Song boosting processing method and device, program product and electronic equipment

Abstract

The invention discloses a song boosting processing method and device, a computer program product and electronic equipment, and relates to the technical field of computers. The song boosting processing method comprises the following steps: in response to a song query operation, determining a target song to be put according to song query information; if the exposure frequency of the target song is in a first preset interval, taking the first boosting strategy as a target boosting strategy; the first boosting strategy is a boosting strategy for determining a putting sorting position based on a first preset sorting position; if the exposure frequency of the target song is in a second preset interval, taking the second boosting strategy as a target boosting strategy; the exposure frequency of the second preset interval is higher than that of the first preset interval, and the second boosting strategy is a boosting strategy for determining a putting sorting position based on an effective click rate; and putting the target song according to the target boosting strategy. According to the invention, song personalized boosting is carried out according to the song exposure times, so that high-quality unpopular songs can be found by users in time and accelerated to grow into popular songs.

Description

Technical Field

[0001] Embodiments of the present disclosure relate to the field of computer technologies, and more particularly, to a song boosting processing method, a song boosting processing apparatus, a computer program product, and an electronic device. Background Art

[0002] This section aims to provide background or context for the embodiments of the invention described in the claims. The description herein is not admitted to be prior art merely by virtue of being included in this section.

[0003] In a music search scenario, there are a large number of song resources that can be distributed to users, and the number of these songs is still increasing rapidly, which easily causes some high-quality unpopular songs to be buried under limited search traffic.

[0004] In related technologies, in order to avoid high-quality unpopular songs from being buried, for songs with insufficient exposure and thus lacking user interaction data, that is, cold start songs, a fixed start position is usually set to provide exposure support, that is, boosting. When a certain support time limit is exceeded or the exposure reaches a certain threshold, an exit mechanism is activated and boosting is no longer performed. This single and uniform cold start method makes it difficult for truly popular songs among users to quickly stand out and grow into popular songs.

[0005] It should be noted that the information disclosed in the above background art section is only used to enhance the understanding of the background of the present disclosure, and thus may include information that does not constitute prior art known to those of ordinary skill in the art. Summary of the Invention

[0006] The present disclosure provides a song boosting processing method, a song boosting processing apparatus, a computer program product, and an electronic device, thereby at least to some extent solving the problem that truly popular songs among cold start songs in related technologies cannot quickly grow into popular songs.

[0007] Other features and advantages of the present disclosure will become apparent through the following detailed description, or be learned in part through the practice of the present disclosure.

[0008] According to a first aspect of the present disclosure, there is provided a method for processing song boosting, the method including: in response to a song query operation, determining a target song to be launched according to song query information; if the exposure times of the target song are within a first preset interval, taking a first boosting strategy as the target boosting strategy; the first boosting strategy is a boosting strategy for determining a launch sorting position based on a first preset sorting position; if the exposure times of the target song are within a second preset interval, taking a second boosting strategy as the target boosting strategy; the exposure times in the second preset interval are higher than those in the first preset interval, and the second boosting strategy is a boosting strategy for determining a launch sorting position based on the effective click-through rate; launching the target song according to the target boosting strategy.

[0009] In an exemplary embodiment of the present disclosure, the launching the target song according to the target boosting strategy includes: if the target boosting strategy is the first boosting strategy, taking a first preset sorting position as the launch sorting position of the target song; launching the target song according to the launch sorting position of the target song.

[0010] In an exemplary embodiment of the present disclosure, the launching the target song according to the target boosting strategy includes: if the target boosting strategy is the second boosting strategy, determining whether the effective click-through rate of the target song meets a preset position rising condition; if the effective click-through rate of the target song meets the preset position rising condition, on the basis of a second preset sorting position, obtaining the launch sorting position of the target song by iteratively presetting a rising position; wherein the second preset sorting position is higher than the first preset sorting position; if the effective click-through rate of the target song does not meet the preset position rising condition, taking the second preset sorting position as the launch sorting position of the target song; launching the target song according to the launch sorting position of the target song.

[0011] In an exemplary embodiment of the present disclosure, the preset position rising condition includes: the effective click-through rate of the target song is higher than that of the previous song by a preset ratio.

[0012] In an exemplary embodiment of the present disclosure, when the effective click-through rate of the target song meets the preset condition for ranking improvement, based on the second preset ranking position, the placement ranking position of the target song is obtained by iteratively presetting the ascending ranking, including: if the effective click-through rate of the target song is higher than the effective click-through rate of the previous song by a preset ratio, taking the second preset ranking position as the initial placement ranking position of the target song, and updating the placement ranking position of the target song by iteratively presetting the ascending ranking until the effective click-through rate of the target song is not higher than the effective click-through rate of the previous song by the preset ratio or the updated placement ranking position of the target song reaches the third preset ranking position.

[0013] In an exemplary embodiment of the present disclosure, the method further includes: if the target boost strategy of the target song is the first boost strategy, real-time monitoring the exposure times of the target song; in response to the exposure times of the target song entering the second preset interval from the first preset interval, controlling to switch the target boost strategy of the target song from the first boost strategy to the second boost strategy; wherein, the first preset interval and the second preset interval are adjacent intervals.

[0014] In an exemplary embodiment of the present disclosure, the controlling to switch the target boost strategy of the target song from the first boost strategy to the second boost strategy in response to the exposure times of the target song entering the second preset interval from the first preset interval includes: in response to the exposure times of the target song entering the second preset interval from the first preset interval, determining whether the effective click-through rate of the target song is higher than the preset click-through rate threshold; if the effective click-through rate of the target song is higher than the preset click-through rate, controlling to switch the target boost strategy of the target song from the first boost strategy to the second boost strategy.

[0015] In an exemplary embodiment of the present disclosure, the method further includes: if the target boost strategy of the target song is the second boost strategy, real-time monitoring the exposure times of the target song; in response to the exposure times of the target song being greater than the preset exposure times, controlling the target song to exit the second boost strategy, where the preset exposure times is greater than or equal to the maximum value of the second preset interval.

[0016] In an exemplary embodiment of the present disclosure, the determining the target song to be placed according to the song query information in response to the song query operation includes: in response to the song query operation, extracting the target song that matches the song query information from the candidate song library according to the ranking of the songs in the candidate song library.

[0017] In an exemplary embodiment of the present disclosure, the method further includes: identifying the song quality of each song in the candidate song library; determining the sorting weight of each song in the candidate song library according to the song quality of each song in the candidate song library; the song quality is positively correlated with the sorting weight; based on the sorting weight of each song in the candidate song library, sorting each song in the candidate song library to obtain the sorting of the songs in the candidate song library.

[0018] In an exemplary embodiment of the present disclosure, the identifying the song quality of each song in the candidate song library includes: extracting multi-modal features of the song to be identified, where the song to be identified is any song in the candidate song library; determining the song quality of each song in the candidate song library based on the multi-modal features of the song to be identified.

[0019] In an exemplary embodiment of the present disclosure, the multi-modal features include any one or more of the following: song cover image features, song audio features, lyric text features, artist features.

[0020] According to a second aspect of the present disclosure, there is provided a song boosting processing device, the device includes: a song recall module, configured to respond to a song query operation and determine a target song to be delivered according to song query information; a first boosting module, configured to use a first boosting strategy as the target boosting strategy if the exposure times of the target song are within a first preset interval; the first boosting strategy is a boosting strategy for determining a delivery sorting position based on a first preset sorting position; a second boosting module, configured to use a second boosting strategy as the target boosting strategy if the exposure times of the target song are within a second preset interval; the exposure times in the second preset interval are higher than those in the first preset interval, and the second boosting strategy is a boosting strategy for determining a delivery sorting position based on the effective click-through rate; a song delivery module, configured to deliver the target song according to the target boosting strategy.

[0021] According to a third aspect of the embodiments of the present disclosure, there is provided a computer program product, including a computer program, which when executed by a processor, implements the song boosting processing method and its possible implementation manners in the first aspect above.

[0022] According to a fourth aspect of the embodiments of the present disclosure, there is provided an electronic device, including: a processor; and a memory for storing executable instructions of the processor; wherein, the processor is configured to execute the executable instructions to execute the song boosting processing method and its possible implementation manners in the first aspect above.

[0023] The technical solution of the present disclosure has the following beneficial effects:

[0024] During the above-mentioned song boosting process, in response to a song query operation, the target song to be launched is determined according to the song query information; if the exposure times of the target song are within the first preset interval, the first boosting strategy is taken as the target boosting strategy; the first boosting strategy is a boosting strategy for determining the launch sorting position based on the first preset sorting position; if the exposure times of the target song are within the second preset interval, the second boosting strategy is taken as the target boosting strategy; the exposure times in the second preset interval are higher than those in the first preset interval, and the second boosting strategy is a boosting strategy for determining the launch sorting position based on the effective click-through rate; the target song is launched according to the target boosting strategy. The present disclosure realizes personalized boosting of songs by integrating multiple boosting strategies. When the exposure times of the song are within the first preset interval, the launch sorting position is determined according to the preset sorting position, so that unpopular songs can obtain a certain initial exposure to be discovered by users in time. When the exposure times of the song are within the second preset interval, the song has obtained a certain amount of exposure. At this time, determining the launch sorting position based on the effective click-through rate can enable the songs that are truly popular among users in the cold start songs to grow into popular songs at an accelerated pace.

[0025] It should be understood that the above general description and the following detailed description are only exemplary and do not limit the present disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] The accompanying drawings herein are incorporated into the specification and constitute a part of this specification, showing embodiments consistent with the present disclosure and used together with the specification to explain the principles of the present disclosure. Obviously, the accompanying drawings in the following description are only some embodiments of the present disclosure, and those of ordinary skill in the art can obtain other drawings based on these drawings without creative efforts.

[0027] Figure 1 A flowchart showing a song boosting processing method in this exemplary embodiment;

[0028] Figure 2 A flowchart showing a song recall in this exemplary embodiment;

[0029] Figure 3 A schematic structural diagram showing a song quality classification model in this exemplary embodiment;

[0030] Figure 4 A schematic structural diagram showing a sorting position boost in this exemplary embodiment;

[0031] Figure 5 A flowchart showing a song boost in this exemplary embodiment;

[0032] Figure 6Shows the system structure diagram of a song cold start rising distribution based on large model content understanding in this exemplary embodiment;

[0033] Figure 7 Shows the structural block diagram of a song boost processing device in this exemplary embodiment;

[0034] Figure 8 Shows an electronic device for implementing the above song boost processing method in this exemplary embodiment.

[0035] In the drawings, the same or corresponding reference numerals indicate the same or corresponding parts. Detailed implementation manners

[0036] Hereinafter, the principles and spirit of the present disclosure will be described with reference to several exemplary embodiments. It should be understood that these embodiments are provided only to enable those skilled in the art to better understand and then implement the present disclosure, rather than limiting the scope of the present disclosure in any way. On the contrary, these embodiments are provided to make the present disclosure more thorough and complete, and to be able to convey the scope of the present disclosure completely to those skilled in the art.

[0037] Those skilled in the art know that the embodiments of the present disclosure can be implemented as a device, equipment, method, or computer program product. Therefore, the present disclosure can be specifically implemented in the following forms: completely hardware, completely software (including firmware, resident software, microcode, etc.), or a combination of hardware and software.

[0038] According to the embodiments of the present disclosure, a song boost processing method, a song boost processing device, a computer program product, and an electronic device are provided.

[0039] The number of any element in the drawings is for illustration rather than limitation, and any naming is only for distinction and does not have any limiting meaning.

[0040] Hereinafter, the principles and spirit of the present disclosure will be elaborated in detail with reference to several representative embodiments of the present disclosure.

[0041] In the related art of the present disclosure, a single uniform cold start method is used for song boost, making it difficult for truly popular songs among users to stand out quickly and grow into popular songs rapidly.

[0042] Based on the above problems, the present disclosure proposes a song boosting processing method, which realizes personalized boosting of songs by integrating multiple boosting strategies. When the exposure times of a song are within the first preset interval, the placement sorting position is determined according to the preset sorting position, so that unpopular songs can obtain a certain initial exposure to be discovered by users in a timely manner. When the exposure times of a song are within the second preset interval, the song has obtained a certain amount of exposure. At this time, the placement sorting position is determined based on the effective click-through rate, which can enable the truly popular songs among the cold-start songs to grow into hot songs at an accelerated pace.

[0043] Overview of Application Scenarios

[0044] It should be noted that the following application scenarios are only shown for the convenience of understanding the spirit and principle of the present disclosure, and the embodiments of the present disclosure are not limited in this regard. On the contrary, the embodiments of the present disclosure can be applied to any applicable scenario. Exemplarily, it can be applied to scenarios related to song placement such as song recommendation and single song search. The present disclosure does not make specific limitations in this regard.

[0045] Exemplary Method

[0046] Next, in combination with the above application scenarios, refer to Figure 1 to describe the song boosting processing method according to an exemplary embodiment of the present disclosure.

[0047] Please refer to Figure 1 , Figure 1 which shows a schematic flowchart of the song boosting processing method according to an exemplary embodiment of the present disclosure. As Figure 1 shown, the song boosting processing method may include:

[0048] Step S110, in response to a song query operation, determine a target song to be placed according to the song query information;

[0049] Step S120, if the exposure times of the target song are within the first preset interval, use the first boosting strategy as the target boosting strategy; the first boosting strategy is a boosting strategy for determining the placement sorting position based on the first preset sorting position;

[0050] Step S130, if the exposure times of the target song are within the second preset interval, use the second boosting strategy as the target boosting strategy; the exposure times in the second preset interval are higher than those in the first preset interval, and the second boosting strategy is a boosting strategy for determining the placement sorting position based on the effective click-through rate;

[0051] Step S140, place the target song according to the target boosting strategy.

[0052] Implement Figure 1The song boosting processing method shown realizes personalized boosting of songs by integrating multiple boosting strategies. When the exposure times of a song are within the first preset interval, the placement sorting position is determined according to the preset sorting position, so that unpopular songs can obtain a certain initial exposure to be discovered by users in time. When the exposure times of a song are within the second preset interval, the song has obtained a certain amount of exposure. At this time, the placement sorting position is determined based on the effective click-through rate, which can enable the truly popular songs among the cold-start songs to accelerate their growth into popular songs.

[0053] The following describes these steps in detail.

[0054] In step S110, in response to a song query operation, the target song to be placed is determined according to the song query information.

[0055] Among them, the song query operation refers to the triggering operation of retrieving songs from the song resource pool, which can be triggered by the user. Optionally, the song query operation can be the information input operation of the user in the song retrieval box provided by the song platform, or the click operation of the user on the song type tags provided by the song platform (for example, tags such as national style, pop, rock, etc.). The present disclosure does not make specific limitations on this. It should be noted that the song resource pool can contain a large number of various types of songs for users to query. These songs can be songs without user interaction data or songs with user interaction data. The present disclosure does not make specific limitations on this. Among them, the user interaction data can include, but is not limited to, interaction data such as likes, collections, and comments.

[0056] Among them, the song query information refers to the retrieval information used when retrieving songs from the song resource pool. For example, the information input by the user in the retrieval box, the information of the song type tags clicked by the user, etc.

[0057] Among them, the target song refers to the song retrieved from the song resource pool that matches the song query information. Optionally, the target song can be a cold-start song, that is, a song with insufficient exposure resulting in a lack of user interaction data.

[0058] In an optional implementation manner, the above-mentioned determining the target song to be placed in response to the song query operation according to the song query information can be implemented through the following steps: In response to the song query operation, the target song that matches the song query information is extracted from the candidate song library according to the sorting of the songs in the candidate song library.

[0059] Among them, the candidate music library can be a music library composed of songs in the song resource pool. The songs in the candidate music library can be sorted according to specific rules so as to be extracted and batch delivered to users in sequence. Exemplarily, for example, the songs in the candidate music library can be roughly sorted according to one or more factors such as song quality and song release time, so as to facilitate subsequent batch extraction of songs for delivery.

[0060] Exemplarily, in response to a song query operation, according to the sorting order of the songs in the candidate music library, target songs that match the preset quantity and song query information can be extracted in sequence.

[0061] In an alternative embodiment, the following steps can also be performed to sort each song in the candidate music library: identify the song quality of each song in the candidate music library; determine the sorting weight of each song in the candidate music library according to the song quality of each song in the candidate music library; the song quality is positively correlated with the sorting weight; based on the sorting weight of each song in the candidate music library, sort each song in the candidate music library to obtain the sorting of the songs in the candidate music library.

[0062] Exemplarily, the song quality of each song in the candidate music library can be identified through a song quality classification model. The song quality classification model can divide the song quality into multiple levels, such as three levels: high quality, average quality, and poor quality. It should be noted that in the actual application process, the song quality can be further divided into more detailed levels, and the present disclosure does not specifically limit this.

[0063] Exemplarily, taking the song quality being divided into three levels: high quality, average quality, and poor quality as an example, if the output of the song quality classification model is 2, it can indicate high song quality; if the output of the song quality classification model is 1, it can indicate average song quality; if the output of the song quality classification model is 0, it can indicate poor song quality.

[0064] Specifically, when determining the sorting weight of each song in the candidate music library according to the song quality of each song in the candidate music library, a higher sorting weight can be configured for high-quality songs, and a lower sorting weight can be configured for low-quality songs. The mapping relationship between song quality and sorting weight can be set in the system in advance, and the present disclosure does not specifically limit this. Exemplarily, the sorting weight of high-quality songs can be configured as a value greater than 1. For example, the sorting weight of poor-quality songs can be configured as 0.8, the sorting weight of average-quality songs can be configured as 1, and the sorting weight of high-quality songs can be configured as 1.2.

[0065] Optionally, sorting the songs in the candidate song library according to the sorting weights of the songs in the candidate song library to obtain the sorting of the songs in the candidate song library can be achieved through the following steps: obtaining the initial rough sorting scores of the songs in the candidate song library, multiplying the initial rough sorting scores of the songs in the candidate song library by the sorting weights respectively to obtain the rough sorting scores of the songs in the candidate song library that incorporate the song quality factor, and sorting the songs in the candidate song library from high to low according to the rough sorting scores of the songs in the candidate song library that incorporate the song quality factor to obtain the sorting of the songs in the candidate song library.

[0066] Among them, the initial rough sorting score refers to the quantitative score of the rough sorting of the song at the initial moment, that is, before incorporating the song quality factor.

[0067] Optionally, the initial rough sorting scores of the songs in the candidate song library can be determined according to one or more factors such as the release time and the number of exposures. The present disclosure does not specifically limit this. Exemplarily, when the number of exposures is the same, the newer the release time, the higher the initial rough sorting score; when the release times are the same, the higher the number of exposures, the higher the initial sorting score. Exemplarily, different weight values can be assigned to the songs according to one or more factors such as their release time and the number of exposures to determine the initial rough sorting scores of the songs through the weight values of the various factors.

[0068] By sorting the songs in the candidate song library by integrating the song quality factor, when querying the candidate song library, the unpopular high-quality songs that match the song query information can be recalled. Among them, the unpopular high-quality songs refer to the high-quality songs that have few exposures under the query and thus do not accumulate enough user interaction data and are ranked relatively low under the general sorting strategy.

[0069] During the process of recalling the target song, introducing a sorting weight greater than 1 for the high-quality songs (i.e., high-quality songs) recognized by the model can enable the unpopular high-quality songs related to the song query information to be recalled when the user inputs the song query information.

[0070] Exemplarily, as Figure 2 shown, a flowchart for song recall is provided, which may specifically include the following steps:

[0071] Step S201, identifying the song quality of each song in the candidate song library;

[0072] Step S202, determining the sorting weights of the songs in the candidate song library according to the song quality of the songs in the candidate song library; the song quality is positively correlated with the sorting weight;

[0073] Step S203, sorting the songs in the candidate song library based on the sorting weights of the songs in the candidate song library to obtain the sorting of the songs in the candidate song library;

[0074] Step S204, in response to the song query operation, extract the target song that matches the song query information from the candidate song library according to the sorting of the songs in the candidate song library.

[0075] In an optional implementation manner, the song quality of each song in the candidate song library can be identified through the following steps: extract the multi-modal features of the song to be identified, where the song to be identified is any song in the candidate song library; based on the multi-modal features of the song to be identified, determine the song quality of each song in the candidate song library.

[0076] Among them, the multi-modal features refer to the performance features of the song in multiple dimensions. Exemplarily, the multi-modal features may include any one or more of the following: song cover image features, song audio features, lyric text features, artist features. It should be noted that in the actual application process, the multi-modal features may also include other features that can reflect the song quality, and the present disclosure does not specifically limit this.

[0077] Exemplarily, the multi-modal features of the song to be identified can be extracted through a song quality classification model, and based on the multi-modal features of the song to be identified, the song quality of each song in the candidate song library can be obtained.

[0078] Optionally, the song quality classification model can be based on the AudioCLIP (Audio Contrastive Language-Image Pretraining) large model structure, and use multi-modal information such as song cover image features, song audio features, lyric text features, and artist features to perform three-classification on the song quality.

[0079] Among them, AudioCLIP is similar to the CLIP (Contrastive Language-Image Pretraining) method. The CLIP method is pre-trained on a large number of image-text pairs, and the model is made to understand and associate information in different modalities through contrastive learning. AudioCLIP is extended to process audio data.

[0080] Exemplarily, as Figure 3 shown, a song quality classification model structure is provided, including the following 4 layers: multi-modal feature extraction layer, multi-modal feature fusion layer, MLP (Multi-Layer Perceptron) layer, activation function layer.

[0081] Among them, the multi-modal feature extraction layer is used for feature extraction. The extracted features may include, but are not limited to, song cover image features, song audio features, lyric text features, artist features, etc. Exemplarily, the song cover image features can be extracted using a ResNet (Residual Neural Network) structure. In ResNet, a residual structure is added to the deep neural network to solve the problem of gradient disappearance in deep network training. The extracted deep features pass through a linear layer to obtain the image feature representation of the song. Exemplarily, the song audio features can be extracted using a Multi-layer 1D CNN (Multi-layer 1D Convolutional Neural Network) and a Transformer network (a neural network architecture based on the attention mechanism). First, the Multi-layer 1D CNN can be used to extract audio convolution features, and then the Transformer network is used to extract deep features to obtain the audio feature representation of the song. Exemplarily, the lyric text features can be extracted using a multi-language pre-trained model - MultiLingual BERT (Bidirectional Encoder Representations from Transformers). Exemplarily, the artist features can be extracted using an MLP (Multi-Layer Perceptron).

[0082] Among them, the multi-modal feature fusion layer is used for feature fusion. Exemplarily, in a splicing manner, multi-modal features such as song cover image features, song audio features, lyric text features, and artist features are characterized as one-dimensional linear features to obtain the multi-modal feature fusion result.

[0083] Among them, the MLP layer uses a multi-layer perceptron, that is, a multi-layer fully connected neural network, to perform representation learning on the multi-modal features to obtain the hidden relationships between different modal features.

[0084] Among them, the activation function layer uses an activation function to perform three-class classification on the hidden features learned by the MLP layer to predict the song quality level. Exemplarily, an output of 2 can indicate high song quality, an output of 1 can indicate average song quality, and an output of 0 can indicate poor song quality.

[0085] By performing representation learning on the multi-modal features through the song quality classification model in the present disclosure, the need for complex feature engineering design in traditional song quality recognition models is avoided, and the problems of insufficient feature richness and low song quality recognition accuracy in single-modal song quality recognition models are solved.

[0086] After recalling the target song, based on the exposure times of the target song, it can be determined whether to execute step S120 or step S130 next. Specifically, if the exposure times of the target song are within the first preset interval, step S120 can be executed. If the exposure times of the target song are within the second preset interval, step S130 can be executed. The exposure times of the target song can be the exposure times within n days, where n is an integer, such as 7 days, 30 days, etc., which can be specifically set by developers, and the present disclosure does not specifically limit this.

[0087] In step S120, if the exposure times of the target song are within the first preset interval, the first boosting strategy is taken as the target boosting strategy; the first boosting strategy is a boosting strategy for determining the placement sorting position based on the first preset sorting position.

[0088] Exemplarily, the first preset interval can be, for example, the interval [0, x_0), where x is the highest exposure threshold for entering the first boosting strategy preset in advance, which can be set by developers according to experience, and the present disclosure does not specifically limit this.

[0089] Exemplarily, as Figure 4 shown, when the exposure times n_day_pv of the target song within n days satisfy n_day_pv < x_0, the first boosting strategy can be taken as the target boosting strategy.

[0090] Among them, the first boosting strategy is a boosting strategy for determining the placement sorting position based on the first preset sorting position, which can achieve the cold start of the song. It should be noted that in the actual application process, the first preset sorting position can be set according to actual needs to ensure that unpopular high-quality songs can obtain a certain initial exposure, such as the 8th position, and the present disclosure does not specifically limit this.

[0091] In step S130, if the exposure times of the target song are within the second preset interval, the second boosting strategy is taken as the target boosting strategy; the exposure times in the second preset interval are higher than those in the first preset interval, and the second boosting strategy is a boosting strategy for determining the placement sorting position based on the effective click-through rate.

[0092] Exemplarily, the second preset interval can be, for example, the interval (x_1, y_1), where x_1 is the lowest exposure threshold for entering the second boosting strategy preset in advance, which can be greater than or equal to x_0, and y_1 is the highest exposure threshold for entering the second boosting strategy preset in advance. x_1 and y_1 can be set by developers according to experience, and the present disclosure does not specifically limit this. Exemplarily, when the exposure times n_day_pv of the target song within n days satisfy x_1 < n_day_pv < y_1, the second boosting strategy can be taken as the target boosting strategy.

[0093] Among them, the second boosting strategy is a boosting strategy for determining the placement sorting position based on the effective click-through rate, enabling songs with a certain amount of exposure to be personalized boosted according to their online performance effects, so as to accelerate the growth rate of popular songs.

[0094] Among them, the effective click-through rate can be the ratio of the number of times a song is clicked to the number of times it is displayed.

[0095] Optionally, if the exposure times of the target song are within the second preset interval and the effective click-through rate is higher than the preset click-through rate threshold, the second boosting strategy can be used as the target boosting strategy, enabling songs with better online performance effects to be further boosted to accelerate their growth rate. Among them, the preset click-through rate threshold is used to limit the minimum click-through rate for entering the second boosting strategy, which is between 0% and 100%. Specifically, it can be set by developers according to experience, and the present disclosure does not specifically limit this.

[0096] After determining the target boosting strategy of the song, step S140 can be continued. It should be noted that in the actual application process, the first preset interval and the second preset interval can be open intervals, closed intervals, left-closed right-open intervals, or left-open right-closed intervals. Specifically, they can be set by developers, and the present disclosure does not specifically limit this.

[0097] In step S140, the target song is placed according to the target boosting strategy.

[0098] In an optional implementation manner, the above-mentioned placing the target song according to the target boosting strategy can be achieved through the following steps: If the target boosting strategy is the first boosting strategy, the first preset sorting position is used as the placement sorting position of the target song; according to the placement sorting position of the target song, the target song is placed.

[0099] Exemplarily, as Figure 4 shown, when the exposure times of the target song within n days, n_day_pv, satisfy n_day_pv < x_0, the first preset sorting position (for example, the 8th position) can be used as the placement sorting position of the target song for placement, so as to ensure that unpopular high-quality songs can obtain a certain amount of initial exposure.

[0100] In an alternative embodiment, the above-mentioned step of promoting the target song according to the target promotion strategy can be achieved through the following steps: If the target promotion strategy is the second promotion strategy, determine whether the effective click-through rate of the target song meets the preset condition for ranking position increase; if the effective click-through rate of the target song meets the preset condition for ranking position increase, based on the second preset sorting position, obtain the placement sorting position of the target song by iteratively presetting the ascending ranking; if the effective click-through rate of the target song does not meet the preset condition for ranking position increase, use the second preset sorting position as the placement sorting position of the target song; according to the placement sorting position of the target song, promote the target song.

[0101] Among them, the second preset sorting position is higher than the first preset sorting position. It should be noted that in actual application, the second preset sorting position can be set according to actual needs to ensure that the corresponding song can be further continuously exposed, such as the 5th position, and the present disclosure does not specifically limit this.

[0102] Among them, the preset ascending ranking refers to the preset ascending ranking. Exemplarily, the preset ascending ranking can be set to 1, and in actual application, it can also be set to other values, and the present disclosure does not specifically limit this.

[0103] Through multiple rounds of ascending boosts, personalized boosting of songs is achieved, which can accelerate the growth rate of high-quality songs.

[0104] In an alternative embodiment, the preset condition for ranking position increase includes: the effective click-through rate of the target song is higher than the effective click-through rate of the previous song by a preset ratio.

[0105] Among them, the preset ratio can be the preset ratio threshold for starting the ranking position increase boost, which can be a ratio value greater than 0%, such as 60%, and the present disclosure does not specifically limit this. Exemplarily, when the effective click-through rate of the target song is greater than 60% of the effective click-through rate of the previous song, the preset condition for ranking position increase is met, and the ranking position increase boost is started, so that the ranking position of the target song continues to increase.

[0106] In an alternative embodiment, if the effective click-through rate of the target song meets the preset condition for ranking position increase, based on the second preset sorting position, obtaining the placement sorting position of the target song by iteratively presetting the ascending ranking can be achieved through the following steps: If the effective click-through rate of the target song is higher than the effective click-through rate of the previous song by a preset ratio, use the second preset sorting position as the initial placement sorting position of the target song, and update the placement sorting position of the target song by iteratively presetting the ascending ranking until the effective click-through rate of the target song is not higher than the effective click-through rate of the previous song or the updated placement sorting position of the target song reaches the third preset sorting position.

[0107] Exemplarily, when the number of exposures n_day_pv of the target song within n days satisfies x_1 < n_day_pv < y_1, it can be determined whether the effective click-through rate of the target song meets the preset position rising condition; if the effective click-through rate of the target song meets the preset position rising condition, on the basis of the second preset sorting position, iterate the preset rising position until the effective click-through rate of the target song no longer meets the preset position rising condition or the updated placement sorting position of the target song reaches the third preset sorting position.

[0108] Exemplarily, as Figure 4 shown, when the number of exposures n_day_pv of the target song within n days satisfies x_1 < n_day_pv < y_1 and n_day_ctr > min_ctr, it can be determined whether the effective click-through rate of the target song meets the preset position rising condition; if the effective click-through rate of the target song meets the preset position rising condition, on the basis of the second preset sorting position, iterate the preset rising position until the effective click-through rate of the target song no longer meets the preset position rising condition or the updated placement sorting position of the target song reaches the third preset sorting position. Where x_1 is the lowest exposure threshold for entering the second boost strategy set in advance, y_1 is the highest exposure threshold for entering the second boost strategy set in advance, n_day_ctr is the effective click-through rate of the target song within n days, and min_ctr is the preset click-through rate threshold.

[0109] Among them, the third preset sorting position refers to the highest boost position set in advance, which is higher than the second preset sorting position. When the placement sorting position of the target song reaches the third preset sorting position, for example, the 3rd position, the boost for the sorting position of the target song can be stopped.

[0110] Optionally, as Figure 4 shown, when determining whether the effective click-through rate of the target song meets the preset position rising condition for the first time, the preset ratio in the preset position rising condition can be configured as w1*pre1_cvr; when determining whether the effective click-through rate of the target song meets the preset position rising condition for the second time, the preset ratio in the preset position rising condition can be w2*pre2_cvr, and so on for the boost of the sorting position. Among them, w1*pre1_cvr and w2*pre2_cvr are used to represent the preset ratios set in advance, which can be the same or different, and the present disclosure does not specifically limit this. In addition, the preset rising positions configured for different rounds can be the same or different, and the present disclosure does not specifically limit this.

[0111] Through multi-round boosting of the ranking position, the present disclosure solves the problem in the uniform cold start method that in traffic allocation, the true differences between contents of different qualities are not considered, and high-quality contents cannot be allocated more traffic, resulting in a slow growth rate of high-quality contents and difficulty in quickly growing into popular contents.

[0112] In an alternative embodiment, if the target boosting strategy for the target song is the first boosting strategy, the exposure times of the target song can be monitored in real time; in response to the exposure times of the target song entering the second preset interval from the first preset interval, control is performed to switch the target boosting strategy of the target song from the first boosting strategy to the second boosting strategy.

[0113] Among them, the first preset interval can be, for example, [0, x_0), the second preset interval can be, for example, (x_1, y_1), and when the first preset interval and the second preset interval are adjacent intervals, x_1 can be equal to x_0.

[0114] Among them, x is the highest exposure threshold preset for entering the first boosting strategy, x_1 is the lowest exposure threshold preset for entering the second boosting strategy, and y_1 is the highest exposure threshold preset for entering the second boosting strategy.

[0115] Taking the exposure times being less than x_0 as the entry condition for the first boosting strategy and the exposure times being greater than x_0 as the exit condition for the first boosting strategy can enable the cold start song to obtain a certain initial exposure.

[0116] Taking the exposure times being within the second preset interval as the entry condition for the second boosting strategy can enable the songs that have already obtained partial exposure to accelerate growth when their online performance is good.

[0117] In an alternative embodiment, the above-mentioned control to switch the target boosting strategy of the target song from the first boosting strategy to the second boosting strategy in response to the exposure times of the target song entering the second preset interval from the first preset interval can be achieved through the following steps: in response to the exposure times of the target song entering the second preset interval from the first preset interval, determine whether the effective click-through rate of the target song is higher than the preset click-through rate threshold; if the effective click-through rate of the target song is higher than the preset click-through rate threshold, control is performed to switch the target boosting strategy of the target song from the first boosting strategy to the second boosting strategy.

[0118] Taking the exposure times of the target song being within the second preset interval and the effective click-through rate of the target song being higher than the preset click-through rate threshold as the entry condition for the second boosting strategy can enable the songs with better online performance to be further accelerated in boosting to speed up their growth rate and quickly become popular songs.

[0119] In an alternative embodiment, if the target boosting strategy for the target song is the second boosting strategy, the exposure count of the target song is monitored in real time; in response to the exposure count of the target song being greater than a preset exposure count, which is greater than or equal to the maximum value of the second preset interval, the target song is controlled to exit the second boosting strategy.

[0120] Among them, the preset exposure count can be the upper limit of the exposure count for exiting the second boosting strategy set in advance, and can be greater than or equal to the maximum value y_1 of the second preset interval.

[0121] Exemplarily, as Figure 4 shown, when the exposure count n_day_pv of the target song within n days satisfies being greater than the preset exposure count n_th, the target song is controlled to exit the second boosting strategy, so that the target song will maintain the original natural sorting content and no longer be boosted.

[0122] Exemplarily, as Figure 5 shown, a flowchart of song boosting is provided, which may specifically include the following steps:

[0123] Step S501, if the exposure count of the target song is within the first preset interval, control the target song to be boosted using the first boosting strategy, and monitor the exposure count of the target song in real time;

[0124] Step S502, in response to the exposure count of the target song entering the second preset interval from the first preset interval and the effective click-through rate of the target song being higher than the preset click-through rate threshold, control the target song to exit the first boosting strategy and be boosted using the second boosting strategy;

[0125] Step S503, in response to the exposure count of the target song being greater than the preset exposure count, control the target song to exit the second boosting strategy.

[0126] It can be understood that after boosting the song using the boosting strategy in the present disclosure, those songs that users really like to consume have accumulated sufficient interaction data. Through sorting by the sorting model, their natural sorting positions will be relatively high and they may no longer need to be boosted; for those songs that users don't like, after boosting, their natural sorting positions are still not high, and boosting them to the front positions will waste sorting positions, so there is no need to perform boosting processing anymore.

[0127] Exemplarily, as Figure 6As shown, a system structure diagram for the cold start and ascending distribution of songs based on large model content understanding is provided. The quality of songs in the candidate song library can be classified in advance through a song quality classification model, and high-quality songs can be screened out. By responding to the song query operation of users, rough ranking and weighting are performed on the high-quality songs identified by the model to recall unpopular high-quality songs related to the song query information. The recalled unpopular high-quality songs related to the song query information can be used as target songs, and the first boosting strategy and the second boosting strategy are used for boosting to achieve the cold start and ascending distribution of songs.

[0128] It should be noted that the rough ranking and weighting here refer to assigning a ranking weight greater than 1 to the identified high-quality songs to magnify the rough ranking score of the high-quality songs by multiples, so as to improve the ranking of high-quality songs in the candidate song library. Unpopular high-quality songs can be high-quality songs identified by the song quality classification model, but songs that have not accumulated enough user interaction data due to low exposure under queries. The stage of using the first boosting strategy for song boosting can be called the uniform quantity guarantee stage. In this stage, the same cold start position, that is, the first preset ranking position, can be given to unpopular high-quality songs to ensure that each unpopular high-quality song can obtain a certain initial exposure. The stage of using the second boosting strategy for song boosting can be called the boosting amplification stage. In this stage, for songs that have obtained partial exposure and the effective click-through rate is greater than the preset click-through rate threshold, multiple rounds of position ascending boosting are performed according to the real-time calculated effective click-through rate, which can accelerate the growth rate of high-quality content.

[0129] Exemplary Device

[0130] After introducing the method of the exemplary embodiment of the present disclosure, next, reference is made to Figure 7 to describe the song boosting processing device of the exemplary embodiment of the present disclosure.

[0131] Please refer to Figure 7 , Figure 7 which shows a structural block diagram of a song boosting processing device according to an exemplary embodiment of the present disclosure. As Figure 7 shown, a song boosting processing device 700 according to an exemplary embodiment of the present disclosure includes: a song recall module 710, a first boosting module 720, a second boosting module 730, and a song placement module 740, where:

[0132] The song recall module 710 is configured to determine target songs to be placed in response to a song query operation according to song query information;

[0133] The first boosting module 720 is configured to use the first boosting strategy as the target boosting strategy if the exposure times of the target song are within the first preset interval; the first boosting strategy is a boosting strategy for determining the placement ranking position based on the first preset ranking position;

[0134] A second boosting module 730, configured to use the second boosting strategy as the target boosting strategy if the exposure count of the target song is within a second preset range; the exposure count in the second preset range is higher than that in the first preset range, and the second boosting strategy is a boosting strategy for determining the placement sorting position based on the effective click-through rate.

[0135] A song placement module 740, configured to place the target song according to the target boosting strategy.

[0136] In an optional implementation manner, based on the foregoing solution, the song placement module 740 may be configured to: if the target boosting strategy is the first boosting strategy, use the first preset sorting position as the placement sorting position of the target song; and place the target song according to the placement sorting position of the target song.

[0137] In an optional implementation manner, based on the foregoing solution, the song placement module 740 includes: a condition determination module, configured to determine whether the effective click-through rate of the target song meets a preset position-ascending condition if the target boosting strategy is the second boosting strategy; a position-ascending iteration module, configured to, if the effective click-through rate of the target song meets the preset position-ascending condition, obtain the placement sorting position of the target song by iteratively presetting an ascending position based on a second preset sorting position, where the second preset sorting position is higher than the first preset sorting position; a sorting position determination module, configured to use the second preset sorting position as the placement sorting position of the target song if the effective click-through rate of the target song does not meet the preset position-ascending condition; and a placement processing module, configured to place the target song according to the placement sorting position of the target song.

[0138] In an optional implementation manner, based on the foregoing solution, the preset position-ascending condition includes: the effective click-through rate of the target song is higher than that of the previous song by a preset ratio.

[0139] In an optional implementation manner, based on the foregoing solution, the position-ascending iteration module may be configured to: if the effective click-through rate of the target song is higher than that of the previous song by a preset ratio, use the second preset sorting position as the initial placement sorting position of the target song, and update the placement sorting position of the target song by iteratively presetting an ascending position until the effective click-through rate of the target song is not higher than that of the previous song by the preset ratio or the updated placement sorting position of the target song reaches a third preset sorting position.

[0140] In an alternative embodiment, based on the foregoing solution, the song boosting processing device 700 may include: a first monitoring module configured to, if the target boosting strategy for the target song is the first boosting strategy, monitor the exposure count of the target song in real time; a boosting strategy switching module configured to, in response to the exposure count of the target song entering a second preset range from a first preset range, control the target boosting strategy of the target song to be switched from the first boosting strategy to the second boosting strategy; wherein the first preset range and the second preset range are adjacent ranges.

[0141] In an alternative embodiment, based on the foregoing solution, the boosting strategy switching module may be configured to: in response to the exposure count of the target song entering a second preset range from a first preset range, determine whether the effective click-through rate of the target song is higher than a preset click-through rate threshold; if the effective click-through rate of the target song is higher than the preset click-through rate threshold, control the target boosting strategy of the target song to be switched from the first boosting strategy to the second boosting strategy.

[0142] In an alternative embodiment, based on the foregoing solution, the song boosting processing device 700 may include: a second monitoring module configured to, if the target boosting strategy for the target song is the second boosting strategy, monitor the exposure count of the target song in real time; a boosting exit module configured to, in response to the exposure count of the target song being greater than a preset exposure count, control the target song to exit the second boosting strategy, where the preset exposure count is greater than or equal to the maximum value of the second preset range.

[0143] In an alternative embodiment, based on the foregoing solution, the song recall module 710 may be configured to: in response to a song query operation, extract a target song that matches the song query information from the candidate song library according to the sorting of the songs in the candidate song library.

[0144] In an alternative embodiment, based on the foregoing solution, the song boosting processing device 700 may include: a song quality identification module configured to identify the song quality of each song in the candidate song library; a sorting weight determination module configured to determine the sorting weight of each song in the candidate song library according to the song quality of each song in the candidate song library; the song quality is positively correlated with the sorting weight; a song rough sorting module configured to sort each song in the candidate song library based on the sorting weight of each song in the candidate song library to obtain the sorting of the songs in the candidate song library.

[0145] In an alternative embodiment, based on the foregoing solution, the song quality identification module may be configured to: extract the multi-modal features of the song to be identified, where the song to be identified is any song in the candidate song library; based on the multi-modal features of the song to be identified, determine the song quality of each song in the candidate song library.

[0146] In an alternative embodiment, based on the foregoing solution, the multimodal features include any one or more of the following: song cover image features, song audio features, lyric text features, and artist features.

[0147] Implementing the embodiments of the present disclosure, through the integration of multiple boosting strategies, personalized boosting of songs is achieved. When the exposure times of a song are within the first preset interval, the placement sorting position is determined according to the preset sorting position, so that unpopular songs can obtain a certain initial exposure to be discovered by users in a timely manner. When the exposure times of a song are within the second preset interval, the song has obtained a certain amount of exposure. At this time, the placement sorting position is determined based on the effective click-through rate, which can enable the songs that are truly popular among users in the cold start to grow into popular songs at an accelerated pace.

[0148] It should be noted that although several modules or units of the song boosting processing device are mentioned in the above detailed description, this division is not mandatory. In fact, according to the embodiments of the present disclosure, the features and functions of the 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.

[0149] Exemplary Program Product

[0150] After introducing the device of the exemplary embodiments of the present disclosure, next, the program product of the exemplary embodiments of the present disclosure will be described.

[0151] The exemplary embodiments of the present disclosure also provide a computer program product. The computer program product includes a computer program, and when the computer program is executed by a processor, the above-mentioned song boosting processing method is implemented.

[0152] 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 signals such as electricity, magnetism, light, electromagnetic, infrared, etc., including but not limited to: random access memory (RAM), read-only memory (ROM), magnetic tape, floppy disk, flash memory (Flash), mechanical hard disk (HDD), solid-state drive (SSD), etc. Exemplarily, the computer program product can be implemented as a non-volatile storage medium storing the computer program, such as read-only memory, Nand Flash, etc.

[0153] In one embodiment, the computer program product can be an intangible product containing a computer program. Exemplarily, the computer program product can be implemented as a virtual digital product, such as an executable file storing the computer program, an installation package and other digital files.

[0154] The code of a computer program can be written in one or more programming languages. Examples of programming languages include C, Java, C++, etc. The program code can be executed entirely on the user's computing device, or partially on the user's computing device, or as an independent software package, or 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 a remote computing device, the remote computing device can be connected to the user's computing device through any type of network, such as a local area network (LAN), wide area network (WAN), etc., or it can be connected to an external computing device (e.g., through an Internet connection provided by an operator).

[0155] A computer program can be carried or transmitted by signals such as electricity, magnetism, light, electromagnetic, infrared, etc. An electronic device can convert the signal carrying the computer program into a digital signal and then run the computer program. When the computer program runs on an electronic device, its code is used to cause the electronic device to execute (more specifically, to cause the processor of the electronic device to execute) the method steps of various exemplary embodiments of the present disclosure. Exemplarily, the following steps may be included:

[0156] In response to a song query operation, determine the target song to be played according to the song query information;

[0157] If the exposure count of the target song is within the first preset range, use the first boosting strategy as the target boosting strategy; the first boosting strategy is a boosting strategy for determining the placement sorting position based on the first preset sorting position;

[0158] If the exposure count of the target song is within the second preset range, use the second boosting strategy as the target boosting strategy; the exposure count in the second preset range is higher than that in the first preset range, and the second boosting strategy is a boosting strategy for determining the placement sorting position based on the effective click-through rate;

[0159] Place the target song according to the target boosting strategy.

[0160] In an alternative embodiment, based on the foregoing solution, the step of placing the target song according to the target boosting strategy can be implemented through the following steps: If the target boosting strategy is the first boosting strategy, use the first preset sorting position as the placement sorting position of the target song; place the target song according to the placement sorting position of the target song.

[0161] In an alternative embodiment, based on the foregoing solution, for the above-mentioned target song to be launched according to the target boosting strategy, it can be achieved through the following steps: If the target boosting strategy is the second boosting strategy, determine whether the effective click-through rate of the target song meets the preset position-rising condition; if the effective click-through rate of the target song meets the preset position-rising condition, based on the second preset sorting position, obtain the launch sorting position of the target song by iteratively presetting the rising position; wherein the second preset sorting position is higher than the first preset sorting position; if the effective click-through rate of the target song does not meet the preset position-rising condition, use the second preset sorting position as the launch sorting position of the target song; and launch the target song according to the launch sorting position of the target song.

[0162] In an alternative embodiment, based on the foregoing solution, the above-mentioned preset position-rising condition includes: the effective click-through rate of the target song is higher than the effective click-through rate of the previous song by a preset ratio.

[0163] In an alternative embodiment, based on the foregoing solution, for the above-mentioned step of obtaining the launch sorting position of the target song by iteratively presetting the rising position based on the second preset sorting position when the effective click-through rate of the target song meets the preset position-rising condition, it can be achieved through the following steps: If the effective click-through rate of the target song is higher than the effective click-through rate of the previous song by a preset ratio, use the second preset sorting position as the initial launch sorting position of the target song, and update the launch sorting position of the target song by iteratively presetting the rising position until the effective click-through rate of the target song is not higher than the effective click-through rate of the previous song by the preset ratio or the updated launch sorting position of the target song reaches the third preset sorting position.

[0164] In an alternative embodiment, based on the foregoing solution, the following steps can also be performed: If the target boosting strategy of the target song is the first boosting strategy, monitor the exposure times of the target song in real time; in response to the exposure times of the target song entering the second preset interval from the first preset interval, control to switch the target boosting strategy of the target song from the first boosting strategy to the second boosting strategy; wherein, the first preset interval and the second preset interval are adjacent intervals.

[0165] In an alternative embodiment, based on the foregoing solution, for the above-mentioned step of controlling to switch the target boosting strategy of the target song from the first boosting strategy to the second boosting strategy in response to the exposure times of the target song entering the second preset interval from the first preset interval, it can be achieved through the following steps: In response to the exposure times of the target song entering the second preset interval from the first preset interval, determine whether the effective click-through rate of the target song is higher than the preset click-through rate threshold; if the effective click-through rate of the target song is higher than the preset click-through rate threshold, control to switch the target boosting strategy of the target song from the first boosting strategy to the second boosting strategy.

[0166] In an alternative embodiment, based on the foregoing solution, the following steps may further be performed: If the target boosting strategy for the target song is the second boosting strategy, the exposure times of the target song are monitored in real time; in response to the exposure times of the target song being greater than the preset exposure times, the target song is controlled to exit the second boosting strategy, and the preset exposure times are greater than or equal to the maximum value of the second preset interval.

[0167] In an alternative embodiment, based on the foregoing solution, in response to a song query operation, the target song to be delivered is determined according to the song query information, which may be implemented by the following steps: In response to the song query operation, the target song matching the song query information is extracted from the candidate song library according to the sorting of the songs in the candidate song library.

[0168] In an alternative embodiment, based on the foregoing solution, the following steps may further be performed: The song quality of each song in the candidate song library is identified; according to the song quality of each song in the candidate song library, the sorting weight of each song in the candidate song library is determined; the song quality is positively correlated with the sorting weight; based on the sorting weight of each song in the candidate song library, each song in the candidate song library is sorted to obtain the sorting of the songs in the candidate song library.

[0169] In an alternative embodiment, based on the foregoing solution, the identification of the song quality of each song in the candidate song library may be implemented by the following steps: The multi-modal features of the song to be identified are extracted, where the song to be identified is any song in the candidate song library; based on the multi-modal features of the song to be identified, the song quality of each song in the candidate song library is determined.

[0170] In an alternative embodiment, based on the foregoing solution, the multi-modal features include any one or more of the following: song cover picture features, song audio features, lyric text features, artist features.

[0171] Implementing the embodiments of the present disclosure realizes personalized boosting of songs by integrating multiple boosting strategies. When the exposure times of the song are within the first preset interval, the delivery sorting position is determined according to the preset sorting position, so that unpopular songs can obtain a certain initial exposure to be discovered by users in time. When the exposure times of the song are within the second preset interval, the song has obtained a certain exposure. At this time, the delivery sorting position is determined based on the effective click-through rate, which can enable the songs that are really popular among users in the cold start songs to accelerate their growth into popular songs.

[0172] Exemplary Electronic Device

[0173] After introducing the song boosting processing method, device and program product of the exemplary embodiments of the present disclosure, next, an electronic device according to another exemplary embodiment of the present disclosure is introduced.

[0174] Exemplary embodiments of the present disclosure also provide an electronic device capable of implementing the above-mentioned song boosting processing method. The electronic device may include a processor and a memory. The memory stores executable instructions of the processor, which may be program code. The processor executes the method in this exemplary embodiment by executing the executable instructions. In addition, the electronic device may further include a display for displaying a graphical user interface.

[0175] Reference is made below Figure 8 to an exemplary description of the electronic device in the form of a general-purpose computing device. It should be understood that Figure 8 the electronic device 800 shown is merely an example and should not impose limitations on the functions and usage scope of the embodiments of the present disclosure.

[0176] As Figure 8 shown, the electronic device 800 may include: a processor 810, a memory 820, a bus 830, an I / O (input / output) interface 840, a network adapter 850, and a display 860.

[0177] The memory 820 may include volatile memory, such as RAM 821 and a cache unit 822, and may also include non-volatile memory, such as ROM 823. The memory 820 may further include one or more program modules 824. Such program modules 824 include, but are not limited to: an operating system, one or more application programs, other program modules, and program data. Each or some combination of these examples may include the implementation of a network environment. For example, the program module 824 may include each module in the above-mentioned device.

[0178] The processor 810 may include one or more processing units. For example, the processor 810 may include 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), etc.

[0179] The processor 810 is available for executing the executable instructions stored in the memory 820, such as executing any one or more method steps in this exemplary embodiment.

[0180] Exemplarily, the processor 810 may execute the following steps:

[0181] In response to a song query operation, determine a target song to be launched according to the song query information;

[0182] If the exposure count of the target song is within a first preset range, use the first boosting strategy as the target boosting strategy; the first boosting strategy is a boosting strategy for determining the launch sorting position based on a first preset sorting position;

[0183] If the exposure count of the target song is within a second preset range, use the second boosting strategy as the target boosting strategy; the exposure count in the second preset range is higher than that in the first preset range, and the second boosting strategy is a boosting strategy for determining the launch sorting position based on the effective click-through rate;

[0184] Launch the target song according to the target boosting strategy.

[0185] In an optional implementation manner, based on the foregoing solution, the above-mentioned launching of the target song according to the target boosting strategy can be implemented through the following steps: If the target boosting strategy is the first boosting strategy, use the first preset sorting position as the launch sorting position of the target song; launch the target song according to the launch sorting position of the target song.

[0186] In an optional implementation manner, based on the foregoing solution, the above-mentioned launching of the target song according to the target boosting strategy can be implemented through the following steps: If the target boosting strategy is the second boosting strategy, determine whether the effective click-through rate of the target song meets the preset position-ascending condition; if the effective click-through rate of the target song meets the preset position-ascending condition, based on the second preset sorting position, obtain the launch sorting position of the target song by iteratively presetting the ascending position; wherein the second preset sorting position is higher than the first preset sorting position; if the effective click-through rate of the target song does not meet the preset position-ascending condition, use the second preset sorting position as the launch sorting position of the target song; launch the target song according to the launch sorting position of the target song.

[0187] In an optional implementation manner, based on the foregoing solution, the above-mentioned preset position-ascending condition includes: the effective click-through rate of the target song is higher than the effective click-through rate of the previous song by a preset ratio.

[0188] In an alternative embodiment, based on the foregoing solution, if the effective click-through rate of the target song meets the preset condition for ascending ranking, based on the second preset sorting position, by iteratively presetting the ascending ranking, the placement sorting position of the target song can be obtained, which can be achieved through the following steps: If the effective click-through rate of the target song is higher than the effective click-through rate of the previous song by a preset ratio, take the second preset sorting position as the initial placement sorting position of the target song, and update the placement sorting position of the target song by iteratively presetting the ascending ranking until the effective click-through rate of the target song is not higher than the effective click-through rate of the previous song by the preset ratio or the updated placement sorting position of the target song reaches the third preset sorting position.

[0189] In an alternative embodiment, based on the foregoing solution, the following steps may also be performed: If the target boost strategy of the target song is the first boost strategy, monitor the exposure count of the target song in real time; in response to the exposure count of the target song entering the second preset range from the first preset range, control to switch the target boost strategy of the target song from the first boost strategy to the second boost strategy; wherein, the first preset range and the second preset range are adjacent ranges.

[0190] In an alternative embodiment, based on the foregoing solution, the control to switch the target boost strategy of the target song from the first boost strategy to the second boost strategy in response to the exposure count of the target song entering the second preset range from the first preset range can be achieved through the following steps: In response to the exposure count of the target song entering the second preset range from the first preset range, determine whether the effective click-through rate of the target song is higher than the preset click-through rate threshold; if the effective click-through rate of the target song is higher than the preset click-through rate threshold, control to switch the target boost strategy of the target song from the first boost strategy to the second boost strategy.

[0191] In an alternative embodiment, based on the foregoing solution, the following steps may also be performed: If the target boost strategy of the target song is the second boost strategy, monitor the exposure count of the target song in real time; in response to the exposure count of the target song being greater than the preset exposure count, control the target song to exit the second boost strategy, where the preset exposure count is greater than or equal to the maximum value of the second preset range.

[0192] In an alternative embodiment, based on the foregoing solution, in response to a song query operation, the target song to be placed can be determined according to the song query information, which can be achieved through the following steps: In response to the song query operation, extract the target song that matches the song query information from the candidate song library according to the sorting of the songs in the candidate song library.

[0193] In an alternative embodiment, based on the foregoing solution, the following steps may further be performed: identifying the song quality of each song in the candidate song library; determining the sorting weight of each song in the candidate song library according to the song quality of each song in the candidate song library; the song quality is positively correlated with the sorting weight; sorting each song in the candidate song library based on the sorting weight of each song in the candidate song library to obtain the sorting of the songs in the candidate song library.

[0194] In an alternative embodiment, based on the foregoing solution, the above-mentioned identifying the song quality of each song in the candidate song library may be implemented by the following steps: extracting the multi-modal features of the song to be identified, where the song to be identified is any song in the candidate song library; determining the song quality of each song in the candidate song library based on the multi-modal features of the song to be identified.

[0195] In an alternative embodiment, based on the foregoing solution, the multi-modal features include any one or more of the following: song cover image features, song audio features, lyric text features, artist features.

[0196] Implementing the embodiments of the present disclosure realizes personalized boosting of songs by integrating multiple boosting strategies. When the exposure times of a song are within the first preset interval, the placement sorting position is determined according to the preset sorting position, so that unpopular songs can obtain a certain initial exposure to be discovered by users in time. When the exposure times of a song are within the second preset interval, the song has obtained a certain amount of exposure. At this time, the placement sorting position is determined based on the effective click-through rate, which can enable the songs that are truly popular among users in the cold start songs to accelerate their growth into popular songs.

[0197] The bus 830 is used to realize the connection between different components of the electronic device 800, and may include a data bus, an address bus, and a control bus.

[0198] The electronic device 800 may communicate with one or more external devices 900 (such as a keyboard, a mouse, an external controller, etc.) through the I / O interface 840.

[0199] The electronic device 800 may communicate with one or more networks through the network adapter 850. For example, the network adapter 850 may provide mobile communication solutions such as 3G / 4G / 5G, or provide wireless communication solutions such as wireless local area network, Bluetooth, and near field communication. The network adapter 850 may communicate with other modules of the electronic device 800 through the bus 830.

[0200] The electronic device 800 may display a graphical user interface, etc. through the display 860.

[0201] Although Figure 8Although not shown in the figure, other hardware and / or software modules may also be provided in the electronic device 800, including but not limited to: a display, microcode, device drivers, redundant processors, external disk drive arrays, RAID (Redundant Arrays of Independent Disks) systems, tape drives, and data backup storage systems, etc.

[0202] It should be noted that although several modules or units of devices for action execution are mentioned in the above detailed description, this division is not mandatory. In fact, according to the exemplary embodiments of the present 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.

[0203] Those skilled in the art can understand that various aspects of the present disclosure can be implemented as a system, a method, or a program product. Therefore, various aspects of the present disclosure can be specifically implemented in the following forms: a complete hardware implementation, a complete software implementation (including firmware, microcode, etc.), or an implementation combining hardware and software aspects, which can be collectively referred to as "circuits", "modules", or "systems" here. After considering the specification and practicing the invention disclosed herein, those skilled in the art will readily think of other embodiments of the present disclosure. This application aims to cover any variations, uses, or adaptations of the present disclosure, which follow the general principles of the present disclosure and include common general knowledge or conventional technical means in the technical field not disclosed in the present disclosure. The specification and embodiments are only regarded as exemplary, and the true scope and spirit of the present disclosure are pointed out by the claims.

[0204] It should be understood that the present disclosure is not limited to the exact structures already described and shown in the figures, and various modifications and changes can be made without departing from its scope. The scope of the present disclosure is only defined by the appended claims.

Claims

1. A song boosting processing method, characterized in that: The method comprises: In response to the song query operation, determining a target song to be played according to the song query information; If the exposure times of the target song are in the first preset interval, the first boosting strategy is used as the target boosting strategy; the first boosting strategy is a boosting strategy that determines the placement ranking position based on the first preset ranking position; If the exposure times of the target song are in the second preset interval, the second boosting strategy is used as the target boosting strategy; if the exposure times of the second preset interval are higher than the first preset interval, the second boosting strategy is a boosting strategy that determines the placement ranking position based on the effective click rate; According to the target boosting strategy, the target song is released.

2. The method according to claim 1, characterized in that The method of placing the target song according to the target boosting strategy includes: If the target boosting strategy is the second boosting strategy, determining whether the effective click rate of the target song meets a preset ranking increase condition; If the effective click rate of the target song meets the preset ranking ascending condition, the placement ranking position of the target song is obtained by iterating the preset ascending ranking on the basis of the second preset ranking position; wherein the second preset ranking position is higher than the first preset ranking position; If the effective click rate of the target song does not meet the preset ranking ascending condition, the second preset ranking position is used as the placement ranking position of the target song; The target song is delivered according to the delivery ranking position of the target song.

3. The method according to claim 2, characterized in that The preset ranking ascending conditions include: The effective click rate of the target song is higher than the effective click rate of the previous song by a preset ratio.

4. The method according to claim 3, characterized in that If the effective click rate of the target song meets the preset ranking ascending condition, based on the second preset ranking position, the placement ranking position of the target song is obtained by iterating the preset ascending ranking, including: If the effective click rate of the target song is higher than that of the previous song by a preset proportion, the second preset sorting position is used as the initial sorting position of the target song, and the sorting position of the target song is updated by iterating the preset ascending order until the effective click rate of the target song is no higher than that of the previous song by a preset proportion or the updated sorting position of the target song reaches the third preset sorting position.

5. The method according to claim 1, characterized in that: In response to the song query operation, determining the target song to be delivered according to the song query information includes: In response to the song query operation, target songs matching the song query information are extracted from the candidate music library according to the ranking of the songs in the candidate music library.

6. The method according to claim 5, characterized in that The method comprises: Identifying the song quality of each song in the candidate music library; Determine the ranking weight of each song in the candidate music library according to the song quality of each song in the candidate music library; the song quality is positively correlated with the ranking weight; Based on the ranking weight of each song in the candidate music library, the songs in the candidate music library are ranked to obtain the ranking of the songs in the candidate music library.

7. The method according to claim 6, characterized in that The identifying the song quality of each song in the candidate music library includes: Extracting multimodal features of a song to be identified, wherein the song to be identified is any song in the candidate song library; Based on the multimodal features of the songs to be identified, the song quality of each song in the candidate music library is determined.

8. A song boosting processing device, characterized in that: The device comprises: A song recall module, for responding to a song query operation and determining a target song to be placed according to the song query information; A first boosting module, configured to use a first boosting strategy as a target boosting strategy if the exposure times of the target song are within a first preset interval; the first boosting strategy is a boosting strategy for determining a placement ranking position based on a first preset ranking position; A second boosting module, configured to use a second boosting strategy as a target boosting strategy if the exposure times of the target song are within a second preset interval; if the exposure times of the second preset interval are higher than those of the first preset interval, the second boosting strategy is a boosting strategy for determining a placement ranking position based on an effective click rate; The song delivery module is used to deliver the target song according to the target boosting strategy.

9. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

10. An electronic device, characterized in that: include: processor; as well as A memory, configured to store executable instructions of the processor; The processor is configured to perform the method of any one of claims 1 to 7 by executing the executable instructions.

Images