CN106611059A - Method and device for recommending multi-media files - Google Patents

Abstract

The disclosure provides a method and device for recommending multimedia files, which belong to the technical field of the Internet. The method includes: acquiring the first tag information of the first multimedia file corresponding to the terminal, the first tag information is extracted based on the first subtitle information of the first multimedia file; according to the first tag information and each The second label information of a second multimedia file, select the second multimedia file whose second label information matches the first label information from the multimedia file library, the second label information of each second multimedia file Extracted based on the second subtitle information of each second multimedia file; sending the identifier of the second multimedia file to the terminal. Since the extracted first tag information and the second tag information of each second multimedia file are more accurate, according to the first tag information and the second tag information of each second multimedia file, recommend the second most The media file can improve the accuracy of recommending the second multimedia file to the user.

Description

Method and device for recommending multimedia files

technical field

The present disclosure relates to the technical field of the Internet, in particular to a method and device for recommending multimedia files.

Background technique

With the advent of the information age, more and more video files are stored in the server; the user can use the terminal to obtain the video file that the user is interested in from the server. In order to improve user experience, the server may also recommend video files of interest to the user.

At present, when recommending video files for users, the server counts the click-through rate of each video file in the server, selects the video file with the highest click-through rate according to the click-through rate of each video file, and recommends the selected video file to the user.

Contents of the invention

In order to overcome the problems existing in related technologies, the present disclosure provides a method and device for recommending multimedia files, and the technical solution is as follows:

According to a first aspect of an embodiment of the present disclosure, a method for recommending a multimedia file is provided, the method comprising:

Acquiring first tag information of a first multimedia file corresponding to the terminal, where the first tag information is extracted based on first subtitle information of the first multimedia file;

According to the first tag information and the second tag information of each second multimedia file in the multimedia file library, select a second multi-media file whose second tag information matches the first tag information from the multimedia file library. For media files, the second tag information of each second multimedia file is extracted based on the second subtitle information of each second multimedia file;

Sending the identifier of the second multimedia file to the terminal.

In the embodiment of the present disclosure, since the first tag information is extracted based on the first subtitle information of the first multimedia file, the second tag information of each second multimedia file is extracted based on the first subtitle information of each second multimedia file. The second subtitle information of the file is extracted, so the extracted first tag information and the second tag information of each second multimedia file are more accurate, so that according to the first tag information and the first tag information of each second multimedia file Two tag information is used to recommend the second multimedia file to the user, which can improve the accuracy of recommending the second multimedia file to the user.

In a possible implementation manner, the acquiring the first label information of the first multimedia file corresponding to the terminal includes:

Obtain first subtitle information of the first multimedia file;

performing word segmentation on the first subtitle information to obtain a first keyword set;

Analyzing each keyword in the first keyword set to obtain the first tag information.

In the embodiment of the present disclosure, the first label information of the first multimedia file is extracted by performing semantic analysis on the first subtitle information of the first multimedia file, thereby improving the accuracy of the first label information.

In a possible implementation manner, the analyzing each keyword in the first keyword set to obtain the first tag information includes:

Acquiring the probability of each keyword in the first subtitle information, and obtaining the probability that each keyword belongs to each theme information in the theme information library, the theme information library is used to store a plurality of preset subject information;

According to the probability of each keyword in the first subtitle information and the probability that each keyword belongs to each topic information, determine the probability that the first multimedia file belongs to each topic information ;

According to the probability that the first multimedia file belongs to each topic information, select a preset number of topic information with the highest probability from each topic information;

Compose the selected preset number of topic information into the first tag information.

In the embodiment of the present disclosure, according to the probability of each keyword in the first subtitle information and the probability of each keyword belonging to each topic information, the probability that the first multimedia file belongs to each topic information is determined; according to the A multimedia file belongs to the probability of each topic information, and select a preset number of topic information with the highest probability from each topic information, thereby improving the accuracy of the first tag information.

In a possible implementation manner, the first multimedia subtitle is determined according to the probability of each keyword in the first subtitle information and the probability that each keyword belongs to each topic information Probability that the document belongs to each of the subject information, including:

Composing the probability of each keyword in the first subtitle information into a first probability matrix, and forming the probability of each keyword belonging to each topic information into a second probability matrix;

multiplying the inverse matrix of the second probability matrix by the first probability matrix to obtain a third probability matrix;

Obtaining the probability that the first multimedia file belongs to each topic information from the third probability matrix.

In the embodiment of the present disclosure, the probability of each keyword in the first subtitle information is formed into a first probability matrix, and the probability of each keyword belonging to each topic information is formed into a second probability matrix. According to the first probability matrix and The second probability matrix determines the probability that the first multimedia file belongs to each topic information, improves the accuracy of determining the probability that the first multimedia file belongs to each topic information, and then improves the accuracy of the first label information sex.

In a possible implementation manner, the obtaining the probability that each keyword belongs to each subject information in the subject information database includes:

For each subject information, obtain a preset keyword set corresponding to the subject information;

Determine that each keyword belongs to the Probability of Topic Information.

In a possible implementation manner, according to the probability of each keyword in the first subtitle information, the preset keyword set, and the number of keywords contained in the preset keyword set, Determining the probability that each keyword belongs to the subject information includes:

If each keyword is included in the preset keyword set, the ratio of the probability of each keyword in the first subtitle information to the number of keywords included in the preset keyword set As the probability that each keyword belongs to the subject information;

If each keyword is not included in the preset keyword set, it is determined that the probability that each keyword belongs to the subject information is zero.

In the embodiment of the present disclosure, the ratio of the probability of each keyword in the first subtitle information to the number of keywords contained in the preset keyword set is used as the probability of each keyword belonging to the topic information, since each The probability of each keyword in the first subtitle information determines the probability that each keyword belongs to the topic information, improves the accuracy of determining the probability that each keyword belongs to the topic information, and then improves the accuracy of the first label information. accuracy.

In a possible implementation manner, the performing word segmentation on the first subtitle information to obtain a first keyword set includes:

performing word segmentation on the first subtitle information, and forming each word segment included in the first subtitle information into a second keyword set;

The keywords of the preset type in the second keyword set are removed to obtain the first keyword set.

In the embodiment of the present disclosure, removing the keywords of the preset type in the second keyword set not only reduces the calculation amount, but also improves the accuracy of the first tag information.

In a possible implementation manner, the first tag information includes at least one of subject information to which the first multimedia file belongs, user information that produced the first multimedia file, and user information that was photographed .

In an embodiment of the present disclosure, the first tag information includes at least one of subject information to which the first multimedia file belongs, user information that makes the first multimedia file, and user information that is photographed. Therefore, in combination with at least one of the subject information to which the first multimedia file belongs, the user information for making the first multimedia file, and the photographed user information, recommending the second multimedia file to the user can improve the ability to recommend the second multimedia file to the user. Accuracy of multimedia files.

According to a second aspect of an embodiment of the present disclosure, there is provided an apparatus for recommending multimedia files, the apparatus comprising:

An acquisition module, configured to acquire first tag information of a first multimedia file corresponding to the terminal, where the first tag information is extracted based on first subtitle information of the first multimedia file;

A selection module, configured to select the second tag information and the first tag information from the multimedia file library according to the first tag information and the second tag information of each second multimedia file in the multimedia file library Matching second multimedia files, the second tag information of each second multimedia file is extracted based on the second subtitle information of each second multimedia file;

A sending module, configured to send the identifier of the second multimedia file to the terminal.

In a possible implementation manner, the acquisition module includes:

an acquiring unit, configured to acquire first subtitle information of the first multimedia file;

a word segmentation unit, configured to perform word segmentation on the first subtitle information to obtain a first set of keywords;

An analyzing unit, configured to analyze each keyword in the first keyword set to obtain the first tag information.

In a possible implementation manner, the analysis unit is further configured to obtain the probability of each keyword in the first subtitle information, and obtain each The probability of each subject information, the subject information library is used to store a plurality of preset subject information, according to the probability of each keyword in the first subtitle information and the probability that each keyword belongs to each subject information probability, determining the probability that the first multimedia file belongs to each subject information, and according to the probability that the first multimedia file belongs to each subject information, from each subject information A preset number of topic information with the highest probability is selected, and the selected preset number of topic information is combined into the first tag information.

In a possible implementation manner, the analysis unit is further configured to form a probability of each keyword in the first subtitle information into a first probability matrix, and form each keyword into each The probability of subject information constitutes a second probability matrix, and the inverse matrix of the second probability matrix is multiplied by the first probability matrix to obtain a third probability matrix, and the first probability matrix is obtained from the third probability matrix The probability that the multimedia file belongs to each subject information.

In a possible implementation manner, the analysis unit is further configured to obtain, for each subject information, a preset keyword set corresponding to the subject information, and according to each keyword in the first subtitle The probability in the information, the preset keyword set and the number of keywords included in the preset keyword set determine the probability that each keyword belongs to the subject information.

In a possible implementation manner, the analysis unit is further configured to calculate the probability of each keyword in the first subtitle information if the preset keyword set contains each keyword and the ratio of the number of keywords included in the preset keyword set as the probability that each keyword belongs to the topic information, if the preset keyword set does not contain each keyword, determine The probability that each keyword belongs to the subject information is zero.

In a possible implementation manner, the word segmentation unit is further configured to segment the first subtitle information, form each word segment included in the first subtitle information into a second keyword set, and combine the second The keywords of the preset type in the keyword set are removed to obtain the first keyword set.

In a possible implementation manner, the first tag information includes at least one of subject information to which the first multimedia file belongs, user information that produced the first multimedia file, and user information that was photographed .

According to a third aspect of an embodiment of the present disclosure, there is provided an apparatus for recommending multimedia files, the apparatus comprising:

processor;

memory for storing processor-executable instructions;

Wherein, the processor is configured to: acquire the first label information of the first multimedia file corresponding to the terminal, the first label information is extracted based on the first subtitle information of the first multimedia file;

According to the first tag information and the second tag information of each second multimedia file in the multimedia file library, select a second multi-media file whose second tag information matches the first tag information from the multimedia file library. For media files, the second tag information of each second multimedia file is extracted based on the second subtitle information of each second multimedia file;

sending the second multimedia file to the terminal.

The technical solutions provided by the embodiments of the present disclosure may include the following beneficial effects:

In the embodiment of the present disclosure, since the first tag information is extracted based on the first subtitle information of the first multimedia file, the second tag information of each second multimedia file is extracted based on the first subtitle information of each second multimedia file. The second subtitle information of the file is extracted, so the extracted first tag information and the second tag information of each second multimedia file are more accurate, so that according to the first tag information and the first tag information of each second multimedia file Two tag information is used to recommend the second multimedia file to the user, which can improve the accuracy of recommending the second multimedia file to the user.

It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present disclosure.

Description of drawings

The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the disclosure and together with the description serve to explain the principles of the disclosure.

Fig. 1 is a schematic diagram of an implementation environment according to an exemplary embodiment;

Fig. 2 is a flow chart showing a method for recommending multimedia files according to an exemplary embodiment;

Fig. 3 is a flow chart showing a method for recommending multimedia files according to an exemplary embodiment;

Fig. 4 is a block diagram of an apparatus for recommending multimedia files according to an exemplary embodiment;

Fig. 5 is a block diagram of an acquisition module according to an exemplary embodiment;

Fig. 6 is a block diagram of an apparatus for recommending multimedia files according to an exemplary embodiment.

detailed description

In order to make the purpose, technical solution and advantages of the present disclosure clearer, the implementation manners of the present disclosure will be further described in detail below in conjunction with the accompanying drawings.

Reference will now be made in detail to the exemplary embodiments, examples of which are illustrated in the accompanying drawings. When the following description refers to the accompanying drawings, the same numerals in different drawings refer to the same or similar elements unless otherwise indicated. The implementations described in the following exemplary examples do not represent all implementations consistent with the present disclosure. Rather, they are merely examples of apparatuses and methods consistent with aspects of the present disclosure as recited in the appended claims.

Fig. 1 is a schematic diagram of an implementation environment provided by an embodiment of the present disclosure. Referring to Fig. 1 , the implementation environment includes: a terminal 101 and a server 102, and the terminal 101 and the server 102 are connected through a communication network.

Wherein, the application associated with the server 102 is run in the terminal 101, and the application can be logged in based on the user ID, so as to interact with the server 102. The application may be a variety of applications such as a video application or an audio application, and the user identifier may be a user account, a phone number, etc., which is not limited in this embodiment of the present disclosure.

The terminal 101 may be a mobile phone terminal, a PAD (portable android device, tablet computer) terminal, or a computer terminal. The server 102 may be one server, or a server cluster composed of several servers, or a cloud computing server center, which is not limited in this embodiment of the present disclosure.

An application is installed on the terminal, and the application may be a video application or an audio application. When the terminal plays a multimedia file through the application, the server can recommend a second multimedia file for the user based on the first multimedia file currently played or historically played by the user. When recommending the second multimedia file, the server obtains the first The first label information of the multimedia file, the first label information is extracted based on the first subtitle information of the first multimedia file, and the first multimedia file can be a multimedia file, or can include multiple multimedia files file; obtain the second label information of each second multimedia file in the multimedia file library, and recommend the second multimedia file to the user according to the first label information and the second label information of each second multimedia file , the second multimedia file may be one multimedia file, or may include multiple multimedia files.

Since the first tag information is extracted based on the first subtitle information of the first multimedia file, the second tag information of each second multimedia file is extracted based on the second subtitle information of each second multimedia file Therefore, the extracted first tag information and the second tag information of each second multimedia file are more accurate, so according to the first tag information and the second tag information of each second multimedia file, recommend The second multimedia file can improve the accuracy of recommending the second multimedia file to the user.

Fig. 2 is a flow chart of a method for recommending multimedia files according to an exemplary embodiment. The method may be executed by a server, as shown in Fig. 2 , including the following steps.

In step S201, the first tag information of the first multimedia file corresponding to the terminal is acquired, and the first tag information is extracted based on the first subtitle information of the first multimedia file.

In step S202, according to the first tag information and the second tag information of each second multimedia file in the multimedia file library, select the second multimedia file whose second tag information matches the first tag information from the multimedia file library. body files, the second tag information of each second multimedia file is extracted based on the second subtitle information of each second multimedia file.

In step S203, the identifier of the second multimedia file is sent to the terminal.

In a possible implementation manner, obtaining the first tag information of the first multimedia file corresponding to the terminal includes:

Obtain first subtitle information of the first multimedia file;

performing word segmentation on the first subtitle information to obtain a first keyword set;

Each keyword in the first keyword set is analyzed to obtain first tag information.

In a possible implementation manner, each keyword in the first keyword set is analyzed to obtain the first tag information, including:

Acquiring the probability of each keyword in the first subtitle information, and obtaining the probability that each keyword belongs to each theme information in the theme information library, the theme information library is used to store a plurality of preset theme information;

According to the probability of each keyword in the first subtitle information and the probability of each keyword belonging to each topic information, determine the probability that the first multimedia file belongs to each topic information;

According to the probability that the first multimedia file belongs to each topic information, select a preset number of topic information with the highest probability from each topic information;

The selected preset number of topic information is combined into the first tag information.

In a possible implementation manner, according to the probability of each keyword in the first subtitle information and the probability of each keyword belonging to each topic information, the probability that the first multimedia file belongs to each topic information is determined, including :

Composing the probability of each keyword in the first subtitle information into a first probability matrix, and forming the probability of each keyword belonging to each topic information into a second probability matrix;

multiplying the inverse matrix of the second probability matrix by the first probability matrix to obtain a third probability matrix;

The probability that the first multimedia file belongs to each subject information is obtained from the third probability matrix.

In a possible implementation manner, obtaining the probability that each keyword belongs to each subject information in the subject information database includes:

For each subject information, obtain a preset keyword set corresponding to the subject information;

According to the probability of each keyword in the first subtitle information, the preset keyword set and the number of keywords contained in the preset keyword set, the probability of each keyword belonging to the topic information is determined.

In a possible implementation manner, according to the probability of each keyword in the first subtitle information, the preset keyword set, and the number of keywords contained in the preset keyword set, it is determined that each keyword belongs to the topic information Probability, including:

If each keyword is included in the preset keyword set, the ratio of the probability of each keyword in the first subtitle information and the number of keywords included in the preset keyword set is used as the value of each keyword belonging to the topic information probability;

If each keyword is not included in the preset keyword set, it is determined that the probability that each keyword belongs to the topic information is zero.

In a possible implementation manner, word segmentation is performed on the first subtitle information to obtain a first keyword set, including:

performing word segmentation on the first subtitle information, and forming a second keyword set from each word segment included in the first subtitle information;

The keywords of the preset type in the second keyword set are removed to obtain the first keyword set.

In a possible implementation manner, the first tag information includes at least one of subject information to which the first multimedia file belongs, user information that produces the first multimedia file, and user information that is photographed.

All the above optional technical solutions may be combined in any way to form optional embodiments of the present disclosure, which will not be repeated here.

Fig. 3 is a flow chart of a method for setting tag information according to an exemplary embodiment. The method may be executed by a server, as shown in Fig. 3 , including the following steps.

In step S301, the server acquires first tag information of a first multimedia file corresponding to the terminal, where the first tag information is extracted based on first subtitle information of the first multimedia file.

When the terminal logs in to the server or obtains a multimedia file from the server, the server obtains the first multimedia file corresponding to the terminal from the historical playback records of the terminal according to the terminal identifier of the terminal; the first multimedia file may be a multimedia file, Multiple multimedia files can also be included. In order to improve the accuracy, the server may acquire a first preset number of multimedia files whose playback time is closest to the current time from the historical playback records of the terminal as the first multimedia files. Wherein, the first tag information includes at least one of subject information to which the first multimedia file belongs, information of a user who made the first multimedia file, and information of a user who was photographed.

The first multimedia file can be a video file or an audio file; when the first multimedia file is a video file, the user information making the first multimedia file can be the director and/or producer of the first multimedia file Filmmaker, the photographed user information may be an actor or the like who photographed the first multimedia file. When the first multimedia file is an audio file, the user information for making the first multimedia file may be a singer or the like. The first preset number can be 1 or 5.

When the first tag information includes the theme information to which the first multimedia file belongs, this step can be implemented through the following steps 3011-3013, including:

3011: The server acquires first subtitle information of the first multimedia file.

The corresponding relationship between the identification of the multimedia file and the subtitle file is stored in the server; correspondingly, this step can be:

According to the identification of the first multimedia file, the server obtains the subtitle file of the first multimedia file from the correspondence between the identification of the multimedia file and the subtitle file, and obtains the first multi-media file from the subtitle file of the first multimedia file. Caption information of the media file.

The first multimedia file may be a video file or an audio file. The identifier of the first multimedia file may be the name or serial number of the first multimedia file. In the embodiment of the present disclosure, the identification of the first multimedia file is not specifically limited.

3012: The server performs word segmentation on the first subtitle information to obtain a first keyword set.

In this step, the server may perform word segmentation on the first subtitle information, and form each word segment included in the first subtitle information into a first keyword set; the server may also obtain the first keyword through the following steps (1)-(2): collection of words, including:

(1): The server performs word segmentation on the first subtitle information, and forms each word segment included in the first subtitle information into a second keyword set.

The server performs word segmentation on the first subtitle information by using a preset word segmentation tool, obtains each word segment included in the first subtitle information, and forms each word segment included in the first subtitle information into a second keyword set.

For example, if the first subtitle information is "The person who knows you best is not your friend, but your enemy." Then use the preset word segmentation tool to segment the first subtitle information to obtain each word segment included in the first subtitle information For "most", "understand", "your", "person", "not", "your", "friend", "but", "your", "enemy", then the second keyword The set is {"most", "know", "your", "person", "not", "your", "friend", "rather", "your", "enemy"}.

The preset word segmentation tool can be StandardAnalyzer (standardized word segmentation tool), ChineseAnalyzer (Chinese word segmentation tool), CJKAnalyzer (CJK word segmentation tool) or IKAnalyzer (IK word segmentation tool). In the embodiment of the present disclosure, there is no specific limitation on the preset word segmentation tool.

Because keywords such as "de", "le", "what", "bar", "ah", and "most" do not play a key role in label information; therefore, in order to reduce the amount of calculation and improve the efficiency of setting label information Accuracy, in this step, the server can also pass the following steps (2) keywords such as " of ", " ", " what ", " bar ", " ah ", " most " from the second key removed from the word set.

(2): The server removes keywords of a preset type in the second keyword set to obtain the first keyword set.

The keywords of the preset type may be modal particles or auxiliary words. Then this step can be: the server marks the part of speech of each keyword in the second keyword set, and searches for a preset type of keyword from the second keyword set according to each keyword in the second keyword set, The keywords of the preset type are removed from the second keyword set to obtain the first keyword set.

For example, the server collects the second key words {"most", "understanding", "your", "person", "not", "your", "friend", "but", "your", "Most", "your", "person", "not" and "but" in "enemy"} are removed, and the first keyword set is obtained as {"knowledge", "friend", "enemy"}.

In a possible implementation, since the first keyword set may contain synonyms or near synonyms, for example, "capital" and "Beijing" are synonyms; therefore, in order to reduce the amount of computation, after the server obtains the first keyword set, it also Multiple synonyms or synonyms in the first keyword set may be combined into one keyword. Since the number of keywords in the first keyword set is reduced, the computation load of the server is reduced, thereby improving the efficiency of obtaining the first tag information.

3013: The server analyzes each keyword in the first keyword set to obtain first tag information.

This step can be realized through the following first or second way; for the first kind of implementation, this step can be realized through the following steps (1)-(3), including:

(1): The server obtains the probability of each keyword in the first subtitle information.

The server acquires the number of occurrences of each keyword in the first subtitle information, calculates the sum of the number of occurrences of each keyword, and determines the ratio of the number of occurrences of each keyword to the sum of the number of occurrences of each keyword as Probability in the first subtitle information.

It should be noted that, if the server combines multiple synonyms or synonyms in the first keyword set into one keyword, when the server obtains the probability of the keyword in the first subtitle information, the server obtains the synonyms or synonyms of the keyword The sum of the number of times that synonyms appear in the first subtitle information, calculate the sum of the number of occurrences of each keyword, and compare the sum of the number of times that synonyms of the keyword or synonyms appear in the first subtitle information with each keyword The ratio of the sum of the occurrence times of is determined as the probability of the keyword in the first subtitle information.

(2): The server obtains the probability that each keyword belongs to each topic information in the topic information database, and the topic information database is used to store a plurality of preset topic information.

The preset theme information may be "friendship", "emotion" and "love". This step can be realized through the following steps (2-1)-(2-2), including:

(2-1): For each subject information, the server acquires a preset keyword set corresponding to the subject information.

For each topic information in the topic information database, the server stores the corresponding relationship between each topic information and the preset keyword set; correspondingly, this step can be:

According to the theme information, the server obtains the preset keyword set corresponding to the theme information from the correspondence relationship between the theme information and the preset keyword set. Wherein, the preset keyword set includes a plurality of preset keywords belonging to the subject information.

For example, the preset keyword set corresponding to the subject information "friendship" acquired by the server is {friend, friendship, loyalty}.

(2-2): The server determines that each keyword belongs to the topic information according to the probability of each keyword in the subtitle information, the preset keyword set and the number of keywords contained in the preset keyword set probability.

For each keyword, the server detects whether the keyword is included in the preset keyword set; if the keyword is included in the preset keyword set, the probability of the keyword in the first subtitle information and the preset The ratio of the number of keywords contained in the keyword set is used as the probability that the keyword belongs to the topic.

If the keyword is not included in the preset keyword set, it is determined that the probability that the keyword belongs to the subject information is zero.

(3): The server determines the probability that the first multimedia file belongs to each topic information according to the probability of each keyword in the first subtitle information and the probability that each keyword belongs to each topic information.

This step can be realized through the following steps (3-1)-(3-3), including:

(3-1): The server forms the probability of each keyword in the first subtitle information into a first probability matrix, and forms the probability of each keyword belonging to each topic information into a second probability matrix.

The server takes the probability of each keyword in the first subtitle information as a row of data to form the first probability matrix; for each keyword, the server takes the probability of the keyword belonging to each topic information as a row of data to form the second probability matrix.

The first probability matrix is a matrix of n×1, and the second probability matrix is a matrix of n×m; wherein, n is the number of keywords included in the first keyword set, and m is a preset included in the subject information database The number of subject information for .

For example, each keyword is A, B, and C respectively; the probabilities of A, _{B, and C in the first subtitle information are PA, P B , and} PC respectively, and each topic information included in the topic information base is Topic 1, Topic 2, Topic 3, and Topic 4; the probabilities of keyword A belonging to each topic information are A1, A2, A3, and A4, and the probabilities of keyword B belonging to each topic information are B1, B2, B3, and B4, the probability that keyword C belongs to each topic information is C1, C2, C3 and C4 respectively.

Then the first probability matrix is The second probability matrix is

(3-2): The server multiplies the inverse matrix of the second probability matrix by the first probability matrix to obtain a third probability matrix.

The server determines an inverse matrix of the second probability matrix according to the second probability matrix; multiplies the inverse matrix of the second probability matrix by the first probability matrix to obtain a third probability matrix. Wherein, the third probability matrix is an m×1 matrix, and each row of data in the third probability matrix is the probability that the first multimedia file belongs to each subject information.

For example, the server obtains the third probability matrix as

(3-3): The server acquires the probability that the first multimedia file belongs to each topic information from the third probability matrix.

Each row of data in the third probability matrix is the probability that the first multimedia file belongs to each subject information. The server can obtain the probability that the first multimedia file belongs to each topic information from the third probability matrix.

For example, the third matrix is Then P ₁ is the probability that the first multimedia file belongs to topic information 1, P ₂ is the probability that this first multimedia file belongs to topic information 2, and P ₃ is the probability that this first multimedia file belongs to topic information 3. Probability, P ₄ is the probability that the first multimedia file belongs to topic information 4.

(4): The server selects a preset number of topic information with the highest probability from each topic information according to the probability that the first multimedia file belongs to each topic information.

In order to facilitate the distinction, the preset number here is called the second preset number, and the second preset number can be set and changed according to needs. In the embodiment of the present disclosure, the second preset number is not specifically limited; for example , the second preset number may be 1 or 2 and so on.

(5): The server forms the selected second preset number of theme information into the first tag information of the first multimedia file.

For example, if the selected theme information is comedy and romance, the first tag information of the first multimedia file is comedy and romance.

For the second implementation, this step can be:

The server obtains the probability of each keyword in the first subtitle information, and selects a third preset number of keywords with the highest probability from each keyword according to the probability of each keyword in the first subtitle information, and obtains the selected The subject information to which the selected keyword belongs, and the subject information to which the selected keyword belongs constitute the label information of the first multimedia file.

When the first tag information includes the information of the user who made the first multimedia file and/or the information of the user who was photographed, the first subtitle information includes the information of the user who made the first multimedia file and/or the information of the user who was photographed , the server may directly extract the information of the user who made the first multimedia file and/or the information of the user who was photographed from the first subtitle information.

The server extracts the information of the user who made the first multimedia file and/or the information of the user who was photographed from the first subtitle information by means of keyword matching.

For example, if the information about the user who made the first multimedia file is the director of the first multimedia file, the server extracts the name after the keyword "director" from the first subtitle information.

It should be noted that the server can obtain the label information of each multimedia file in the server in advance, and establish the corresponding relationship between the identification of each multimedia file and the label information; correspondingly, this step can be:

According to the identifier of the first multimedia file, the server acquires the first tag information of the first multimedia file from the correspondence between the identifier of the multimedia file and the tag information.

Wherein, the step for the server to acquire the label information of each multimedia file in the server in advance is the same as the process for the server to acquire the first label information of the first multimedia file, and will not be repeated here.

In step S302, the server obtains the second label information of each second multimedia file in the multimedia file library, and the second label information of each second multimedia file is based on the first label information of each second multimedia file. Second subtitle information extraction.

For each second multimedia file, the server may obtain the second tag information of the second multimedia file according to the above process of obtaining the first tag information of the first multimedia file. The server can also obtain the label information of each multimedia file in the server in advance, and establish the corresponding relationship between the identification of each multimedia file and the label information; correspondingly, this step can be:

According to the identifier of each second multimedia file, the server obtains the second tag information of each second multimedia file from the correspondence between the identifier of the multimedia file and the tag information.

Wherein, the second tag information includes at least one of subject information to which the second multimedia file belongs, user information that makes the second multimedia file, and user information that is photographed.

In step S303, according to the first tag information and the second tag information of each second multimedia file in the multimedia file library, the server selects the second multi-media file whose second tag information matches the first tag information from the multimedia file library. media files.

The server calculates the matching degree between the first tag information and the second tag information of each second multimedia file in the multimedia file library, according to the first tag information and the first tag information of each second multimedia file in the multimedia file library. For the matching degree between the two tag information, select the second multimedia file corresponding to the second tag information with the highest matching degree or the matching degree exceeding a preset value from the multimedia files.

Because the first tag information includes at least one of subject information to which the first multimedia file belongs, user information that makes the first multimedia file, and user information that is photographed. Therefore, in combination with at least one of the subject information to which the first multimedia file belongs, the user information for making the first multimedia file, and the photographed user information, recommending the second multimedia file to the user can improve the ability to recommend the second multimedia file to the user. Accuracy of multimedia files.

For example, the first tag information includes topic information and user information for making the first multimedia file, and the first tag information is love and romance, Zhang San. Comprise 3 second multimedia files in the multimedia file storehouse, be respectively the second multimedia file 1, the second multimedia file 2 and the second multimedia file 3, the second multimedia file 1 of the second The tag information is: Horror and Thriller, Lee Si. The second label information of the second multimedia file 2 is: love and romance, Zhang San. The second label information of the second multimedia file 3 is: costume and comedy, Zhao Wu. Then the server selects the second multimedia file 2 from the multimedia file library.

In step S304, the server sends the identifier of the second multimedia file to the terminal.

The terminal receives the identifier of the second multimedia file sent by the server, and displays the identifier of the second multimedia file; when the user plays the second multimedia file, the terminal pulls the second multimedia file from the server according to the second multimedia identifier. Two multimedia files. Wherein, the identifier of the second multimedia file may be a play link of the second multimedia file.

Since the first tag information is extracted based on the first subtitle information of the first multimedia file, the second tag information of each second multimedia file is extracted based on the second subtitle information of each second multimedia file Therefore, the extracted first tag information and the second tag information of each second multimedia file are more accurate, so according to the first tag information and the second tag information of each second multimedia file, recommend The second multimedia file can improve the accuracy of recommending the second multimedia file to the user.

Fig. 4 is a block diagram of an apparatus for recommending multimedia files according to an exemplary embodiment. Referring to FIG. 4 , the device includes an acquisition module 401 , a selection module 402 and a sending module 403 .

The obtaining module 401 is configured to obtain the first tag information of the first multimedia file corresponding to the terminal, the first tag information is extracted based on the first subtitle information of the first multimedia file;

The selection module 402 is configured to select the second tag information and the first tag information from the multimedia file library according to the first tag information and the second tag information of each second multimedia file in the multimedia file library. The second multimedia file whose label information matches, the second label information of each second multimedia file is extracted based on the second subtitle information of each second multimedia file;

The sending module 403 is configured to send the second multimedia file to the terminal.

In a possible implementation manner, referring to FIG. 5, the obtaining module 401 includes:

The obtaining unit 4011 is configured to obtain the first subtitle information of the first multimedia file;

The word segmentation unit 4012 is configured to perform word segmentation on the first subtitle information to obtain a first keyword set;

The analysis unit 4013 is configured to analyze each keyword in the first keyword set to obtain the first tag information.

In a possible implementation manner, the analysis unit 4013 is further configured to obtain the probability of each keyword in the first subtitle information, and obtain the probability that each keyword belongs to the subject information database The probability of each topic information, the topic information base is configured to store a plurality of preset topic information, according to the probability of each keyword in the first subtitle information and the probability that each keyword belongs to The probability of each topic information, determine the probability that the first multimedia file belongs to the each topic information, according to the probability that the first multimedia file belongs to the each topic information, from each A preset number of topic information with the highest probability is selected from the topic information, and the selected preset number of topic information is combined into the first tag information.

In a possible implementation manner, the analysis unit 4013 is further configured to form the probability of each keyword in the first subtitle information into a first probability matrix, and combine the probability of each keyword The probability belonging to each topic information forms a second probability matrix, multiplying the inverse matrix of the second probability matrix by the first probability matrix to obtain a third probability matrix, and obtaining the The probability that the first multimedia file belongs to each subject information.

In a possible implementation manner, the analyzing unit 4013 is further configured to, for each subject information, obtain a preset keyword set corresponding to the subject information, and according to each keyword in the A probability in the subtitle information, the preset keyword set and the number of keywords included in the preset keyword set determine the probability that each keyword belongs to the subject information.

In a possible implementation manner, the analysis unit 4013 is further configured to include each keyword in the first subtitle information if the preset keyword set contains each keyword The ratio of the probability of the keyword to the number of keywords contained in the preset keyword set is used as the probability that each keyword belongs to the topic information, if the preset keyword set does not contain the each keyword , determining that the probability that each keyword belongs to the subject information is zero.

In a possible implementation manner, the word segmentation unit 4012 is further configured to perform word segmentation on the first subtitle information, form each word segment included in the first subtitle information into a second keyword set, and combine the The keywords of the preset type in the second keyword set are removed to obtain the first keyword set.

In a possible implementation manner, the first tag information includes at least one of subject information to which the first multimedia file belongs, user information that produced the first multimedia file, and user information that was photographed .

Since the first tag information is extracted based on the first subtitle information of the first multimedia file, the second tag information of each second multimedia file is extracted based on the second subtitle information of each second multimedia file Therefore, the extracted first tag information and the second tag information of each second multimedia file are more accurate, so according to the first tag information and the second tag information of each second multimedia file, recommend The second multimedia file can improve the accuracy of recommending the second multimedia file to the user.

All the above optional technical solutions may be combined in any way to form optional embodiments of the present disclosure, which will not be repeated here.

It should be noted that: when the device for recommending multimedia files provided by the above-mentioned embodiments recommends multimedia files, the division of the above-mentioned functional modules is used as an example for illustration. In practical applications, the above-mentioned functions can be assigned to different functional modules according to needs To complete means to divide the internal structure of the device into different functional modules to complete all or part of the functions described above. In addition, the device for recommending multimedia files provided by the above embodiment is based on the same idea as the embodiment of the method for recommending multimedia files. For the specific implementation process, refer to the method embodiment for details, and will not be repeated here.

Fig. 6 is a block diagram of an apparatus 600 for recommending multimedia files according to an exemplary embodiment. For example, the apparatus 600 may be provided as a server. Referring to FIG. 6 , apparatus 600 includes processing component 622 , which further includes one or more processors, and a memory resource represented by memory 632 for storing instructions executable by processing component 622 , such as application programs. The application program stored in memory 632 may include one or more modules each corresponding to a set of instructions. In addition, the processing component 622 is configured to execute instructions to perform the above-mentioned method for recommending multimedia files.

Device 600 may also include a power component 626 configured to perform power management of device 600 , a wired or wireless network interface 650 configured to connect device 600 to a network, and an input-output (I/O) interface 658 . The apparatus 600 can operate based on an operating system stored in the memory 632, such as Windows Server ^™ , Mac OS X ^™ , Unix ^™ , Linux ^™ , FreeBSD ^™ or the like.

Other embodiments of the present disclosure will be readily apparent to those skilled in the art from consideration of the specification and practice of the invention disclosed herein. This application is intended to cover any modification, use or adaptation of the present disclosure, and these modifications, uses or adaptations follow the general principles of the present disclosure and include common knowledge or conventional technical means in the technical field not disclosed in the present disclosure . The specification and examples are to be considered exemplary only, with a true scope and spirit of the disclosure being indicated by the following claims.

It should be understood that the present disclosure is not limited to the precise constructions which have been described above and shown in the drawings, and various modifications and changes may be made without departing from the scope thereof. The scope of the present disclosure is limited only by the appended claims.

Claims (17)

1. it is a kind of recommend multimedia file method, it is characterised in that methods described includes：

The first label information of corresponding first multimedia file of terminal is obtained, first label information is based on described first What the first caption information of multimedia file was extracted；

According to the second label information of each the second multimedia file in first label information and multimedia file storehouse, from institute The second multimedia file for selecting the second label information to match with first label information in stating multimedia file storehouse, it is described every Second label information of individual second multimedia file is that the second caption information based on each second multimedia file is extracted 's；

The mark of second multimedia file is sent to the terminal.

2. method according to claim 1, it is characterised in that the of corresponding first multimedia file of the acquisition terminal One label information, including：

Obtain the first caption information of first multimedia file；

Participle is carried out to first caption information, the first keyword set is obtained；

Each key word in first keyword set is analyzed, first label information is obtained.

3. method according to claim 2, it is characterised in that described crucial to each in first keyword set Word is analyzed, and obtains first label information, including：

Probability of each key word in first caption information described in obtaining, and, obtain described each key word and belong to The probability of each subject information in subject information storehouse, the subject information storehouse are used to store multiple default subject informations；

Probability and described each key word according to described each key word in first caption information belongs to each theme The probability of information, determines that first multimedia file belongs to the probability of each subject information；

The probability of each subject information according to first multimedia file belongs to, selects from described each subject information The preset number subject information of maximum probability；

The preset number subject information for selecting is constituted into first label information.

4. method according to claim 3, it is characterised in that each key word described in the basis is in first captions Probability and described each key word in information belongs to the probability of each subject information, determines that first multimedia file belongs to The probability of each subject information, including：

Probability by described each key word in first caption information constitutes the first probability matrix, and, will be described every Individual key word belongs to the probability of each subject information and constitutes the second probability matrix；

The inverse matrix of second probability matrix is multiplied with first probability matrix, the 3rd probability matrix is obtained；

The probability that first multimedia file belongs to each subject information is obtained from the 3rd probability matrix.

5. method according to claim 3, it is characterised in that each key word belongs to subject information storehouse described in the acquisition In each subject information probability, including：

For described each subject information, the corresponding preset critical set of words of the subject information is obtained；

Probability, the preset critical set of words according to described each key word in first caption information and described default The number of the key word that keyword set is included, it is determined that described each key word belongs to the probability of the subject information.

6. method according to claim 5, it is characterised in that each key word described in the basis is in first captions The number of the key word that the probability, the preset critical set of words and the preset critical set of words in information is included, determines institute The probability that each key word belongs to the subject information is stated, including：

If described each key word is included in the preset critical set of words, by described each key word in first captions The ratio of the number of the key word that the probability and the preset critical set of words in information is included belongs to as described each key word In the probability of the subject information；

If not including described each key word in the preset critical set of words, it is determined that described each key word belongs to the master The probability of topic information is zero.

7. method according to claim 2, it is characterised in that described to carry out participle to first caption information, obtains First keyword set, including：

Participle is carried out to first caption information, each participle that first caption information is included constitutes the second key word Set；

The key word of preset kind in second keyword set is removed, first set of keywords is obtained.

8. method according to claim 1, it is characterised in that first label information includes the first multimedia text Belonging to part in subject information, the user profile for making first multimedia file and the user profile being taken at least one It is individual.

9. it is a kind of recommend multimedia file device, it is characterised in that described device includes：

Acquisition module, for obtaining the first label information of corresponding first multimedia file of terminal, first label information It is the first caption information extraction based on first multimedia file；

Selecting module, for second according to each the second multimedia file in first label information and multimedia file storehouse Label information, from the multimedia file storehouse, select the second label information to match with first label information matchmaker more than second Body file, the second label information of each second multimedia file are based on each second multimedia file second What caption information was extracted；

Sending module, for the mark of second multimedia file is sent to the terminal.

10. device according to claim 9, it is characterised in that the acquisition module, including：

Acquiring unit, for obtaining the first caption information of first multimedia file；

Participle unit, for carrying out participle to first caption information, obtains the first keyword set；

Analytic unit, for being analyzed to each key word in first keyword set, obtains first label Information.

11. devices according to claim 10, it is characterised in that

The analytic unit, is additionally operable to obtain the probability of each key word in first caption information, and, obtain Described each key word belongs to the probability of each subject information in subject information storehouse, and the subject information storehouse is used to store multiple Default subject information, the probability and described each key word according to described each key word in first caption information belong to In the probability of each subject information, determine that first multimedia file belongs to the probability of each subject information, according to institute The probability that the first multimedia file belongs to each subject information is stated, maximum from select probability in described each subject information The preset number subject information for selecting is constituted first label information by preset number subject information.

12. devices according to claim 11, it is characterised in that

The analytic unit, is additionally operable to the probability by described each key word in first caption information and constitutes the first probability Matrix, and, the probability that described each key word belongs to each subject information is constituted into the second probability matrix, it is general by described second The inverse matrix of rate matrix is multiplied with first probability matrix, obtains the 3rd probability matrix, is obtained from the 3rd probability matrix Take the probability that first multimedia file belongs to each subject information.

13. devices according to claim 11, it is characterised in that

The analytic unit, is additionally operable to for described each subject information, obtains the corresponding predetermined keyword of the subject information Set, probability, the preset critical set of words according to described each key word in first caption information and described pre- If the number of the key word that keyword set is included, it is determined that described each key word belongs to the probability of the subject information.

14. devices according to claim 13, it is characterised in that

The analytic unit, if in being additionally operable to the preset critical set of words include described each key word, by it is described each The ratio of the number of the key word that probability of the key word in first caption information and the preset critical set of words are included Belong to the probability of the subject information as described each key word, if not comprising described every in the preset critical set of words Individual key word, it is determined that the probability that described each key word belongs to the subject information is zero.

15. devices according to claim 10, it is characterised in that

The participle unit, is additionally operable to carry out participle to first caption information, by first caption information include it is every Individual participle constitutes the second keyword set, and the key word of preset kind in second keyword set is removed, and obtains described First set of keywords.

16. devices according to claim 9, it is characterised in that first label information includes first multimedia In subject information, the user profile for making first multimedia file and the user profile being taken belonging to file at least One.

17. a kind of devices for recommending multimedia file, it is characterised in that include：

Processor；

For storing the memorizer of processor executable；

Wherein, the processor is configured to：Obtain the first label information of corresponding first multimedia file of terminal, described the One label information is that the first caption information based on first multimedia file is extracted；

According to the second label information of each the second multimedia file in first label information and multimedia file storehouse, from institute The second multimedia file for selecting the second label information to match with first label information in stating multimedia file storehouse, it is described every Second label information of individual second multimedia file is that the second caption information based on each second multimedia file is extracted 's；

Second multimedia file is sent to the terminal.