[0015] The result corresponding to the content request is displayed.
[0016] The third aspect of the present application provides a quota adjustment device for recall results, including:
[0017] A result set acquisition module, used to acquire a recall result set of a target recall branch in a current cycle and a first screening result set of the current cycle; the recall result set is obtained in a candidate result set based on a recall strategy of the target recall branch and a current recall result quota of the target recall branch; the first screening result set includes a plurality of screening results further screened from the recall result set;
[0018] A retention rate acquisition module, configured to obtain, based on the distribution information of the multiple screening results in the recall result set, the screening result retention rate corresponding to configuring different recall result quotas for the target recall branch; the screening result retention rate is used to indicate the proportion of the multiple screening results that are still selected into the second screening result set; the second screening result set is the screening result set obtained by configuring different recall result quotas for the target recall branch;
[0019] A quota determination module, configured to determine a target recall result quota from the different recall result quotas based on the screening result retention rates corresponding to the different recall result quotas;
[0020] The quota adjustment module is used to adjust the current recall result quota of the target recall branch to the target recall result quota if the target recall result quota is inconsistent with the recall result quota corresponding to the recall result set.
[0021] A fourth aspect of the present application provides a result display device based on a content request, comprising:
[0022] A recall result set screening module, used to screen the recall result sets of different recall branches from the candidate result set based on the content request, the recall strategies of different recall branches and the recall result quotas of the different recall branches; at least one recall result quota of the different recall branches is obtained according to the quota adjustment method of the recall result provided in the first aspect;
[0023] A refined sorting result set screening module, used to screen the recall result sets of the different recall branches to obtain a rough sorting result set, and to screen the rough sorting result set to obtain a refined sorting result set;
[0024] A result acquisition module, used for breaking up the refined ranking results in the refined ranking result set to obtain the result corresponding to the content request;
[0025] The result display module is used to display the result corresponding to the content request.
[0026] A fifth aspect of the present application provides a computer device, the computer device comprising a processor and a memory:
[0027] The memory is used to store a computer program and transmit the computer program to the processor;
[0028] The processor is used to execute the quota adjustment method for recall results provided in the first aspect, or execute the result display method based on content request provided in the second aspect according to the instructions in the computer program.
[0029] In a sixth aspect, the present application provides a computer-readable storage medium, which is used to store a computer program. When the computer program is executed by a computer device, it implements the quota adjustment method for recall results provided in the first aspect, or implements the result display method based on content request provided in the second aspect.
[0030] The seventh aspect of the present application provides a computer program product, including a computer program, which, when executed by a computer device, implements the quota adjustment method for recall results provided in the first aspect, or implements the result display method based on content request provided in the second aspect.
[0031] It can be seen from the above technical solutions that the embodiments of the present application have the following advantages:
[0032] The embodiment of the present application provides a quota adjustment method for recall results. In this method, the recall result set of the target recall branch in the current cycle and the first screening result set of the current cycle are first obtained. The recall result set is obtained by screening from the candidate result set based on the recall strategy of the target recall branch and its current recall result quota. The first screening result set includes multiple screening results obtained by further screening from the corresponding recall result set. That is, the recall result set and the first screening result set can be used as data for analyzing whether the current recall result quota of the target recall branch is appropriate. The multiple screening results in the first screening result set correspond to the multiple recall results in the recall result set. In order to determine the target recall result quota that better matches the target recall branch, the screening result retention rate corresponding to the different recall result quotas configured for the target recall branch can be obtained based on the distribution information of the multiple screening results in the recall result set. The screening result retention rate is used to indicate the proportion of results that are still selected into the second screening result set among the multiple screening results. The second screening result set is the screening result set obtained by configuring different recall result quotas for the target recall branch. Then, based on the retention rates of the screening results corresponding to different recall result quotas, a target recall result quota that better matches the target recall branch is determined from the different recall result quotas. If the target recall result quota is inconsistent with the recall result quota corresponding to the recall result set, the current recall result quota of the target recall branch can be adjusted to the target recall result quota.
[0033] It can be seen that the present application can clarify the specific positions of the multiple recall results that are filtered into the first filtering result set in the recall result set through the distribution information of the multiple filtering results of the first filtering result set in the recall result set. Then, according to the retention rate of the filtering results corresponding to different recall result quotas, the effective utilization rate of the machine cost occupied by the target recall branch can be determined, so as to make corresponding adjustments to the recall result quota of the target recall branch with a reduced effective utilization rate. Therefore, the present application can automatically adjust the recall result quota of the target recall branch, while ensuring that the search effect and recommendation effect of the recall system are not affected, and can reduce the computing resources occupied by the recall system, thereby saving the machine cost used by the system and improving the effective utilization rate of the machine cost. BRIEF DESCRIPTION OF THE DRAWINGS
[0034] Figure 1 A scenario architecture diagram of a quota adjustment method for recall results provided in an embodiment of the present application;
[0035] Figure 2 A flowchart of a quota adjustment method for a recall result provided in an embodiment of the present application;
[0036] Figure 3 A schematic diagram of adding a mark to a recall result in a recall result set provided in an embodiment of the present application;
[0037] Figure 4 A schematic diagram of obtaining the screening result retention rate corresponding to different recall result quotas provided in an embodiment of the present application;
[0038] Figure 5 A schematic diagram of a recall system screening process provided in an embodiment of the present application;
[0039] Figure 6 A schematic diagram of a quota adjustment method for a recall result provided in an embodiment of the present application;
[0040] Figure 7 A flowchart of another method for adjusting quota of recall results provided in an embodiment of the present application;
[0041] Figure 8 A flowchart of a result display method based on content request provided in an embodiment of the present application;
[0042] Fig. 9 A schematic diagram of a page for displaying results corresponding to a content request provided in an embodiment of the present application;
[0043] Fig.10 A schematic diagram of the structure of a quota adjustment device for recall results provided in an embodiment of the present application;
[0044] Fig.11A schematic diagram of the structure of a result display device based on content request provided in an embodiment of the present application;
[0045] Fig.12 A schematic diagram of the structure of the server in the embodiment of the present application;
[0046] Fig.13 A schematic diagram of the structure of a terminal device in an embodiment of the present application. DETAILED DESCRIPTION
[0047] At present, the recall system corresponding to the search system and recommendation system usually includes four modules: recall, rough ranking, fine ranking and scattering. When the recall system receives a user request, it first uses the recall module to filter from the candidate result set to obtain multiple recall results of multiple recall branches. Each recall branch corresponds to a recall strategy and a recall result quota; then, the rough ranking module is used to further filter the multiple recall results of multiple recall branches to obtain multiple rough ranking results; then, the fine ranking module is used to further filter the multiple rough ranking results to obtain multiple fine ranking results; finally, the scattering module is used to scatter and re-rank the multiple fine ranking results to obtain the final result, so as to return the result corresponding to the user request to the user.
[0048] However, after research, it was found that during the evolution and upgrade of the recall system, recall strategies will be gradually added to return better results to users, and the recall branches of the recall module will continue to increase. However, the quota of coarse sorting results of the coarse sorting module is usually maintained unchanged. The addition of new recall branches may reduce the number of recall results that can enter the coarse sorting module in the old recall branches, but the quota of recall results of the old recall branches remains unchanged, and the old recall branches still use the same machine cost. Therefore, as the number of recall branches increases, the recall system will occupy more and more computing resources, resulting in the continuous expansion of the machine cost used, while the effective utilization rate of the machine cost may gradually decrease.
[0049] In view of the above problems, a quota adjustment method for recall results, a result display method and related products are provided in the present application, with the aim of reducing the computing resources occupied by the recall system and saving the cost of using the machine while ensuring that the search effect and recommendation effect of the recall system are not reduced. In the technical solution provided in the present application, the retention rate of the screening results corresponding to the different recall result quotas configured for the target recall branch is obtained based on the distribution information of the multiple screening results in the first screening result set in the current cycle, and then the target recall result quota is determined and adjusted accordingly. The present application can automatically adjust the recall result quota of the recall branch, which greatly saves the cost of the machine used by the recall system.
[0050] First, several terms that may be involved in the embodiments of the present application are explained below.
[0051] Recall system: An important module in the search system and recommendation system. It is responsible for quickly filtering out relevant and high-quality results from a set of candidate results in the millions, tens of millions, or hundreds of millions according to user requests, and returning the results to the search system and recommendation system. The recall system includes four modules: recall, rough sorting, fine sorting, and scattering.
[0052] Recall module: responsible for the first round of screening from the candidate result set to obtain the recall results, and send the recall results to the rough sorting module. The recall module includes multiple recall branches, each of which has a corresponding recall strategy and recall result quota.
[0053] Rough sorting module: responsible for further screening the recall results obtained from the recall module to obtain rough sorting results, and sending the rough sorting results to the fine sorting module.
[0054] Fine sorting module: responsible for further screening the rough sorting results obtained from the rough sorting module to obtain fine sorting results, and send the fine sorting results to the scattering module.
[0055] Scattering module: responsible for scattering and re-arranging the refined sorting results obtained by the refined sorting module to obtain the final results and return them to the user.
[0056] The execution subject of the quota adjustment method for recall results provided in the embodiment of the present application may be a terminal device with data processing capabilities. For example, on a terminal device with an application such as a browser installed, a user can search through the browser to obtain search results. Then the terminal device can return the search results to the user through the recall system in response to the user's search operation, so the terminal device can obtain the recall result set and the first screening result set of the target recall branch in the current cycle. As an example, the terminal device may specifically include but is not limited to mobile phones, desktop computers, tablet computers, notebook computers, PDAs, intelligent voice interaction devices, smart home appliances, vehicle terminals, aircraft, etc. The embodiment of the present invention can be applied to various scenarios, including but not limited to advertising, recommendation, and search. The execution subject of the quota adjustment method for recall results provided in the embodiment of the present application may also be a server, that is, the server may obtain the recall result set of the target recall branch in the current cycle and the first screening result set of the current cycle. In addition, the quota adjustment method for recall results provided in the embodiment of the present application may also be executed collaboratively by the terminal device and the server. For example, the terminal device first obtains the recall result set of the target recall branch in the current cycle and the first screening result set of the current cycle, and then sends the recall result set and the first screening result set to the server so that the server determines the target recall result quota and returns it to the terminal device. Finally, if the terminal device determines that the target recall result quota is inconsistent with the recall result quota corresponding to the recall result set, the recall result quota of the target recall branch is adjusted to the target recall result quota.
[0057] The execution subject of the result display method based on content request provided in the embodiment of the present application can be a terminal device. Or the result display method based on content request can also be executed collaboratively by a terminal device and a server. For example, the terminal device can send a content request to the server; the server filters the recall result set of different recall branches from the candidate result set based on the content request, the recall strategy of different recall branches and the recall result quota of different recall branches; then the server filters the recall result set of different recall branches to obtain a rough sorting result set, and filters the refined sorting result set from the rough sorting result set; then the server breaks up the refined sorting results in the refined sorting result set to obtain the results corresponding to the content request; finally, the server can send the results to the terminal device so that the terminal device displays the results corresponding to the content request to the user.
[0058] Therefore, the embodiments of the present application do not limit the implementation subject of the technical solution of the present application. It should be noted that when the terminal device and the server jointly execute the quota adjustment method of the recall result or the result display method based on the content request, the terminal device and the server can be directly or indirectly connected via wired or wireless communication, and the present application does not limit this.
[0059] Figure 1 A scenario architecture diagram of a quota adjustment method for recall results provided in an embodiment of the present application, including a server and various forms of terminal devices. Figure 1 The server shown can be an independent physical server, or a server cluster or distributed system composed of multiple physical servers. In addition, the server can also be a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms. It should be noted that Figure 1 It may also be a scenario architecture diagram of the result display method based on content request provided in an embodiment of the present application.
[0060] Next, taking the terminal device as the execution subject, a quota adjustment method for a recall result provided in an embodiment of the present application is described in detail in combination with the accompanying drawings.
[0061] See also Figure 2 , which is a flow chart of a quota adjustment method for a recall result provided in an embodiment of the present application, and specifically includes the following steps:
[0062] S201: Obtain the recall result set of the target recall branch in the current cycle and the first screening result set of the current cycle.
[0063] In an embodiment of the present application, the target recall branch in the recall module can filter out the recall result set from the candidate result set based on the recall strategy of the target recall branch and the current recall result quota of the target recall branch, and the value of the number of recall results in the recall result set is the same as the value of the recall result quota of the target recall branch. The rough sorting module of the recall system can further filter the recall results in the recall result set to obtain a first screening result set including multiple screening results, and the multiple screening results in the first screening result set correspond to the multiple recall results in the recall result set.
[0064] It should be noted that, in actual applications, the rough sorting module of the recall system screens the recall results of multiple recall branches of the recall module to obtain multiple rough sorting results, that is, the multiple rough sorting results correspond to the multiple recall results in the multiple recall branches. In the embodiment of the present application, only the result obtained by the rough sorting module screening the recall results of the target recall branch is obtained. Therefore, in order to distinguish them, the present application records the rough sorting results in the target recall branch that are screened into the rough sorting module as the screening results in the first screening set. That is, the multiple screening results in the first screening result set only correspond to the multiple recall results of the target recall branch.
[0065] In the embodiment of the present application, the target recall branch does not specifically refer to a certain recall branch, but refers to any one of the multiple recall branches of the recall module in the recall system. The recall strategy of the target recall branch refers to the strategy of how the target recall branch recalls from the candidate result set. For example, in the recall system corresponding to the video recommendation system. The recall strategy may include: "hot videos", "director recall", "actor recall", "recent releases", "popular trends" and "well-received recalls", and other recall strategies, which are not limited in this application. The current cycle refers to a fixed time period, such as 12 hours, a day or a week, etc., which is not limited in this application. Among them, the specific length of the current cycle can be determined by the technician based on the specific business scenario and the recall situation of the comprehensive recall module, which is not limited in this application. For example, the target recall branch of a video playback application can be obtained, and the recall result set and the first screening result set within one day, that is, one day is taken as the current cycle.
[0066] S202: Based on the distribution information of the plurality of screening results in the recall result set, the screening result retention rates corresponding to different recall result quotas configured for the target recall branch are obtained.
[0067] If multiple screening results in the first screening result set correspond to multiple recall results in the recall result set, the distribution information of the multiple screening results in the recall result set can be obtained. Based on the distribution information of the multiple screening results in the recall result set, it can be obtained that when different recall result quotas are configured for the target recall branch, the multiple recall results in the recall result set corresponding to the first screening result set can still enter the recall results of the second screening result set, thereby obtaining the screening result retention rate corresponding to the different recall result quotas configured for the target recall branch.
[0068] In the embodiment of the present application, the screening result retention rate is used to indicate the proportion of the results that are still selected into the second screening result set among the multiple screening results; the second screening result set is the screening result set obtained by configuring different recall result quotas for the target recall branch. It should be noted that the second screening result set is also obtained by screening the recall results of only the target recall branch.
[0069] It is understandable that when different recall result quotas are configured for the target recall branch, the number of recall results obtained by the recall module will also vary with the different recall result quotas, which will also cause the number of screening results obtained from the target recall branch in the rough sorting module to change. In order to obtain the screening result retention rate corresponding to the different recall result quotas configured for the target recall branch, you can first count the distribution information of the multiple recall results in the recall result set corresponding to the multiple screening results in the first screening result set, and when different recall result quotas are configured for the target recall branch, the corresponding multiple recall results are still selected into the second screening result set; finally, according to the ratio of the number of screening results in the second screening result set to the number of screening results in the first screening result set, the screening result retention rate corresponding to the different recall result quotas is obtained.
[0070] Therefore, in a possible implementation of the present application, S202 may specifically include: based on the distribution information of multiple screening results in the recall result set, obtaining the number of screening results of the second screening result set obtained by configuring different recall result quotas for the target recall branch; according to the ratio of the number of screening results of the second screening result set to the number of screening results of the first screening result set, obtaining the screening result retention rate corresponding to the different recall result quotas. It can be seen that according to the screening result retention rate corresponding to different recall result quotas, the number of screening results that are still retained when different recall result quotas are configured for the target recall branch can be clearly determined, which is conducive to obtaining a more appropriate target recall result quota.
[0071] It should be noted that the different recall result quotas configured for the target recall branch can be any value that is less than or equal to the current recall result quota, and this application does not limit this.
[0072] In order to more conveniently calculate the retention rate of screening results corresponding to different recall result quotas, the following steps can be used to determine the distribution information of multiple screening results in the first screening set in the recall result set. First, a first mark can be added to the multiple recall results in the recall result set corresponding to the multiple screening results in the first screening result set, and a second mark can be added to the remaining recall results in the recall result set. Then, in the recall result set, the distribution of multiple screening results in the recall result set can be intuitively seen based on the position difference of the first mark and the second mark. Therefore, in a possible implementation of the present application, the distribution information of the above-mentioned multiple screening results in the recall result set can be specifically determined by the following steps: add a first mark to the multiple recall results in the recall result set that are screened into the first screening result set, and add a second mark to the recall results in the recall result set that are not screened into the first screening result set; based on the position information of the first mark and the second mark in the recall result set, determine the distribution information of multiple screening results in the recall result set.
[0073] The present application does not limit the first mark and the type and number of the first mark. For example, the first mark and the second mark may include color marks; or the first mark and the second mark may include color marks and digital marks at the same time.
[0074] As an example, see Figure 3 , which is a schematic diagram of adding a mark to a recall result in a recall result set provided by an embodiment of the present application. Figure 3 As shown, the recall result set includes 6 recall results, and the first screening result set includes 4 screening results. The first mark and the second mark are color marks, and black marks are added to the 4 recall results in the recall result set that are screened into the first screening result set, and gray marks are added to the two recall results in the recall result set that are not screened into the first screening result set.
[0075] It is understandable that in the current cycle, the recall system may receive multiple user requests, then the recall module will filter from the candidate result set based on each user request, and each screening will obtain a recall result set corresponding to the target recall branch. In other words, the target recall branch may have multiple recall result sets in the current cycle, and each recall result set will correspond to a first screening result set. Therefore, based on the distribution information of the multiple screening results of each first screening result set in the corresponding recall result set, the screening result retention rate corresponding to different recall result quotas configured for the target recall branch can be calculated. The total number of first screening results can be obtained by first adding up the number of screening results in multiple first screening result sets of the target recall branch within the current cycle; when a recall result quota is configured for the target recall branch, multiple second screening result sets corresponding to the multiple recall result sets can be obtained, and the total number of second screening results can be obtained by adding up the screening results in multiple second screening result sets, and then the total number of second screening results corresponding to different recall result quotas configured for the target recall branch can be obtained; finally, the ratio of the total number of different second screening results to the total number of first screening results is determined as the screening result retention rate corresponding to different recall result quotas.
[0076] As an example, see Figure 4 , which is a schematic diagram of obtaining the screening result retention rate corresponding to different recall result quotas provided by an embodiment of the present application. Figure 4 As shown, the target recall branch has three recall result sets in the current cycle, namely, recall result set 1, recall result set 2, and recall result set 3, corresponding to the three first screening result sets, namely, first screening result set 1, first screening result set 2, and first screening result set 3; the first mark is specifically a black mark and a digital mark 1, and the second mark is specifically a gray mark and a digital mark 0. The digital mark is helpful for calculating the total number of the first screening results and the total number of the second screening results. Figure 4 As shown, extracting the digital labels of the three recall result sets can obtain three arrays [1 0 1 1 1 0], [1 0 1 0 1 0], and [1 1 1 0 0 0]. Vertically superimposing these three arrays can obtain [3 1 3 1 2 0]. Then, all the numbers in [3 1 3 1 2 0] can be added together to calculate the total number of first screening results of the three first screening result sets corresponding to the target recall branch, which is 10. Then, the total number of different second screening results corresponding to the configuration of 6 different recall result quotas from 1 to 6 for the target recall branch is calculated. As shown Figure 4As shown in the figure, when the recall result quota configured for the target recall branch is 2, the first two numbers in [3 1 3 1 2 0] can be added to calculate the total number of second screening results of the three second screening result sets corresponding to the three recall result sets, which is 4; when the recall result quota configured for the target recall branch is 3, the first three numbers in [3 1 3 1 2 0] can be added to calculate the total number of second screening results corresponding to 7; similarly, when the recall result quotas configured for the target recall branch are 1 to 6, the total number of second screening results obtained is 3, 4, 7, 8, 10 and 10 respectively. Finally, according to the ratio of different total number of second screening results to the total number of first screening results 10, the screening result retention rates corresponding to the recall result quotas 1 to 6 can be obtained, which are 30%, 40%, 70%, 80%, 100% and 100% respectively.
[0077] S203: Determine a target recall result quota from the different recall result quotas based on the screening result retention rates corresponding to the different recall result quotas.
[0078] In the embodiment of the present application, the target recall result quota refers to the recall result quota that meets the recall system's requirement for retaining the screening results in the rough sorting module, determined based on the retention rates of the screening results corresponding to the different recall result quotas. It should be noted that if the current recall result quota of the target recall branch may exist among the multiple recall result quotas in the different recall result quotas, the target recall result quota may still be the current recall result quota.
[0079] It is understandable that the number of recall branches in the recall system will gradually increase, and the quota of coarse sorting results obtained by the recall system in the coarse sorting module will usually be maintained at a stable number, so the number of results in the recall result set corresponding to the old recall branch that can enter the coarse sorting result set may decrease. At the same time, considering that in actual applications, even if new recall branches are not added, the number of results that can enter the coarse sorting module for each recall branch in the recall system may be adjusted according to the recall strategy proportion of the recall branch and the computing resources occupied. For recall branches with fewer results entering the coarse sorting module, it is necessary to adjust their recall result quotas to save the cost of the machines they use and improve the effective use of machine costs.
[0080] After obtaining the screening result retention rates corresponding to different recall result quotas, the different screening result retention rates can be compared with the preset retention rate to determine the screening result retention rate that is less than or the same as the preset retention rate, that is, determine the first recall result quota whose screening result retention rate is greater than or equal to the preset retention rate. However, there may be one or more screening result retention rates greater than or equal to the preset retention rate, that is, the first recall result quota may be one first recall result quota or may be multiple first recall result quotas. When there is only one first recall result quota among different recall result quotas, the first recall result quota is the target recall result quota; when there are multiple first recall results among different recall result quotas, the values of the multiple first recall result quotas can be compared, and the first recall result quota with the smallest value can be determined as the target recall result quota. Therefore, in a possible implementation of the present application, S203 may specifically include: determining a first recall result quota from different recall result quotas; the screening result retention rate corresponding to the first recall result quota is greater than or equal to a preset retention rate; if there is one first recall result quota among different recall result quotas, the first recall result quota is determined as the target recall result quota; if there are multiple first recall result quotas among different recall result quotas, the first recall result quota with the smallest value among the multiple first recall result quotas is determined as the target recall result quota.
[0081] Among them, the preset retention rate can be pre-set by the technician based on experience, or it can be pre-set by the terminal device according to the change in the proportion of the recall strategy of the target recall branch and the computing resources occupied. This application does not limit this. For example, the preset retention rate can be set to 100%, that is, when it is determined that multiple screening results in the first screening result set can enter the second screening result set, it is necessary to configure a target recall result quota for the target recall branch; it can also be that the recall strategy of the target recall branch has a small proportion, but relative to other recall branches with a larger recall strategy proportion, the target recall branch occupies more computing resources, then the preset retention rate can be set to 80%, and the target recall result quota that can appropriately reduce the computing resources occupied by the target recall branch is determined.
[0082] As an example, as mentioned above Figure 3 As shown, if all screening results of the target recall branch need to be retained, the preset retention rate can be set to 100%, then there are two first recall result quotas corresponding to different recall result quotas, and the minimum recall result quota 5 of these two first recall result quotas can be determined as the target recall result quota.
[0083] S204: If the target recall result quota is inconsistent with the recall result quota corresponding to the recall result set, the current recall result quota of the target recall branch is adjusted to the target recall result quota.
[0084] In an embodiment of the present application, the target recall result quota determined from different recall result quotas may be consistent with the current recall result quota of the target recall branch, or may be inconsistent. Therefore, if they are consistent, the recall result quota of the target recall branch may not be adjusted, and it can remain the current recall result quota; if they are inconsistent, it is necessary to adjust the current recall result quota of the target recall branch to the target recall result quota. So that after the current cycle ends, the target recall branch of the recall system can filter from the candidate result set based on the target recall result quota to obtain the recall result set.
[0085] As an example, see Figure 5 , which is a schematic diagram of the recall system screening process. Figure 5 As shown, the recall system includes n recall branches. First, in the recall module, the candidate result set is screened based on the recall strategies and recall result quotas of the n recall branches to obtain the recall result set; then, the recall result set is further screened in the rough sorting module to obtain the rough sorting result set; then, the rough sorting result set is further screened in the fine sorting module to obtain the fine sorting result set; finally, the fine sorting results in the fine sorting result set are scattered to obtain the final result.
[0086] As an example, see Figure 6 , which is a schematic diagram of a quota adjustment method for recall results provided in an embodiment of the present application. Figure 6 As shown, the process of the recall system screening and returning results layer by layer may specifically include the following steps:
[0087] S1: The terminal device front end sends a user request to the terminal device back end, and the scheduling module of the recall system in the terminal device back end receives the user request.
[0088] S2: The scheduling module sends the user request to the recall module. Each recall branch in the recall module obtains a recall result set by filtering from the candidate result set through the corresponding recall strategy and recall result quota.
[0089] S3: The recall module returns the recall result set to the scheduling module.
[0090] S4: The scheduling module sends the recall result to the coarse sorting module, and the coarse sorting model in the coarse sorting module filters the recall result set to obtain the coarse sorting result set.
[0091] S5: The coarse sorting module returns the coarse sorting result set to the scheduling module.
[0092] S6: The scheduling module sends the rough sorting result set to the fine sorting module, and the fine sorting model in the fine sorting module filters the rough sorting result set to obtain the fine sorting result set.
[0093] S7: The fine sorting module returns the fine sorting result set to the scheduling module.
[0094] S8: The scheduling module sends the refined sorting result set to the scattering module, and the scattering module scatters the refined sorting results in the refined sorting result set to obtain the final result.
[0095] S9: The scattering module sends the final result to the scheduling module.
[0096] S10: The scheduling module returns the final result to the terminal device front end so that the terminal device front end displays the result corresponding to the user request to the user.
[0097] Combination Figure 6 As shown, after the current cycle ends, the terminal device background can obtain the recall result set and the first screening result set of the target recall branch in the current cycle from the recall system, and then determine the target recall result quota of the target recall branch, and finally adjust the recall result quota of the target recall branch. So that after the current cycle ends, the target recall branch of the recall system in the terminal device background can filter the recall result set from the candidate result set according to the current recall result quota and recall strategy.
[0098] In actual applications, in order to ensure that the recall result set corresponding to the target recall branch can have recall results that are more in line with the requirements of the coarse sorting module and be filtered into the first screening result set, the recall result quota of the target recall branch may be set to a larger value, so that the coarse sorting module can be used to filter from a larger number of recall results. At the same time, abnormal situations may also occur, resulting in the recall result quota of the recall module being set to a larger value in order to filter out the recall results that meet the requirements of the coarse sorting module. That is, taking into account the noise points caused by individual long-tail situations or abnormal situations of the target recall branch, the recall result quota of the target recall branch is set to a larger value.
[0099] However, the problem that this brings is that, perhaps only for a few individual long-tail situations or abnormal situations, the target recall branch needs to use the same amount of computing resources for each recall result set obtained by screening, which is equivalent to doing useless work in many cases, resulting in a reduction in the effective utilization rate of machine costs and a waste of machine usage costs. In this application, the specific locations where multiple recall results corresponding to multiple screening results are closely distributed in the recall result set, and the specific locations where they are sparsely distributed, can be determined first, and then it can be determined whether the recall result quota of the target recall branch needs to be adjusted accordingly.
[0100] Therefore, in view of the above problems, another method for adjusting the quota of recall results is provided in an embodiment of the present application. The purpose is to further calculate the distribution density of the multiple filtering results in the first filtering result set in the recall result set after obtaining the filtering result retention rate corresponding to different recall result quotas, so as to adjust the recall result quota of the target recall branch accordingly.
[0101] See also Figure 7 , which is a flow chart of another method for adjusting quota of recall results provided by an embodiment of the present application. Figure 7 As shown, the quota adjustment method for the recall result specifically includes:
[0102] S701: Obtain the recall result set of the target recall branch in the current cycle and the first screening result set of the current cycle.
[0103] In the current cycle, the terminal device obtains a recall result set corresponding to a target recall branch in the recall module, and obtains a first screening result set including a plurality of screening results obtained by further screening the recall result set.
[0104] S702: Based on the distribution information of the plurality of screening results in the recall result set, the screening result retention rates corresponding to the different recall result quotas configured for the target recall branch are obtained.
[0105] The terminal device calculates the proportion of results that can still enter the second filtering result set when different recall result quotas are configured for the target recall branch based on the distribution information of multiple recall results corresponding to the multiple filtering results in the first filtering result set, and obtains the corresponding filtering result retention rate.
[0106] S703: Determine recall result quota intervals corresponding to different retention rate intervals based on the screening result retention rates corresponding to different recall result quotas.
[0107] In an embodiment of the present application, different retention rate intervals refer to preset retention rate intervals of different screening results; the recall result quota intervals corresponding to different retention rate intervals include the recall result quotas determined from different recall result quotas. Among them, different retention rate intervals can be uniformly set in advance by technical personnel based on experience, or they can be set by the terminal device based on the recall strategy proportion of the target recall branch and the machine occupied resources. For example, different retention rate intervals can be set to four retention rate intervals of [80%, 85%], [85%, 90%], [90%, 95%] and [95%, 100%]. It should be noted that the present application does not limit the number of retention rate intervals, the interval endpoints of the retention rate intervals, and the interval endpoint differences.
[0108] As an example, the current recall result quota of the target recall branch can be 100. Then, according to the distribution information of the multiple screening results in the first screening result set corresponding to the target recall branch in the current cycle, the screening result retention rate corresponding to the recall result quota between 1 and 100 for the target recall branch can be obtained. See Table 1 for details, which is a schematic table of screening result retention rates corresponding to different recall result quotas provided in an embodiment of the present application. In combination with Table 1, when the recall result quota of the target recall branch is 1 to 100, the corresponding screening result retention rates can be obtained respectively. If the different retention rate intervals are specifically the four retention rate intervals in the above example, including [80%, 85%], [85%, 90%], [90%, 95%] and [95%, 100%]. Then, based on the screening result retention rates corresponding to the 100 different recall result quotas from 1 to 100 in Table 1, the recall result quota interval corresponding to each retention rate interval can be determined. Specifically, for the retention rate interval [80%, 85%], the left endpoint of the retention rate interval 80% corresponds to the recall result quota 20, but there is no recall result quota corresponding to the right endpoint of the retention rate interval 85%. Then, the recall result quota 21 corresponding to the minimum screening result retention rate 86% from multiple screening result retention rates greater than 85% can be determined as the recall result quota corresponding to the right endpoint of the retention rate interval 85%. Then the recall result quota interval corresponding to the retention rate interval [80%, 85%] is [20, 21]. Similarly, the recall result quota interval corresponding to the retention rate interval [85%, 90%] is [21, 28], the recall result quota interval corresponding to the retention rate interval [90%, 95%] is [28, 30], and the recall result quota interval corresponding to the retention rate interval [95%, 100%] is [30, 100].
[0109] Table 1
[0110]
[0111] As shown in the above example, in actual applications, the filter result retention rates corresponding to different recall result quotas may not have the same value as the interval endpoint of the retention rate interval. Therefore, for each interval endpoint of the retention rate interval, the recall result quota corresponding to the minimum filter result retention rate among the filter result retention rates corresponding to different recall result quotas that are greater than the interval endpoint of the retention rate interval can be determined as the interval endpoint of the recall result quota interval corresponding to the retention rate interval.
[0112] S704: Based on the recall result quota intervals corresponding to the different retention rate intervals and the different retention rate intervals, determine the screening result distribution density corresponding to the different recall result quota intervals.
[0113] In an embodiment of the present application, the distribution density of the screening results is the distribution density of the screening results corresponding to different recall result quota intervals in the recall result set. For example, as shown in Table 1 in the above example, the distribution density of the screening results corresponding to the recall result quota interval [30, 100] corresponding to the retention rate interval [80%, 85%] in the recall result set. That is, the distribution density of the recall results selected into the first screening result set among the recall results from the 30th position to the 100th position of the recall result set.
[0114] Among them, in order to obtain the distribution density of screening results corresponding to different recall result quota intervals, the interval endpoint difference of the retention rate interval and the interval endpoint difference of the recall result quota interval corresponding to the retention rate interval can be calculated first; then the ratio of the two interval endpoint differences is determined as the distribution density of screening results corresponding to the recall result quota interval. Therefore, in a possible implementation of the present application, S704 may specifically include: determining the ratio of the interval endpoint difference of different retention rate intervals to the interval endpoint difference of the recall result quota interval corresponding to the same retention rate interval as the distribution density of screening results corresponding to different recall result quota intervals.
[0115] In addition, considering that the interval endpoint difference of the retention rate interval is a percentage, the calculated distribution density of the screening results is a very small value. Therefore, in order to facilitate the setting of the preset distribution density threshold in the subsequent steps, the value obtained by multiplying the interval endpoint difference of different retention rate intervals by the preset value can be used as the ratio of the interval endpoint difference of the recall result quota interval corresponding to the same retention rate interval to determine the screening result distribution density corresponding to the different recall result quota intervals. It should be noted that the present application does not limit the preset value, and it is sufficient to facilitate the setting of the preset distribution density threshold.
[0116] As an example, the screening result distribution density can be expressed as result_density, and the preset value can be 100. Based on Table 1 in the above example and the four retention rate intervals, four recall result quota intervals [20, 21], [21, 28], [28, 30], and [30, 100] corresponding to [80%, 85%], [85%, 90%], [90%, 95%], and [95%, 100%] can be obtained. Then the calculation formula of the screening result distribution density result_density [80%, 85%] corresponding to the recall result quota interval [20, 21] can be shown as the following formula 1:
[0117]
[0118] Similarly, the calculation formulas for result_density[85%, 90%], result_density[90%, 95%], and result_density[95%, 100%] can be shown in Formula 2, Formula 3, and Formula 4, respectively:
[0119]
[0120]
[0121]
[0122] From the above, we can get result_density[80%, 85%] is 5, result_density[85%, 90%] is 0.71, result_density[90%, 95%] is 2.5, and result_density[95%, 100%] is 0.07.
[0123] S705: Determine a target recall result quota from different recall result quotas according to a comparison result of screening result distribution densities corresponding to different recall result quota intervals and a preset distribution density threshold corresponding to the same recall result quota interval.
[0124] In the embodiment of the present application, the preset distribution density threshold refers to a preset distribution density threshold for comparison with the screening result distribution density. The preset distribution density thresholds of different recall result quota intervals may be the same or different, and the present application does not limit this.
[0125] In addition, it should be noted that, usually, corresponding preset distribution density thresholds are set for different retention rate intervals. After obtaining the recall result quota intervals corresponding to different retention rate intervals, the preset distribution density thresholds corresponding to the different recall result quota intervals can be determined. Among them, the specific value of the preset distribution density threshold can be set by the technician based on experience, and this application does not impose any restrictions on this.
[0126] For example, different retention rate intervals may be specifically four retention rate intervals of [80%, 85%], [85%, 90%], [90%, 95%], and [95%, 100%], and the preset distribution density threshold may be uniformly 2. A preset distribution density threshold may also be set for each retention rate interval, and the preset distribution density threshold corresponding to the retention rate interval [80%, 85%] may be 2.5, the preset distribution density threshold corresponding to the retention rate interval [85%, 90%] may be 2, the preset distribution density threshold corresponding to the retention rate interval [90%, 95%] may be 1.5, and the preset distribution density threshold corresponding to the retention rate interval [95%, 100%] may be 1.
[0127] It is understandable that when the distribution density of screening results corresponding to different recall result quota intervals is compared with the preset distribution density threshold corresponding to the same recall result quota interval, it is necessary to proceed in order from the smaller interval endpoints of the recall result quota interval. For example, recall result quota interval 1 is a recall result quota interval with a larger interval endpoint, and recall result quota interval 2 is a recall result quota interval with a smaller interval endpoint. If the distribution density of screening results corresponding to recall result quota interval 1 is greater than or equal to the preset distribution density threshold, it can be determined that the distribution density of screening results within recall result quota interval 1 is relatively tight. At this time, regardless of whether the distribution density of screening results within recall result quota interval 2 is tight or sparse, the recall result quota will be determined from recall result quota interval 1. However, if recall result quota interval 2 is compared first, the target recall result interval must also be determined based on the comparison result of recall result quota interval 1.
[0128] Therefore, in a possible implementation of the present application, S705 may specifically include: based on the order of the interval endpoints of different recall result quota intervals from large to small, comparing the screening result distribution densities corresponding to different recall result quota intervals with the preset distribution density threshold corresponding to the same recall result quota interval; if there is a target recall result quota interval in different recall result quota intervals, then from the different recall result quotas, the maximum recall result quota corresponding to the target recall result quota interval is determined as the target recall result quota; the target recall result quota interval is a recall result quota interval whose screening result distribution density is greater than or equal to the corresponding preset distribution density threshold determined for the first time.
[0129] In addition, considering that in actual applications, there may be a situation where the target recall result quota does not exist in different recall result quota intervals. That is to say, the distribution density of the screening results corresponding to each recall result quota interval is less than the corresponding preset distribution density threshold, which indicates that the distribution density of the screening results in each recall result quota interval is very sparse, and the minimum recall result quota in the different recall result quota intervals can be determined as the target recall result quota. Therefore, in a possible implementation of the present application, the quota adjustment method for the recall result may also include: if there is no target recall result quota interval in different recall result quota intervals, determining the minimum recall result quota interval from the different recall result quota intervals; from the different recall result quotas, determining the minimum recall result quota corresponding to the minimum recall result quota interval as the target recall result quota.
[0130] As an example, the preset distribution density threshold can be expressed as threshold, and the four preset distribution density thresholds corresponding to the four recall result quota intervals in the above example can be expressed as threshold[80%, 85%], threshold[85%, 90%], threshold[90%, 95%] and threshold[95%, 100%]. Based on the above example, it can be concluded that the four screening result distribution densities corresponding to the four recall result quota intervals are result_density[80%, 85%] is 5, result_density[85%, 90%] is 0.71, result_density[90%, 95%] is 2.5, and result_density[95%, 100%] is 0.07. Therefore, the specific process of determining the target recall result quota from different recall result quotas can be shown in the following steps 1 to 6:
[0131] Step 1: If result_density[95%, 100%] is greater than or equal to threshold[95%, 100%], the recall result quota interval [30, 100] corresponding to result_density[95%, 100%] is determined as the target recall result quota interval, and step 5 is executed; otherwise, step 2 is executed;
[0132] Step 2: If result_density[90%, 95%] is greater than or equal to threshold[90%, 95%], the recall result quota interval [28, 30] corresponding to result_density[95%, 100%] is determined as the target recall result quota interval, and step 5 is executed; otherwise, step 3 is executed;
[0133] Step 3: If result_density[85%, 90%] is greater than or equal to threshold[85%, 90%], the recall result quota interval [21, 28] corresponding to result_density[85%, 90%] is determined as the target recall result quota interval, and step 5 is executed; otherwise, step 4 is executed;
[0134] Step 4: If result_density[80%, 85%] is greater than or equal to threshold[80%, 85%], the recall result quota interval [20, 21] corresponding to result_density[80%, 85%] is determined as the target recall result quota interval, and step 5 is executed; otherwise, step 6 is executed;
[0135] Step 5: From different recall result quotas, determine the maximum recall result quota corresponding to the target recall result quota range as the target recall result quota;
[0136] Step 6: From different recall result quotas, determine the minimum recall result quota corresponding to the recall result quota interval [20, 21] as the target recall result quota.
[0137] Specifically, if threshold[80%, 85%], threshold[85%, 90%], threshold[90%, 95%] and threshold[95%, 100%] are set to 2.5, 2, 1.5 and 1 respectively, and then steps 1 to 6 in the above steps are executed, [28, 30] can be determined as the target recall result quota interval, and 30 can be determined as the target recall result quota.
[0138] S706: If the target recall result quota is inconsistent with the recall result quota corresponding to the recall result set, the current recall result quota of the target recall branch is adjusted to the target recall result quota.
[0139] When the terminal device determines that the target recall result quota is inconsistent with the current recall result quota, the recall result quota of the target recall branch can be adjusted. For example, if the recall result quota corresponding to the recall result set is 100 and the determined target recall result quota is 30, the recall quota of the target recall branch can be adjusted from 100 to 30.
[0140] In addition, in a possible implementation of the present application, the preset distribution density threshold can be determined by the following steps: according to the click-through rate of multiple screening results in the first screening result set of the target recall branch in the current cycle, the preset distribution density threshold is determined. The multiple screening results in the first screening result set include results that need to be displayed to the user, and the preset distribution density threshold can be determined based on the click-through rate of the results displayed to the user. For example, if the click-through rate of the result is high, it indicates that the recommendation effect or search effect of the result obtained through the target recall branch is good, then more screening results need to be retained for the target recall branch in the rough sorting module. Therefore, the preset distribution density can be set to a smaller value in order to obtain a target recall result quota with a larger value. Among them, the click-through rate of the result can be calculated by the user's interactive behavior with the result. For example, interactive behavior can include article reading behavior, video playback behavior, or collection behavior, etc., which is not limited in the present application.
[0141] It should be noted that the embodiments of this application involve operations related to obtaining user personal information such as interactive behaviors. When applying them to specific products or technologies, it is necessary to obtain the informed consent or separate consent of the subject of personal information. And within the scope of authorization of laws and regulations and the subject of personal information, the subsequent data collection, use and processing of user personal information shall be carried out.
[0142] As an example, when a user registers a personal account in a certain browser application, a usage statement page for interactive behavior data will pop up. This page includes a prompt to collect the behavior data generated by the user's use of this browser application, and the user can choose to agree or disagree. It should be emphasized that if the user chooses not to agree, it will not affect the user's normal use of the browser application.
[0143] Next, taking the terminal device as the execution subject, a result display method based on content request provided by an embodiment of the present application is described in detail in combination with the accompanying drawings.
[0144] See also Figure 8 , which is a flowchart of a result display method based on content request provided in an embodiment of the present application.
[0145] like Figure 8 As shown, the result display method based on content request specifically includes:
[0146] S801: Based on the content request, the recall strategies of different recall branches and the recall result quotas of different recall branches, the recall result sets of different recall branches are obtained by screening from the candidate result sets.
[0147] In the embodiments of the present application, a content request refers to a request corresponding to a user's operation. For example, a content request may include a search request corresponding to a user's search operation in multiple applications such as a browser or a video playback platform, or a recommendation request generated by multiple applications in response to the user's operation when the user operates to open multiple applications, and the present application does not limit this.
[0148] It should be noted that the recall result quota corresponding to at least one recall branch among different recall branches is obtained according to a quota adjustment method for recall results described in the aforementioned embodiments.
[0149] S802: Filter the recall result sets of different recall branches to obtain a rough ranking result set, and filter the rough ranking result set to obtain a fine ranking result set.
[0150] According to the rough sorting model of the rough sorting module in the recall system, the recall results in the recall result sets of different recall branches are screened to obtain a rough sorting result set including multiple rough sorting results. Then, according to the fine sorting model of the fine sorting module in the recall system, the multiple rough sorting results in the rough sorting result set are screened to obtain a fine sorting result set.
[0151] S803: Scatter the refined ranking results in the refined ranking result set to obtain results corresponding to the content request.
[0152] In the breakup module of the recall system, the refined ranking results in the refined ranking result set are broken up to obtain the results corresponding to the content request. The results corresponding to the content request may include: the search results corresponding to the search request obtained by the terminal device in response to the user's search operation; the recommendation results corresponding to the recommendation request obtained by the terminal device in response to the user's start operation, which are not limited in this application.
[0153] S804: Display the result corresponding to the content request.
[0154] The results corresponding to the content request are displayed to the user to satisfy the user's content request.
[0155] As an example, see Fig. 9 , which is a schematic diagram of a page showing the result corresponding to a content request provided by an embodiment of the present application. Fig. 9 As shown in FIG. 1 , the content request is specifically a search request "delicious restaurants in xx shopping mall" entered by the user in the search box of a browser application. The recall system in the terminal device screens the search request layer by layer and finally obtains the search result corresponding to the user's search request. Fig. 9 As shown, the page of the browser application displays multiple search results corresponding to the search request "delicious restaurants in xx shopping mall" to the user, including "Haoyouji Restaurant", "Sichuan Restaurant", "Roast Duck Restaurant" and so on, so that the user can view them.
[0156] It can be seen that in the result display method based on content request provided in the embodiment of the present application, among the multiple recall branches of the recall module, the recall result quota of at least one recall branch can be automatically adjusted, which can reduce the computing resources occupied by the recall module, thereby saving the machine cost used by the recall system.
[0157] Based on the quota adjustment method of the recall result provided in the above embodiment, the present application also provides a quota adjustment device of the recall result. The quota adjustment device of the recall result provided in the embodiment of the present application is specifically introduced below.
[0158] See also Fig.10 , which is a schematic diagram of the structure of a quota adjustment device for recall results provided in an embodiment of the present application. Fig.10 As shown, the quota adjustment device 100 for the recall result may specifically include:
[0159] The result set acquisition module 110 is used to acquire the recall result set of the target recall branch in the current cycle and the first screening result set of the current cycle; the recall result set is obtained in the candidate result set based on the recall strategy of the target recall branch and the current recall result quota of the target recall branch; the first screening result set includes multiple screening results further screened from the recall result set;
[0160] The retention rate acquisition module 120 is used to obtain the screening result retention rate corresponding to configuring different recall result quotas for the target recall branch based on the distribution information of the multiple screening results in the recall result set; the screening result retention rate is used to indicate the proportion of the results still selected into the second screening result set among the multiple screening results; the second screening result set is the screening result set obtained by configuring different recall result quotas for the target recall branch;
[0161] A quota determination module 130, configured to determine a target recall result quota from different recall result quotas based on the screening result retention rates corresponding to the different recall result quotas;
[0162] The quota adjustment module 140 is configured to adjust the current recall result quota of the target recall branch to the target recall result quota if the target recall result quota is inconsistent with the recall result quota corresponding to the recall result set.
[0163] As an implementation manner, the retention rate acquisition module 120 may specifically include:
[0164] A result quantity acquisition unit, configured to obtain the number of screening results of the second screening result set obtained by configuring different recall result quotas for the target recall branch based on the distribution information of the multiple screening results in the recall result set;
[0165] The result retention rate acquisition unit is used to obtain the screening result retention rates corresponding to different recall result quotas according to the ratio of the number of screening results in the second screening result set to the number of screening results in the first screening result set.
[0166] As an implementation manner, the quota determination module 130 may specifically include:
[0167] A first quota determination unit, configured to determine a first recall result quota from different recall result quotas; a screening result retention rate corresponding to the first recall result quota is greater than or equal to a preset retention rate;
[0168] a second quota determination unit, configured to determine the first recall result quota as the target recall result quota if there is a first recall result quota among the different recall result quotas;
[0169] The third quota determination unit is configured to determine the first recall result quota with the smallest value among the multiple first recall result quotas as the target recall result quota if there are multiple first recall result quotas in the different recall result quotas.
[0170] As an implementation manner, the quota determination module 130 may specifically include:
[0171] A quota interval determination unit, configured to determine the recall result quota intervals corresponding to different retention rate intervals based on the screening result retention rates corresponding to different recall result quotas; the different retention rate intervals are preset different screening result retention rate intervals; the recall result quota intervals include the recall result quotas corresponding to the different retention rate intervals;
[0172] A distribution density determination unit, used to determine the distribution density of screening results corresponding to different recall result quota intervals based on the recall result quota intervals corresponding to different retention rate intervals and the different retention rate intervals; the screening result distribution density is the distribution density of the screening results corresponding to different recall result quota intervals in the recall result set;
[0173] The fourth quota determination unit is used to determine the target recall result quota from different recall result quotas according to the comparison result of the screening result distribution density corresponding to different recall result quota intervals and the preset distribution density threshold corresponding to the same recall result quota interval.
[0174] As an implementation manner, the fourth quota determination unit may specifically include:
[0175] A distribution density comparison subunit, for comparing the screening result distribution densities corresponding to different recall result quota intervals with a preset distribution density threshold corresponding to the same recall result quota interval based on the interval endpoints of different recall result quota intervals in descending order;
[0176] The first quota determination subunit is used to determine the maximum recall result quota corresponding to the target recall result quota interval as the target recall result quota from the different recall result quotas if there is a target recall result quota interval in different recall result quota intervals; the target recall result quota interval is a recall result quota interval in which the distribution density of the screening results determined for the first time is greater than or equal to the corresponding preset distribution density threshold.
[0177] As an implementation manner, the fourth quota determination unit may further include:
[0178] A quota interval determination subunit, used for determining a minimum recall result quota interval from among different recall result quota intervals if the target recall result quota interval does not exist among the different recall result quota intervals;
[0179] The second quota determination subunit is used to determine the minimum recall result quota corresponding to the minimum recall result quota range as the target recall result quota from different recall result quotas.
[0180] As an implementation manner, the distribution density determination unit may be specifically used for:
[0181] The ratio of the interval endpoint difference of different retention rate intervals to the interval endpoint difference of the recall result quota interval corresponding to the same retention rate interval is determined as the screening result distribution density corresponding to the different recall result quota intervals.
[0182] As an implementation manner, the preset distribution density threshold in the quota determination module 130 may be determined by the following units:
[0183] The threshold determination unit is used to determine a preset distribution density threshold according to the click rate of multiple screening results of the target recall branch in the first screening result set of the current cycle.
[0184] As an implementation manner, the distribution information of the multiple screening results in the retention rate acquisition module 120 in the recall result set can be specifically determined by the following units:
[0185] a tag adding unit, configured to add a first tag to a plurality of recall results in the recall result set that are filtered into the first filtering result set, and to add a second tag to a plurality of recall results in the recall result set that are not filtered into the first filtering result set;
[0186] The distribution information determining unit is used to determine the distribution information of multiple screening results in the recall result set based on the position information of the first mark and the second mark in the recall result set.
[0187] Based on the result display method based on content request provided in the above embodiment, the present application also provides a result display device based on content request. The result display device based on content request provided in the embodiment of the present application is specifically introduced below.
[0188] See also Fig.11 , which is a schematic diagram of the structure of a result display device based on content request provided by an embodiment of the present application. Fig.11 As shown, the result display device 1100 based on content request specifically includes:
[0189] The recall result set screening module 1101 is used to screen the recall result sets of different recall branches from the candidate result set based on the content request, the recall strategies of different recall branches and the recall result quotas of different recall branches; at least one recall result quota of different recall branches is obtained according to a recall result quota adjustment method described in each of the above embodiments;
[0190] A refined sorting result set screening module 1102 is used to screen the recall result sets of different recall branches to obtain a rough sorting result set, and to screen the rough sorting result set to obtain a refined sorting result set;
[0191] The result acquisition module 1103 is used to break up the refined ranking results in the refined ranking result set to obtain the results corresponding to the content request;
[0192] The result display module 1104 is used to display the result corresponding to the content request.
[0193] An embodiment of the present application provides a computer device, which may be a server. Fig.12 : is a schematic diagram of a server structure provided in an embodiment of the present application. The server 900 may have relatively large differences due to different configurations or performances, and may include one or more central processing units (CPU) 922 (for example, one or more processors) and memory 932, and one or more storage media 930 (for example, one or more mass storage devices) storing application programs 942 or data 944. Among them, the memory 932 and the storage medium 930 can be short-term storage or permanent storage. The program stored in the storage medium 930 may include one or more modules (not shown in the figure), and each module may include a series of instruction operations on the server. Furthermore, the central processing unit 922 can be configured to communicate with the storage medium 930 to execute a series of instruction operations in the storage medium 930 on the server 900.
[0194] The server 900 may also include one or more power supplies 926, one or more wired or wireless network interfaces 950, one or more input and output interfaces 958, and / or one or more operating systems 941, such as Windows Server 2000. TM , Mac OS X TM , Unix TM ,Linux TM , FreeBSD TM etc.
[0195] The CPU 922 is used to execute the following steps:
[0196] Obtaining the recall result set of the target recall branch in the current cycle and the first screening result set of the current cycle; the recall result set is obtained in the candidate result set based on the recall strategy of the target recall branch and the current recall result quota of the target recall branch; the first screening result set includes multiple screening results further screened from the recall result set;
[0197] Based on the distribution information of multiple screening results in the recall result set, the screening result retention rate corresponding to different recall result quotas configured for the target recall branch is obtained; the screening result retention rate is used to indicate the proportion of the results still selected into the second screening result set among the multiple screening results; the second screening result set is the screening result set obtained by configuring different recall result quotas for the target recall branch;
[0198] Based on the screening result retention rates corresponding to different recall result quotas, determine the target recall result quota from the different recall result quotas;
[0199] If the target recall result quota is inconsistent with the recall result quota corresponding to the recall result set, the current recall result quota of the target recall branch is adjusted to the target recall result quota.
[0200] Alternatively, the CPU 922 may also be configured to execute the following steps:
[0201] Based on the content request, the recall strategies of different recall branches and the recall result quotas of different recall branches, the recall result sets of different recall branches are obtained by screening from the candidate result sets; at least one recall result quota of different recall branches is obtained according to a recall result quota adjustment method described in each of the above embodiments;
[0202] A rough ranking result set is obtained by filtering the recall result sets of different recall branches, and a fine ranking result set is obtained by filtering the rough ranking result set;
[0203] Scatter the refined ranking results in the refined ranking result set to obtain the results corresponding to the content request;
[0204] Displays the results corresponding to the content request.
[0205] The present application embodiment also provides another computer device, which may be a terminal device. Fig.13 For the sake of convenience, only the parts related to the embodiments of the present application are shown. For specific technical details not disclosed, please refer to the method part of the embodiments of the present application. Take the terminal device as a mobile phone as an example:
[0206] Fig.13 The block diagram shows a partial structure of a mobile phone provided in an embodiment of the present application. Fig.13 The mobile phone includes: a radio frequency (RF) circuit 1010, a memory 1020, an input unit 1030, a display unit 1040, a sensor 1050, an audio circuit 1060, a wireless fidelity (WiFi) module 1070, a processor 1080, and a power supply 1090. Those skilled in the art can understand that Fig.13 The mobile phone structure shown in the figure does not constitute a limitation on the mobile phone, and may include more or fewer components than shown in the figure, or combine certain components, or arrange the components differently.
[0207] Combine the following Fig.13 A detailed introduction to the various components of the mobile phone:
[0208] The RF circuit 1010 can be used for receiving and sending signals during information transmission or calls. In particular, after receiving the downlink information of the base station, it is sent to the processor 1080 for processing; in addition, the designed uplink data is sent to the base station. Generally, the RF circuit 1010 includes but is not limited to an antenna, at least one amplifier, a transceiver, a coupler, a low noise amplifier (full name: LowNoiseAmplifier, English abbreviation: LNA), a duplexer, etc. In addition, the RF circuit 1010 can also communicate with the network and other devices through wireless communication. The above-mentioned wireless communications may use any communication standard or protocol, including but not limited to Global System of Mobile communications (Global System of Mobile communication, English abbreviation: GSM), General Packet Radio Service (General Packet Radio Service, GPRS), Code Division Multiple Access (Code Division Multiple Access, English abbreviation: CDMA), Wideband Code Division Multiple Access (WCDMA), Long Term Evolution (Long Term Evolution, English abbreviation: LTE), e-mail, Short Messaging Service (SMS), etc.
[0209] The memory 1020 can be used to store software programs and modules. The processor 1080 executes various functional applications and data processing of the mobile phone by running the software programs and modules stored in the memory 1020. The memory 1020 can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, an application required for at least one function (such as a sound playback function, an image playback function, etc.), etc.; the data storage area can store data created according to the use of the mobile phone (such as audio data, a phone book, etc.), etc. In addition, the memory 1020 can include a high-speed random access memory, and can also include a non-volatile memory, such as at least one disk storage device, a flash memory device, or other volatile solid-state storage devices.
[0210] The input unit 1030 can be used to receive input digital or character information, and to generate key signal input related to the user settings and function control of the mobile phone. Specifically, the input unit 1030 may include a touch panel 1031 and other input devices 1032. The touch panel 1031, also known as a touch screen, can collect the user's touch operation on or near it (such as the user's operation on the touch panel 1031 or near the touch panel 1031 using any suitable object or accessory such as a finger, stylus, etc.), and drive the corresponding connection device according to a pre-set program. Optionally, the touch panel 1031 may include two parts: a touch detection device and a touch controller. Among them, the touch detection device detects the user's touch orientation, detects the signal brought by the touch operation, and transmits the signal to the touch controller; the touch controller receives the touch information from the touch detection device, converts it into contact coordinates, and then sends it to the processor 1080, and can receive and execute commands sent by the processor 1080. In addition, the touch panel 1031 can be implemented in various types such as resistive, capacitive, infrared, and surface acoustic waves. In addition to the touch panel 1031, the input unit 1030 may further include other input devices 1032. Specifically, the other input devices 1032 may include but are not limited to one or more of a physical keyboard, function keys (such as volume control keys, switch keys, etc.), a trackball, a mouse, a joystick, and the like.
[0211] The display unit 1040 can be used to display information input by the user or information provided to the user and various menus of the mobile phone. The display unit 1040 may include a display panel 1041. Optionally, the display panel 1041 may be configured in the form of a liquid crystal display (full name in English: Liquid Crystal Display, English abbreviation: LCD), an organic light-emitting diode (full name in English: Organic Light-Emitting Diode, English abbreviation: OLED), etc. Further, the touch panel 1031 may cover the display panel 1041. When the touch panel 1031 detects a touch operation on or near it, it is transmitted to the processor 1080 to determine the type of touch event. Subsequently, the processor 1080 provides a corresponding visual output on the display panel 1041 according to the type of touch event. Although in Fig.13 In the embodiment, the touch panel 1031 and the display panel 1041 are used as two independent components to realize the input and output functions of the mobile phone, but in some embodiments, the touch panel 1031 and the display panel 1041 can be integrated to realize the input and output functions of the mobile phone.
[0212] The mobile phone may also include at least one sensor 1050, such as a light sensor, a motion sensor, and other sensors. Specifically, the light sensor may include an ambient light sensor and a proximity sensor, wherein the ambient light sensor may adjust the brightness of the display panel 1041 according to the brightness of the ambient light, and the proximity sensor may turn off the display panel 1041 and / or the backlight when the mobile phone is moved to the ear. As a type of motion sensor, the accelerometer sensor can detect the magnitude of acceleration in all directions (generally three axes), and can detect the magnitude and direction of gravity when stationary. It can be used for applications that identify the posture of the mobile phone (such as horizontal and vertical screen switching, related games, magnetometer posture calibration), vibration recognition related functions (such as pedometer, tapping), etc.; as for other sensors that can be configured in the mobile phone, such as gyroscopes, barometers, hygrometers, thermometers, infrared sensors, etc., they will not be repeated here.
[0213] The audio circuit 1060, the speaker 1061, and the microphone 1062 can provide an audio interface between the user and the mobile phone. The audio circuit 1060 can transmit the received audio data to the speaker 1061 after converting the received audio data into an electrical signal, which is converted into a sound signal for output; on the other hand, the microphone 1062 converts the collected sound signal into an electrical signal, which is received by the audio circuit 1060 and converted into audio data, and then the audio data is output to the processor 1080 for processing, and then sent to another mobile phone through the RF circuit 1010, or the audio data is output to the memory 1020 for further processing.
[0214] WiFi is a short-range wireless transmission technology. The mobile phone can help users send and receive emails, browse web pages and access streaming media through the WiFi module 1070. It provides users with wireless broadband Internet access. Fig.13 A WiFi module 1070 is shown, but it is understandable that it is not an essential component of the mobile phone and can be omitted as needed without changing the essence of the invention.
[0215] The processor 1080 is the control center of the mobile phone. It uses various interfaces and lines to connect various parts of the entire mobile phone. By running or executing software programs and / or modules stored in the memory 1020, and calling data stored in the memory 1020, it executes various functions of the mobile phone and processes data, thereby collecting overall data and information of the mobile phone. Optionally, the processor 1080 may include one or more processing units; preferably, the processor 1080 may integrate an application processor and a modem processor, wherein the application processor mainly processes the operating system, user interface, and application programs, and the modem processor mainly processes wireless communications. It is understandable that the above-mentioned modem processor may not be integrated into the processor 1080.
[0216] The mobile phone also includes a power supply 1090 (such as a battery) for supplying power to various components. Preferably, the power supply can be logically connected to the processor 1080 through a power management system, so that the power management system can manage functions such as charging, discharging, and power consumption.
[0217] Although not shown, the mobile phone may also include a camera, a Bluetooth module, etc., which will not be described in detail here.
[0218] In the embodiment of the present application, the processor 1080 included in the mobile phone also has the following functions:
[0219] Obtaining the recall result set of the target recall branch in the current cycle and the first screening result set of the current cycle; the recall result set is obtained in the candidate result set based on the recall strategy of the target recall branch and the current recall result quota of the target recall branch; the first screening result set includes multiple screening results further screened from the recall result set;
[0220] Based on the distribution information of multiple screening results in the recall result set, the screening result retention rate corresponding to different recall result quotas configured for the target recall branch is obtained; the screening result retention rate is used to indicate the proportion of the results still selected into the second screening result set among the multiple screening results; the second screening result set is the screening result set obtained by configuring different recall result quotas for the target recall branch;
[0221] Based on the screening result retention rates corresponding to different recall result quotas, determine the target recall result quota from the different recall result quotas;
[0222] If the target recall result quota is inconsistent with the recall result quota corresponding to the recall result set, the current recall result quota of the target recall branch is adjusted to the target recall result quota.
[0223] Alternatively, the processor 1080 included in the terminal mobile phone may also have the following functions:
[0224] Based on the content request, the recall strategies of different recall branches and the recall result quotas of different recall branches, the recall result sets of different recall branches are obtained by screening from the candidate result sets; at least one recall result quota of different recall branches is obtained according to a recall result quota adjustment method described in each of the above embodiments;
[0225] A rough ranking result set is obtained by filtering the recall result sets of different recall branches, and a fine ranking result set is obtained by filtering the rough ranking result set;
[0226] Scatter the refined ranking results in the refined ranking result set to obtain the results corresponding to the content request;
[0227] Displays the results corresponding to the content request.
[0228] An embodiment of the present application also provides a computer-readable storage medium for storing a computer program. When the computer program is executed on a computer device, the computer device executes any implementation of a quota adjustment method for recall results described in the aforementioned embodiments, or executes any implementation of a result display method based on content request.
[0229] An embodiment of the present application also provides a computer program product including a computer program. When the computer program product is run on a computer device, the computer device executes any implementation of a quota adjustment method for recall results described in the aforementioned embodiments, or executes any implementation of a result display method based on content request.
[0230] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems and devices described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.
[0231] In the several embodiments provided in the present application, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the system embodiments described above are only schematic. For example, the division of the system is only a logical function division. There may be other division methods in actual implementation, such as multiple systems can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or units, which can be electrical, mechanical or other forms.
[0232] The systems described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
[0233] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.
[0234] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (full name in English: Read-Only Memory, English abbreviation: ROM), random access memory (full name in English: Random Access Memory, English abbreviation: RAM), disk or optical disk and other media that can store computer programs.
[0235] The above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.
Claims
1. A quota adjustment method for recall results, characterized in that: include: Obtaining a recall result set of a target recall branch in a current cycle and a first screening result set of the current cycle; The recall result set is obtained in the candidate result set based on the recall strategy of the target recall branch and the current recall result quota of the target recall branch; The first screening result set includes a plurality of screening results further screened from the recall result set; Based on the distribution information of the multiple screening results in the recall result set, obtaining the screening result retention rates corresponding to configuring different recall result quotas for the target recall branch; The screening result retention rate is used to indicate the proportion of the results among the multiple screening results that are still selected into the second screening result set; The second screening result set is a screening result set obtained by configuring different recall result quotas for the target recall branch; Based on the screening result retention rates corresponding to the different recall result quotas, determining a target recall result quota from the different recall result quotas; If the target recall result quota is inconsistent with the recall result quota corresponding to the recall result set, the current recall result quota of the target recall branch is adjusted to the target recall result quota.
2. The method according to claim 1, characterized in that The obtaining, based on the distribution information of the multiple screening results in the recall result set, the screening result retention rate corresponding to configuring different recall result quotas for the target recall branch specifically includes: Based on the distribution information of the multiple screening results in the recall result set, obtaining the number of screening results of the second screening result set obtained by configuring different recall result quotas for the target recall branch; The screening result retention rates corresponding to the different recall result quotas are obtained according to the ratio of the number of screening results in the second screening result set to the number of screening results in the first screening result set.
3. The method according to claim 1 or 2, characterized in that: The determining of a target recall result quota from the different recall result quotas based on the screening result retention rates corresponding to the different recall result quotas specifically includes: Determining a first recall result quota from the different recall result quotas; the screening result retention rate corresponding to the first recall result quota is greater than or equal to a preset retention rate; If there is a first recall result quota among the different recall result quotas, determining the first recall result quota as the target recall result quota; If there are multiple first recall result quotas in the different recall result quotas, the first recall result quota with the smallest value among the multiple first recall result quotas is determined as the target recall result quota.
4. The method according to claim 1 or 2, characterized in that: The determining of a target recall result quota from the different recall result quotas based on the screening result retention rates corresponding to the different recall result quotas specifically includes: Based on the screening result retention rates corresponding to the different recall result quotas, determining the recall result quota intervals corresponding to the different retention rate intervals; the different retention rate intervals are preset different screening result retention rate intervals; the recall result quota intervals include the recall result quotas corresponding to the different retention rate intervals; Based on the recall result quota intervals corresponding to the different retention rate intervals and the different retention rate intervals, determining the screening result distribution density corresponding to the different recall result quota intervals; the screening result distribution density is the distribution density of the screening results corresponding to the different recall result quota intervals in the recall result set; According to the comparison result of the screening result distribution density corresponding to the different recall result quota intervals and the preset distribution density threshold corresponding to the same recall result quota interval, the target recall result quota is determined from the different recall result quotas.
5. The method according to claim 4, characterized in that The step of determining a target recall result quota from the different recall result quotas based on a comparison result of the screening result distribution densities corresponding to the different recall result quota intervals with a preset distribution density threshold corresponding to the same recall result quota interval specifically includes: Based on the order of the interval endpoints of the different recall result quota intervals from large to small, the screening result distribution densities corresponding to the different recall result quota intervals are compared with the preset distribution density thresholds corresponding to the same recall result quota interval; If there is a target recall result quota interval among the different recall result quota intervals, the maximum recall result quota corresponding to the target recall result quota interval is determined as the target recall result quota from the different recall result quotas; the target recall result quota interval is a recall result quota interval in which the screening result distribution density determined for the first time is greater than or equal to the corresponding preset distribution density threshold.
6. The method according to claim 5, characterized in that Also includes: If the target recall result quota interval does not exist in the different recall result quota intervals, determining a minimum recall result quota interval from the different recall result quota intervals; From the different recall result quotas, a minimum recall result quota corresponding to the minimum recall result quota interval is determined as the target recall result quota.
7. The method according to claim 4, characterized in that The determining, based on the recall result quota intervals corresponding to the different retention rate intervals and the different retention rate intervals, the screening result distribution density corresponding to the different recall result quota intervals specifically includes: The ratio of the interval endpoint difference of the different retention rate intervals to the interval endpoint difference of the recall result quota interval corresponding to the same retention rate interval is determined as the screening result distribution density corresponding to the different recall result quota intervals.
8. The method according to claim 4 or 5, characterized in that: The preset distribution density threshold is determined by the following steps: The preset distribution density threshold is determined according to the click rates of the plurality of screening results of the target recall branch in the first screening result set of the current cycle.
9. The method according to claim 1, characterized in that: The distribution information of the multiple screening results in the recall result set is determined by the following steps: adding a first mark to a plurality of recall results in the recall result set that are filtered into the first screening result set, and adding a second mark to a plurality of recall results in the recall result set that are not filtered into the first screening result set; Based on the position information of the first mark and the second mark in the recall result set, distribution information of the multiple screening results in the recall result set is determined.
10. A result display method based on content request, characterized in that: include: Based on the content request, the recall strategies of different recall branches and the recall result quotas of the different recall branches, the recall result sets of the different recall branches are obtained by screening from the candidate result sets; at least one recall result quota of the different recall branches is obtained according to the quota adjustment method of the recall result according to any one of claims 1 to 9; Screening the recall result sets of the different recall branches to obtain a rough sorting result set, and screening the rough sorting result set to obtain a fine sorting result set; Scattering the refined ranking results in the refined ranking result set to obtain the results corresponding to the content request; The result corresponding to the content request is displayed.
11. A quota adjustment device for recall results, characterized in that: include: A result set acquisition module, used to acquire the recall result set of the target recall branch in the current cycle and the first screening result set of the current cycle; The recall result set is obtained in the candidate result set based on the recall strategy of the target recall branch and the current recall result quota of the target recall branch; The first screening result set includes a plurality of screening results further screened from the recall result set; A retention rate acquisition module, configured to obtain, based on distribution information of the multiple screening results in the recall result set, screening result retention rates corresponding to different recall result quotas configured for the target recall branch; The screening result retention rate is used to indicate the proportion of the results among the multiple screening results that are still selected into the second screening result set; The second screening result set is a screening result set obtained by configuring different recall result quotas for the target recall branch; A quota determination module, configured to determine a target recall result quota from the different recall result quotas based on the screening result retention rates corresponding to the different recall result quotas; The quota adjustment module is used to adjust the current recall result quota of the target recall branch to the target recall result quota if the target recall result quota is inconsistent with the recall result quota corresponding to the recall result set.
12. A result display device based on content request, characterized in that: include: A recall result set screening module, used for screening the recall result sets of different recall branches from the candidate result set based on the content request, the recall strategies of different recall branches and the recall result quotas of the different recall branches; at least one recall result quota of the different recall branches is obtained according to the quota adjustment method of the recall result according to any one of claims 1 to 9; A refined sorting result set screening module, used to screen the recall result sets of the different recall branches to obtain a rough sorting result set, and to screen the rough sorting result set to obtain a refined sorting result set; A result acquisition module, used for breaking up the refined ranking results in the refined ranking result set to obtain the result corresponding to the content request; The result display module is used to display the result corresponding to the content request.
13. A computer device, characterized in that: The computer device comprises a processor and a memory: The memory is used to store a computer program and transmit the computer program to the processor; The processor is used to execute the quota adjustment method for recall results described in any one of claims 1 to 9 according to the instructions in the computer program, or to execute the result display method based on content request described in claim 10.
14. A computer-readable storage medium, characterized in that: The computer-readable storage medium is used to store a computer program, and when the computer program is executed by a computer device, it implements the quota adjustment method for recall results described in any one of claims 1 to 9, or implements the result display method based on content request described in claim 10.
15. A computer program product, characterized in that It includes a computer program, which, when executed by a computer device, implements the quota adjustment method for recall results described in any one of claims 1 to 9, or implements the result display method based on content request described in claim 10.
