Multimedia content intelligent recommendation display control method and system

Abstract

The invention provides a multimedia content intelligent recommendation display control method and system, and relates to the technical field of multimedia, and the method comprises the steps: obtaining a recommendation list, type information and a recommendation coefficient; acquiring environment audio and position information; obtaining the watching duration of the played multiple pieces of multimedia content; determining an environment content adjustment coefficient according to the environment audio, the position information and the type information; determining a watching time length adjustment coefficient according to the watching time length, the content time length of the multimedia content and the type information; and according to the environment content adjustment coefficient, the watching duration adjustment coefficient and the recommendation coefficient, determining an adjusted recommendation coefficient and a recommendation list. According to the method and the device, the recommendation coefficient of the multimedia content which is not suitable for playing can be reduced, and the recommendation amount and the playing amount of the multimedia content which is not suitable for playing can be reduced, so that the possibility of privacy leakage of a user is reduced, and the playing safety and the recommendation accuracy of the multimedia content are improved.

Description

Technical Field

[0001] The present invention relates to the field of multimedia technologies, and in particular, to an intelligent recommendation display control method and system for multimedia content. Background Art

[0002] In the related art, CN119357476A provides a method, apparatus, storage medium, and electronic device for recommending multimedia resources. The method includes: predicting a first account feature of a target account and a reference feature of each multimedia resource included in the first browsing information according to the first browsing information; predicting a second account feature of the target account according to the second browsing information and the reference feature; predicting a target browsing feature of the target account according to the first account feature and the second account feature; and recommending a target multimedia resource matching the target browsing feature to the target account. Through this solution, the problem that the matching degree between the multimedia resources recommended for the target account and the current resource preference of the target account for multimedia resources is relatively low is solved, and the matching degree between the multimedia resources recommended for the target account and the current resource preference of the target account for multimedia resources is improved.

[0003] CN119179816A discloses an intelligent content recommendation system, which relates to the technical field of recommendation methods, and includes an information scraping module, an emotion analysis module, a data processing module, a recommendation module, a login module, a detection module, an exception handling module, and a display module. The information scraping module scrapes user data and data of content types to be recommended; the emotion analysis module is used to obtain the current emotion type of the user; the data processing module is used for data processing and uploading content keywords to the total content keyword list. This solution classifies the user emotion type by using an emotion analysis module, increases the diversity of user samples on the basis of traditional user browsing content information and historical behavior data, expands the application range of traditional collaborative filtering algorithms, recommends content to users according to user emotion changes, combines a detection module, a recommendation module, and a screening module, details and eliminates content that users are not interested in, and more carefully meets the search needs of users, thereby improving the accuracy of recommendation preferences for users.

[0004] CN119293340A discloses a content recommendation method, device, equipment, storage medium and product, which relates to the technical field of large models, and includes obtaining multiple feature information of a target user; performing natural language splicing on the multiple feature information to obtain a user portrait description text; inputting the user portrait description text into a feature fusion model to obtain the fusion feature of the target user; the feature fusion model is a BERT model; inputting the fusion feature into a content recommendation model to obtain a content recommendation prediction result. This solution can fully capture the context semantic information of natural language text through multi-dimensional feature screening and natural language splicing, and combined with the BERT model for semantic encoding, so that the trained feature fusion model can better understand the user's needs and preferences, so that the output fusion feature provides personalized and contextual processing for users, making the content recommendation using the fusion feature more in line with the user's preferences, and realizing more accurate and user-specific recommendation.

[0005] Therefore, in the related art, although it is possible to recommend content that users are interested in, when recommending, it does not consider whether the environment where the user is located is suitable for playing this content. For example, some content related to medicine may disclose the user's privacy such as diseases if played in public. Therefore, although the user is interested in this content, it is not convenient to play this content in the environment where the user is located, and the related art does not consider the issues of the user's environment and user privacy.

[0006] The information disclosed in the background art part of this application is only intended to deepen the understanding of the general background art of this application, and should not be regarded as an admission or any form of implication that this information constitutes the prior art known to those skilled in the art. Summary of the Invention

[0007] The present invention provides a method and system for intelligent recommendation display control of multimedia content, which can solve the technical problem that the related art does not consider whether the environment where the user is located is suitable for playing the recommended content.

[0008] According to the first aspect of the present invention, a method for intelligent recommendation display control of multimedia content is provided, including:

[0009] When starting a multimedia playback application on a terminal device, obtain a recommendation list of multimedia content, as well as the type information and recommendation coefficients of multiple multimedia content in the recommendation list;

[0010] Through the audio receiving component of the terminal device, obtain the ambient audio of the environment where the terminal device is located within a preset time period;

[0011] Through the positioning component of the terminal device, obtain the location information of the terminal device when the preset time period ends;

[0012] Within the preset time period, obtain the viewing durations of multiple multimedia contents that have been played;

[0013] Determine an environmental content adjustment coefficient according to the environmental audio, the location information, and the type information;

[0014] Determine a viewing duration adjustment coefficient according to the viewing durations, the content durations of the multiple multimedia contents that have been played, and the type information;

[0015] Determine an adjusted recommendation coefficient for the multimedia contents that have not been played according to the environmental content adjustment coefficient, the viewing duration adjustment coefficient, and the recommendation coefficient;

[0016] Obtain an adjusted multimedia content recommendation list according to the adjusted recommendation coefficient.

[0017] According to a second aspect of the present invention, there is provided an intelligent recommendation display control system for multimedia contents, including:

[0018] An acquisition module, configured to obtain a recommendation list of multimedia contents, as well as the type information and recommendation coefficient of multiple multimedia contents in the recommendation list, when starting a multimedia playback application on a terminal device;

[0019] An environmental audio module, configured to obtain the environmental audio of the environment where the terminal device is located within a preset time period through an audio receiving component of the terminal device;

[0020] A location information module, configured to obtain the location information of the terminal device at the end of the preset time period through a positioning component of the terminal device;

[0021] A viewing duration module, configured to obtain the viewing durations of multiple multimedia contents that have been played within the preset time period;

[0022] An environmental content adjustment coefficient module, configured to determine an environmental content adjustment coefficient according to the environmental audio, the location information, and the type information;

[0023] A viewing duration adjustment coefficient module, configured to determine a viewing duration adjustment coefficient according to the viewing durations, the content durations of the multiple multimedia contents that have been played, and the type information;

[0024] An adjustment module, configured to determine an adjusted recommendation coefficient for the multimedia contents that have not been played according to the environmental content adjustment coefficient, the viewing duration adjustment coefficient, and the recommendation coefficient;

[0025] A recommendation list module, configured to obtain an adjusted multimedia content recommendation list according to the adjusted recommendation coefficient.

[0026] By adopting the above technical solution, the present invention can achieve the following technical effects:

[0027] According to the present invention, after obtaining the recommendation list, based on the environmental audio and location information of the terminal device's location, as well as parameters such as the viewing duration of the played multimedia content, etc., it can be determined whether a specific type of multimedia content is suitable for playing in the current environment, and a recommendation coefficient after the recommendation of the multimedia content can be obtained, so as to reduce the recommendation coefficient of the multimedia content that is not suitable for playing, reduce the recommended quantity and playback volume of the multimedia content that is not suitable for playing, thereby reducing the possibility of user privacy leakage, and improving the playback security and recommendation accuracy of the multimedia content. When determining the environmental privacy coefficient, the audio type annotation vector in the case of the user being alone can be set, so that the probability of the user being alone can be determined through vector operations, and the environmental solitude probability information can also be determined through the user's location information, so as to comprehensively determine the environmental privacy coefficient of the current environment from multiple aspects, so as to accurately and objectively describe the privacy degree of the user's current environment and the possibility of the user being alone. When training the audio recognition model, different weights can be set through a conditional function in the process of determining the loss function, so as to more objectively reflect the importance of the recognition error of the audio recognition model for the probability that the user is in a alone environment, and the importance of the recognition error of the audio recognition model for the probability that the user is in a non-alone environment, and improve the training efficiency and the performance of the audio recognition model. When determining the viewing duration adjustment coefficient, the viewing duration adjustment coefficient can be determined through the average playback ratio of the same type of multimedia content in the current environment and the average playback ratio of the same type of multimedia content without the risk of privacy leakage, so as to accurately describe the privacy leakage risk of playing this type of multimedia content and improve the objectivity and accuracy of the viewing duration adjustment coefficient.

[0028] It should be understood that the above general description and the following detailed description are only exemplary and explanatory, and do not limit the present invention. According to the following detailed description of the exemplary embodiments with reference to the accompanying drawings, other features and aspects of the present invention will be clearer. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other embodiments can also be obtained based on these drawings;

[0030] Figure 1 Exemplarily shows a schematic flowchart of a method for intelligent recommendation display control of multimedia content according to an embodiment of the present invention;

[0031] Figure 2 Exemplarily shown is a block diagram of a multimedia content intelligent recommendation display control system according to an embodiment of the present invention. Specific embodiments

[0032] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Apparently, the described embodiments are only some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0033] The technical solutions of the present invention will be described in detail below with specific embodiments. These specific embodiments may be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments.

[0034] Figure 1 Exemplarily shown is a flowchart of a multimedia content intelligent recommendation display control method according to an embodiment of the present invention. The method includes:

[0035] Step S101, when starting a multimedia playback application on a terminal device, obtain a recommendation list of multimedia content, as well as type information and recommendation coefficients of multiple multimedia contents in the recommendation list;

[0036] Step S102, obtain environmental audio of the environment where the terminal device is located within a preset time period through an audio receiving component of the terminal device;

[0037] Step S103, obtain location information where the terminal device is located at the end of the preset time period through a positioning component of the terminal device;

[0038] Step S104, within the preset time period, obtain the viewing durations of multiple multimedia contents that have been played;

[0039] Step S105, determine an environmental content adjustment coefficient according to the environmental audio, the location information, and the type information;

[0040] Step S106, determine a viewing duration adjustment coefficient according to the viewing durations, the content durations of multiple multimedia contents that have been played, and the type information;

[0041] Step S107, determine an adjusted recommendation coefficient of the multimedia content that has not been played according to the environmental content adjustment coefficient, the viewing duration adjustment coefficient, and the recommendation coefficient;

[0042] Step S108: Obtain an adjusted multimedia content recommendation list according to the adjusted recommendation coefficient.

[0043] According to the multimedia content intelligent recommendation display control method of an embodiment of the present invention, after obtaining the recommendation list, based on the environmental audio and location information of the environment where the terminal device is located, as well as parameters such as the viewing duration of the played multimedia content, it can be determined whether a specific type of multimedia content is suitable for playing in the current environment, and the recommendation coefficient after the recommendation of the multimedia content can be obtained, so as to reduce the recommendation coefficient of the multimedia content that is not suitable for playing, reduce the recommended amount and playback amount of the multimedia content that is not suitable for playing, thereby reducing the possibility of user privacy leakage and improving the playback security and recommendation accuracy of the multimedia content.

[0044] According to an embodiment of the present invention, in step S101, the terminal device may include mobile devices such as mobile phones and tablets, and the multimedia playback application may include a video playback application program, an audio playback application program, etc. When the multimedia playback application is started, the multimedia playback application may generate a recommendation list according to the user's preferences. For example, based on factors such as the user's past playback records, search records, and the viewing duration of various types of multimedia content, a recommendation coefficient for the multimedia content in the database is generated for the user, and the multimedia content with a higher recommendation coefficient in the database is searched as the multimedia content that the user may be interested in, and the searched multimedia content is sorted according to the level of the recommendation coefficient to obtain a recommendation list. At the same time, the type information and recommendation coefficient of the multimedia content in the recommendation list can also be obtained.

[0045] According to an embodiment of the present invention, in step S102, the preset time period may be set to 3 minutes, 5 minutes, etc., and the present invention does not limit the duration of the preset time period. The preset time period can be used as an environmental test time period for detecting the environment where the terminal device is located. For example, the environmental audio can be received through the audio receiving component (such as a microphone) of the terminal device within the preset time period, so as to analyze the environmental audio to determine the environment where the user is located, and it can also be used to analyze whether the environment where the user is located is suitable for playing the recommended multimedia content and whether it may leak the user's privacy.

[0046] According to an embodiment of the present invention, in step S103, the positioning component may include a GPS positioning component. At the end of the preset time period, the location information of the terminal device can be determined, so that based on this location information, it can be judged whether some multimedia content is suitable for playing at the location information at the end of the preset time period and whether it may leak the user's privacy.

[0047] According to an embodiment of the present invention, in step S104, within a preset time period, the user may have watched a part of the multimedia content. For the multimedia content that the user is interested in and suitable for playing in the user's current environment, the viewing time may be longer, while for the content that the user is not interested in or not suitable for playing in the user's current environment, the viewing duration is shorter.

[0048] According to an embodiment of the present invention, in step S105, the environmental content adjustment coefficient can be a coefficient less than 1 and greater than 0, which can be used to adjust the recommendation coefficient of the unplayed multimedia content. For example, when the environmental content adjustment coefficient is small, a certain multimedia content is not suitable for playing in the user's current environment, so the environmental content adjustment coefficient is small. By multiplying the environmental content adjustment coefficient by the recommendation coefficient, a smaller adjusted recommendation coefficient can be obtained, and then this multimedia content will be removed from the recommendation list or adjusted to a more posterior position in the recommendation list, reducing the probability of playing in the current environment.

[0049] According to an embodiment of the present invention, determining the environmental content adjustment coefficient according to the environmental audio, the location information, and the type information includes: inputting the environmental audio into a trained audio recognition model to obtain audio type recognition information; determining the environmental privacy coefficient of the current environment according to the audio type recognition information and the location information; determining the playback suitability coefficient according to the location information and the type information of the unplayed multimedia content; and determining the environmental content adjustment coefficient of the unplayed multimedia content according to the environmental privacy coefficient and the playback suitability coefficient.

[0050] According to an embodiment of the present invention, the audio recognition model is a deep learning neural network model, such as a recurrent neural network model, etc., which can be used to determine the audio type recognition information of the environmental audio collected within a preset time period, that is, to identify the type of the environment where the environmental audio is located. For example, it is in a private bedroom, in a private vehicle, in a public vehicle, in a shopping mall, in an office, etc.

[0051] According to an embodiment of the present invention, the environmental privacy coefficient can be used to describe the privacy of the user's current environment. The closer the current environment is to a solitary environment, the higher the environmental privacy coefficient, otherwise, the lower the environmental privacy coefficient.

[0052] According to an embodiment of the present invention, determining an environmental privacy coefficient of the environment where the user is located based on the audio type recognition information and the location information includes: determining an audio type annotation vector in the case of being alone according to the audio type recognition information; determining an environmental category of the environment where the user is located according to the location information; determining environmental solitude probability information according to the environmental category; and determining the environmental privacy coefficient of the environment where the user is located according to the audio type recognition information, the audio type annotation vector, and the environmental solitude probability information.

[0053] According to an embodiment of the present invention, the audio type recognition information may be information in vector form, and each component in the vector may represent the probability that the environment where the user is located belongs to the type corresponding to the component. For example, the probability that the user is in a private bedroom is 0.6, the probability that the user is in a private vehicle is 0.3, the probability that the user is in a public vehicle is 0.01, the probability that the user is in a shopping mall is 0.01, and the probability that the user is in an office is 0.08. Then the audio type recognition information is The audio type annotation vector is similar in form to the audio type recognition information, but only the annotation value of the component in the case of the user being alone is 1, and the rest are 0. For example, the components of the user in the private bedroom and the private vehicle are 1, and the rest of the components are 0. That is, the audio type annotation vector is

[0054] According to an embodiment of the present invention, in addition to judging the environment where the user is located through the audio type recognition information, comprehensive judgment can also be carried out through the location information. For example, if the location information of the user is in a certain shopping mall, then the environmental category is a shopping mall, and the environmental solitude probability information is 0. If the location information of the user is in a private residence, then the environmental category is a private residence, and the environmental solitude probability information is 1. If the location information of the user is on the street, then the user may be in a private vehicle or a public vehicle, and the environmental solitude probability is 0.5.

[0055] According to an embodiment of the present invention, by combining the above various information, the environmental privacy coefficient of the environment where the user is located can be determined. Determining the environmental privacy coefficient of the environment where the user is located according to the audio type recognition information, the audio type annotation vector, and the environmental solitude probability information includes: determining the environmental privacy coefficient EP of the environment where the user is located according to formula (1).

[0056] EP = w1AT T AA + w2P al (1)

[0057] where, AT is the audio type recognition information, AT T is the transposed vector of AT, AA is the audio type annotation vector, and P alis the environmental solitude probability information, and w1 and w2 are preset weights.

[0058] According to an embodiment of the present invention, in formula (1), AT T AA is the probability that the user is alone determined based on the audio type recognition information. As in the above example, the audio type recognition information is The audio type annotation vector is Then AT T AA = 0.9, that is, the probability that the user is alone determined based on the audio type recognition information is 0.9. P al is the environmental solitude probability information, that is, the probability that the user is alone determined based on the location information. By comprehensively judging with these two types of information, that is, performing weighted summation on the probability that the user is alone determined based on the audio type recognition information and the probability that the user is alone determined based on the location information, the environmental privacy coefficient of the environment where the user is located is obtained, that is, the comprehensive probability that the user is alone, which can describe the privacy degree of the environment where the user is located and the possibility that the user is alone from multiple aspects.

[0059] In this way, by setting the audio type annotation vector in the case where the user is alone, the probability that the user is alone can be determined through vector operations. The environmental solitude probability information can also be determined through the user's location information, so as to comprehensively determine the environmental privacy coefficient of the environment where the user is located from multiple aspects, so as to accurately and objectively describe the privacy degree of the environment where the user is located and the possibility that the user is alone.

[0060] According to an embodiment of the present invention, the above audio recognition model can be trained before use. The training steps of the audio recognition model include: obtaining first sample audio within a preset time period indoors and second sample audio within a preset time period outdoors; scaling the amplitude of the second sample audio to obtain multiple third sample audio; superimposing the third sample audio and the first sample audio to obtain multiple fourth sample audio, and annotating the audio type of the fourth sample audio to obtain annotated audio type recognition information; inputting the fourth sample audio into the audio recognition model to obtain sample audio type recognition information; determining the loss function of the audio recognition model according to the annotated audio type recognition information, the sample audio type recognition information, and the audio type annotation vector; training the audio recognition model according to the loss function of the audio recognition model to obtain the trained audio recognition model.

[0061] According to an embodiment of the present invention, a quiet indoor environment (for example, only including sounds made by a person alone, such as the sound of eating, drinking water, etc.) can be selected to record the first sample audio within a preset time period. An outdoor environment with more noise (such as in a public transportation vehicle, in a shopping mall, on a street, etc.) can be selected to record the second sample audio within a preset time period.

[0062] According to an embodiment of the present invention, to simulate various environments, for example, being indoors with good room sound insulation; being indoors but with poor room sound insulation and being able to hear outdoor sounds; being outdoors and other various environments, the amplitude of the second sample audio can be scaled, for example, the amplitude of the second sample audio can be overall scaled to 0.8 times, 0.6 times, 0.4 times, 0.2 times, etc., to obtain a plurality of third sample audios, and the third sample audios are superimposed with the first sample audio to obtain a fourth sample audio, and the obtained audio type recognition information can be marked, and this information is used to describe in which of the above three situations the fourth sample audio is obtained.

[0063] According to an embodiment of the present invention, the fourth sample audio is input into an audio recognition model to obtain sample audio type recognition information, and the form of the sample audio type recognition information is similar to the form of the audio type recognition information, which will not be elaborated here.

[0064] According to an embodiment of the present invention, according to the marked audio type recognition information, the sample audio type recognition information and the audio type annotation vector, the loss function of the audio recognition model is determined, including: determining the loss function LOSS of the audio recognition model according to formula (2),

[0065]

[0066] where, ATA i,j is the marked audio type recognition information of the fourth sample audio obtained by superimposing the i-th first sample audio and the j-th third sample audio, is the transposed vector of ATA i,j ATS i,j is the sample audio type recognition information of the fourth sample audio obtained by superimposing the i-th first sample audio and the j-th third sample audio, is the transposed vector of ATS i,j AA is the audio type annotation vector, A1 is a column vector with all component values being 1 and the number of components being equal to the number of components of AA, w3 and w4 are preset weights, and w3>w4, α1 is a preset probability threshold, if is a conditional function, n i is the number of third sample audios superimposed with the i-th first sample audio, n is the number of first sample audios, j≤n i , i≤n, and j, n i 、i and n are all positive integers.

[0067] According to an embodiment of the present invention, in formula (2), the conditional function means that in the case of, the conditional function value is Otherwise, the conditional function value is In case, the labeled audio type recognition information indicates that the user is in a solitary environment, that is, the sum of the probabilities of being in a solitary environment is greater than α1. In this case, the importance of the probability indicating that the user is in a solitary environment in the sample audio type recognition information is higher. Let be the probability indicating that the user is in a solitary environment in the labeled audio type recognition information, and Let be the probability indicating that the user is in a non - solitary environment in the labeled audio type recognition information. The difference between the two is the probability error of the user being in a non - solitary environment calculated by the audio recognition model. (A1 - AA) is the audio type annotation vector in the non - solitary environment. For example, Let be the probability indicating that the user is in a non - solitary environment in the sample audio type recognition information. The difference between the two is the probability error of the user being in a non - solitary environment calculated by the audio recognition model. And, when the labeled audio type recognition information indicates that the user is in a solitary environment, the importance of the recognition error of the probability that the user is in a solitary environment by the audio recognition model is higher, that is, reducing this error is more helpful for improving the judgment accuracy of the audio recognition model for the user being in a solitary environment. Therefore, in this case, can be given a higher weight w3, and can be given a lower weight w4, and the weighted sum of the two can obtain the conditional function value in this case. On the contrary, if the labeled audio type recognition information indicates that the user is in a non - solitary environment, the importance of the recognition error of the probability that the user is in a non - solitary environment by the audio recognition model is higher, that is, reducing this error is more helpful for improving the judgment accuracy of the audio recognition model for the user being in a non - solitary environment. Therefore, in this case, can be given a lower weight w4, and

[0068] can be given a higher weight w3, and the weighted sum of the two can obtain the conditional function value in this case. Further, the conditional function values calculated based on the fourth sample audio corresponding to multiple first - sample audios are summed to obtain the loss function of the audio recognition model.

[0068] According to an embodiment of the present invention, based on the above loss function, the audio recognition model is trained, that is, by means of backpropagation, the audio recognition model is adjusted based on the loss function to minimize the loss function, and after multiple trainings, the trained audio recognition model is obtained.

[0069] In this way, during the process of determining the loss function, different weights can be set through a conditional function, so as to more objectively reflect the importance of the recognition error of the audio recognition model for the probability that the user is in a solitary environment, and the importance of the recognition error of the audio recognition model for the probability that the user is in a non-solitary environment, improving the training efficiency and the performance of the audio recognition model.

[0070] According to an embodiment of the present invention, after determining the environmental privacy coefficient, the playback suitability coefficient can also be determined to describe whether the multimedia content that has not been played in the recommended list is suitable for playback in the user's environment and whether it may disclose the user's privacy.

[0071] According to an embodiment of the present invention, the playback suitability coefficient is determined according to the location information and the type information of the unplayed multimedia content, including: determining the environmental type of the environment where the terminal device is located according to the location information; screening, through the server side, the first playback records of sample terminal devices that are in the same environmental type of environment and have the multimedia playback application enabled; determining the first historical playback duration and the first total duration of the multimedia content of multiple types of information according to the first playback records; determining the average value of the ratio of the first historical playback duration to the first total duration of the multiple multimedia content of the kth type of information as the playback suitability coefficient of the kth type of information; and determining the playback suitability coefficient of each unplayed multimedia content according to the type information of the unplayed multimedia content in the recommended list and the playback suitability coefficients of various types.

[0072] According to an embodiment of the present invention, the average value of the ratio of the first historical playback duration of multiple sample terminal devices playing the same type of multimedia content in the same environment to the first total duration of the multimedia content can be used as the playback suitability coefficient of this type. For example, in an environment suitable for playback, since the interests of the users of each sample terminal device are different, the ratio of the first historical playback duration to the first total duration is uncertain. However, in an environment unsuitable for playback, regardless of whether the user is interested in this type of multimedia content, the ratio of the first historical playback duration to the first total duration is relatively small. Therefore, the average value of the ratio of the first historical playback duration to the first total duration of the multimedia content can be used as the playback suitability coefficient of this type of information to describe whether this type of multimedia content is suitable for playback in this environment. If it is suitable, the playback suitability coefficient is relatively high; if it is not suitable, the playback suitability coefficient is at a relatively low level. Further, after determining the playback suitability coefficients of various types, the type information of the unplayed multimedia content can be determined, and the playback suitability coefficient corresponding to this type information can be determined as the playback suitability coefficient of the multimedia content.

[0073] According to an embodiment of the present invention, determining the environmental content adjustment coefficient of the unplayed multimedia content according to the environmental privacy coefficient and the playback suitability coefficient includes: if the playback suitability coefficient of the x-th unplayed multimedia content is greater than or equal to the preset coefficient threshold, then the environmental content adjustment coefficient of the x-th multimedia content is determined to be 1; if the playback suitability coefficient of the x-th unplayed multimedia content is less than the preset coefficient threshold, then the environmental content adjustment coefficient of the x-th multimedia content is determined to be the environmental privacy coefficient. That is, if the playback suitability coefficient of the x-th unplayed multimedia content is greater than or equal to the preset coefficient threshold, then the x-th multimedia content is suitable for playback in the current environment. In this case, the environmental content adjustment coefficient of the x-th multimedia content can be determined to be 1. On the contrary, if the playback suitability coefficient of the x-th unplayed multimedia content is less than the preset coefficient threshold, then the x-th multimedia content is not suitable for playback in the current environment. In this case, the environmental content adjustment coefficient of the x-th multimedia content can be determined to be the environmental privacy coefficient. In other words, if the multimedia content is not suitable for playback in the current environment, the privacy of the current environment is represented by the environmental privacy coefficient, and then it is determined whether the environment is private and whether it is not easy to leak user privacy. If the environmental privacy coefficient is low, the environmental content adjustment coefficient is low, and the recommendation coefficient of the x-th multimedia content can be adjusted to a lower value through the environmental content adjustment coefficient, so that it is removed from the recommendation list or moved to a lower position in the recommendation list, thereby reducing the probability of the x-th multimedia content being played and leaking user privacy.

[0074] According to an embodiment of the present invention, in step S106, determining a viewing duration adjustment coefficient according to the viewing duration, the content durations of a plurality of played multimedia contents, and the type information includes: obtaining a second historical play record of the terminal device in an environment where the environmental solitude probability information is higher than or equal to a preset probability threshold; determining, according to the second historical play record, the second historical play duration and the second total duration of multimedia contents of various type information; determining the viewing duration adjustment coefficient ED of the x-th multimedia content that has not been played according to formula (3) x ,

[0075]

[0076] where Δt p,x,z is the play duration of the z-th multimedia content with the same type information as the x-th multimedia content among the plurality of played multimedia contents, and Δt total,x,z is the content duration of the z-th multimedia content with the same type information as the x-th multimedia content among the plurality of played multimedia contents, n c,x is the number of multimedia contents with the same type information as the x-th multimedia content among the plurality of played multimedia contents, Δt 2,p,x,y is the second historical play duration of the y-th multimedia content with the same type information as the x-th multimedia content in the second historical play record, and Δt 2,total,x,y is the second total duration of the y-th multimedia content with the same type information as the x-th multimedia content in the second historical play record, n 2,x is the number of multimedia contents with the same type information as the x-th multimedia content in the second historical play record, z ≤ n c,x , y ≤ n 2,x , and z, n c,x , y, and n 2,x are all positive integers.

[0077] According to an embodiment of the present invention, in formula (3), represents the ratio of the play duration to the content duration of multimedia contents with the same type information as the x-th multimedia content in a preset time period, that is, the play ratio of the same type of multimedia contents in the preset time period. Therefore, is the average play ratio of the same type of multimedia contents in the preset time period. Similarly, is the average play ratio of the same type of multimedia contents in the case of the user being alone, which is used to represent the average play ratio of the same type of multimedia contents without the risk of privacy leakage. The ratio of the two is the viewing duration adjustment coefficient, which can be used to represent the privacy leakage risk of playing the same type of multimedia contents in the current environment, that is, The higher it is, the closer the average playback ratio of the same type of multimedia content in the environment is to the average playback ratio of the same type of multimedia content without the risk of privacy leakage. That is, the lower the risk of privacy leakage. On the contrary, The lower it is, the greater the gap between the average playback ratio of the same type of multimedia content in the environment and the average playback ratio of the same type of multimedia content without the risk of privacy leakage. That is, the higher the risk of privacy leakage.

[0078] In this way, the viewing duration adjustment coefficient can be determined through the average playback ratio of the same type of multimedia content in the environment and the average playback ratio of the same type of multimedia content without the risk of privacy leakage, so as to accurately describe the privacy leakage risk of playing this type of multimedia content and improve the objectivity and accuracy of the viewing duration adjustment coefficient.

[0079] According to an embodiment of the present invention, in step S107, the environmental content adjustment coefficient and the viewing duration adjustment coefficient of each unplayed multimedia content can be weighted and averaged to obtain an adjustment coefficient, and the adjustment coefficient is used to multiply the recommendation coefficient to obtain the adjusted recommendation coefficient of the unplayed multimedia content. Thus, when the environment is not suitable for playing multimedia content, the adjusted recommendation coefficient of the multimedia content is reduced, and then the inappropriate multimedia content can be recommended less or not recommended at all, reducing the probability of leaking user privacy.

[0080] According to an embodiment of the present invention, in step S108, the adjusted recommendation coefficients of multiple multimedia contents can be solved in the database to obtain an adjusted multimedia content recommendation list, so that the multimedia contents in the adjusted multimedia content recommendation list are suitable for playing in the user's environment, reducing the risk of leaking user privacy.

[0081] The intelligent recommendation display control method for multimedia content according to an embodiment of the present invention, after obtaining a recommendation list, can determine whether a specific type of multimedia content is suitable for playing in the environment based on the ambient audio and location information of the terminal device, as well as parameters such as the viewing duration of the played multimedia content, and obtain the recommendation coefficient after the recommendation of the multimedia content, so as to reduce the recommendation coefficient of the multimedia content that is not suitable for playing, reduce the recommended amount and playback amount of the multimedia content that is not suitable for playing, thereby reducing the possibility of user privacy leakage, and improving the playback security and recommendation accuracy of the multimedia content. When determining the environmental privacy coefficient, the audio type annotation vector in the case of the user being alone can be set, so that the probability of the user being alone can be determined through vector operations, and the environmental solitude probability information can also be determined through the user's location information, so as to comprehensively determine the environmental privacy coefficient of the environment where the user is located from multiple aspects, so as to accurately and objectively describe the privacy degree of the environment where the user is located and the possibility of the user being alone. When training the audio recognition model, different weights can be set through a conditional function in the process of determining the loss function, so as to more objectively reflect the importance of the recognition error of the audio recognition model for the probability that the user is in a solitary environment, and the importance of the recognition error of the audio recognition model for the probability that the user is in a non-solitary environment, and improve the training efficiency and the performance of the audio recognition model. When determining the viewing duration adjustment coefficient, the viewing duration adjustment coefficient can be determined through the average playback ratio of the same type of multimedia content in the environment where the user is located and the average playback ratio of the same type of multimedia content in the case of no privacy leakage risk, so as to accurately describe the privacy leakage risk of playing this type of multimedia content and improve the objectivity and accuracy of the viewing duration adjustment coefficient.

[0082] Figure 2 Exemplarily, a block diagram of a multimedia content intelligent recommendation display control system according to an embodiment of the present invention is shown. The system includes:

[0083] An acquisition module, configured to obtain a recommendation list of multimedia content, as well as the type information and recommendation coefficients of multiple multimedia contents in the recommendation list, when starting a multimedia playback application on a terminal device;

[0084] An ambient audio module, configured to obtain the ambient audio of the environment where the terminal device is located within a preset time period through an audio receiving component of the terminal device;

[0085] A location information module, configured to obtain the location information of the terminal device at the end of the preset time period through a positioning component of the terminal device;

[0086] A viewing duration module, configured to obtain the viewing durations of multiple played multimedia contents within the preset time period;

[0087] An environmental content adjustment coefficient module, configured to determine an environmental content adjustment coefficient according to the environmental audio, the location information, and the type information;

[0088] A viewing duration adjustment coefficient module, configured to determine a viewing duration adjustment coefficient according to the viewing duration, the content durations of multiple multimedia contents that have been played, and the type information;

[0089] An adjustment module, configured to determine an adjusted recommendation coefficient of an unplayed multimedia content according to the environmental content adjustment coefficient, the viewing duration adjustment coefficient, and the recommendation coefficient;

[0090] A recommendation list module, configured to obtain an adjusted multimedia content recommendation list according to the adjusted recommendation coefficient.

[0091] Those skilled in the art should understand that the embodiments of the present invention described above and shown in the accompanying drawings are only examples and do not limit the present invention. The object of the present invention has been fully and effectively achieved. The function and structural principle of the present invention have been shown and described in the embodiments. Without departing from the principle, any deformation or modification can be made to the embodiments of the present invention.

[0092] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements on some or all of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A multimedia content intelligent recommendation display control method, characterized in that: include: When a multimedia playback application is started on a terminal device, a recommendation list of multimedia contents and type information and recommendation coefficients of multiple multimedia contents in the recommendation list are obtained; Acquire the ambient audio of the environment in which the terminal device is located within a preset time period through the audio receiving component of the terminal device; Obtaining, through a positioning component of the terminal device, location information of the terminal device at the end of the preset time period; Acquire the viewing duration of the plurality of multimedia contents that have been played within the preset time period; Determining an environmental content adjustment coefficient according to the environmental audio, the position information and the type information; Determining a viewing time adjustment coefficient according to the viewing time, the content durations of the multiple multimedia contents that have been played, and the type information; Determining an adjusted recommendation coefficient of the unplayed multimedia content according to the environment content adjustment coefficient, the viewing time adjustment coefficient, and the recommendation coefficient; According to the adjusted recommendation coefficient, an adjusted multimedia content recommendation list is obtained.

2. The multimedia content intelligent recommendation display control method according to claim 1, characterized in that: Determining an environmental content adjustment coefficient according to the environmental audio, the position information, and the type information includes: Inputting the environmental audio into the trained audio recognition model to obtain audio type recognition information; Determining an environmental privacy coefficient of the environment according to the audio type identification information and the location information; Determining a playback suitability coefficient according to the location information and type information of the unplayed multimedia content; An environmental content adjustment coefficient of the unplayed multimedia content is determined according to the environmental privacy coefficient and the playback suitability coefficient.

3. The multimedia content intelligent recommendation display control method according to claim 2, characterized in that: Determining an environmental privacy coefficient of the environment according to the audio type identification information and the location information includes: Determining an audio type label vector for a solitary audio situation according to the audio type identification information; Determining the environment category of the environment according to the location information; Determining environmental solitude probability information according to the environmental category; An environmental privacy coefficient of the environment is determined according to the audio type identification information, the audio type annotation vector and the environmental solitude probability information.

4. The method for intelligent recommendation and display control of multimedia content according to claim 3, characterized in that: Determining an environmental privacy coefficient of the environment according to the audio type identification information, the audio type annotation vector, and the environmental solitude probability information includes: According to the formula EP=w1AT T AA+w2P al Determine the environmental privacy coefficient EP of the environment, where AT is the audio type identification information, AT T is the transposed vector of AT, AA is the audio type label vector, P al is the probability information of being alone in the environment, and w1 and w2 are preset weights.

5. The multimedia content intelligent recommendation display control method according to claim 2, characterized in that: The training steps of the audio recognition model include: Acquire a first sample audio for a preset time period indoors, and acquire a second sample audio for a preset time period outdoors; Scaling the amplitude of the second audio sample to obtain a plurality of third audio sample; The third sample audio and the first sample audio are superimposed to obtain a plurality of fourth sample audios, and the audio types of the fourth sample audios are marked to obtain marked audio type identification information; Inputting the fourth sample audio into the audio recognition model to obtain sample audio type recognition information; Determining a loss function of an audio recognition model according to the annotated audio type identification information, the sample audio type identification information, and the audio type annotated vector; The audio recognition model is trained according to the loss function of the audio recognition model to obtain a trained audio recognition model.

6. The multimedia content intelligent recommendation display control method according to claim 5, characterized in that: Determining a loss function of an audio recognition model according to the annotated audio type identification information, the sample audio type identification information, and the audio type annotated vector includes: According to the formula Determine the loss function LOSS of the audio recognition model, where ATA i,j is the annotated audio type identification information of the fourth sample audio obtained by superimposing the i-th first sample audio and the j-th third sample audio, For ATA i,j The transposed vector of ATS i,j is sample audio type identification information of a fourth sample audio obtained by superimposing the i-th first sample audio and the j-th third sample audio, For ATS i,j , AA is the audio type label vector, A1 is a column vector whose component values ​​are all 1 and the number of components is equal to the number of components of AA, w3 and w4 are preset weights, and w3>w4, α1 is a preset probability threshold, if is a conditional function, n i is the number of third audio samples superimposed on the i-th first audio sample, n is the number of first audio samples, j≤n i , i≤n, and j, n i , i and n are all positive integers.

7. The multimedia content intelligent recommendation display control method according to claim 2, characterized in that: Determining a playback suitability coefficient according to the location information and the type information of the unplayed multimedia content includes: Determining the environment type of the environment where the terminal device is located according to the location information; Filtering, through the server side, first playback records of multiple sample terminal devices in the same environment type and with multimedia playback applications turned on; Determining, according to the first playback record, a first historical playback duration of multimedia content of the plurality of types of information and a first total duration of the multimedia content; Determine an average value of the ratios of the first historical play durations of the plurality of multimedia contents of the k-th type of information to the first total duration of the multimedia contents as a play suitability coefficient of the k-th type of information; The play suitability coefficient of each unplayed multimedia content is determined according to the type information of the unplayed multimedia content in the recommendation list and the play suitability coefficients of various types.

8. The multimedia content intelligent recommendation display control method according to claim 2, characterized in that: Determining an environmental content adjustment coefficient of the unplayed multimedia content according to the environmental privacy coefficient and the play suitability coefficient includes: If the playback suitability coefficient of the unplayed x-th multimedia content is greater than or equal to the preset coefficient threshold, the environmental content adjustment coefficient of the x-th multimedia content is determined to be 1; If the playback suitability coefficient of the unplayed x-th multimedia content is less than the preset coefficient threshold, the environmental content adjustment coefficient of the x-th multimedia content is determined as the environmental privacy coefficient.

9. The method for intelligent recommendation and display control of multimedia content according to claim 3, characterized in that: Determining a viewing time adjustment coefficient according to the viewing time, the content time of the plurality of multimedia contents that have been played, and the type information includes: Acquire a second historical playback record of the terminal device being in an environment where the probability information of the environment being alone is higher than or equal to a preset probability threshold; Determine, according to the second historical playback record, a second historical playback duration and a second total duration of the multimedia content of the plurality of types of information; According to the formula Determine the viewing time adjustment factor ED of the xth multimedia content that has not been played x , where Δt p,x,z is the playing time of the zth multimedia content with the same type information as the xth multimedia content among the multiple multimedia contents that have been played, Δt total,x,z is the content duration of the zth multimedia content having the same type information as the xth multimedia content among the multiple multimedia contents that have been played, n c,x is the number of multimedia contents with the same type information as the xth multimedia content among the multiple multimedia contents that have been played, Δt 2,p,x,y is the second historical playback duration of the yth multimedia content with the same type information as the xth multimedia content in the second historical playback record, Δt 2,total,x,y is the second total duration of the yth multimedia content having the same type information as the xth multimedia content in the second historical playback record, n 2,x is the number of multimedia contents with the same type information as the x-th multimedia content in the second historical playback record, z≤n c,x , y≤n 2,x , and z, n c,x , y and n 2,x All are positive integers.

10. A multimedia content intelligent recommendation and display control system, characterized in that: include: An acquisition module, used to acquire a recommendation list of multimedia contents, and type information and recommendation coefficients of multiple multimedia contents in the recommendation list when a multimedia playback application is started on a terminal device; The ambient audio module is used to obtain the ambient audio of the environment in which the terminal device is located within a preset time period through the audio receiving component of the terminal device; A location information module, used to obtain the location information of the terminal device at the end of the preset time period through a positioning component of the terminal device; A viewing duration module, used to obtain the viewing duration of the multiple multimedia contents played within the preset time period; An environmental content adjustment coefficient module, used to determine an environmental content adjustment coefficient according to the environmental audio, the position information and the type information; A viewing time adjustment coefficient module, used to determine a viewing time adjustment coefficient according to the viewing time, the content duration of the multiple multimedia contents played, and the type information; an adjustment module, configured to determine an adjusted recommendation coefficient of the unplayed multimedia content according to the environment content adjustment coefficient, the viewing time adjustment coefficient and the recommendation coefficient; The recommendation list module is used to obtain an adjusted multimedia content recommendation list according to the adjusted recommendation coefficient.

Images