Method and apparatus for sorting shuffled services again based on similarity, and electronic device
Patent Information
- Application Number
- PCT/CN2024/091188
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- Priority Date
- 2023-05-08
- Filing Date
- 2024-05-06
- Publication Date
- 2024-11-14
Smart Images

Figure CN2024091188_14112024_PF_FP_ABST
Abstract
Description
Similarity-based service reordering method, device, and electronic device Technical Field
[0001] The present invention relates to the field of personalized recommendation technology, and in particular to a similarity-based service reordering method, device, and electronic device. Background Art
[0002] Vector feature information is a floating-point array consisting of 128 floating-point numbers. It is generated by processing reported user behaviors or items through machine learning models. It typically includes item title vector features and user click sequence vector features.
[0003] The architecture of a recommendation system is broadly divided into the following modules: recall, rough ranking, refined ranking, and re-ranking. During the refined ranking phase, the goal is to obtain a globally optimal solution for the candidate ranking queue. However, in practice, the refined ranking model outputs a score for each user and candidate item, and each candidate item influences the others. Therefore, the re-ranking module primarily considers information related to the item styles displayed on the same screen and uses this information to re-rank the refined candidate item queue. Therefore, the ranking model focuses on improving the effectiveness of refined ranking, while the re-ranking strategy handles the rest. Re-ranking algorithms or strategies are applied based on actual business needs to ensure that the ranking results are less rigid, thereby achieving overall optimization. This refined ranking-first, re-ranking approach decouples the ranking model from the specific business operations. Academia focuses on exploring more effective ranking models, while industry designs specific re-ranking strategies based on its own needs. This decoupling of the two improves overall efficiency.
[0004] Summary of the Invention
[0005] The present application provides a method, device and electronic device for reordering services by similarity scattering, so as to solve the above-mentioned technical problem of poor ordering effect in the prior art.
[0006] According to one aspect of the present application, an embodiment provides a method for reordering services by similarity scattering, including:
[0007] Select a base material a1 in the recall pool mix1, and select other materials a2, b1, a3, etc. after the base material a1 as candidate materials, where the mix1 sequence is: [a1, a2, b1, a3, etc.];
[0008] Calculate the vector similarity between the reference material a1 and the candidate material a2 adjacent to the reference material a1; if the similarity between the reference material a1 and the candidate material a2 is within the scattering threshold interval, then select the next candidate material b1 in order to calculate the vector similarity until all the candidate materials in the recall pool mix1 are traversed; if the similarity between the reference material a1 and the candidate material a2 is higher than the upper limit of the scattering threshold interval, then select the next material a3 to compare the similarity with the reference material a1. Material a3 and the reference material a1 are from the same content library. If the similarity between material a3 and the reference material a1 is lower than the scattering threshold interval, then swap the positions of material a3 and material a2 in the recall pool mix1 to form the mix2 sequence: [a1, a3, b1, a2…];
[0009] Select the next material a3 in the mix2 sequence as the reference material, and cyclically calculate the similarity in the mix2 sequence; wherein the cyclic similarity calculation includes calculating the vector similarity between the reference material a3 and the material b1 adjacent to the reference material a3;
[0010] The final sorting result is obtained until the materials in at least one content library in the recall pool mix1 are exhausted.
[0011] In one embodiment, after the positions of the material a3 and the material a2 are swapped, the cycle using the material a1 as the reference material is completed.
[0012] In one embodiment, the similarity calculation is set to cosine similarity calculation.
[0013] In one embodiment, the scattering threshold interval is [75%, 95%).
[0014] In one embodiment, the similarity-breaking business reordering method further includes mixing multiple materials to form a recall pool mix1.
[0015] In one embodiment, the recall pool mix1 comes from at least two original recommended material recall pools; and the content libraries of any two of the original recommended material recall pools are different.
[0016] In one embodiment, the original recommended material recall pool includes a recall pool ar and a recall pool br.
[0017] According to one aspect of the present application, an embodiment provides a similarity-based service reordering device, including:
[0018] The material selection module is used to select a reference material a1 in the recall pool mix1, and select other materials a2, b1, a3... after the reference material a1 as candidate materials, where the mix1 sequence is: [a1, a2, b1, a3...];
[0019] The first processing module is configured to calculate the vector similarity between the reference material a1 and the adjacent candidate material a2 of the reference material a1; if the similarity between the reference material a1 and the candidate material a2 is within the scattering threshold interval, the next candidate material b1 is selected in sequence to calculate the vector similarity until all candidate materials in the recall pool mix1 are traversed; if the similarity between the reference material a1 and the candidate material a2 is higher than the upper limit of the scattering threshold interval, the next material a3 is selected to compare the similarity with the reference material a1, and the material a3 and the reference material a1 are from the same content library. If the similarity between the material a3 and the reference material a1 is lower than the scattering threshold interval, the positions of the materials a3 and a2 in the recall pool mix1 are swapped to form the mix2 sequence: [a1, a3, b1, a2…];
[0020] The second processing module is used to select the next material a3 in the mix2 sequence as the reference material and cyclically calculate the similarity in the mix2 sequence;
[0021] The loop processing module is used for loop calculation of similarity, wherein the loop calculation of similarity includes calculating vector similarity between the reference material a3 and the adjacent material b1 of the reference material a3; until the materials in at least one content library in the recall pool mix1 are exhausted.
[0022] According to one aspect of the present application, an embodiment provides an electronic device, including:
[0023] Memory;
[0024] and
[0025] processor;
[0026] The memory is used to store one or more computer instructions; the one or more computer instructions are executed by the processor to implement any one of the similarity-breaking service reordering methods described above.
[0027] According to one aspect of the present application, an embodiment provides a readable storage medium having computer instructions stored thereon; wherein, when the computer instructions are executed by a processor, the similarity-breaking service reordering method as described above is implemented.
[0028] The technical solution of the above embodiment of the present application can significantly improve the sorting efficiency of various algorithm-recommended indicators by more than 5% based on the title vector information of the recommended materials. Among them, the above indicators include revenue, click-through rate, and ARPU value. After that, the materials can be recommended more targeted and personalized. BRIEF DESCRIPTION OF THE DRAWINGS
[0029] FIG1 is a flowchart of a method for reordering services by similarity scattering in an embodiment;
[0030] FIG2 is a flow chart of a method for quantitatively predicting tidal channel sandstone in an embodiment;
[0031] FIG3 is a schematic structural diagram of a device for reordering services by breaking up similarity in an embodiment. DETAILED DESCRIPTION
[0032] It should be noted that, in the absence of conflict, the embodiments and features of the embodiments in this application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.
[0033] In order to enable those skilled in the art to better understand the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts should fall within the scope of protection of this application.
[0034] It should be noted that the terms "first", "second" etc. in this application are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequential order. It should be understood that the data used in this way can be interchanged in appropriate circumstances, so that the embodiments of the application described here. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, the process, method, system, product or equipment that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or equipment.
[0035] It should be understood that when an element (such as a layer, film, region, or substrate) is described as being "on" another element, the element may be directly on the other element or intervening elements may be present. Moreover, in this application, when an element is described as being "connected" to another element, the element may be "directly connected" to the other element or "connected" to the other element through a third element.
[0036] Example 1
[0037] Referring to FIG1 , an embodiment provides a method for reordering services by similarity scattering, including the following steps:
[0038] S1. Select a base material a1 in the recall pool mix1, and select other materials a2, b1, a3, etc. after the base material a1 as candidate materials, where the mix1 sequence is: [a1, a2, b1, a3, etc.];
[0039] S2. Calculate the vector similarity between the reference material a1 and the candidate material a2 adjacent to the reference material a1; if the similarity between the reference material a1 and the candidate material a2 is within the scattering threshold interval, then select the next candidate material b1 in order to calculate the vector similarity until all candidate materials in the recall pool mix1 are traversed; if the similarity between the reference material a1 and the candidate material a2 is higher than the upper limit of the scattering threshold interval, then select the next material a3 to compare the similarity with the reference material a1. Material a3 and the reference material a1 are from the same content library. If the similarity between material a3 and the reference material a1 is lower than the scattering threshold interval, then swap the positions of material a3 and material a2 in the recall pool mix1 to form the mix2 sequence: [a1, a3, b1, a2...]; at this point, the cycle with material a1 as the reference material is completed; in the above, the scattering threshold interval can be set to [75%, 95%);
[0040] S3. Select the next material a3 in the mix2 sequence as the reference material, and cyclically calculate the similarity in the mix2 sequence; wherein the cyclic similarity calculation includes calculating the vector similarity between the reference material a3 and the material b1 adjacent to the reference material a3;
[0041] S4. Until the materials in at least one content library in the recall pool mix1 are exhausted, the final sorting result is obtained.
[0042] There are many ways to calculate the similarity. In one embodiment, the similarity is calculated by cosine similarity calculation.
[0043] In one embodiment, the similarity-breaking business reordering method further includes mixing multiple materials to form a recall pool mix1.
[0044] In one embodiment, the recall pool mix1 comes from at least two original recommended material recall pools; and the content libraries of any two of the original recommended material recall pools are different.
[0045] In one embodiment, the original recommended material recall pool includes a recall pool ar and a recall pool br. Preferably, the number of materials in the recall pool ar and the recall pool br are equal.
[0046] Example 2
[0047] Referring to FIG. 2 , an embodiment provides a similarity-based business reordering method, which is a similarity-based business reordering strategy based on material title vector information. The specific steps are as follows:
[0048] 1. The original recommended material recall pool has a recall pool (ar): [a1, a2, a3, a4, a5], b recall pool (br): [b1, b2, b3, b4, b5];
[0049] 2. Assume that some materials are selected from recall pool a and recall pool b and mixed to form a new recall pool mix1 to enter the reordering process. Assume that the sequence of the new mixed material recall pool mix1 is: [a1, a2, b1, a3, b2, a4, b3, b4, a5, b5];
[0050] 3. In mix1, calculate the cosine similarity between the vectors of a1 and a2; a1 is called the reference material and a2 is called the candidate material;
[0051] (1) If the calculated similarity between vectors a1 and a2 is within the scattering threshold range (e.g. [75%, 95%)), the candidate material is set as the next-ranked material b1 in mix1; If the vector similarity between a1 and b1 is still within the scattering threshold range, a3 is used as the candidate material; and so on, until mix1 is traversed;
[0052] (2) If the vector similarity between a1 and a2 is higher than the upper limit of the scattering threshold, the next-ranked material a3 is selected from the a recall pool that belongs to the same content library as a1 to compare the similarity with a1; if the similarity between a1 and a3 is lower than the scattering threshold, the positions of a3 and a2 in mix1 are swapped. After the swap, the sequence of mix1 becomes mix2: [a1, a3, b1, a2, b2, a4, b3, b4, a5, b5]
[0053] At this point, the cycle with a1 as the base material is complete; use the next material a3 in mix1 as the base material and continue with step 3, that is, compare a3 and b1 in mix2;
[0054] 4. Until the materials in a certain content library are used up (for example, all the materials in recall pool a are used up), the backup materials will eventually be used to back up the materials. The backup materials are independent of the material pool in the process, and are generally selected by a random algorithm. The backup materials are prepared separately to avoid the situation where the recommendation results are empty and to ensure that the recommendation results must have data.
[0055] Example 3
[0056] Referring to FIG3 , an embodiment provides a similarity-based service reordering device, including:
[0057] 1. Material selection module 10
[0058] The material selection module 10 is used to select a reference material a1 in the recall pool mix1, and select other materials a2, b1, a3, etc. after the reference material a1 as candidate materials, where the mix1 sequence is: [a1, a2, b1, a3, etc.];
[0059] 2. First processing module 20
[0060] The first processing module 20 is used to calculate the vector similarity between the reference material a1 and the candidate material a2 adjacent to the reference material a1; wherein, if the similarity between the reference material a1 and the candidate material a2 is within the scattering threshold interval, the next candidate material b1 is selected in sequence to calculate the vector similarity until all the candidate materials in the recall pool mix1 are traversed; if the similarity between the reference material a1 and the candidate material a2 is higher than the upper limit of the scattering threshold interval, the next material a3 is selected to compare the similarity with the reference material a1, and the material a3 and the reference material a1 are from the same content library. If the similarity between the material a3 and the reference material a1 is lower than the scattering threshold interval, the positions of the materials a3 and a2 in the recall pool mix1 are swapped to form the mix2 sequence: [a1, a3, b1, a2...];
[0061] 3. Second processing module 30
[0062] The second processing module 30 is used to select the next material a3 in the mix2 sequence as the reference material, and cyclically calculate the similarity in the mix2 sequence;
[0063] 4. Loop processing module 40
[0064] The loop processing module 40 is used for loop calculation of similarity, which includes calculating vector similarity between the reference material a3 and the adjacent material b1 of the reference material a3 ; until the materials in at least one content library in the recall pool mix1 are exhausted.
[0065] In one embodiment, the similarity-breaking service reordering device further includes a mixing module 50; the mixing module 50 is configured to combine multiple materials into a recall pool mix1. Generally, the recall pool mix1 is derived from at least two original recommended material recall pools; the content libraries of any two of the original recommended material recall pools are different. Preferably, the original recommended material recall pools include a recall pool ar and a recall pool br.
[0066] Example 4
[0067] Based on the same inventive concept, an embodiment of the present application provides an electronic device, comprising: a memory and a processor; wherein the memory is used to store one or more computer instructions; the one or more computer instructions are executed by the processor according to the similarity scrambling service reordering method described in any one of the above embodiments.
[0068] Example 5
[0069] Based on the same inventive concept, an embodiment of the present application provides a readable storage medium, on which computer instructions are stored; wherein, when the computer instructions are executed by a processor, the similarity-breaking service reordering method described in any one of the above embodiments is implemented.
[0070] The one or more computer instructions mentioned above may be formed into a program.
[0071] The above program can be executed in a processor or stored in a memory (or computer-readable medium), which includes permanent and non-permanent, removable and non-removable media that can implement information storage by any method or technology. Information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassettes, tape, disk storage or other magnetic storage devices, or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer-readable media does not include transitory computer-readable media such as modulated data signals and carrier waves.
[0072] These computer programs can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce computer-implemented processing, so that the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in one or more processes in the flowchart and / or one or more boxes in the block diagram, and corresponding different steps can be implemented through different modules.
[0073] The above description is merely a preferred embodiment of the present application and is not intended to limit the present application. Various modifications and variations are possible for those skilled in the art. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present application shall be included within the scope of protection of the present application.
Claims
1. A similarity-dispersed service reordering method, characterized in that: include: Select a base material a1 in the recall pool mix1, and select other materials a2, b1, a3, etc. after the base material a1 as candidate materials, where the mix1 sequence is: [a1, a2, b1, a3, etc.]; Calculate the vector similarity between the reference material a1 and the candidate material a2 adjacent to the reference material a1; if the similarity between the reference material a1 and the candidate material a2 is within the scattering threshold interval, then select the next candidate material b1 in order to calculate the vector similarity until all the candidate materials in the recall pool mix1 are traversed; if the similarity between the reference material a1 and the candidate material a2 is higher than the upper limit of the scattering threshold interval, then select the next material a3 to compare the similarity with the reference material a1, and the material a3 and the reference material a1 are from the same content library. If the similarity between the material a3 and the reference material a1 is lower than the scattering threshold interval, then swap the positions of the material a3 and the material a2 in the recall pool mix1 to form the mix2 sequence: [a1, a3, b1, a2…]; Select the next material a3 in the mix2 sequence as the reference material, and cyclically calculate the similarity in the mix2 sequence; wherein the cyclic calculation of the similarity includes calculating the vector similarity between the reference material a3 and the material b1 adjacent to the reference material a3; Until the materials in at least one content library in the recall pool mix1 are exhausted, the final sorting result is obtained.
2. A similarity-dispersed service reordering method according to claim 1, characterized in that: After the positions of the material a3 and the material a2 are swapped, the cycle with the material a1 as the reference material is completed.
3. A similarity-dispersed service reordering method according to claim 1, characterized in that: The similarity calculation is set to be cosine similarity calculation.
4. A similarity-dispersed service reordering method according to claim 1, characterized in that: The scattering threshold interval is [75%, 95%).
5. A similarity-dispersed service reordering method according to any one of claims 1 to 4, characterized in that: It also includes mixing multiple materials into a recall pool mix1.
6. A similarity-dispersed service reordering method according to claim 5, characterized in that: The recall pool mix1 comes from at least two original recommended material recall pools; the content libraries of any two of the original recommended material recall pools are different.
7. A similarity-dispersed service reordering method according to claim 6, characterized in that: The original recommended material recall pool includes recall pool ar and recall pool br.
8. A similarity-dispersed service reordering device, characterized in that: include: The material selection module is used to select a reference material a1 in the recall pool mix1, and select other materials a2, b1, a3... after the reference material a1 as candidate materials, where the mix1 sequence is: [a1, a2, b1, a3...]; The first processing module is used to calculate the vector similarity between the reference material a1 and the candidate material a2 adjacent to the reference material a1; wherein, if the similarity between the reference material a1 and the candidate material a2 is within the scattering threshold interval, the next candidate material b1 is selected in order to calculate the vector similarity until the candidate materials in the recall pool mix1 are traversed; if the similarity between the reference material a1 and the candidate material a2 is higher than the upper limit of the scattering threshold interval, the next-ranked material a3 is selected to compare the similarity with the reference material a1, and the material a3 and the reference material a1 are from the same content library. If the similarity between the material a3 and the reference material a1 is lower than the scattering threshold interval, the positions of the material a3 and the material a2 in the recall pool mix1 are swapped to form mix2. Sequence: [a1, a3, b1, a2…]; The second processing module is used to select the next material a3 in the mix2 sequence as the reference material, and cyclically calculate the similarity in the mix2 sequence; The loop processing module is used for loop calculation of similarity, wherein the loop calculation of similarity includes calculating the vector similarity between the reference material a3 and the material b1 adjacent to the reference material a3; until the materials in at least one content library in the recall pool mix1 are exhausted.
9. An electronic device, characterized in that: include: Memory; and processor; Wherein, the memory is used to store one or more computer instructions; the one or more computer instructions are executed by the processor to implement the similarity scattering service reordering method described in any one of claims 1 to 7.
10. A readable storage medium, characterized in that: The readable storage medium stores computer instructions; wherein, when the computer instructions are executed by the processor, the similarity scattering service reordering method described in any one of claims 1 to 7 is implemented.
Citation Information
Patent Citations
Neighbor vector recall method and device, computer equipment and storage medium
CN111209953A
Video data sorting method and device, computer equipment and storage medium
CN112528071A
Similarity scattering service reordering method and device and electronic equipment
CN116610948A
Multistage feed ranking system with methodology providing recall approximation at scale
US20200410025A1
Cited By
Content recommendation method and device based on dynamic weight and cyclic sorting, terminal and medium
CN120596721A
