A topic label recommendation method and device, electronic equipment and storage medium

Abstract

The present disclosure relates to a topic label recommendation method and device, electronic equipment and storage medium. The method comprises: obtaining audio information of audio data in to-be-published multimedia content; the audio information is an attribute feature of the audio data; the audio information has uniqueness; obtaining an audio topic corresponding relationship; wherein the audio topic corresponding relationship comprises a plurality of reference audio information and at least one associated topic label corresponding to each reference audio information; from the plurality of reference audio information, determine the target audio information matched with the audio information; at least one associated topic label corresponding to the target audio information is taken as a recommended topic label set of the to-be-published multimedia content; the recommended topic label set is used to push to the publishing client corresponding to the to-be-published multimedia content. The topic label recommendation method of the present disclosure has simple algorithm and requires less computing resources. When the publishing demand is large, the topic label recommendation efficiency can be improved.

Description

Technical Field

[0001] This disclosure relates to the field of Internet technology, and in particular to a method, apparatus, electronic device, and storage medium for recommending topic tags. Background Technology

[0002] As the short video industry continues to develop, the consumption and creation of short videos have become part of the daily lives of the general public. Among these, topic elements have become an important requirement in the creation of short videos. Good topics not only help to better convey the content of the work, but also can form a linkage with other works on the same topic on the platform, thereby achieving better dissemination results.

[0003] In short video production scenarios, non-professional ordinary users often lack the ability to formulate good topics or actively edit topics. Therefore, it is often necessary to provide accurate topic recommendations for users in the publishing page scenario. Summary of the Invention

[0004] This disclosure provides a method, apparatus, electronic device, and storage medium for recommending topic tags. The technical solution of this disclosure is as follows:

[0005] According to a first aspect of the present disclosure, a topic tag recommendation method is provided, comprising:

[0006] Retrieve audio information from audio data in multimedia content to be published; audio information is an attribute characteristic of audio data; audio information is unique;

[0007] Obtain the audio topic correspondence; wherein, the audio topic correspondence includes multiple reference audio information and at least one associated topic tag corresponding to each reference audio information;

[0008] From multiple reference audio information, determine the target audio information that matches the audio information;

[0009] At least one associated topic tag corresponding to the target audio information is used as a set of recommended topic tags for the multimedia content to be published; the set of recommended topic tags is used to push to the publishing client corresponding to the multimedia content to be published.

[0010] In some possible embodiments, the method further includes:

[0011] Retrieve multiple published historical multimedia content;

[0012] The historical audio information of the audio data in each historical multimedia content is used as the reference audio information; the historical audio information of the audio data in each historical multimedia content is obtained by extracting audio fingerprints from the audio data in each historical multimedia content.

[0013] Use the historical topic tags carried by each historical multimedia content as topic tags to be associated;

[0014] Based on multiple topic tags to be associated involving multiple historical multimedia content and multiple audio information to be referenced involving multiple historical multimedia content, construct the audio topic correspondence relationship;

[0015] The audio topic correspondence is updated at preset time intervals.

[0016] In some possible embodiments, an audio topic correspondence is constructed based on multiple unassociated topic tags involved in multiple historical multimedia content and multiple reference audio information involved in multiple historical multimedia content, including:

[0017] Based on the audio information to be referenced, multiple historical multimedia contents are classified to obtain multiple sets of historical multimedia contents; each set of historical multimedia contents corresponds one-to-one with each set of audio information to be referenced.

[0018] Based on the topic tags to be associated carried by each historical multimedia content in each historical multimedia content set, a set of associated topic tags corresponding to each audio information to be referenced is obtained; each associated topic tag in the set of associated topic tags corresponding to each audio information to be referenced indicates a topic tag that is associated with each audio information to be referenced.

[0019] Based on the set of associated topic tags corresponding to each audio information to be referenced, the audio topic correspondence is obtained.

[0020] In some possible embodiments, based on the topic tags to be associated carried by each historical multimedia content in each historical multimedia content set, a set of associated topic tags corresponding to each audio information to be referenced is obtained, including:

[0021] Based on each topic tag to be associated, multiple historical multimedia contents are classified to obtain the number of historical multimedia contents corresponding to each topic tag to be associated.

[0022] Each historical multimedia content in each historical multimedia content set carries a topic tag to be associated, which is then used as the set of topic tags to be associated for each audio information to be referenced.

[0023] Based on the number of historical multimedia contents corresponding to each topic tag to be associated, determine the associated topic tag set corresponding to each audio information to be referenced from the set of topic tags to be associated corresponding to each audio information to be referenced; the number of historical multimedia contents corresponding to each associated topic tag in the associated topic tag set corresponding to each audio information to be referenced is greater than or equal to the first preset number.

[0024] In some possible embodiments, after determining the set of associated topic tags corresponding to each audio information from the set of associated topic tags corresponding to each audio information based on the quantity of historical multimedia content corresponding to each topic tag, the method further includes:

[0025] Determine the number of associated topic tags in the associated topic tag set corresponding to each audio information to be referenced;

[0026] Audio information with a number of associated topic tags less than or equal to the second preset number in the associated topic tag set is identified as audio information to be matched.

[0027] Based on the similarity value between the audio information to be matched and each audio information to be referenced and the number of associated topic tags corresponding to each audio information to be referenced, the target reference audio information is determined from multiple audio information to be referenced; the similarity value between the target reference audio information and the audio information to be matched is greater than or equal to the preset similarity value, and the number of associated topic tags corresponding to the target reference audio information is greater than the third preset number.

[0028] The set of associated topic tags corresponding to the target reference audio information is used as the set of associated topic tags corresponding to the audio information to be matched.

[0029] In some possible embodiments, determining target audio information that matches the audio information from a plurality of reference audio information includes:

[0030] Determine the degree of matching between the audio information and each reference audio information;

[0031] If among multiple reference audio information, there is a reference audio information whose matching degree value is greater than or equal to a preset value, the reference audio information whose matching degree value is greater than or equal to the preset value will be used as the target audio information.

[0032] In some possible embodiments, after including at least one associated topic tag corresponding to the target audio information as a set of recommended topic tags for the multimedia content to be published, the method further includes:

[0033] Get the account click count for each recommended topic tag in the recommended topic tag set;

[0034] Based on the number of clicks per account for each recommended topic tag, the recommended topic tags in the recommended topic tag set are sorted to obtain the sorted recommended topic tag set.

[0035] According to a second aspect of the present disclosure, a topic tag recommendation device is provided, comprising:

[0036] The first acquisition module is configured to acquire audio information from the audio data in the multimedia content to be published; audio information is an attribute feature of audio data; audio information is unique.

[0037] The second acquisition module is configured to acquire audio topic correspondences; wherein, the audio topic correspondences include multiple reference audio information and at least one associated topic tag corresponding to each reference audio information;

[0038] The first determining module is configured to determine the target audio information that matches the audio information from multiple reference audio information.

[0039] The second determining module is configured to use at least one associated topic tag corresponding to the target audio information as a set of recommended topic tags for the multimedia content to be published; the set of recommended topic tags is used to push to the publishing client corresponding to the multimedia content to be published.

[0040] In some possible embodiments, the apparatus further includes a construction module configured to perform the following actions: acquiring multiple published historical multimedia content; using historical audio information of audio data in each historical multimedia content as reference audio information; the historical audio information of audio data in each historical multimedia content is obtained by extracting audio fingerprints from the audio data in each historical multimedia content; using historical topic tags carried by each historical multimedia content as topic tags to be associated; constructing an audio topic correspondence based on the multiple topic tags to be associated and the multiple reference audio information involved in the multiple historical multimedia content; and updating the audio topic correspondence at preset time intervals.

[0041] In some possible embodiments, the construction module is further configured to perform the following actions: classifying multiple historical multimedia contents based on each audio information to be referenced, resulting in multiple sets of historical multimedia contents; each set of historical multimedia contents corresponds one-to-one with each audio information to be referenced; based on the topic tags to be associated carried by each historical multimedia content in each set of historical multimedia contents, obtaining a set of associated topic tags corresponding to each audio information to be referenced; each associated topic tag in the set of associated topic tags corresponding to each audio information to be referenced indicates a topic tag that is associated with each audio information to be referenced; and obtaining the audio topic correspondence based on the set of associated topic tags corresponding to each audio information to be referenced.

[0042] In some possible embodiments, the construction module is further configured to perform the following: classify multiple historical multimedia contents based on each topic tag to be associated, and obtain the number of historical multimedia contents corresponding to each topic tag to be associated; use the topic tags to be associated carried by each historical multimedia content in each historical multimedia content set as the topic tag set to be associated corresponding to each audio information to be referenced; determine the associated topic tag set corresponding to each audio information to be referenced from the topic tag set corresponding to each audio information to be referenced based on the number of historical multimedia contents corresponding to each topic tag to be associated; and ensure that the number of historical multimedia contents corresponding to each associated topic tag in the associated topic tag set corresponding to each audio information to be referenced is greater than or equal to a first preset number.

[0043] In some possible embodiments, the construction module is further configured to perform the following: determine the number of associated topic tags in the associated topic tag set corresponding to each audio information to be referenced; determine audio information to be matched whose number of associated topic tags in the associated topic tag set is less than or equal to a second preset number; determine target reference audio information from multiple audio information to be referenced based on the similarity value between the audio information to be matched and each audio information to be referenced and the number of associated topic tags corresponding to each audio information to be referenced; the similarity value between the target reference audio information and the audio information to be matched is greater than or equal to a preset similarity value, and the number of associated topic tags corresponding to the target reference audio information is greater than a third preset number; and use the associated topic tag set corresponding to the target reference audio information as the associated topic tag set corresponding to the audio information to be matched.

[0044] In some possible embodiments, the first determining module is further configured to determine the matching degree value between the audio information and each reference audio information; if there is a reference audio information among the multiple reference audio information whose matching degree value is greater than or equal to a preset value, the reference audio information whose matching degree value is greater than or equal to the preset value is taken as the target audio information.

[0045] In some possible embodiments, the apparatus further includes a sorting module configured to perform the following operations: obtaining the number of account clicks for each recommended topic tag in the recommended topic tag set; and sorting the recommended topic tags in the recommended topic tag set according to the number of account clicks for each recommended topic tag to obtain a sorted set of recommended topic tags.

[0046] According to a third aspect of the present disclosure, an electronic device is provided, comprising:

[0047] processor;

[0048] Memory used to store processor-executable instructions;

[0049] The processor is configured to execute instructions to implement the topic tag recommendation method of the first aspect of the present disclosure.

[0050] According to a fourth aspect of the present disclosure, a computer-readable storage medium is provided, which, when executed by a processor of an electronic device, enables the electronic device to perform the topic tag recommendation method of the first aspect of the present disclosure.

[0051] According to a fifth aspect of the present disclosure, a computer program product is provided, comprising a computer program stored in a readable storage medium, wherein at least one processor of a computer device reads from the readable storage medium and executes the computer program, causing the computer device to perform the topic tag recommendation method of the first aspect of the present disclosure.

[0052] The technical solutions provided by the embodiments of this disclosure have at least the following beneficial effects:

[0053] The method involves obtaining audio topic correspondences, where each audio topic correspondence includes multiple reference audio information and at least one associated topic tag corresponding to each reference audio information. For the audio information in the audio data of the multimedia content to be published, at least one associated topic tag is matched. This at least one associated topic tag serves as a recommended topic tag set for the multimedia content to be published, and is used to push it to the publishing client corresponding to the multimedia content. The topic tag recommendation method of this embodiment is simple in algorithm, requires relatively few computational resources, and can improve the efficiency of topic tag recommendation when the publishing demand is large.

[0054] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this disclosure. Attached Figure Description

[0055] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this disclosure and, together with the description, serve to explain the principles of this disclosure, and are not intended to unduly limit this disclosure.

[0056] Figure 1 This is a schematic diagram illustrating an application environment according to an exemplary embodiment;

[0057] Figure 2 This is a flowchart illustrating a topic tag recommendation method according to an exemplary embodiment;

[0058] Figure 3 This is a flowchart illustrating an example of constructing an audio topic correspondence;

[0059] Figure 4This is a flowchart illustrating an example of constructing an audio topic correspondence;

[0060] Figure 5 This is a flowchart illustrating a method for determining a set of associated topic tags according to an exemplary embodiment;

[0061] Figure 6 This is a flowchart illustrating a topic tag recommendation method according to an exemplary embodiment;

[0062] Figure 7 This is a flowchart illustrating a method for determining target audio information according to an exemplary embodiment;

[0063] Figure 8 This is a flowchart illustrating a method for sorting a set of recommended topic tags according to an exemplary embodiment;

[0064] Figure 9 This is a block diagram illustrating a topic tag recommendation device according to an exemplary embodiment;

[0065] Figure 10 This is a block diagram illustrating an electronic device for recommending hashtags, according to an exemplary embodiment. Detailed Implementation

[0066] To enable those skilled in the art to better understand the technical solutions of this disclosure, the technical solutions in the embodiments of this disclosure will be clearly and completely described below with reference to the accompanying drawings.

[0067] It should be noted that the terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this disclosure are used to distinguish similar first objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this disclosure described herein can be implemented in orders other than those illustrated or described herein. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this disclosure. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this disclosure as detailed in the appended claims.

[0068] It should be noted that the user information involved in this disclosure (including but not limited to user device information, user personal information, etc.) is all information authorized by the user or fully authorized by all parties.

[0069] This disclosure provides a topic tag recommendation method that can be applied to scenarios where users add topic tags when publishing content. For example, when a user shoots a short video and wants to publish it on a platform, the topic tag recommendation method provided in this disclosure can automatically generate matching topic tags for the short video and send the topic tags to the publishing page, thereby providing the user with the function of selecting topic tags. Below, we will briefly introduce topic tags and related scenarios.

[0070] Hashtags are textual descriptions of videos. For example, hashtags are text messages with "#" added by users during the video creation process to describe the video content. Examples of hashtags include "#dance" and "#funny". In the context of publishing short videos, platforms aggregate and distribute content based on the hashtags associated with the videos. Therefore, accurate hashtags play a significant positive role in the understanding, aggregation, and distribution of short videos.

[0071] In related technologies, topic tag recommendation in short video publishing scenarios mainly involves using models to learn from video footage, predicting topic tags related to the video content, and also associating them with some users' historical topic tags to recommend topics in the publishing context. This approach has complex algorithms, requires significant computational resources, and results in low efficiency for topic tag recommendation.

[0072] This disclosure provides a topic tag recommendation method. By identifying audio information in the audio data of a short video to be published, and based on a pre-built audio-topic correspondence, it quickly determines a set of recommended topic tags for the short video to be published. The topic tag recommendation method of this disclosure has a simple algorithm, requires relatively few computational resources, and can improve the efficiency of topic tag recommendation when the demand for publishing is large.

[0073] The following describes the application environment of the topic tag recommendation method according to embodiments of this disclosure. Please refer to... Figure 1 , Figure 1 A schematic diagram of an application environment is shown according to an exemplary embodiment, such as... Figure 1 As shown, it includes a client 110 and a server 120, which can be connected via a wired network or a wireless network, without limitation herein.

[0074] Client 110 can display a short video publishing interface, including a display area for short videos to be published and a display area for a set of recommended topic tags, which can be obtained from server 120. Client 110 can be, but is not limited to, smartphones, desktop computers, tablets, laptops, smart speakers, digital assistants, augmented reality (AR) / virtual reality (VR) devices, smart wearable devices, etc. It can also be software running on the aforementioned clients, such as an application (app). An application can be a standalone application or a subroutine within an application. For example, an application can be a news application, a live streaming application, or a video application. Optionally, the operating system running on client 110 can include, but is not limited to, Android, iOS, Linux, Windows, Unix, etc.

[0075] Server 120 can be a server providing backend services for applications in client 110. Specifically, the service provided by server 120 can be a topic tag recommendation service, determining a set of recommended topic tags for short videos to be published on the publishing interface of client 110. Server 120 can be a standalone physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN (Content Delivery Network), and big data and artificial intelligence platforms. The operating system running on server 120 can include, but is not limited to, Android, iOS, Linux, Windows, Unix, etc.

[0076] It should be noted that, Figure 1 The application environment shown is merely an example. In practical applications, the topic tag recommendation method of this disclosure embodiment can be executed by the client and server in cooperation, or it can be executed by the client or server independently. This disclosure embodiment does not limit the specific application environment.

[0077] Figure 2 This is a flowchart illustrating a topic tag recommendation method according to an exemplary embodiment, such as... Figure 2 As shown, the topic tag recommendation method can be applied to the server, including the following steps:

[0078] In step S201, audio information of the audio data in the multimedia content to be published is obtained; audio information is an attribute feature of audio data; audio information is unique.

[0079] In this embodiment, the multimedia content is content that matches the actual application scenario. In some possible application scenarios, the multimedia content may include music, video, advertisements, news information, etc. Before the multimedia content is officially published, it will go through several different stages or states. For example, initially, the multimedia content is triggered by the user to enter the editing state. Specifically, the user triggers the short video shooting button to enter the editing state, shoots short video content, and selects background music. Then, the user triggers the editing completion button to enter the pre-publishing state. The client displays the corresponding publishing interface and sends the multimedia content publishing request to the server. Thus, the aforementioned pre-published multimedia content is the multimedia content in the pre-publishing state. For the multimedia content in the pre-publishing state, the user can further edit it through the client on the relevant interface, such as editing text descriptions and selecting topic tags. The topic tags can be obtained from the server. On the server side, the server obtains the audio information of the audio data in the pre-published multimedia content and automatically generates a set of recommended topic tags for the pre-published multimedia content for the user to choose from.

[0080] Audio data can refer to the background music used in the multimedia content to be published. Audio information is unique, a characteristic that distinguishes one audio data point from another. The form of audio information can include at least one of letters, numbers, and special symbols. Specifically, audio information can be obtained by extracting audio fingerprints from the audio data in the multimedia content to be published using audio fingerprinting technology. Audio fingerprinting technology refers to extracting unique numerical features from a segment of audio data using a specific algorithm, thus audio information can also be interpreted as the audio fingerprint of the audio data.

[0081] In step S203, the audio topic correspondence is obtained; wherein, the audio topic correspondence includes multiple reference audio information and at least one associated topic tag corresponding to each reference audio information.

[0082] In this embodiment of the disclosure, the server pre-constructs an audio topic correspondence, which includes multiple reference audio information and at least one associated topic tag corresponding to each of the multiple reference audio information.

[0083] In some possible embodiments, the server constructs audio topic correspondences by performing data mining on historical multimedia content. Therefore, the topic tag recommendation method of this disclosure also includes, for example... Figure 3The following steps are shown:

[0084] In step S301, multiple historical multimedia contents that have been published are obtained.

[0085] In one specific embodiment, the server can obtain multiple historical multimedia contents published within a historical time period, where the historical time period includes, but is not limited to, the past 1 hour, 2 hours, 3 hours, etc.; for example, the server obtains all historical multimedia contents published in the past 1 hour.

[0086] In another specific embodiment, the server may also obtain a preset number of recently published historical multimedia content, wherein the preset number includes, but is not limited to, thousands or tens of thousands; for example, the server obtains 10,000 recently published historical multimedia content.

[0087] In the two specific embodiments described above, the server can filter based on the publication timestamp information carried by each historical multimedia content to obtain all historical multimedia content published in the past hour or 10,000 most recently published historical multimedia content.

[0088] In step S303, the historical audio information of the audio data in each historical multimedia content is used as the audio information to be referenced; the historical audio information of the audio data in each historical multimedia content is obtained by extracting audio fingerprints from the audio data in each historical multimedia content.

[0089] In this step, the server retrieves the historical audio information of the audio data in each historical multimedia content and uses this historical audio information as the reference audio information.

[0090] In one specific embodiment, obtaining historical audio information from audio data in each historical multimedia content can specifically include: extracting audio fingerprints from audio data in each historical multimedia content to obtain historical audio fingerprints corresponding to each historical multimedia content, i.e., historical audio information.

[0091] Generally, only when the audio data is the same in two historical multimedia contents, such as when the same background music is used, will the server obtain the same historical audio information corresponding to the two historical multimedia contents. In this case, the historical audio information corresponding to either of the two historical multimedia contents can be used as the audio information to be referenced.

[0092] In step S305, the historical topic tags carried by each historical multimedia content are used as topic tags to be associated.

[0093] In this step, the server uses the historical topic tags carried by each historical multimedia content as topic tags to be associated.

[0094] In a specific example, suppose the server obtains multimedia content published in the past hour, namely Video 1 to Video 8; after audio fingerprint extraction, the historical audio information of the audio data in each video is as follows: the historical audio information of Video 1 is m1, the historical audio information of Video 2 is m2, the historical audio information of Video 3 is m3, the historical audio information of Video 4 is m4, the historical audio information of Video 5 is m3, the historical audio information of Video 6 is m2, the historical audio information of Video 7 is m5, and the historical audio information of Video 8 is m6; the above m1, m2, m3, m4, m5, and m6 are used as the audio information to be referenced.

[0095] The historical hashtags carried by each video are as follows: Video 1 carries #A, Video 2 carries #A#B, Video 3 carries #C, Video 4 carries #D, Video 5 carries #E#C, Video 6 carries #B, Video 7 carries #E, and Video 8 carries #D; the above #A, #B, #C, #D, and #E will be used as hashtags to be associated with.

[0096] In step S307, an audio topic correspondence is constructed based on multiple topic tags to be associated involving multiple historical multimedia content and multiple audio information to be referenced involving multiple historical multimedia content.

[0097] In this step, an audio topic correspondence is constructed based on multiple topic tags to be associated involving multiple historical multimedia content and multiple audio information to be referenced involving multiple historical multimedia content; wherein, the relationship between the audio information to be referenced and the topic tags to be associated can be one-to-one, one-to-many, or many-to-one.

[0098] Referring back to the example above, this involves establishing a correspondence between the audio information to be referenced (m1, m2, m3, m4, m5, m6) and the topic tags to be associated (#A, #B, #C, #D, #E). The process of constructing this audio topic correspondence will be explained in detail below, and will not be elaborated upon here.

[0099] In step S309, the audio topic correspondence is updated according to a preset time interval.

[0100] In this step, the server updates the audio topic mapping relationship every preset time by setting a scheduled task. Each update executes steps S301-S307 to construct a new audio topic mapping relationship and replace the historical one, thus completing the update. The preset time can be set according to the frequency of multimedia content release in actual applications. For example, if the server considers the current multimedia content release to be frequent, the preset time can be set to 30 minutes.

[0101] Based on the example above, the historical audio topic correspondence is the audio topic correspondence between m1, m2, m3, m4, m5, m6 and #A, #B, #C, #D, #E. 30 minutes after the server completes the above historical audio topic correspondence, it triggers a scheduled task to update the above historical audio topic correspondence. Specifically, starting from step S301, it retrieves the historical multimedia content published in the past hour, and then executes subsequent steps S303 to S307 in sequence to obtain the new audio topic correspondence.

[0102] In the above embodiments, the server associates historical topic tags and historical audio information based on historical multimedia content to construct an audio-topic correspondence, and updates the audio-topic correspondence regularly to ensure an effective correspondence between audio information and topic tags. This helps to achieve accurate topic tag recommendations based on the audio of multimedia content to be published.

[0103] The following details the process of constructing audio topic correspondences. In some possible embodiments, the above-mentioned construction of audio topic correspondences is based on multiple topic tags to be associated involving multiple historical multimedia content and multiple audio information to be referenced involving multiple historical multimedia content. Specifically, this may include, for example... Figure 4 The following steps are shown:

[0104] In step S401, based on each audio information to be referenced, multiple historical multimedia contents are classified to obtain multiple sets of historical multimedia contents; each set of historical multimedia contents corresponds one-to-one with each audio information to be referenced.

[0105] In this step, considering that in practical applications, audio topic correspondence is constructed based on massive amounts of historical multimedia content, there is a high probability that a large amount of historical multimedia content uses the same audio data. Therefore, the server first classifies the massive amount of historical multimedia content from the audio dimension. After classification, a set of historical multimedia content corresponding to each audio information to be referenced is obtained.

[0106] Referring back to the example above, after classifying the videos 1 through 8 from the audio dimension, the historical multimedia content set corresponding to the audio information to be referenced, m1, is {Video 1}; the historical multimedia content set corresponding to the audio information to be referenced, m2, is {Video 2, Video 6}; the historical multimedia content set corresponding to the audio information to be referenced, m3, is {Video 3, Video 5}; the historical multimedia content set corresponding to the audio information to be referenced, m4, is {Video 4}; the historical multimedia content set corresponding to the audio information to be referenced, m5, is {Video 7}; and the historical multimedia content set corresponding to the audio information to be referenced, m6, is {Video 8}.

[0107] In step S403, based on the topic tags to be associated carried by each historical multimedia content in each historical multimedia content set, a set of associated topic tags corresponding to each audio information to be referenced is obtained.

[0108] Among them, each associated topic tag in the set of associated topic tags corresponding to each audio information to be referenced indicates the topic tag that is associated with each audio information to be referenced.

[0109] In this step, after the server obtains the set of historical multimedia content corresponding to each audio information to be referenced, it takes the topic tags to be associated carried by each historical multimedia content in the set as the associated topic tags corresponding to the corresponding audio information to be referenced, thereby obtaining the set of associated topic tags corresponding to each audio information to be referenced, that is, obtaining the audio topic correspondence.

[0110] For example, in the example above, the set of historical multimedia content corresponding to the audio information m1 to be referenced is {Video 1}, and Video 1 carries #A. Thus, the server can determine that the set of associated topic tags corresponding to the audio information m1 to be referenced is {#A}. Correspondingly, the set of associated topic tags corresponding to the audio information m2 to be referenced is {#A,#B}, the set of associated topic tags corresponding to the audio information m3 to be referenced is {#C,#E}, the set of associated topic tags corresponding to the audio information m4 to be referenced is {#D}, the set of associated topic tags corresponding to the audio information m5 to be referenced is {#E}, and the set of associated topic tags corresponding to the audio information m6 to be referenced is {#D}.

[0111] Furthermore, considering that in practical applications, the number of associated topic tags in the set of associated topic tags corresponding to each audio information to be referenced is not only one or two as in the example above, but hundreds or thousands; and in recommendation scenarios, multimedia publishing interfaces do not recommend hundreds or thousands of associated topic tags. Therefore, when constructing the audio topic correspondence, the number of associated topic tags in the set of associated topic tags corresponding to each audio information to be referenced can be limited. For example, the number of associated topic tags in the set of associated topic tags can be 10, 20, etc.

[0112] Therefore, for each audio information to be referenced, the server filters and selects the related topic tags carried by each historical multimedia content in the historical multimedia content set corresponding to each audio information to obtain a set of related topic tags corresponding to each audio information; that is, in a specific embodiment, the above step S403 may include, for example, Figure 5 The following steps are shown:

[0113] In step S501, based on each topic tag to be associated, multiple historical multimedia contents are classified to obtain the number of historical multimedia contents corresponding to each topic tag to be associated.

[0114] In this step, the multiple unassociated topic tags refer to the multiple unassociated topic tags involved in the multiple historical multimedia content mentioned in step S301 above. Based on each of these multiple unassociated topic tags, the multiple historical multimedia content is classified to obtain the number of historical multimedia content corresponding to each unassociated topic tag. That is, similar to the classification of multiple historical multimedia content from the audio dimension above, considering that a large number of historical multimedia content uses the same historical topic tags in actual applications, here, the server classifies multiple historical multimedia content from the dimension of topic tags, and after classification, obtains the number of historical multimedia content corresponding to each unassociated topic tag.

[0115] Referring back to the example above, after classifying videos 1 to 8 from the perspective of topic tags, we find that there are 2 pieces of historical multimedia content corresponding to the topic tag #A, 2 pieces of historical multimedia content corresponding to the topic tag #B, 2 ​​pieces of historical multimedia content corresponding to the topic tag #C, 2 pieces of historical multimedia content corresponding to the topic tag #D, and 2 pieces of historical multimedia content corresponding to the topic tag #E.

[0116] In step S503, the topic tags to be associated carried by each historical multimedia content in each historical multimedia content set are used as the topic tag set to be associated corresponding to each audio information to be referenced.

[0117] In this step, as described above, the server, for each audio information to be referenced, firstly, directly uses the topic tags to be associated carried by each historical multimedia content in the historical multimedia content set corresponding to the audio information to be referenced as the topic tags to be associated corresponding to the audio information to be referenced, thus obtaining the set of topic tags to be associated corresponding to the audio information to be referenced; here, as mentioned before, in actual applications, the number of topic tags to be associated in the set of topic tags to be associated may be in the tens of thousands.

[0118] In step S505, based on the number of historical multimedia contents corresponding to each topic tag to be associated, a set of associated topic tags corresponding to each audio information to be referenced is determined from the set of topic tags to be associated corresponding to each audio information to be referenced; the number of historical multimedia contents corresponding to each associated topic tag in the set of associated topic tags corresponding to each audio information to be referenced is greater than or equal to a first preset number.

[0119] In this step, it can be understood that the more historical multimedia content corresponding to the topic tag to be associated, the more popular the topic tag to be associated is under the corresponding audio information to be referenced. Therefore, for each audio information to be referenced, the server filters the set of topic tags to be associated corresponding to the audio information based on the number of historical multimedia content corresponding to each associated topic tag in the set of associated topic tags to be referenced. The number of historical multimedia content corresponding to each associated topic tag in the set of associated topic tags obtained after filtering is greater than or equal to the first preset number.

[0120] The first preset quantity mentioned above can be a fixed quantity, for example, it can be 10; thus, only the topic tags to be associated that have more than 10 corresponding historical multimedia contents in the topic tag set to be associated are retained, and the remaining topic tags to be associated are deleted, thus obtaining the associated topic tag set. In the example above, the set of associated topic tags corresponding to the audio information m1 to be referenced is {#A}. However, as mentioned earlier in practical applications, the set of associated topic tags that the audio information m1 to be referenced may correspond to is {#A,#B,#C,#D,#E,#F,#G,#H}. Therefore, it is necessary to filter them: determine the number of historical multimedia contents corresponding to each of #A,#B,#C,#D,#E,#F,#G,#H in multiple historical multimedia contents. Assume that the number of historical multimedia contents using #A is 11, the number of historical multimedia contents using #B is 13, the number of historical multimedia contents using #C is 4, the number of historical multimedia contents using #D is 6, the number of historical multimedia contents using #E is 10, the number of historical multimedia contents using #F is 8, the number of historical multimedia contents using #G is 7, and the number of historical multimedia contents using #H is 1. Thus, the set of associated topic tags obtained after retaining only those with more than 10 corresponding historical multimedia contents is {#A,#B,#E}.

[0121] The aforementioned first preset quantity can also be dynamically adjusted according to the different audio information to be referenced. Specifically, based on the number of historical multimedia contents corresponding to each topic tag to be associated in the set of topic tags to be associated, the topic tags to be associated in the set of topic tags to be associated are sorted in descending order, and then the top N topic tags to be associated are used as associated topic tags to obtain the set of associated topic tags; at this time, the first preset quantity is the number of historical multimedia contents corresponding to the Nth topic tag to be associated. Where N is a natural number greater than or equal to 1, for example, N can be 5; corresponding to the example above, based on the number of historical multimedia contents corresponding to each topic tag in the set of topic tags to be associated {#A,#B,#C,#D,#E,#F,#G,#H} corresponding to the audio information to be referenced m1, #A,#B,#C,#D,#E,#F,#G,#H are sorted in descending order, and the result is #B,#A,#E,#F,#G,#D,#C,#H; thus, the set of associated topic tags obtained after keeping the first five is {#B,#A,#E,#F,#G}.

[0122] In step S405, the audio topic correspondence is obtained based on the set of associated topic tags corresponding to each audio information to be referenced.

[0123] In this step, after the server obtains the set of associated topic tags corresponding to each audio information to be referenced, it marks each audio information to be referenced as reference audio information, and then generates an audio topic correspondence; the audio topic correspondence includes the set of associated topic tags corresponding to each reference audio information.

[0124] In the above embodiments, for each audio information to be referenced, the server counts the number of historical multimedia contents corresponding to each topic tag to be associated under the audio information to be referenced. Based on the number of historical multimedia contents corresponding to each topic tag to be associated, the server further filters out the popular topic tags under the audio information to be referenced. On the one hand, this can reduce the amount of data in the final generated audio topic correspondence, which is convenient for retrieval during subsequent recommendations. On the other hand, it ensures that each reference audio information in the correspondence corresponds to popular topic tags that are widely used by users under each reference audio information, which helps to improve the accuracy of subsequent topic tag recommendations.

[0125] In practical applications, the number of associated topic tags corresponding to each reference audio information in the audio information correspondence relationship should not be too small; otherwise, the number of recommended topic tags available for users to choose from will be limited, thus affecting the recommendation effect. Therefore, when the number of associated topic tags in the set of associated topic tags corresponding to the audio information to be referenced is small, appropriate associated topic tags can be added by combining other audio information to be referenced. Thus, in some possible embodiments, before step S505 above, the topic tag recommendation method of this disclosure embodiment may further include the following: Figure 6 The following steps are shown:

[0126] In step S601, the number of associated topic tags in the associated topic tag set corresponding to each audio information to be referenced is determined.

[0127] In step S603, audio information to be referenced that has a number of associated topic tags less than or equal to a second preset number is determined as audio information to be matched.

[0128] In the above steps, for each piece of audio information to be referenced, the server determines the number of associated topic tags in the associated topic tag set corresponding to each piece of audio information to be referenced; then, audio information to be referenced with a number of associated topic tags in the associated topic tag set that is less than or equal to a second preset number is identified and used as audio information to be matched. The second preset number can be a fixed value; for example, the second preset number can be 3 or 5.

[0129] In step S605, the target reference audio information is determined from multiple audio information to be referenced based on the similarity value between the audio information to be matched and each audio information to be referenced and the number of associated topic tags corresponding to each audio information to be referenced.

[0130] Among them, the similarity value between the target reference audio information and the audio information to be matched is greater than or equal to the preset similarity value, and the number of associated topic tags corresponding to the target reference audio information is greater than the third preset number.

[0131] In this step, the server sequentially calculates the similarity value between the audio information to be matched and each of the other audio information to be referenced. Based on the similarity value between the audio information to be matched and each audio information to be referenced and the number of associated topic tags corresponding to each audio information to be referenced, the target reference audio information is determined from the multiple audio information to be referenced.

[0132] Specifically, the server selects the audio information from multiple reference audio information sets whose similarity value to the audio information to be matched is greater than or equal to a preset similarity value as the quasi-target reference audio information. A higher similarity value indicates that the two audio information sets have similar audio fingerprints, meaning the corresponding audio data are similar. Here, the preset similarity value can be a fixed value, for example, 0.99; alternatively, it can be non-fixed. Based on the similarity value between the actual audio information to be matched and each reference audio information set, the reference audio information sets are sorted in descending order. The first reference audio information set in the sorted result is the quasi-target reference audio information, and its similarity value is the preset similarity value. Then, if the number of associated topic tags corresponding to the quasi-target reference audio information is greater than the third preset number, the quasi-target reference audio information is used as the target reference audio information. Here, the third preset number can be a fixed value, for example, the third preset number can be 3 or 5. The third preset number can also be non-fixed. For example, when there are multiple quasi-target reference audio information, the quasi-target reference audio information can be sorted in descending order based on the number of associated topic tags corresponding to each quasi-target reference audio information. The quasi-target reference audio information ranked first in the sorting result is the target reference audio information. Correspondingly, the number of associated topic tags corresponding to the quasi-target reference audio information ranked first is the third preset number.

[0133] In step S607, the set of associated topic tags corresponding to the target reference audio information is used as the set of associated topic tags corresponding to the audio information to be matched.

[0134] In this step, after the server determines the target reference audio information that matches the audio information to be matched, it uses the set of associated topic tags corresponding to the target reference audio information as the set of associated topic tags corresponding to the audio information to be matched. This can make up for the situation where there are too few associated topic tags for the audio information to be matched.

[0135] Furthermore, if the number of associated topic tags corresponding to the quasi-target reference audio information is less than or equal to the third preset number, the quasi-target reference audio information is re-determined. Specifically, from the sorting results obtained based on the above similarity values, the second-ranked audio information to be referenced is taken as the quasi-target reference audio information. Then, the number of associated topic tags corresponding to the quasi-target reference audio information is determined. If it is still not satisfied, the similarity value sorting results are traversed until a target reference audio information that meets the above judgment conditions is found, and its corresponding associated topic tags are taken as the associated topic tags corresponding to the audio information to be matched.

[0136] In the above embodiments, when the number of associated topic tags in the associated topic tag set corresponding to the audio information to be referenced is small, appropriate supplementation of associated topic tags by combining other audio information to be referenced can avoid the situation where there are too few associated topic tags corresponding to each reference audio information in the audio information correspondence, resulting in insufficient recommended topic tags available for users to choose from, and can improve the topic tag recommendation effect.

[0137] In step S205, target audio information that matches the audio information is determined from multiple reference audio information.

[0138] In step S207, at least one associated topic tag corresponding to the target audio information is used as a set of recommended topic tags for the multimedia content to be published; the set of recommended topic tags is used to push to the publishing client corresponding to the multimedia content to be published.

[0139] In this embodiment, the server indexes the audio topic correspondence and finds the target audio information that matches the audio information of the multimedia content to be published from multiple reference audio information. Then, at least one associated topic tag corresponding to the target audio information is used as a recommended topic tag set for the multimedia content to be published. The recommended topic tag set is used to push to the publishing client corresponding to the multimedia content to be published. The server can send the recommended topic tag set to the client, and after receiving it, the client displays it on the corresponding publishing interface for the user to select.

[0140] In some possible embodiments, determining the target audio information that matches the audio information from a plurality of reference audio information may specifically include, for example: Figure 7 The following steps are shown:

[0141] In step S701, the matching degree value between the audio information and each reference audio information is determined.

[0142] In this step, the server determines the degree of matching between each reference audio piece of information and the audio information of the multimedia content to be published.

[0143] In step S703, if there is a reference audio information among the multiple reference audio information whose matching degree value is greater than or equal to a preset value, then the reference audio information whose matching degree value is greater than or equal to the preset value is taken as the target audio information.

[0144] In this step, the server can set an initial preset value of 1. The server first determines whether there is audio information of the multimedia content to be published among multiple reference audio information. If it exists, it is directly determined as the target audio information. If it does not exist, the server attenuates the initial preset value according to a fixed attenuation value to obtain an intermediate preset value, until the server determines a reference audio information that is greater than or equal to the intermediate preset value among multiple reference audio information.

[0145] Specifically, the aforementioned fixed attenuation value can be 0.02. When the server determines that there is no audio information for the multimedia content to be published among the multiple reference audio information, the initial preset value is attenuated according to the attenuation value of 0.02, and the intermediate preset value is determined to be 0.98. Then, the server determines whether there is any reference audio information greater than or equal to 0.98 among the multiple reference audio information. If not, the server determines the intermediate preset value to be 0.96, and then continues to determine whether there is any reference audio information greater than or equal to 0.96 among the multiple reference audio information. This continues until a reference audio information that meets the intermediate preset value is found, and it is taken as the target audio information.

[0146] In the above embodiments, the server retrieves the corresponding topic tags directly from the pre-built audio topic correspondence based on the audio information of the multimedia to be published. Compared with the method of determining topic tags through online real-time training models in related technologies, this can greatly speed up the production speed and efficiency of recommendation results.

[0147] In some possible embodiments, after using at least one associated topic tag corresponding to the target audio information as a set of recommended topic tags for the multimedia content to be published, the topic tag recommendation method of this disclosure may further include, for example: Figure 8 The following steps are shown:

[0148] In step S801, the number of account clicks for each recommended topic tag in the recommended topic tag set is obtained.

[0149] Specifically, the server can obtain the number of account clicks for each recommended topic tag in the recommended topic tag set. The number of account clicks is calculated by the server based on the actual clicks of users in the historical recommendation data.

[0150] In step S803, the recommended topic tags in the recommended topic tag set are sorted according to the number of account clicks for each recommended topic tag, resulting in a sorted recommended topic tag set.

[0151] In this step, the server combines the number of clicks per account for each recommended topic tag and sorts the recommended topic tags in the recommended topic tag set in descending order to obtain a sorted recommended topic tag set. The sorted recommended topic tag set is then sent to the client, which can display them sequentially in the corresponding publishing interface.

[0152] In the above embodiments, the server sorts the recommended topic tags in the recommended topic tag set based on the actual click data of users in the historical recommendation data, so that the client can display the recommended topic tags with a higher click probability in a prominent position at the front, making it easier for users to find them, thereby helping to improve the effect of topic tag recommendation.

[0153] In summary, the topic tag recommendation method of this disclosure can quickly determine the recommended topic tag set for the short video to be published by identifying the audio information in the audio data of the short video to be published and based on the pre-constructed audio topic correspondence. The topic tag recommendation method of this disclosure has a simple algorithm, requires relatively few computational resources, and can improve the efficiency of topic tag recommendation when the demand for publishing is large.

[0154] Figure 9 This is a block diagram illustrating a topic tag recommendation device according to an exemplary embodiment. (Refer to...) Figure 9 The device includes a first acquisition module 901, a second acquisition module 902, a first determination module 903, and a second determination module 904.

[0155] The first acquisition module 901 is configured to acquire audio information from the audio data in the multimedia content to be published; audio information is an attribute feature of audio data; audio information is unique.

[0156] The second acquisition module 902 is configured to acquire audio topic correspondence; wherein, the audio topic correspondence includes multiple reference audio information and at least one associated topic tag corresponding to each reference audio information;

[0157] The first determining module 903 is configured to determine the target audio information that matches the audio information from multiple reference audio information;

[0158] The second determining module 904 is configured to use at least one associated topic tag corresponding to the target audio information as a set of recommended topic tags for the multimedia content to be published; the set of recommended topic tags is used to push to the publishing client corresponding to the multimedia content to be published.

[0159] In some possible embodiments, the apparatus further includes a construction module configured to perform the following actions: acquiring multiple published historical multimedia content; using historical audio information of audio data in each historical multimedia content as reference audio information; the historical audio information of audio data in each historical multimedia content is obtained by extracting audio fingerprints from the audio data in each historical multimedia content; using historical topic tags carried by each historical multimedia content as topic tags to be associated; constructing an audio topic correspondence based on the multiple topic tags to be associated and the multiple reference audio information involved in the multiple historical multimedia content; and updating the audio topic correspondence at preset time intervals.

[0160] In some possible embodiments, the construction module is further configured to perform the following actions: classifying multiple historical multimedia contents based on each audio information to be referenced, resulting in multiple sets of historical multimedia contents; each set of historical multimedia contents corresponds one-to-one with each audio information to be referenced; based on the topic tags to be associated carried by each historical multimedia content in each set of historical multimedia contents, obtaining a set of associated topic tags corresponding to each audio information to be referenced; each associated topic tag in the set of associated topic tags corresponding to each audio information to be referenced indicates a topic tag that is associated with each audio information to be referenced; and obtaining the audio topic correspondence based on the set of associated topic tags corresponding to each audio information to be referenced.

[0161] In some possible embodiments, the construction module is further configured to perform the following: classify multiple historical multimedia contents based on each topic tag to be associated, and obtain the number of historical multimedia contents corresponding to each topic tag to be associated; use the topic tags to be associated carried by each historical multimedia content in each historical multimedia content set as the topic tag set to be associated corresponding to each audio information to be referenced; determine the associated topic tag set corresponding to each audio information to be referenced from the topic tag set corresponding to each audio information to be referenced based on the number of historical multimedia contents corresponding to each topic tag to be associated; and ensure that the number of historical multimedia contents corresponding to each associated topic tag in the associated topic tag set corresponding to each audio information to be referenced is greater than or equal to a first preset number.

[0162] In some possible embodiments, the construction module is further configured to perform the following: determine the number of associated topic tags in the associated topic tag set corresponding to each audio information to be referenced; determine audio information to be matched whose number of associated topic tags in the associated topic tag set is less than or equal to a second preset number; determine target reference audio information from multiple audio information to be referenced based on the similarity value between the audio information to be matched and each audio information to be referenced and the number of associated topic tags corresponding to each audio information to be referenced; the similarity value between the target reference audio information and the audio information to be matched is greater than or equal to a preset similarity value, and the number of associated topic tags corresponding to the target reference audio information is greater than a third preset number; and use the associated topic tag set corresponding to the target reference audio information as the associated topic tag set corresponding to the audio information to be matched.

[0163] In some possible embodiments, the first determining module 903 is further configured to determine the matching degree value between the audio information and each reference audio information; if there is a reference audio information among the multiple reference audio information whose matching degree value is greater than or equal to a preset value, the reference audio information whose matching degree value is greater than or equal to the preset value is taken as the target audio information.

[0164] In some possible embodiments, the apparatus further includes a sorting module configured to perform the following operations: obtaining the number of account clicks for each recommended topic tag in the recommended topic tag set; and sorting the recommended topic tags in the recommended topic tag set according to the number of account clicks for each recommended topic tag to obtain a sorted set of recommended topic tags.

[0165] Regarding the apparatus in the above embodiments, the specific manner in which each module performs its operation has been described in detail in the embodiments related to the method, and will not be elaborated upon here.

[0166] Figure 10 This is a block diagram illustrating an electronic device for topic tag recommendation according to an exemplary embodiment. The electronic device may be a terminal, and its internal structure diagram may be as follows: Figure 10As shown, the electronic device includes a processor, memory, network interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage medium. The network interface is used to communicate with external terminals via a network connection. When the computer program is executed by the processor, it implements a topic tag recommendation method. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad mounted on the device's casing, or an external keyboard, touchpad, or mouse.

[0167] Those skilled in the art will understand that Figure 10 The structure shown is merely a block diagram of a portion of the structure related to the present disclosure and does not constitute a limitation on the electronic device to which the present disclosure is applied. A specific electronic device may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0168] In an exemplary embodiment, an electronic device is also provided, including: a processor; and a memory for storing processor-executable instructions; wherein the processor is configured to execute the instructions to implement the topic tag recommendation method as described in the embodiments of this disclosure.

[0169] In an exemplary embodiment, a computer-readable storage medium is also provided, wherein when the instructions in the storage medium are executed by a processor of an electronic device, the electronic device is enabled to perform the topic tag recommendation method in the embodiments of this disclosure.

[0170] In an exemplary embodiment, a computer program product including instructions is also provided. The computer program product includes a computer program stored in a readable storage medium. At least one processor of a computer device reads from the readable storage medium and executes the computer program, causing the computer device to perform the topic tag recommendation method of the present disclosure embodiments.

[0171] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. This computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and RAMbus dynamic RAM (RDRAM), etc.

[0172] Other embodiments of this disclosure will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not disclosed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of this disclosure are indicated by the following claims.

[0173] It should be understood that this disclosure is not limited to the precise structures described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of this disclosure is limited only by the appended claims.

Claims

1. A topic tag recommendation method, characterized in that, include: Obtain audio data from the multimedia content to be published, and extract audio fingerprints from the audio data to obtain the audio information of the multimedia content to be published. The audio information is an attribute feature of the audio data; The audio information is unique; Based on the audio information, a pre-constructed audio topic correspondence is retrieved; wherein, the audio topic correspondence includes multiple reference audio information and a set of associated topic tags corresponding to each reference audio information, the multiple reference audio information being determined based on audio data corresponding to audio data in multiple historical multimedia content; the audio information corresponding to each historical multimedia content is obtained by extracting audio fingerprints from the audio data in the historical multimedia content; From the plurality of reference audio information, determine the target audio information that matches the audio information; At least one associated topic tag corresponding to the target audio information is used as a set of recommended topic tags for the multimedia content to be published; the set of recommended topic tags is used to push the multimedia content to be published to the publishing client. The process of constructing the audio topic correspondence includes: Based on the aforementioned reference audio information, the historical multimedia content is grouped to obtain multiple sets of historical multimedia content; wherein, each set of historical multimedia content corresponds to one set of reference audio information. For each set of historical multimedia content, based on the distribution of topic tags carried by each historical multimedia content in the set, a set of associated topic tags for the audio information to be referenced corresponding to the set of historical multimedia content is determined. If there is audio information to be referenced that does not meet the preset number of tags in the set of associated topic tags, the set of associated topic tags corresponding to the audio information to be referenced is supplemented based on the similarity between the audio information to be referenced and other audio information to be referenced, and the number of tags in the set of associated topic tags corresponding to the other audio information to be referenced. Based on each of the reference audio information and its corresponding set of associated topic tags, the audio topic correspondence is constructed.

2. The topic tag recommendation method according to claim 1, characterized in that, The process of constructing the audio topic correspondence also includes: Retrieve multiple published historical multimedia content; The historical audio information of the audio data in each of the historical multimedia contents is used as the audio information to be referenced; the historical audio information of the audio data in each of the historical multimedia contents is obtained by extracting audio fingerprints from the audio data in each of the historical multimedia contents. The historical topic tags carried by each of the aforementioned historical multimedia contents will be used as topic tags to be associated. Based on the multiple topic tags to be associated involved in the multiple historical multimedia content and the multiple audio information to be referenced involved in the multiple historical multimedia content, the audio topic correspondence is constructed; The audio topic correspondence is updated at preset time intervals.

3. The topic tag recommendation method according to claim 2, characterized in that, The step of constructing the audio topic correspondence based on multiple unassociated topic tags involved in the multiple historical multimedia content and multiple reference audio information involved in the multiple historical multimedia content includes: Based on the audio information to be referenced, the multiple historical multimedia contents are classified to obtain multiple sets of historical multimedia contents; each set of historical multimedia contents corresponds one-to-one with each set of audio information to be referenced. Based on the topic tags to be associated carried by each historical multimedia content in each set of historical multimedia content, a set of associated topic tags corresponding to each audio information to be referenced is obtained; each associated topic tag in the set of associated topic tags corresponding to each audio information to be referenced indicates a topic tag that is associated with each audio information to be referenced. Based on the set of associated topic tags corresponding to each of the audio information to be referenced, the audio topic correspondence is obtained.

4. The topic tag recommendation method according to claim 3, characterized in that, The process involves obtaining a set of associated topic tags corresponding to each set of audio information to be referenced, based on the topic tags carried by each historical multimedia content in each set of historical multimedia content, including: Based on each of the topic tags to be associated, the multiple historical multimedia contents are classified to obtain the number of historical multimedia contents corresponding to each of the topic tags to be associated. Each historical multimedia content in each of the aforementioned historical multimedia content sets carries a topic tag to be associated, which is then used as the topic tag set to be associated corresponding to each of the aforementioned audio information to be referenced. Based on the number of historical multimedia contents corresponding to each of the aforementioned topic tags to be associated, a set of associated topic tags corresponding to each of the aforementioned audio information to be referenced is determined from the set of associated topic tags corresponding to each of the aforementioned audio information to be referenced; the number of historical multimedia contents corresponding to each associated topic tag in the set of associated topic tags corresponding to each of the aforementioned audio information to be referenced is greater than or equal to a first preset number.

5. The topic tag recommendation method according to claim 4, characterized in that, In the case where there is audio information to be referenced that does not meet the preset tag quantity condition in the set of associated topic tags, the set of associated topic tags corresponding to the audio information to be referenced is supplemented based on the similarity between the audio information to be referenced and other audio information to be referenced, and the number of tags in the set of associated topic tags corresponding to the other audio information to be referenced. This includes: Determine the number of associated topic tags in the associated topic tag set corresponding to each of the aforementioned audio information to be referenced; Audio information with a number of associated topic tags less than or equal to the second preset number in the associated topic tag set is identified as audio information to be matched. Based on the similarity value between the audio information to be matched and each of the audio information to be referenced, and the number of associated topic tags corresponding to each of the audio information to be referenced, a target reference audio information is determined from the plurality of audio information to be referenced; the similarity value between the target reference audio information and the audio information to be matched is greater than or equal to a preset similarity value, and the number of associated topic tags corresponding to the target reference audio information is greater than a third preset number. The set of associated topic tags corresponding to the target reference audio information is used as the set of associated topic tags corresponding to the audio information to be matched.

6. The topic tag recommendation method according to claim 1, characterized in that, The step of determining the target audio information that matches the audio information from the plurality of reference audio information includes: Determine the degree of matching between the audio information and each of the reference audio information; If among the plurality of reference audio information, there is reference audio information whose matching degree value with the audio information is greater than or equal to a preset value, the reference audio information whose matching degree value is greater than or equal to the preset value is taken as the target audio information.

7. The topic tag recommendation method according to claim 1, characterized in that, After using the set of associated topic tags corresponding to the target audio information as the recommended topic tag set for the multimedia content to be published, the method further includes: Get the number of account clicks for each recommended topic tag in the recommended topic tag set; Based on the number of account clicks for each recommended topic tag, the recommended topic tags in the recommended topic tag set are sorted to obtain a sorted recommended topic tag set.

8. A topic tag recommendation device, characterized in that, include: The first acquisition module is configured to acquire audio data from the multimedia content to be published, and extract audio fingerprints from the audio data to obtain the audio information of the multimedia content to be published. The audio information is an attribute feature of the audio data; The audio information is unique; The second acquisition module is configured to retrieve a pre-constructed audio topic correspondence based on the audio information; wherein, the audio topic correspondence includes multiple reference audio information and a set of associated topic tags corresponding to each reference audio information, the multiple reference audio information being determined based on audio data corresponding to audio data in multiple historical multimedia content; the audio information corresponding to each historical multimedia content is obtained by extracting audio fingerprints from the audio data in the historical multimedia content; The first determining module is configured to determine target audio information that matches the audio information from the plurality of reference audio information; The second determining module is configured to use the set of associated topic tags corresponding to the target audio information as a set of recommended topic tags for the multimedia content to be published; the set of recommended topic tags is used to push the multimedia content to be published to the publishing client. The building module is configured to perform the process of building the audio topic correspondence; The process of constructing the audio topic correspondence includes: Based on the aforementioned audio information to be referenced, the multiple historical multimedia contents are grouped to obtain multiple sets of historical multimedia contents; wherein, each set of historical multimedia contents corresponds to one set of audio information to be referenced. For each set of historical multimedia content, based on the distribution of topic tags carried by each historical multimedia content in the set, determine the set of associated topic tags for the audio information to be referenced corresponding to the set of historical multimedia content. If there is audio information to be referenced that does not meet the preset number of tags in the set of associated topic tags, the set of associated topic tags corresponding to the audio information to be referenced is supplemented based on the similarity between the audio information to be referenced and other audio information to be referenced, and the number of tags in the set of associated topic tags corresponding to the other audio information to be referenced. Based on each of the reference audio information and its corresponding set of associated topic tags, the audio topic correspondence is constructed.

9. An electronic device, characterized in that, include: processor; Memory used to store the processor's executable instructions; The processor is configured to execute the instructions to implement the topic tag recommendation method as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, When the instructions in the computer-readable storage medium are executed by the processor of the electronic device, the electronic device is able to perform the topic tag recommendation method as described in any one of claims 1-7.

Images