Method, apparatus, program product, and electronic device for recommending objects

Abstract

The present disclosure provides a method, device, program product and electronic device for recommending objects, relating to the technical field of computer. The method comprises: determining a to-be-recommended object; determining a candidate recommendation set according to attribute information of the to-be-recommended object; selecting a target recommendation set matched with the to-be-recommended object from the candidate recommendation set; and inserting the to-be-recommended object into the target recommendation set to recommend the to-be-recommended object based on the target recommendation set. The present disclosure inserts an object into an object set by determining a recommendation set matched with the object, thereby improving the accuracy of recommendation to a certain extent.

Description

Technical Field

[0001] Embodiments of this disclosure relate to the field of computer technology, and more specifically, embodiments of this disclosure relate to a method of recommendation, an apparatus of recommendation, a computer program product, and an electronic device. Background Technology

[0002] Currently, when recommending various objects (such as songs, short videos, etc.) in related technologies, the methods generally involve directly analyzing user behavior data related to these objects to recommend suitable objects, or recommending objects based on their own attributes. Summary of the Invention

[0003] However, the object recommendation methods provided in related technologies cannot achieve accurate object recommendations, thus reducing the effectiveness of object recommendation. Furthermore, the computational load of the analysis may be high, affecting processing efficiency.

[0004] In view of this, the present disclosure provides a method, apparatus, computer program product, and electronic device for recommending objects, so as to improve the accuracy of information recommendation to a certain extent.

[0005] According to a first aspect of this disclosure, a method for recommending objects is provided, the method comprising:

[0006] Identify the candidates to be recommended;

[0007] Based on the attribute information of the objects to be recommended, a candidate recommendation set is determined;

[0008] From the candidate recommendation set, select the target recommendation set that matches the object to be recommended;

[0009] The object to be recommended is inserted into the target recommendation set, so that the object to be recommended is recommended based on the target recommendation set.

[0010] In one possible implementation, determining the candidate recommendation set based on the attribute information of the object to be recommended includes:

[0011] Based on the attribute information of the object to be recommended, determine the growth potential value corresponding to the object to be recommended;

[0012] Based on the growth potential value, a target recommendation candidate pool is selected;

[0013] A candidate recommendation set is determined from the target recommendation candidate pool.

[0014] In one possible implementation, selecting the target recommendation candidate pool based on the growth potential value includes:

[0015] If the growth potential value is within a first preset range, then the first type of recommendation candidate pool is selected as the target recommendation candidate pool;

[0016] If the growth potential value is not within the first preset range, then the second type of recommendation candidate pool is selected as the target recommendation candidate pool;

[0017] The first type of recommendation candidate pool and the second type of recommendation candidate pool correspond to different recommendation channels.

[0018] In one possible implementation, if the target recommendation candidate pool is the first type of recommendation candidate pool, determining the candidate recommendation set from the target recommendation candidate pool includes:

[0019] Determine the recommendation growth value of each recommendation set in the target recommendation candidate pool;

[0020] The growth potential value corresponding to the object to be recommended is compared with the recommendation growth value of each recommendation set in the target recommendation candidate pool to obtain multiple first comparison results;

[0021] Based on the multiple first comparison results, a candidate recommendation set is determined.

[0022] In one possible implementation, selecting a target recommendation set that matches the object to be recommended from the candidate recommendation set includes:

[0023] Based on the attribute information of the object to be recommended, determine the multiple dimensional feature information corresponding to the object to be recommended;

[0024] Determine multiple dimensional feature information for each of the candidate recommendation sets;

[0025] The multiple dimensional feature information corresponding to the object to be recommended is matched with the multiple dimensional feature information of each recommendation set in the candidate recommendation set to obtain multiple matching values;

[0026] Based on the multiple matching values ​​and preset filtering conditions, a target recommendation set matching the object to be recommended is determined.

[0027] In one possible implementation, determining the multi-dimensional feature information of each candidate recommendation set includes:

[0028] Perform the following operations on each of the candidate recommendation sets:

[0029] The overall style characteristics of the candidate recommendation set are determined based on the style information of each object in the candidate recommendation set.

[0030] Based on the emotional information of each object in the candidate recommendation set, determine the emotional characteristics of the candidate recommendation set;

[0031] The scene features of the candidate recommendation set are determined based on the scene information of each object in the candidate recommendation set.

[0032] Based on the overall style features, the emotional features, and the scene features, the candidate recommendation set is obtained with multiple dimensions of feature information.

[0033] In one possible implementation, obtaining multi-dimensional feature information of the candidate recommendation set based on the overall style features, the emotion features, and the scene features includes:

[0034] The overall style features, the emotional features, and the scene features are cross-analyzed to obtain multiple hybrid features;

[0035] The overall style features, the emotional features, the scene features, and the multiple mixed features are used as multiple dimensional feature information of the recommendation set.

[0036] In one possible implementation, the cross-analysis of the overall style features, the emotional features, and the scene features yields multiple mixed features, including:

[0037] The overall style features and the emotional features are cross-analyzed to obtain the first hybrid feature;

[0038] The emotional features and the scene features are cross-analyzed to obtain a second hybrid feature;

[0039] The overall style features and the scene features are cross-analyzed to obtain a third hybrid feature;

[0040] Multiple mixed features are obtained based on the first mixed feature, the second mixed feature, and the third mixed feature.

[0041] In one possible implementation, if the target recommendation candidate pool is the second type of recommendation candidate pool, determining the candidate recommendation set from the target recommendation candidate pool includes:

[0042] Determine the traffic growth value of each recommendation set in the target recommendation candidate pool;

[0043] The traffic potential value corresponding to the object to be recommended is compared with the traffic growth value of each recommendation set in the target recommendation candidate pool to obtain multiple second comparison results;

[0044] Based on the multiple second comparison results, a candidate recommendation set is determined.

[0045] In one possible implementation, selecting a target recommendation set that matches the object to be recommended from the candidate recommendation set includes:

[0046] Determine the traffic gap information for each of the candidate recommendation sets;

[0047] Determine the traffic level information of the object to be recommended;

[0048] Based on the traffic gap information, the traffic level information, and the preset matching rules, the target recommendation set for matching the object to be recommended is determined.

[0049] In one possible implementation, the method further includes:

[0050] If the target recommendation set is a test target recommendation set, then the updated growth potential value of the object to be recommended and the updated recommendation growth value of the target recommendation set are determined within the first period; the first period is the recommendation observation period for the object to be recommended.

[0051] When it is determined that the updated growth potential value is within a first preset range and the updated growth potential value is not less than the updated recommended growth value, the updated target recommendation set of the object to be recommended is redefined, so as to recommend the object to be recommended based on the updated target recommendation set.

[0052] In one possible implementation, the method further includes:

[0053] Determine the performance data information of the target recommendation set;

[0054] When it is determined that the performance data of the target recommendation set does not meet the preset performance conditions, the object to be recommended will be removed from the target recommendation set.

[0055] Determine the updated attribute information of the object to be recommended, and determine the updated candidate recommendation set based on the updated attribute information of the object to be recommended;

[0056] From the updated candidate recommendation set, select the updated target recommendation set that matches the object to be recommended;

[0057] The object to be recommended is inserted into the updated target recommendation set, so that the object to be recommended is recommended based on the updated target recommendation set.

[0058] According to a second aspect of this disclosure, an apparatus for recommending a target is provided, the apparatus comprising:

[0059] The first determining unit is used to determine the objects to be recommended.

[0060] The second determining unit is used to determine a candidate recommendation set based on the attribute information of the object to be recommended;

[0061] A filtering unit is used to select a target recommendation set that matches the object to be recommended from the candidate recommendation set;

[0062] The recommendation unit is used to insert the object to be recommended into the target recommendation set, so as to recommend the object to be recommended based on the target recommendation set.

[0063] In one possible implementation, the second determining unit is configured to:

[0064] Based on the attribute information of the object to be recommended, determine the growth potential value corresponding to the object to be recommended;

[0065] Based on the growth potential value, a target recommendation candidate pool is selected;

[0066] A candidate recommendation set is determined from the target recommendation candidate pool.

[0067] In one possible implementation, the second determining unit is configured to:

[0068] If the growth potential value is within a first preset range, then the first type of recommendation candidate pool is selected as the target recommendation candidate pool;

[0069] If the growth potential value is not within the first preset range, then the second type of recommendation candidate pool is selected as the target recommendation candidate pool;

[0070] The first type of recommendation candidate pool and the second type of recommendation candidate pool correspond to different recommendation channels.

[0071] In one possible implementation, the second determining unit is configured to:

[0072] Determine the recommendation growth value of each recommendation set in the target recommendation candidate pool;

[0073] The growth potential value corresponding to the object to be recommended is compared with the recommendation growth value of each recommendation set in the target recommendation candidate pool to obtain multiple first comparison results;

[0074] Based on the multiple first comparison results, a candidate recommendation set is determined.

[0075] In one possible implementation, the filtering unit is used for:

[0076] Based on the attribute information of the object to be recommended, determine the multiple dimensional feature information corresponding to the object to be recommended;

[0077] Determine multiple dimensional feature information for each of the candidate recommendation sets;

[0078] The multiple dimensional feature information corresponding to the object to be recommended is matched with the multiple dimensional feature information of each recommendation set in the candidate recommendation set to obtain multiple matching values;

[0079] Based on the multiple matching values ​​and preset filtering conditions, a target recommendation set matching the object to be recommended is determined.

[0080] In one possible implementation, the filtering unit is used for:

[0081] Perform the following operations on each of the candidate recommendation sets:

[0082] The overall style characteristics of the candidate recommendation set are determined based on the style information of each object in the candidate recommendation set.

[0083] Based on the emotional information of each object in the candidate recommendation set, determine the emotional characteristics of the candidate recommendation set;

[0084] The scene features of the candidate recommendation set are determined based on the scene information of each object in the candidate recommendation set.

[0085] Based on the overall style features, the emotional features, and the scene features, the candidate recommendation set is obtained with multiple dimensions of feature information.

[0086] In one possible implementation, the filtering unit is used for:

[0087] The overall style features, the emotional features, and the scene features are cross-analyzed to obtain multiple hybrid features;

[0088] The overall style features, the emotional features, the scene features, and the multiple mixed features are used as multiple dimensional feature information of the recommendation set.

[0089] In one possible implementation, the filtering unit is used for:

[0090] The overall style features and the emotional features are cross-analyzed to obtain the first hybrid feature;

[0091] The emotional features and the scene features are cross-analyzed to obtain a second hybrid feature;

[0092] The overall style features and the scene features are cross-analyzed to obtain a third hybrid feature;

[0093] Multiple mixed features are obtained based on the first mixed feature, the second mixed feature, and the third mixed feature.

[0094] In one possible implementation, the second determining unit is configured to:

[0095] Determine the traffic growth value of each recommendation set in the target recommendation candidate pool;

[0096] The traffic potential value corresponding to the object to be recommended is compared with the traffic growth value of each recommendation set in the target recommendation candidate pool to obtain multiple second comparison results;

[0097] Based on the multiple second comparison results, a candidate recommendation set is determined.

[0098] In one possible implementation, the filtering unit is used for:

[0099] Determine the traffic gap information for each of the candidate recommendation sets;

[0100] Determine the traffic level information of the object to be recommended;

[0101] Based on the traffic gap information, the traffic level information, and the preset matching rules, the target recommendation set for matching the object to be recommended is determined.

[0102] In one possible implementation, the recommendation unit is further configured to:

[0103] If the target recommendation set is a test target recommendation set, then the updated growth potential value of the object to be recommended and the updated recommendation growth value of the target recommendation set are determined within the first period; the first period is the recommendation observation period for the object to be recommended.

[0104] When it is determined that the updated growth potential value is within a first preset range and the updated growth potential value is not less than the updated recommended growth value, the updated target recommendation set of the object to be recommended is redefined, so as to recommend the object to be recommended based on the updated target recommendation set.

[0105] In one possible implementation, the recommendation unit is further configured to:

[0106] Determine the performance data information of the target recommendation set;

[0107] When it is determined that the performance data of the target recommendation set does not meet the preset performance conditions, the object to be recommended will be removed from the target recommendation set.

[0108] Determine the updated attribute information of the object to be recommended, and determine the updated candidate recommendation set based on the updated attribute information of the object to be recommended;

[0109] From the updated candidate recommendation set, select the updated target recommendation set that matches the object to be recommended;

[0110] The object to be recommended is inserted into the updated target recommendation set, so that the object to be recommended is recommended based on the updated target recommendation set.

[0111] According to a third aspect of this disclosure, a computer program product is provided, comprising a computer program that, when executed by a processor, implements the method of the first aspect described above and possible implementations thereof.

[0112] According to a fourth aspect of this disclosure, an electronic device is provided, comprising: a processor; and a memory for storing executable instructions of the processor; wherein the processor is configured to perform the method of the first aspect and possible implementations thereof by executing the executable instructions.

[0113] The technical solution disclosed herein has the following beneficial effects:

[0114] In this embodiment, a target recommendation object can be determined, and then a candidate recommendation set can be determined based on the attribute information of the target recommendation object. The candidate recommendation set includes multiple objects belonging to the same category as the target recommendation object. Further, a target recommendation set matching the target recommendation object can be selected from the candidate recommendation set, and the target recommendation object can be inserted into the target recommendation set for recommendation. In other words, in this embodiment, a target recommendation set can be obtained by inserting matching target recommendation objects into a carefully designed recommendation set, and then recommending the target recommendation object. This method can match the target recommendation object to a suitable target recommendation set, thereby achieving precise delivery of recommended content through the thematic and emotional resonance of the recommendation set. Therefore, it can improve the accuracy of recommendations to a certain extent. Furthermore, since the recommendation set can strongly combine the user's emotional content with the song content, it has strong dissemination and dissemination value. Therefore, recommending the target recommendation object based on the target recommendation set can improve the dissemination of the target recommendation object to a certain extent, thereby improving the recommendation effect. In addition, since the recommendation is directly based on the target recommendation set, the processing of related data of the target recommendation object is reduced, thereby improving the recommendation efficiency to a certain extent.

[0115] Other features and advantages of this disclosure will be set forth in the following description and will be apparent in part from the description or may be learned by practicing the disclosure. The objects and other advantages of this disclosure may be realized and obtained by means of the structures particularly pointed out in the written description, claims, and drawings. Attached Figure Description

[0116] To more clearly illustrate the technical solutions of the embodiments of this disclosure, the accompanying drawings used in the embodiments of this disclosure will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this disclosure. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0117] Figure 1 This illustration shows a schematic diagram of an application scenario in this exemplary embodiment;

[0118] Figure 2 A flowchart illustrating a method for recommending objects in this exemplary embodiment is shown;

[0119] Figure 3 A flowchart illustrating another method for recommending objects in this exemplary embodiment is shown.

[0120] Figure 4 A schematic diagram of the structure of an apparatus for recommending a candidate in this exemplary embodiment is shown;

[0121] Figure 5 A schematic diagram of the structure of an electronic device in this exemplary embodiment is shown. Detailed Implementation

[0122] To make the objectives, technical solutions, and advantages of this disclosure clearer, the technical solutions in the embodiments of this disclosure will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this disclosure, and not all of them. Based on the embodiments of this disclosure, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this disclosure. Unless otherwise specified, the embodiments and features in the embodiments of this disclosure can be arbitrarily combined with each other. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be performed in a different order than that shown here.

[0123] The term "comprising" and any variations thereof in the specification and claims of this disclosure are intended to cover non-exclusive protection. For example, a process, method, system, product, or apparatus that comprises a series of steps or units is not limited to the steps or units listed, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to such process, method, product, or apparatus.

[0124] In this disclosure, there are one or more embodiments; "multiple" refers to two or more. "And / or" describes the relationship between the associated objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, or B alone, where A and B can be singular or plural. The character " / " generally indicates that the preceding and following associated objects are in an "or" relationship. "At least one of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one of a, b, or c can represent: a, b, c, ab, ac, bc, or abc, where a, b, and c can be single or multiple.

[0125] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this disclosure are used to distinguish similar objects and are not necessarily used to describe a specific order, sequence, size, or priority. For example, the first comparison result and the second comparison result in the embodiments of this disclosure are merely used to distinguish different comparison results. It should be understood that such data can be interchanged where appropriate so that the embodiments of this disclosure described herein can be implemented in orders other than those illustrated or described herein. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this disclosure. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this disclosure as detailed in the appended claims.

[0126] The exemplary embodiments of this disclosure are described below with reference to the accompanying drawings, which are schematic illustrations of this disclosure and are not necessarily drawn to scale. Some block diagrams shown in the drawings may be functional entities and do not necessarily correspond to physically or logically independent entities. These functional entities may be implemented in software, in hardware modules or integrated circuits, or in networks, processors, or microcontrollers. Implementations can be carried out in various forms and should not be construed as limited to the examples set forth herein. The features, structures, or characteristics described in this disclosure can be combined in any suitable manner in one or more implementations. Numerous specific details are provided in the following description to give a thorough description of the embodiments of this disclosure. However, those skilled in the art will recognize that one or more specific details may be omitted when implementing the technical solutions of this disclosure, or other methods, components, devices, steps, etc., may be used to replace one or more specific details. It should be noted that in the embodiments of this disclosure, the collection, dissemination, use, and display of data all comply with relevant national laws and regulations. Invention Overview

[0128] Currently, when recommending various objects (such as songs, short videos, publications, etc.) in related technologies, the methods generally involve directly analyzing user behavior data related to these objects to recommend suitable objects, or recommending objects based on their own attributes.

[0129] For example, by analyzing a user's listening history, favorites, likes, shares, skips, and other behaviors, the recommendation system can infer the user's music preferences and recommend similar or related songs. Alternatively, it can recommend songs based on their genre.

[0130] However, the aforementioned method of recommending objects may consistently recommend the same type of object to the user or present a chaotic mix of objects, leading to decreased user interest and negatively impacting the recommendation effectiveness. Furthermore, it requires processing large amounts of user behavior data, potentially resulting in high computational demands and reduced processing efficiency.

[0131] It is evident that the relevant technologies cannot achieve accurate object recommendations, which reduces the effectiveness of object recommendations and may result in low processing efficiency.

[0132] In view of this, embodiments of this disclosure provide a method for recommending objects. This method determines objects to be recommended, and then determines a candidate recommendation set based on the attribute information of the objects to be recommended. The candidate recommendation set includes multiple objects belonging to the same category as the objects to be recommended. Further, a target recommendation set matching the objects to be recommended can be selected from the candidate recommendation set, allowing the objects to be recommended to be inserted into the target recommendation set for recommendation. In other words, embodiments of this disclosure can obtain a target recommendation set by inserting matching objects into a carefully designed recommendation set, and then recommend the objects to be recommended. This approach matches suitable target recommendation sets to the objects to be recommended, enabling precise delivery of recommended content through the thematic and emotional resonance of the recommendation sets. Therefore, it can improve the accuracy of recommendations to a certain extent. Furthermore, since recommendation sets can strongly combine user emotional content with song content, they have strong dissemination and dissemination value. Therefore, recommending objects to be recommended based on target recommendation sets can, to a certain extent, increase the dissemination of the objects to be recommended, thereby improving the recommendation effect. Furthermore, since the recommendations are made directly based on the target recommendation set, the processing of related data of the recommended objects is reduced, thereby improving the recommendation efficiency to a certain extent.

[0133] After introducing the basic principles of this disclosure, various non-limiting embodiments of this disclosure will be described in detail below.

[0134] Application Scenarios Overview

[0135] To better understand the technical solutions provided in the embodiments of this disclosure, the following is a brief introduction to the application scenarios applicable to the technical solutions provided in the embodiments of this disclosure. It should be noted that the application scenarios described below are only for illustrating the embodiments of this disclosure and are not intended to limit the scope. In specific implementation, the technical solutions provided in the embodiments of this disclosure can be flexibly applied according to actual needs.

[0136] To provide a clearer description of the embodiments of this disclosure, some key terms are introduced below:

[0137] A recommendation set is a collection of multiple objects organized according to specific emotions, themes, styles, or other rules. Examples include playlists, short dramas, funny video collections, music collections from previous years, publication collections, and novel collections. Playlists, in particular, typically combine songs of different styles, moods, or scenarios to provide a specific musical experience and meet the user's listening needs in a particular context. A playlist is more than just a simple list of songs; it carries thematic elements, emotional resonance, and interactivity with the user. Playlists can be categorized in several ways, such as themed playlists. Themed playlists can be understood as playlists designed based on specific themes (such as "Rainy Day Mood Playlist" or "Classic Oldies"), attracting a specific user group through the overall style or theme of the songs.

[0138] Object Insertion System: A technical system that supports the insertion of objects into recommendation sets. It can automatically or manually insert objects to be recommended into appropriate recommendation sets according to certain strategies. For example, it can insert songs into specific recommendation sets according to certain strategies to increase the exposure and traffic of songs among the target user group, thereby improving the recommendation effect of songs.

[0139] Insertion timing: The optimal time to insert an object into the recommendation set at different stages of its lifecycle, such as when to insert a song into a playlist, to ensure that the song gets the maximum exposure and traffic.

[0140] Automated insertion: The algorithm-based automatic recommendation and insertion function automatically inserts objects into appropriate recommendations based on data such as object matching degree and popularity.

[0141] Manual insertion: The platform operator or the creator of the recommendation set manually inserts the object into a specific recommendation set based on the object's emotional, stylistic and other attributes and the target user group.

[0142] Recommended Sets and Recommended Objects: By carefully designing recommended sets and strategically inserting objects, the dissemination and exposure of objects can be promoted.

[0143] Recommendation Set Creation System: A system that allows users or platforms to create recommendation sets and set themes, tags, descriptions, and other features, and supports the management and updating of recommendation set content.

[0144] Personalized recommendation: This method of recommending songs based on users' historical behavior data (such as listening history, favorites, likes, etc.) is commonly found in recommendation algorithms of streaming media platforms.

[0145] Object lifecycle: refers to the entire process from the release of an object to its eventual disappearance from the market, including the stages of object release, promotion, popularity, and decline.

[0146] Song title: This is the most direct identifier of a song, used to distinguish different musical works.

[0147] Creator information: including lyricist, composer, arranger, and information on each instrumentalist.

[0148] Singer: The person who presents a song to the audience. Different singers can give the song different expressiveness and emotional color through their voices, singing styles and so on.

[0149] Release information: This includes the release date and the issuing company. The release date reflects the historical context of the song, while the issuing company is responsible for its promotion and distribution.

[0150] Musical styles include pop, rock, folk, classical, jazz, and electronic music. Musical style can be understood as a classification based on elements such as melody, rhythm, harmony, and instrumentation, which determines the overall musical characteristics and target audience of a song.

[0151] Song duration: refers to the length of time a song takes from start to finish, usually expressed in minutes and seconds.

[0152] Language: The language in which the song is sung, such as Chinese, English, Japanese, etc. The choice of language will affect the song's reach and the audience's level of understanding.

[0153] Rhythm: is the regular combination of the length and intensity of notes in a song. It can create different rhythms and emotional atmospheres, such as cheerful rhythms, soothing rhythms, and dynamic rhythms.

[0154] Melody: A line composed of a series of notes, it is the part of a song that is most easily remembered by the listener and can express the emotions and themes of the song.

[0155] Harmony: refers to the sound effect created by multiple notes being sung simultaneously. It can enhance the richness and layering of a song, making the music fuller.

[0156] Song theme: The core idea or emotion expressed in a song, such as love, friendship, family affection, inspiration, homesickness, etc., is the soul of the song.

[0157] Copyright information: Information regarding the ownership and licensing of copyrights for songs, short videos, publications (such as novels), etc.

[0158] Play count: Records the number of times a song, short video, novel, or other content has been played.

[0159] Downloads: This reflects the number of times users download songs, short videos, publications (such as novels), etc., to their local devices, indicating their level of liking for these items and their desire to listen to them offline.

[0160] Comment volume: The number of comments posted by users on songs, short videos, publications (such as novels), etc. The content of the comments can reflect the listeners' feelings, understanding, evaluation, and expression of personal emotions towards songs, short videos, publications (such as novels), etc., and reflect the degree of emotional resonance between songs and listeners.

[0161] Likes: A way for users to express their liking and approval of songs, short videos, publications (such as novels), etc. The more likes, the more positive feedback the songs, short videos, publications (such as novels) receive.

[0162] Shares (or forwards): In addition to forwards, this also includes the number of times users share songs, short videos, publications (such as novels), etc., to different social platforms or other functional areas within the app. This reflects the spread of songs, short videos, publications (such as novels), etc. on social networks and users' willingness to actively promote songs, short videos, publications (such as novels), etc.

[0163] Lyrics View Count: This indicates how many users viewed the lyrics while listening to the song. This can reflect the appeal and clarity of the lyrics, as well as the listeners' desire to explore the song's meaning.

[0164] Playlist creation frequency: The number of times a song is added to a user's self-created playlist. Songs that are frequently added to various playlists indicate that they have high applicability and popularity in different music scenarios and user preferences.

[0165] Radio play count: If a song is played on radio or other broadcasting channels, the number of times it is played is an additional attribute that reflects the song's exposure in traditional broadcast media and its popularity with radio programs.

[0166] Music chart rankings: The changes in a song's position on various charts on major music platforms, such as the New Song Chart, Hot Song Chart, and Rising Chart, are important indicators for measuring a song's popularity and competitiveness over a certain period of time.

[0167] User profile data includes information such as the age, gender, region, and interests of users who listen to songs. User profile data can also be categorized, such as user profile data for liking, commenting, sharing, and saving.

[0168] Song revenue: For some paid music platforms or songs with commercial partnerships, the revenue is also an additional attribute, such as paid listening revenue, advertising revenue sharing, copyright licensing fees, etc., which reflects the commercial value of the song.

[0169] Video title: It is a concise description of the short video content, which serves to attract the audience's attention and convey the core information.

[0170] Creator information: This includes the creator's username, account identifier, and bio, helping viewers understand the creator of the video and establishing a connection between the creator and the audience.

[0171] Release time: This records the specific time when the short video was released on the platform, reflecting the video's timeliness. The release time can be used to analyze the video's dissemination patterns and popularity trends.

[0172] Video duration: refers to the length of time a short video takes from start to finish. Generally, short videos are relatively short, usually ranging from a few seconds to a few minutes.

[0173] Video resolution: This indicates the clarity and detail of the video image. Common resolutions include 720p, 1080p, and 4K. The higher the resolution, the clearer the image, but it also consumes more storage space and network bandwidth.

[0174] Video formats, such as MP4, AVI, MOV, etc., vary in encoding methods, compression ratios, and compatibility, which affect the storage and playback quality of the videos.

[0175] Content tags: Keywords or tags added by creators to videos to describe the content theme, type, style, etc., making it easier for platforms to categorize and recommend content and for users to search for it.

[0176] Music information: If music is used in the video, relevant information includes the music title, artist, copyright information, etc.

[0177] Filming Equipment: The equipment used to record the video, such as a mobile phone, camera, or camcorder, along with the model number. Videos shot with professional equipment may offer superior image quality and color reproduction, and are more likely to be recommended by users.

[0178] Filming location: The specific geographical location where the video was filmed.

[0179] Video theme: The core content or theme of the short video, such as comedy, food, travel, technology, education, etc.

[0180] Collection Count: The number of times a user adds songs, short videos, publications (such as novels), etc., to their favorites or a specific list. This collection behavior indicates a user's appreciation and enjoyment of the video content, making it convenient for them to watch it again later.

[0181] Bullet comment volume: If the platform supports the bullet comment function, the bullet comment volume reflects the number of comments users post in real time while watching songs, short videos, publications (such as novels), etc.

[0182] Completion rate: refers to the percentage of users who watch a song, short video, or publication (such as a novel) in its entirety. It is one of the important indicators for evaluating the quality and appeal of such content.

[0183] Video Revenue: For short videos that monetize through advertising, paid memberships, or tipping, revenue is an important additional attribute that reflects the video's commercial value.

[0184] Topic Relevance: The degree to which a short video relates to trending or specific topics on the platform. Videos that are relevant to trending topics are likely to gain more exposure and traffic.

[0185] Platform recommendation level: The platform classifies videos into recommendation levels based on various data and quality factors.

[0186] Follower Growth: This refers to the increase in the number of followers on a creator's account after the release of songs, short videos, publications (such as novels), etc. High-quality short videos can attract new followers, reflecting the video's positive impact on the creator's personal brand or influence.

[0187] Number of imitations or derivative works: If the content, such as songs, short videos, or publications (e.g., novels), is highly creative and influential, it may inspire other users to imitate or create derivative works. The number of imitations or derivative works can reflect the video's reach and creative value.

[0188] Traffic refers to a collective term for data metrics such as the number of views and uses of a song, short video, or publication (e.g., a novel). It can be used to measure the popularity and user affinity of a song, short video, or publication (e.g., a novel).

[0189] In this embodiment of the disclosure, the technology for recommending objects can be applied to various business scenarios for recommending objects, such as business scenarios for recommending songs or short videos. This embodiment of the disclosure does not limit this application.

[0190] Please see Figure 1 As shown, Figure 1 This is an application scenario to which the technical solution of the present disclosure embodiment can be applied. The schematic diagram of this scenario includes a terminal device 110 and an electronic device 120. The terminal device 110 can be one or more. Figure 1 One example is shown below. Terminal device 110 and electronic device 120 are connected via network 130.

[0191] In this embodiment, the terminal device 110 may be equipped with a recommendation system, which can be accessed and used by users with different identities using corresponding accounts and passwords. For example, ordinary users can log in to the recommendation system through their corresponding terminal device 110 to view recommended objects and recommendation sets. The copyright holder of the object to be recommended can log in to the recommendation system through their corresponding terminal device 110 to trigger a recommendation request for the object, thereby receiving the target recommendation set sent by the electronic device 110 and various post-recommendation performance data generated by the target recommendation set, and implementing corresponding optimization and adjustment strategies based on the performance data to better recommend the object.

[0192] In this embodiment, the copyright holder or owner of the object to be recommended can log in to the recommendation system through terminal device 110, select the object to be promoted, and send a recommendation request to electronic device 120. Electronic device 120 can then determine the object to be recommended; determine a candidate recommendation set based on the attribute information of the object to be recommended; select a target recommendation set that matches the object to be recommended from the candidate recommendation set; and insert the object to be recommended into the target recommendation set to recommend it based on the target recommendation set. Furthermore, when ordinary users log in to the recommendation system through terminal device 110, they can see the target recommendation set and perform interactive operations such as viewing, browsing, playing, liking, collecting, and creating derivative works on the target recommendation set and the objects within it.

[0193] In this embodiment of the disclosure, Figure 1The electronic device 120 in the document can also be an independent physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server or cloud server cluster that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDN), and big data and artificial intelligence platforms, but is not limited to these.

[0194] Optionally, the electronic device 120 may be equipped with a recommendation system, a recommendation set creation system, and an object insertion system. The recommendation system may work in conjunction with the recommendation set creation system and the object insertion system to recommend objects, or the recommendation system may include both the recommendation set creation system and the object insertion system; this is not limited in this embodiment.

[0195] In this embodiment of the disclosure, Figure 1 The terminal device 110 can be a mobile phone, tablet computer (PAD), personal computer, smart TV, smart watch, smart speaker, smart in-vehicle device, and wearable device, but is not limited to these.

[0196] Of course, the methods provided in this disclosure are not limited to... Figure 1 The application scenarios shown can also be used in other possible application scenarios, and this disclosure does not limit the scope of the embodiments.

[0197] Exemplary methods

[0198] To further illustrate the technical solutions provided by the embodiments of this disclosure, a detailed description is provided below in conjunction with the accompanying drawings and specific implementation methods. Although the embodiments of this disclosure provide method operation steps as shown in the following embodiments or drawings, the method may include more or fewer operation steps based on conventional or non-inventive methods. In steps where there is no logically necessary causal relationship, the execution order of these steps is not limited to the execution order provided by the embodiments of this disclosure. In actual processing or when the device executes the method, it may be executed sequentially or in parallel according to the method shown in the embodiments or drawings.

[0199] Please see Figure 2 , Figure 2 This is a flowchart illustrating a method for recommending an object according to an embodiment of this disclosure. The method flow can be, for example, provided by an electronic device, such as... Figure 1 The electronic device 120 in the process executes the method, and the specific implementation flow is as follows:

[0200] Step 201: Determine the candidates to be recommended.

[0201] In this embodiment of the disclosure, the electronic device can determine the recommended object based on the identification information carried in the recommendation request triggered by the terminal device, or it can determine the recommended object itself based on the filtering rules. This embodiment of the disclosure does not limit this.

[0202] In this embodiment of the disclosure, the object to be recommended can be understood as the object that the copyright holder or owner wants to promote, such as song A, short video 1, publication X, etc. This embodiment of the disclosure does not limit this.

[0203] Step 202: Determine the candidate recommendation set based on the attribute information of the object to be recommended.

[0204] In this embodiment of the disclosure, the electronic device can pre-create recommendation sets through a recommendation set creation system, and each recommendation set includes multiple objects that belong to the same category as the object to be recommended.

[0205] For example, suppose an electronic device pre-creates playlists, each containing multiple songs. Or, suppose an electronic device pre-creates short dramas, each containing multiple videos. Or, suppose an electronic device pre-creates collections of publications, each containing multiple publications.

[0206] Step 203: Select the target recommendation set that matches the object to be recommended from the candidate recommendation set.

[0207] In this embodiment of the disclosure, the electronic device may use, but is not limited to, the following steps to obtain a candidate recommendation set.

[0208] Step A: Determine the growth potential value of the object to be recommended based on its attribute information.

[0209] In this embodiment of the disclosure, the electronic device can first determine the attribute information of the object to be recommended, wherein the attribute information of the object to be recommended includes basic attribute information and additional attribute information. The basic attribute information can be understood as the information of the object to be recommended itself, while the additional attribute information is determined based on the propagation and user interaction data of the object to be recommended in the platform or network environment.

[0210] For example, assuming the recommended item is a song, the basic attribute information includes at least one of the following: song title, creator information, singer, release information, music style, song duration, language, rhythm, melody, harmony, song theme, and copyright information. Additional attribute information includes at least one of the following: play count, download count, comment count, like count, share count, lyrics view count, playlist creation count, radio play count, music chart ranking, user profile data, and song revenue.

[0211] For example, assuming the recommended content is a short video, the basic attribute information includes at least one of the following: video title, creator information, publication time, video length, video resolution, video format, content tags, music information, shooting equipment, shooting location, shooting theme, and copyright information. Additional attribute information includes at least one of the following: number of views, number of comments, number of shares, number of favorites, number of bullet comments, completion rate, user profile of those who liked the video, video revenue, topic relevance, platform recommendation level, number of followers gained, and number of imitations or derivative works.

[0212] For example, assuming the object to be recommended is a publication, the basic attribute information includes at least one of the following: title, author, genre, subject matter, and story background. Additional attribute information includes at least one of the following: word count, publisher, cover design information, illustration design information, creation date, literary genre, awards, sales performance, interaction status, and adaptation status. The adaptation status can include information on whether the publication has been adapted into other art forms such as film, theater, animation, or games.

[0213] In this embodiment, after the electronic device obtains the basic attribute information and additional attribute information of the object to be recommended, it can filter the basic attribute information and additional attribute information to obtain parameter information related to growth potential. Based on the parameter information and its corresponding weight value, the growth potential value of the object to be recommended is determined. Alternatively, the growth potential value of the object to be recommended can be determined directly based on its basic attribute information and its corresponding first weight, and its additional attribute information and its corresponding second weight. The aforementioned weight values, first weight, and second weight can all be set and adjusted based on actual implementation, and this embodiment does not limit this.

[0214] Step B: Select a target recommendation candidate pool based on the growth potential value.

[0215] In this embodiment, after the electronic device determines the growth potential value, it can compare the growth potential value with a first preset range. If the growth potential value of the object to be recommended is within the first preset range, a first type of recommendation candidate pool is selected as the target recommendation candidate pool; if the growth potential value is not within the first preset range, a second type of recommendation candidate pool is selected as the target recommendation candidate pool. The first and second types of recommendation candidate pools correspond to different recommendation channels. The first preset range can be updated based on actual implementation; for example, the first preset range can be updated according to different recommendation times. This embodiment does not limit this.

[0216] For example, assuming song A's growth potential value is within a first preset range, then the first type of recommendation candidate pool is selected as the target recommendation candidate pool. If song B's growth potential value is not within the first preset range, then the second type of recommendation candidate pool is selected as the target recommendation candidate pool. The first type of recommendation candidate pool could be, for example, a daily incubation recommendation channel, and the second type of recommendation candidate pool could be, for example, a channel aimed at creating viral hits.

[0217] For example, assuming the growth potential of short video 1 falls within a first preset range, then the first type of recommendation candidate pool is selected as the target recommendation candidate pool. If the growth potential of short video 2 does not fall within the first preset range, then the second type of recommendation candidate pool is selected as the target recommendation candidate pool. Here, the first type of recommendation candidate pool represents, for example, a first-level recommendation channel, and the second type of recommendation candidate pool represents, for example, a second-level recommendation channel. The recommendation traffic of the first-level recommendation channel is less than that of the second-level recommendation channel.

[0218] For example, assuming the growth potential of publication 1 falls within a first preset range, then the first type of recommendation candidate pool is selected as the target recommendation candidate pool. If the growth potential of publication 2 does not fall within the first preset range, then the second type of recommendation candidate pool is selected as the target recommendation candidate pool. The first type of recommendation candidate pool could be, for example, a new book recommendation channel, and the second type of recommendation candidate pool could be, for example, a high-scoring recommendation channel.

[0219] Step C: Determine the candidate recommendation set from the target recommendation candidate pool.

[0220] In this embodiment of the disclosure, when the target recommendation candidate pool is a first type of recommendation candidate pool, the recommendation growth value of each recommendation set in the target recommendation candidate pool can be determined.

[0221] In this embodiment, the recommendation growth value of a recommendation set in the target recommendation candidate pool can be determined based on the first attribute information and the second attribute information of the recommendation set, as well as a first preset rule. The first attribute information can be understood as basic information about the recommendation set, such as at least one of the following: the creation time of the recommendation set, the creator, the list, and the name of the recommendation set. The second attribute information can be understood as additional information about the recommendation set, such as at least one of the following: a brief description of the recommendation set, the number of objects, the number of plays, the number of favorites, the update time, and the tags. The first preset rule can be understood as a rule for calculating the growth potential of the recommendation set.

[0222] For example, first determine the first value by multiplying the creator's name and their corresponding weight, the second value by multiplying the number of objects and their corresponding weight, the third value by multiplying the number of plays and their corresponding weight, and the fourth value by multiplying the number of favorites and their corresponding weight. Then, use the sum of the first, second, third, and fourth values ​​as the recommendation growth value of the recommendation set.

[0223] In this embodiment of the disclosure, after determining the recommendation growth value of each recommendation set in the target recommendation candidate pool, the growth potential value corresponding to the object to be recommended can be compared with the recommendation growth value of each recommendation set in the target recommendation candidate pool to obtain multiple first comparison results. Then, based on the multiple first comparison results, the candidate recommendation set is determined.

[0224] For example, the first type of recommendation candidate pool has 5 recommendation sets, namely recommendation set 1, recommendation set 2, recommendation set 3, recommendation set 4 and recommendation set 5. It is determined that the recommendation growth value corresponding to recommendation set 1 is 60, the recommendation growth value corresponding to recommendation set 2 is 35, the recommendation growth value corresponding to recommendation set 3 is 40, the recommendation growth value corresponding to recommendation set 4 is 50 and the recommendation growth value corresponding to recommendation set 5 is 20. It is also determined that the growth potential value corresponding to the object to be recommended is 30. Then multiple first comparison results can be determined. Based on the multiple first comparison results, the candidate recommendation sets include recommendation set 2 and recommendation set 3.

[0225] In this embodiment of the disclosure, after determining the candidate recommendation set, the electronic device can determine multiple dimensional feature information of each recommendation set within the candidate recommendation set.

[0226] In this embodiment of the disclosure, the electronic device may perform the following operations on each recommendation set: determine the overall style characteristics of the recommendation set based on the style information of each object in the recommendation set; determine the emotional characteristics of the recommendation set based on the emotional information of each object in the recommendation set; determine the scene characteristics of the recommendation set based on the scene information of each object in the recommendation set; and obtain multi-dimensional feature information of the recommendation set based on the overall style characteristics, emotional characteristics, and scene characteristics.

[0227] For example, the following section uses a recommended playlist as an example to introduce a scheme for determining the multi-dimensional feature information of a playlist.

[0228] In this embodiment of the disclosure, the electronic device can determine the music type, style and rhythm characteristics of each song in the playlist, and determine the style information of the song based on the music type, style and rhythm characteristics of each song, and then analyze and process the style information of each song in the playlist to determine the overall style characteristics of the playlist.

[0229] In this embodiment, the electronic device can perform lyric sentiment analysis and audio sentiment extraction processing on each song to obtain the corresponding sentiment information for each song. Each sentiment information can indicate the emotion expressed by the song, such as happiness, sadness, encouragement, or relaxation. After obtaining the sentiment information for each song in the playlist, sentiment aggregation processing can be performed based on the sentiment information of each song and the behavioral feedback information generated by the user's interactive operations on the song, thereby obtaining the sentiment characteristics of the playlist.

[0230] In this embodiment of the disclosure, the electronic device can also determine the usage scenario of each song in the playlist, such as work, exercise, party, sleep, etc., thereby determining the scenario information of each song, and then determining the scenario characteristics of the playlist based on the scenario information of each song.

[0231] In this embodiment of the disclosure, after obtaining the overall style features, emotional features, and scene features of the recommendation set, the overall style features, emotional features, and scene features can be directly used as the dimensional feature information of the recommendation set.

[0232] In this embodiment of the disclosure, the electronic device can also perform cross-analysis processing on the overall style features, emotional features and scene features to obtain multiple mixed features, and then use the overall style features, emotional features, scene features and multiple mixed features as multiple dimension feature information of the recommendation set.

[0233] In this embodiment of the disclosure, the electronic device can perform cross-analysis of overall style features and emotional features to obtain a first mixed feature; perform cross-analysis of emotional features and scene features to obtain a second mixed feature; perform cross-analysis of overall style features and scene features to obtain a third mixed feature; and obtain multiple mixed features based on the first mixed feature, the second mixed feature, and the third mixed feature.

[0234] For example, continuing with the playlist as an example, electronic devices can analyze the match between the songs in the playlist in terms of mood and style. For instance, uplifting songs often have a faster tempo and stronger melody, while relaxing songs may have a slower tempo and gentler melody. In other words, electronic devices can perform cross-analysis of overall style characteristics and emotional characteristics to obtain a first mixed feature. This first mixed feature, obtained through cross-analysis, can ensure that the songs in the playlist complement each other in terms of mood and style, thus identifying more suitable playlists for the target audience and improving the accuracy of song recommendations. Furthermore, electronic devices can perform cross-analysis of emotional features and scene characteristics to obtain a second mixed feature. This means that the scene and emotional state of the playlist recommendations can be optimized based on the user's activity scene and emotional state. Thus, when a song is inserted into the target recommended playlist, if it is determined that the user is currently in an active scene, songs that both invigorate and meet the user's emotional needs can be recommended. Furthermore, electronic devices can combine scene data with song genres to ensure that recommended playlists not only match users' interests in style but also meet their listening needs in specific activity scenarios. For example, in a sports setting, upbeat and energetic rock or electronic songs might be recommended; in a relaxation setting, gentle classical or light music might be suggested.

[0235] For example, the following section uses a recommendation set as a short video set to illustrate a scheme for determining multiple dimensions of feature information for a short video set.

[0236] In this embodiment of the disclosure, the electronic device can determine the content tag, music information, and shooting theme of each short video in the short video set, and determine the style information of each short video based on the content tag, music information, and shooting theme of each short video, and then analyze and process the style information of each short video in the short video set to determine the overall style characteristics of the short video set.

[0237] In this embodiment, the electronic device can perform text sentiment analysis and audio-visual sentiment extraction on each short video to obtain the sentiment information corresponding to each short video. Each sentiment information can indicate the overall emotion expressed by the short video, such as happiness, sadness, encouragement, or relaxation. After obtaining the sentiment information corresponding to each short video in the short video set, sentiment aggregation processing can be performed based on the sentiment information of each short video and the behavioral feedback information generated by the user's interactive operations on that short video, thereby obtaining the sentiment characteristics of the short video set.

[0238] In this embodiment of the disclosure, the electronic device can also determine the usage scenario of each short video in the short video set, such as work, sharing session, teaching, exercise, sleep, etc., thereby determining the scene information of each short video, and then determining the scene features of the short video set based on the scene information of each short video.

[0239] In this embodiment of the disclosure, after obtaining the overall style features, emotional features, and scene features of the recommendation set, the overall style features, emotional features, and scene features can be directly used as the dimensional feature information of the recommendation set.

[0240] In this embodiment of the disclosure, the electronic device can also perform cross-analysis processing on the overall style features, emotional features and scene features to obtain multiple mixed features, and then use the overall style features, emotional features, scene features and multiple mixed features as multiple dimension feature information of the recommendation set.

[0241] In this embodiment of the disclosure, the electronic device can perform cross-analysis of overall style features and emotional features to obtain a first mixed feature; perform cross-analysis of emotional features and scene features to obtain a second mixed feature; perform cross-analysis of overall style features and scene features to obtain a third mixed feature; and obtain multiple mixed features based on the first mixed feature, the second mixed feature, and the third mixed feature.

[0242] For example, the following section uses a recommendation set as an example to introduce a scheme for determining multiple dimensions of feature information of a publication set.

[0243] In this embodiment of the disclosure, the electronic device can determine the genre, subject matter, and story background of each publication in the publication collection, and based on the genre, subject matter, and story background of each publication, determine the style information of the publication, and then analyze and process the style information of each publication in the publication collection to determine the overall style characteristics of the publication collection.

[0244] In this embodiment, the electronic device can perform illustration sentiment analysis and text sentiment extraction processing on each publication to obtain the sentiment information corresponding to each publication. The sentiment information for each publication can indicate the overall sentiment expressed by the publication, such as happiness, sadness, encouragement, or relaxation. After obtaining the sentiment information corresponding to each publication within the publication set, sentiment aggregation processing can be performed based on the sentiment information of each publication and the behavioral feedback information generated by the user's interactive operations on that publication, thereby obtaining the sentiment characteristics of the publication set.

[0245] In this embodiment of the disclosure, the electronic device can also determine the usage scenario of each publication in the publication collection, such as work, book club, teaching, exam, sleep, etc., thereby determining the scenario information of each publication, and then determining the scenario characteristics of the publication collection based on the scenario information of each publication.

[0246] In this embodiment of the disclosure, after obtaining the overall style features, emotional features, and scene features of the recommendation set, the overall style features, emotional features, and scene features can be directly used as the dimensional feature information of the recommendation set.

[0247] In this embodiment of the disclosure, the electronic device can also perform cross-analysis processing on the overall style features, emotional features and scene features to obtain multiple mixed features, and then use the overall style features, emotional features, scene features and multiple mixed features as multiple dimension feature information of the recommendation set.

[0248] In this embodiment of the disclosure, the electronic device can perform cross-analysis of overall style features and emotional features to obtain a first mixed feature; perform cross-analysis of emotional features and scene features to obtain a second mixed feature; perform cross-analysis of overall style features and scene features to obtain a third mixed feature; and obtain multiple mixed features based on the first mixed feature, the second mixed feature, and the third mixed feature.

[0249] As can be seen, in determining the characteristics of the recommendation set in this embodiment, not only is the information determined by the direct information of the objects in the recommendation set considered, but also cross-analysis is performed based on the aforementioned information to determine multiple dimensional feature information that can more comprehensively describe the characteristics of the recommendation set.

[0250] In this embodiment of the disclosure, the electronic device can also determine multiple dimensional feature information corresponding to the object to be recommended based on the attribute information of the object to be recommended. Optionally, the electronic device can determine multiple dimensional feature information corresponding to the object to be recommended based on the basic attribute information and additional attribute information of the object to be recommended.

[0251] For example, assuming the recommended object is a song, we can determine the song's style, language, mood, and context information based on multiple factors, including song title, creator information, singer, release information, music style, song length, language, rhythm, melody, harmony, song theme, copyright information, play count, download count, comment count, like count, share count, lyric view count, playlist creation count, radio play count, music chart ranking, user profile data, and song revenue. We can also determine multiple comprehensive information based on the song's style, language, mood, and context information, thus using the song's style, language, mood, context, and multiple comprehensive information as multi-dimensional feature information of the song.

[0252] For example, assuming the recommended content is a short video, we can determine the style, language, emotional, and scene information of the short video based on multiple factors, including video title, language, background music, creator information, release time, video length, video resolution, video format, content tags, music information, shooting equipment, shooting location, shooting theme, copyright information, number of views, number of comments, number of shares, number of favorites, number of bullet comments, completion rate, user profile of likes, video revenue, topic relevance, platform recommendation level, number of followers, and number of imitations or derivative works. We can also determine multiple comprehensive information based on the style, language, emotional, and scene information of the short video, thus using the style, language, emotional, scene, and multiple comprehensive information of the song as multi-dimensional feature information of the song.

[0253] For example, assuming the object to be recommended is a publication, its style, language, emotional content, and contextual information can be determined based on multiple factors such as name, language, author, genre, subject matter, story background, word count, publisher, cover design information, illustration design information, creation time, literary school, awards, sales performance, interaction, and adaptation details. Furthermore, multiple comprehensive information can be determined based on the publication's style, language, emotional content, and contextual information, thus making the song's style, language, emotional content, contextual information, and multiple comprehensive information into multiple dimensions of the publication's characteristics.

[0254] As can be seen, in this embodiment of the disclosure, even if the object to be recommended is a newly published object and no additional attribute information such as play count, read count, forward count, collection count, etc. is generated, multiple dimension feature information of the object to be recommended can still be determined, thereby achieving accurate matching between the object to be recommended and the recommendation set, avoiding situations such as mismatch caused by the object to be recommended being a newly published object, thus solving the cold start problem, that is, by combining social network data and content recommendation, accurate recommendations are provided for new users and newly published objects.

[0255] In this embodiment of the disclosure, after obtaining the multiple dimension feature information of the object to be recommended and the multiple dimension feature information of each recommendation set, the multiple dimension feature information corresponding to the object to be recommended can be matched with the multiple dimension feature information of each recommendation set in the candidate recommendation set to obtain multiple matching values. Then, based on the multiple matching values ​​and preset filtering conditions, the target recommendation set matching the object to be recommended is determined.

[0256] In this embodiment of the disclosure, the preset filtering condition is, for example, filtering the highest value among multiple matching values, thereby using the recommendation set corresponding to the highest value as the target recommendation set. Alternatively, the preset filtering condition is, for example, filtering the recommendation set with matching values ​​greater than a preset value as the target recommendation set; this embodiment of the disclosure does not limit this to the latter.

[0257] In this embodiment of the disclosure, if the target recommendation candidate pool is a second type of recommendation candidate pool, the electronic device can determine the traffic growth value of each recommendation set in the target recommendation candidate pool; compare the traffic potential value corresponding to the object to be recommended with the traffic growth value of each recommendation set in the target recommendation candidate pool to obtain multiple second comparison results; and determine the candidate recommendation set based on the multiple second comparison results.

[0258] In this embodiment of the disclosure, the electronic device can determine the traffic growth value of the recommendation set based on the attribute information of the recommendation set. For example, the electronic device can comprehensively determine the traffic growth value of the recommendation set based on various traffic-related information such as the creator, name, indirect information, number of objects, number of plays, number of favorites, number of reposts, number of comments, number of secondary creations, etc.

[0259] In this embodiment of the disclosure, the electronic device can comprehensively determine the traffic potential value of the object to be recommended based on its playback volume, download volume, comment volume, collection volume, likes, shares, sales volume, discussion volume, and revenue information.

[0260] In this embodiment of the disclosure, after the electronic device determines the traffic potential value corresponding to the object to be recommended and the traffic growth value of each recommendation set in the target recommendation candidate pool, it can compare the traffic potential value corresponding to the object to be recommended with the traffic growth value of each recommendation set in the target recommendation candidate pool to obtain multiple second comparison results, and then determine the candidate recommendation set based on the multiple second comparison results.

[0261] For example, the first type of candidate pool has 5 recommendation sets, namely recommendation set A, recommendation set B, recommendation set C, recommendation set D and recommendation set F. It is determined that the recommendation growth value corresponding to recommendation set A is 10, the recommendation growth value corresponding to recommendation set B is 4, the recommendation growth value corresponding to recommendation set C is 80, the recommendation growth value corresponding to recommendation set D is 40 and the recommendation growth value corresponding to recommendation set F is 60. It is also determined that the growth potential value corresponding to the object to be recommended is 70. Then multiple second comparison results can be determined. Based on the multiple second comparison results, the candidate recommendation sets include recommendation set B and recommendation set F.

[0262] In this embodiment of the disclosure, after determining the candidate recommendation set, the electronic device can determine the traffic gap information of each candidate recommendation set and the traffic level information of the object to be recommended. Then, based on the traffic gap information, the traffic level information, and a preset matching rule, it determines the target recommendation set to match the object to be recommended. The preset matching rule, for example, matches recommendation sets with higher traffic level information with objects indicating larger traffic gaps, and matches recommendation sets with lower traffic level information with objects indicating smaller traffic gaps.

[0263] For example, if the traffic level information of song A indicates that the recommended object is a high-traffic potential object (i.e., a relatively high level), then the recommendation set with a large traffic gap can be determined from the candidate recommendation set as the target recommendation set. If the traffic level information of song B indicates that the recommended object is a low-traffic growth object (i.e., a relatively low level), then the recommendation set with a small traffic gap can be determined from the candidate recommendation set as the target recommendation set.

[0264] Optionally, in this embodiment of the disclosure, after obtaining the candidate recommendation set, the following operations can be performed on each candidate recommendation set: determining the overall style characteristics of the candidate recommendation set based on the style information of each object in the candidate recommendation set; determining the emotional characteristics of the candidate recommendation set based on the emotional information of each object in the candidate recommendation set; determining the scene characteristics of the candidate recommendation set based on the scene information of each object in the candidate recommendation set; and obtaining multi-dimensional feature information of the candidate recommendation set based on the overall style characteristics, emotional characteristics, and scene characteristics.

[0265] Optionally, the electronic device can perform cross-analysis of overall style features, emotional features, and scene features to obtain multiple mixed features; and use the overall style features, emotional features, scene features, and multiple mixed features as multiple dimensional feature information of the recommendation set.

[0266] Optionally, the electronic device can perform cross-analysis of overall style features and emotional features to obtain a first mixed feature; perform cross-analysis of emotional features and scene features to obtain a second mixed feature; perform cross-analysis of overall style features and scene features to obtain a third mixed feature; and obtain multiple mixed features based on the first mixed feature, the second mixed feature, and the third mixed feature.

[0267] In this embodiment of the disclosure, multiple dimensional feature information corresponding to the object to be recommended can be matched with multiple dimensional feature information of each recommendation set in the candidate recommendation set to obtain multiple matching values. Then, based on the multiple matching values ​​and preset filtering conditions, a secondary candidate recommendation set matching the object to be recommended can be determined.

[0268] In this embodiment of the disclosure, after obtaining the secondary candidate recommendation set, the traffic gap information of each secondary candidate recommendation set can be determined, and the traffic level information of the object to be recommended can be determined. Then, based on the traffic gap information, the traffic level information and the preset matching rules, the target recommendation set matching the object to be recommended can be determined.

[0269] As can be seen, in this embodiment, the recommendation set can be initially screened based on traffic matching, and then a second screening can be performed based on feature matching. Alternatively, the initial screening can be performed first based on feature matching, followed by a second screening based on traffic matching; this embodiment does not limit the approach. In this way, the final target recommendation set not only matches the target object in terms of traffic recommendation direction, but also its own characteristics are well-matched with the target object, thereby improving the accuracy of recommendations based on the target recommendation set.

[0270] Step 204: Insert the object to be recommended into the target recommendation set, so as to make recommendations based on the object to be recommended in the target recommendation set.

[0271] In this embodiment of the disclosure, once the target recommendation set is determined, the object to be recommended can be inserted into the target recommendation set so as to recommend the object to be recommended based on the target recommendation set.

[0272] In this embodiment of the disclosure, deep learning algorithms (such as neural networks and graph embedding models) can be used to automatically extract and match features of the object to be recommended and the recommendation set, so as to recommend the object to be recommended based on the target recommendation set. Alternatively, a team of music experts can be used to manually evaluate and match the object to be recommended and the recommendation set, so as to recommend the object to be recommended based on the target recommendation set. Of course, it is also possible to combine automatic and manual evaluation and matching of the object to be recommended and the recommendation set. This embodiment of the disclosure does not limit this.

[0273] In other words, the solution provided in this disclosure can also solve the problem of recommendation data bias in related technologies. Specifically, it can reduce the impact of internal and external data bias by integrating multi-dimensional data and multi-level recommendation mechanisms, thereby improving the fairness and accuracy of recommendations.

[0274] In one possible implementation, when the object to be recommended has not been inserted into the recommendation set, an exploratory strategy can be adopted. That is, a test target recommendation set is determined for the object to be recommended, and the updated growth potential value of the object to be recommended and the updated recommendation growth value of the target recommendation set are determined within a first period. The first period is the recommendation observation period for the object to be recommended. When it is determined that the updated growth potential value is within a first preset range and the updated growth potential value is not less than the updated recommendation growth value, the updated target recommendation set for the object to be recommended is re-determined, so as to recommend the object to be recommended based on the updated target recommendation set.

[0275] As can be seen, in this embodiment of the disclosure, when the object to be recommended has not been inserted into the recommendation set, a trial strategy can be adopted to first insert it into the playlist. After the first period (e.g., 3 days) of observation, the object to be recommended is re-evaluated and the subsequent promotion path is determined, that is, the updated target recommendation set is determined, and the object to be recommended is recommended based on the updated target recommendation set.

[0276] In this embodiment of the disclosure, the electronic device can determine the performance data information of the target recommendation set; when it is determined that the performance data information of the target recommendation set does not meet the preset performance conditions, the object to be recommended is removed from the target recommendation set; and, determine the updated attribute information of the object to be recommended, and determine the updated candidate recommendation set based on the updated attribute information of the object to be recommended; select the updated target recommendation set that matches the object to be recommended from the updated candidate recommendation set; insert the object to be recommended into the updated target recommendation set, so as to recommend the object to be recommended based on the updated target recommendation set.

[0277] In this embodiment, a preset performance condition is, for example, that the current indicator value of the target recommendation set is not less than the standard indicator value. The current indicator value is then determined based on the performance data information of the target recommendation set, and compared with the standard indicator value. If the current indicator value is determined to be less than the standard indicator value, then the performance data information of the target recommendation set is determined to not meet the preset performance condition. The standard indicator can be determined based on the attribute information of the object to be recommended; for example, a funnel value. This embodiment does not limit the specific standard indicator.

[0278] For example, if the recommended song is A, and after inserting song A into the target playlist, the impact of song A on the overall performance data of the playlist (such as click-through rate, like rate, completion rate, etc.) is continuously monitored. If the monitored overall performance data indicates that the standard indicators of the target playlist (such as funnel value) are significantly lowered, the inserted song needs to be reviewed. Based on the review results, unsuitable inserted song A can be adjusted, and the unsuitable inserted song A can be reassigned to other suitable playlists (i.e., the target recommendation set can be updated), or song A can be removed from the promotion channel (i.e., song A is not recommended).

[0279] As can be seen, in this embodiment, by analyzing the style, language, emotional, and contextual information of the objects to be recommended from multiple dimensions, and combining this with the thematic characteristics of the recommendation set, a high degree of matching between the objects to be recommended and the recommendation set is ensured, thereby improving the accuracy of promotion and user experience. In this embodiment, a dynamic matching strategy based on the traffic gap in the recommendation set is also used to rationally allocate promotional resources for objects to be recommended at different traffic levels, maximizing traffic efficiency and avoiding resource waste. Furthermore, by continuously monitoring the data after the objects to be recommended are inserted into the recommendation set, unsuitable objects to be recommended can be adjusted in a timely manner, ensuring the stability of the overall data indicators of the recommendation set and improving the consistency of promotional effects.

[0280] In this embodiment of the disclosure, to better illustrate the recommendation scheme provided by this disclosure, the following example uses a song as the object to be recommended. Figure 3 As shown, the solution provided in this disclosure is presented.

[0281] Step 301: Determine if the song is eligible for inclusion in the playlist support.

[0282] In this embodiment, the electronic device can determine whether a song can be included in a playlist for support based on whether the song has special marking information. The special marking information can be determined by the recommendation system based on promotional information input by the user, such as words indicating support or promotion, or patterns indicating support or promotion, etc., and this embodiment does not limit the scope of the information.

[0283] Step 302: Once it is determined that a song can be included in the playlist support, determine the song's growth potential value. Based on the song's growth potential value, determine whether the song is suitable for the daily incubation song promotion channel.

[0284] Step 303: Once it is determined that the song matches the daily incubation song promotion channel, the song is evaluated based on the preset model to determine whether to insert the song into the target playlist.

[0285] In this embodiment, the preset model can determine the recommendation growth value of each playlist in the target recommendation candidate pool corresponding to the daily incubation song promotion channel; compare the growth potential value corresponding to the song with the recommendation growth value of each playlist in the target recommendation candidate pool to obtain multiple first comparison results; and determine candidate playlists based on the multiple first comparison results. Furthermore, based on the song's attribute information, determine multiple dimensional feature information corresponding to the song; determine multiple dimensional feature information for each candidate playlist; match the multiple dimensional feature information corresponding to the song with the multiple dimensional feature information of each playlist in the candidate playlist to obtain multiple matching values; and determine the target playlist matching the song based on the multiple matching values ​​and preset filtering conditions.

[0286] Step 304: Once it is determined that a song will be inserted into the target playlist, the recommended order of the song in the target playlist is determined, and the song is recommended based on this recommended order.

[0287] Step 305: Once it is determined that a song will be inserted into the test playlist, if the performance of the song in the test playlist meets the preset recommendation conditions, the song will be inserted into the updated target playlist. The recommendation order of the song in the updated target playlist will be determined, and the song will be recommended based on this recommendation order.

[0288] Step 306: Once the song is matched with the promotion channel for hit songs, determine the recommendation order of the song in the target playlist, and recommend the song based on this recommendation order.

[0289] Step 307: Determine the performance data of the target playlist, and replace the songs according to the performance data of the target playlist to insert the songs into the updated target playlist, so as to recommend songs based on the updated target playlist.

[0290] As can be seen, in this embodiment, by analyzing the style, language, emotional, and scene information of songs from multiple dimensions, and combining this with the theme characteristics of the playlist, a high degree of matching between songs and playlists is ensured, thereby improving the accuracy of promotion and user experience. This embodiment also employs a dynamic matching strategy based on playlist traffic gaps to rationally allocate promotional resources for songs of different traffic levels, maximizing traffic efficiency and avoiding resource waste. Furthermore, by continuously monitoring playlist data after song insertion, unsuitable songs can be adjusted in a timely manner, ensuring the stability of overall playlist data metrics and improving the consistency of promotional effects.

[0291] Exemplary device

[0292] Exemplary embodiments of this disclosure also provide an apparatus for a recommended object. (See reference...) Figure 4 As shown, the device 400 for recommending objects includes the following program units:

[0293] The first determining unit 401 is used to determine the object to be recommended;

[0294] The second determining unit 402 is used to determine a candidate recommendation set based on the attribute information of the object to be recommended;

[0295] The filtering unit 403 is used to select a target recommendation set that matches the object to be recommended from the candidate recommendation set;

[0296] The recommendation unit 403 is used to insert the object to be recommended into the target recommendation set, so as to recommend the object to be recommended based on the target recommendation set.

[0297] In one possible implementation, the second determining unit 402 is configured to:

[0298] Based on the attribute information of the object to be recommended, determine the growth potential value corresponding to the object to be recommended;

[0299] Based on the growth potential value, a target recommendation candidate pool is selected;

[0300] A candidate recommendation set is determined from the target recommendation candidate pool.

[0301] In one possible implementation, the second determining unit 402 is configured to:

[0302] If the growth potential value is within a first preset range, then the first type of recommendation candidate pool is selected as the target recommendation candidate pool;

[0303] If the growth potential value is not within the first preset range, then the second type of recommendation candidate pool is selected as the target recommendation candidate pool;

[0304] The first type of recommendation candidate pool and the second type of recommendation candidate pool correspond to different recommendation channels.

[0305] In one possible implementation, the second determining unit 402 is configured to:

[0306] Determine the recommendation growth value of each recommendation set in the target recommendation candidate pool;

[0307] The growth potential value corresponding to the object to be recommended is compared with the recommendation growth value of each recommendation set in the target recommendation candidate pool to obtain multiple first comparison results;

[0308] Based on the multiple first comparison results, a candidate recommendation set is determined.

[0309] In one possible implementation, the filtering unit 403 is used for:

[0310] Based on the attribute information of the object to be recommended, determine the multiple dimensional feature information corresponding to the object to be recommended;

[0311] Determine multiple dimensional feature information for each of the candidate recommendation sets;

[0312] The multiple dimensional feature information corresponding to the object to be recommended is matched with the multiple dimensional feature information of each recommendation set in the candidate recommendation set to obtain multiple matching values;

[0313] Based on the multiple matching values ​​and preset filtering conditions, a target recommendation set matching the object to be recommended is determined.

[0314] In one possible implementation, the filtering unit 403 is used for:

[0315] Perform the following operations on each of the candidate recommendation sets:

[0316] The overall style characteristics of the candidate recommendation set are determined based on the style information of each object in the candidate recommendation set.

[0317] Based on the emotional information of each object in the candidate recommendation set, determine the emotional characteristics of the candidate recommendation set;

[0318] The scene features of the candidate recommendation set are determined based on the scene information of each object in the candidate recommendation set.

[0319] Based on the overall style features, the emotional features, and the scene features, the candidate recommendation set is obtained with multiple dimensions of feature information.

[0320] In one possible implementation, the filtering unit 403 is used for:

[0321] The overall style features, the emotional features, and the scene features are cross-analyzed to obtain multiple hybrid features;

[0322] The overall style features, the emotional features, the scene features, and the multiple mixed features are used as multiple dimensional feature information of the recommendation set.

[0323] In one possible implementation, the filtering unit 403 is used for:

[0324] The overall style features and the emotional features are cross-analyzed to obtain the first hybrid feature;

[0325] The emotional features and the scene features are cross-analyzed to obtain a second hybrid feature;

[0326] The overall style features and the scene features are cross-analyzed to obtain a third hybrid feature;

[0327] Multiple mixed features are obtained based on the first mixed feature, the second mixed feature, and the third mixed feature.

[0328] In one possible implementation, the second determining unit 402 is configured to:

[0329] Determine the traffic growth value of each recommendation set in the target recommendation candidate pool;

[0330] The traffic potential value corresponding to the object to be recommended is compared with the traffic growth value of each recommendation set in the target recommendation candidate pool to obtain multiple second comparison results;

[0331] Based on the multiple second comparison results, a candidate recommendation set is determined.

[0332] In one possible implementation, the filtering unit 403 is used for:

[0333] Determine the traffic gap information for each of the candidate recommendation sets;

[0334] Determine the traffic level information of the object to be recommended;

[0335] Based on the traffic gap information, the traffic level information, and the preset matching rules, the target recommendation set for matching the object to be recommended is determined.

[0336] In one possible implementation, the recommendation unit 404 is further configured to:

[0337] If the target recommendation set is a test target recommendation set, then the updated growth potential value of the object to be recommended and the updated recommendation growth value of the target recommendation set are determined within the first period; the first period is the recommendation observation period for the object to be recommended.

[0338] When it is determined that the updated growth potential value is within a first preset range and the updated growth potential value is not less than the updated recommended growth value, the updated target recommendation set of the object to be recommended is redefined, so as to recommend the object to be recommended based on the updated target recommendation set.

[0339] In one possible implementation, the recommendation unit 404 is further configured to:

[0340] Determine the performance data information of the target recommendation set;

[0341] When it is determined that the performance data of the target recommendation set does not meet the preset performance conditions, the object to be recommended will be removed from the target recommendation set.

[0342] Determine the updated attribute information of the object to be recommended, and determine the updated candidate recommendation set based on the updated attribute information of the object to be recommended;

[0343] From the updated candidate recommendation set, select the updated target recommendation set that matches the object to be recommended;

[0344] The object to be recommended is inserted into the updated target recommendation set, so that the object to be recommended is recommended based on the updated target recommendation set.

[0345] The specific details of each part of the above-mentioned device have been described in detail in the method section of the implementation plan. For any undisclosed details, please refer to the implementation plan of the method section, and therefore will not be repeated here.

[0346] It should be noted that although several modules or units for the device used to perform actions have been mentioned in the detailed description above, this division is not mandatory. In fact, according to exemplary embodiments of this disclosure, the features and functions of two or more modules or units described above can be embodied in one module or unit. Conversely, the features and functions of one module or unit described above can be further divided and embodied by multiple modules or units.

[0347] Exemplary program product

[0348] Exemplary embodiments of this disclosure also provide a computer program product. The computer program product includes a computer program that, when executed by a processor, implements the method of the recommended object described above.

[0349] In one embodiment, the computer program product can be a tangible product containing a computer program, such as a computer-readable storage medium storing the computer program. The readable storage medium can be a storage medium based on electrical, magnetic, optical, electromagnetic, infrared, or other signals, including but not limited to: random access memory (RAM), read-only memory (ROM), magnetic tape, floppy disk, flash memory, hard disk drive (HDD), solid-state drive (SSD), etc. For example, the computer program product can be implemented as a non-volatile storage medium storing the computer program, such as read-only memory, NAND flash memory, etc.

[0350] In one implementation, the computer program product can be an intangible product containing a computer program. For example, the computer program product can be implemented as a virtual digital product, such as an executable file, installation package, or other digital file storing the computer program.

[0351] Computer program code can be written in one or more programming languages. Examples of programming languages ​​include C, Java, and C++. Program code can execute entirely on the user's computing device, partially on the user's computing device, or as a standalone software package. It can also execute partially on the user's computing device and partially on a remote computing device, or entirely on a remote computing device or server. In cases involving remote computing devices, the remote computing device can be connected to the user's computing device via any type of network, such as a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computing device (e.g., via an internet connection provided by a mobile network operator).

[0352] Computer programs can be carried or transmitted via signals such as electricity, magnetism, light, electromagnetic radiation, and infrared rays. Electronic devices can convert signals carrying computer programs into digital signals, thereby running the computer programs. When a computer program runs on an electronic device, its code causes the electronic device to execute (more specifically, its processor) the method steps of various exemplary embodiments of this disclosure, such as the method of the above-described recommended object, which includes the following steps:

[0353] Step 201: Determine the object to be recommended; Step 202: Determine the candidate recommendation set based on the attribute information of the object to be recommended; Step 203: Select the target recommendation set that matches the object to be recommended from the candidate recommendation set; Step 204: Insert the object to be recommended into the target recommendation set to recommend the object based on the target recommendation set.

[0354] By implementing the above method steps through a computer program, the object to be recommended can be determined. Then, based on the attribute information of the object to be recommended, a candidate recommendation set can be determined, wherein the candidate recommendation set includes multiple objects belonging to the same category as the object to be recommended. Further, a target recommendation set matching the object to be recommended can be selected from the candidate recommendation set, thereby inserting the object to be recommended into the target recommendation set for recommendation based on the target recommendation set. In other words, in this embodiment of the disclosure, a target recommendation set can be obtained by inserting matching objects to be recommended into a carefully designed recommendation set, and then recommending the object to be recommended. This method can match the object to be recommended with a suitable target recommendation set, thereby achieving accurate push of the recommended content through the thematic and emotional resonance of the recommendation set. Therefore, it can improve the accuracy of the recommendation to a certain extent. Moreover, since the recommendation set can strongly combine the user's emotional content with the song content, it has strong dissemination and dissemination value. Therefore, recommending the object to be recommended based on the target recommendation set can improve the dissemination of the object to be recommended to a certain extent, thereby improving the recommendation effect. In addition, since the object to be recommended is directly recommended based on the target recommendation set, the processing of the relevant data of the object to be recommended is reduced, thereby improving the recommendation efficiency to a certain extent.

[0355] Exemplary electronic devices

[0356] Exemplary embodiments of this disclosure also provide an electronic device. The electronic device may include a processor and a memory. The memory stores executable instructions for the processor, such as computer programs. The processor executes these executable instructions to perform the method steps of various exemplary embodiments of this disclosure.

[0357] The following is for reference. Figure 5 The electronic device is illustrated by way of a general-purpose computing device. It should be understood that... Figure 5 The electronic device 500 shown is merely an example and should not be construed as limiting the functionality and scope of use of the embodiments disclosed herein.

[0358] like Figure 5 As shown, the electronic device 500 may include: a processor 510, a memory 520, a bus 530, an I / O (input / output) interface 540, and a network adapter 550.

[0359] Memory 520 may include volatile memory, such as RAM 521 and cache unit 522, and may also include non-volatile memory, such as ROM 523. Memory 520 may also include one or more program modules 524, including but not limited to: operating system, one or more application programs, other program modules, and program data. Each or some combination of these examples may include an implementation of a network environment. For example, program module 524 may include the units in the above-described apparatus.

[0360] The processor 510 may include one or more processing units, such as an AP (Application Processor), a modem processor, a GPU (Graphics Processing Unit), an ISP (Image Signal Processor), a controller, an encoder, a decoder, a DSP (Digital Signal Processor), a baseband processor, and / or an NPU (Neural-Network Processing Unit).

[0361] The processor 510 can be used to execute executable instructions stored in the memory 520, such as the method for recommending objects described above, which includes the following steps: Step 201: Determine the object to be recommended; Step 202: Determine a candidate recommendation set based on the attribute information of the object to be recommended; Step 203: Select a target recommendation set that matches the object to be recommended from the candidate recommendation set; Step 204: Insert the object to be recommended into the target recommendation set to recommend the object based on the target recommendation set.

[0362] By executing the above method steps through processor 510, the object to be recommended can be determined. Then, based on the attribute information of the object to be recommended, a candidate recommendation set is determined, wherein the candidate recommendation set includes multiple objects belonging to the same category as the object to be recommended. Further, a target recommendation set matching the object to be recommended can be selected from the candidate recommendation set, thereby inserting the object to be recommended into the target recommendation set for recommendation based on the target recommendation set. In other words, in this embodiment of the present disclosure, a target recommendation set can be obtained by inserting matching objects to be recommended into a carefully designed recommendation set, and then recommending the object to be recommended. This method can match the object to be recommended with a suitable target recommendation set, thereby achieving accurate push of the recommended content through the thematic and emotional resonance of the recommendation set. Therefore, it can improve the accuracy of the recommendation to a certain extent. Moreover, since the recommendation set can strongly combine the user's emotional content with the song content, it has strong dissemination and dissemination value. Therefore, recommending the object to be recommended based on the target recommendation set can improve the dissemination of the object to be recommended to a certain extent, thereby improving the recommendation effect. In addition, since the object to be recommended is directly recommended based on the target recommendation set, the processing of the relevant data of the object to be recommended is reduced, thereby improving the recommendation efficiency of the object to be recommended to a certain extent.

[0363] Bus 530 is used to connect different components of electronic device 500 and may include a data bus, an address bus and a control bus.

[0364] Electronic device 500 can communicate with one or more external devices 600 (such as keyboard, mouse, external controller, etc.) through I / O interface 540.

[0365] Electronic device 500 can communicate with one or more networks via network adapter 550. For example, network adapter 550 can provide mobile communication solutions such as 3G / 4G / 5G, or wireless communication solutions such as wireless LAN, Bluetooth, and near-field communication. Network adapter 550 can communicate with other modules of electronic device 500 via bus 530.

[0366] although Figure 5 As not shown in the diagram, other hardware and / or software modules may also be configured in the electronic device 500, including but not limited to: microcode, device drivers, redundant processors, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0367] As can be seen from the above, the technical solutions disclosed herein can be implemented as methods, apparatus, systems, computer program products, storage media, electronic devices, etc. Those skilled in the art will understand that various aspects of this disclosure can be specifically implemented in the following forms: a completely hardware implementation, a completely software implementation (including firmware, microcode, etc.), or a combination of hardware and software implementations, which may be referred to as "circuit," "module," or "system," respectively.

[0368] It should be understood that this disclosure is not limited to the specific methods, steps, or structures described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. Those skilled in the art will readily conceive of other embodiments based on the specific implementations provided in this disclosure. Therefore, the specific implementations provided in this disclosure are merely exemplary, and the scope and spirit of this disclosure are indicated by the claims, and should cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary technical means in the art not disclosed in this disclosure.

Claims

1. A method for recommending objects, characterized in that, The method includes: Identify the candidates to be recommended; Based on the attribute information of the objects to be recommended, a candidate recommendation set is determined; From the candidate recommendation set, select the target recommendation set that matches the object to be recommended; The object to be recommended is inserted into the target recommendation set, so that the object to be recommended is recommended based on the target recommendation set.

2. The method according to claim 1, characterized in that, The step of determining the candidate recommendation set based on the attribute information of the object to be recommended includes: Based on the attribute information of the object to be recommended, determine the growth potential value corresponding to the object to be recommended; Based on the growth potential value, a target recommendation candidate pool is selected; A candidate recommendation set is determined from the target recommendation candidate pool.

3. The method according to claim 2, characterized in that, The step of selecting the target recommendation candidate pool based on the growth potential value includes: If the growth potential value is within a first preset range, then the first type of recommendation candidate pool is selected as the target recommendation candidate pool; If the growth potential value is not within the first preset range, then the second type of recommendation candidate pool is selected as the target recommendation candidate pool; The first type of recommendation candidate pool and the second type of recommendation candidate pool correspond to different recommendation channels.

4. The method according to claim 3, characterized in that, If the target recommendation candidate pool is the first type of recommendation candidate pool, the step of determining the candidate recommendation set from the target recommendation candidate pool includes: Determine the recommendation growth value of each recommendation set in the target recommendation candidate pool; The growth potential value corresponding to the object to be recommended is compared with the recommendation growth value of each recommendation set in the target recommendation candidate pool to obtain multiple first comparison results; Based on the multiple first comparison results, a candidate recommendation set is determined.

5. The method according to claim 1, characterized in that, The step of selecting a target recommendation set that matches the object to be recommended from the candidate recommendation set includes: Based on the attribute information of the object to be recommended, determine the multiple dimensional feature information corresponding to the object to be recommended; Determine multiple dimensional feature information for each of the candidate recommendation sets; The multiple dimensional feature information corresponding to the object to be recommended is matched with the multiple dimensional feature information of each recommendation set in the candidate recommendation set to obtain multiple matching values; Based on the multiple matching values ​​and preset filtering conditions, a target recommendation set matching the object to be recommended is determined.

6. The method according to claim 5, characterized in that, The determination of multiple dimensional feature information for each candidate recommendation set includes: Perform the following operations on each of the candidate recommendation sets: The overall style characteristics of the candidate recommendation set are determined based on the style information of each object in the candidate recommendation set. Based on the emotional information of each object in the candidate recommendation set, determine the emotional characteristics of the candidate recommendation set; The scene features of the candidate recommendation set are determined based on the scene information of each object in the candidate recommendation set. Based on the overall style features, the emotional features, and the scene features, the candidate recommendation set is obtained with multiple dimensions of feature information.

7. The method according to claim 6, characterized in that, The step of obtaining multi-dimensional feature information of the candidate recommendation set based on the overall style features, the emotion features, and the scene features includes: The overall style features, the emotional features, and the scene features are cross-analyzed to obtain multiple hybrid features; The overall style features, the emotional features, the scene features, and the multiple mixed features are used as multiple dimensional feature information of the recommendation set.

8. An apparatus for recommending objects, characterized in that, The device includes: The first determining unit is used to determine the objects to be recommended. The second determining unit is used to determine a candidate recommendation set based on the attribute information of the object to be recommended; A filtering unit is used to select a target recommendation set that matches the object to be recommended from the candidate recommendation set; The recommendation unit is used to insert the object to be recommended into the target recommendation set, so as to recommend the object to be recommended based on the target recommendation set.

9. An electronic device, characterized in that, include: processor; Memory for storing the executable instructions of the processor; The processor is configured to perform the method of any one of claims 1-7 by executing the executable instructions.

10. A computer program product storing a computer program thereon, characterized in that, When the computer program is executed by a processor, it implements the method described in any one of claims 1-7.

Images