Video bullet screen generation method and device, electronic equipment and storage medium

Abstract

The invention provides a video bullet screen generation method and device, electronic equipment and a storage medium. The method comprises the following steps: acquiring a video plot text of a video file in a specified video playing time period; obtaining semantic feature information of video keywords in the video plot text; performing coding processing on the video plot text and the semantic feature information to obtain a text coding vector of the video plot text and a semantic feature vector corresponding to the semantic feature information; according to the semantic feature vector and the text coding vector, determining a target video keyword of which the similarity index with the video plot text is greater than a threshold value in the video keywords; generating a video bullet screen text according to the target video keyword; and associatively storing the video file identifier of the video file, the specified video playing time period and the video bullet screen text. According to the method and the device, the bullet screen word bank related to the video plot can be generated, so that the user can express the watching feeling more accurately when inputting the bullet screen, and the bullet screen interaction experience is improved.

Description

Technical Field

[0001] The present application relates to the technical field of video barrage processing, and in particular to a method, device, electronic device and storage medium for generating video barrage. Background Art

[0002] With the rapid development of Internet technology, barrage as a unique form of interaction has become increasingly popular on video sharing platforms and social media.

[0003] During the playback of video files, barrage interaction on video sharing platforms has become an important way for users to watch videos. However, the existing barrage input methods are mostly preset templates or user-defined input, which lacks understanding and analysis of video content and cannot provide a personalized barrage experience closely related to the video plot. Summary of the invention

[0004] The purpose of the embodiments of the present application is to provide a method, device, electronic device and storage medium for generating video bullet screens, so as to generate a bullet screen vocabulary library related to the video plot, so that users can express their viewing experience more accurately when inputting bullet screens, and improve the experience of bullet screen interaction. The specific technical solution is as follows:

[0005] In a first aspect of the implementation of the present application, a method for generating a video bullet screen is first provided, comprising:

[0006] Get the video plot text of the video file within the specified video playback period;

[0007] Obtaining semantic feature information of video keywords in the video plot text;

[0008] Encoding the video plot text and the semantic feature information respectively to obtain a text encoding vector of the video plot text and a semantic feature vector corresponding to the semantic feature information;

[0009] Determine, according to the semantic feature vector and the text encoding vector, a target video keyword whose similarity index with the video plot text is greater than a threshold value among the video keywords;

[0010] Generate video barrage text for the video file during the specified video playback period according to the target video keyword;

[0011] The video file identifier of the video file, the designated video play time period and the video barrage text are stored in association.

[0012] In a second aspect of the implementation of the present application, a video barrage generating device is provided, including:

[0013] A video plot text acquisition module is used to acquire the video plot text of a video file within a specified video playback period;

[0014] A semantic feature information acquisition module, used to acquire semantic feature information of video keywords in the video plot text;

[0015] A text semantic vector acquisition module, used to encode the video plot text and the semantic feature information respectively, to obtain a text encoding vector of the video plot text and a semantic feature vector corresponding to the semantic feature information;

[0016] A target keyword determination module, used to determine, according to the semantic feature vector and the text encoding vector, a target video keyword whose similarity index with the video plot text is greater than a threshold value among the video keywords;

[0017] A video barrage text generation module, used to generate a video barrage text of the video file in the specified video playback period according to the target video keyword;

[0018] The video barrage text storage module is used to associate and store the video file identifier of the video file, the specified video playback time period and the video barrage text.

[0019] In another aspect of the present application, there is also provided an electronic device, including a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other via the communication bus;

[0020] Memory, used to store computer programs;

[0021] The processor is used to implement any of the above-mentioned video barrage generation methods when executing the program stored in the memory.

[0022] In another aspect of the implementation of the present application, a computer-readable storage medium is provided, in which instructions are stored. When the computer-readable storage medium is run on a computer, the computer executes any of the above-mentioned methods for generating video barrage.

[0023] In another aspect of the implementation of the present application, a computer program product comprising instructions is also provided, which, when executed on a computer, enables the computer to execute any of the above-described methods for generating video barrages.

[0024] The solution provided by the embodiment of the present application is to obtain the video plot text of the video file within the specified video playback period. Obtain the semantic feature information of the video keywords in the video plot text. Encode the video plot text and the semantic feature information respectively to obtain the text encoding vector of the video plot text and the semantic feature vector corresponding to the semantic feature information. According to the semantic feature vector and the text encoding vector, determine the target video keyword in the video keyword whose similarity index with the video plot text is greater than the threshold. According to the target video keyword, generate the video barrage text of the video file in the specified video playback period. The video file identifier, the specified video playback period and the video barrage text of the video file are stored in association. The embodiment of the present application analyzes and understands the video plot text of the video file within the specified video playback period to generate a specific video barrage vocabulary related to the video plot, so that users can express their viewing experience more accurately when inputting barrage, and improve the experience of barrage interaction. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art are briefly introduced below.

[0026] Figure 1 A flowchart of a method for generating video bullet screen provided in an embodiment of the present application;

[0027] Figure 2 A flowchart of a method for obtaining video plot text provided in an embodiment of the present application;

[0028] Figure 3 A flowchart of a method for acquiring semantic feature information provided in an embodiment of the present application;

[0029] Figure 4 A flowchart of a method for selecting keywords for a target video provided in an embodiment of the present application;

[0030] Figure 5 A flowchart of a method for obtaining video barrage text provided in an embodiment of the present application;

[0031] Figure 6 A flowchart of another method for obtaining video barrage text provided in an embodiment of the present application;

[0032] Figure 7 A flowchart of a method for recommending target video barrage text provided in an embodiment of the present application;

[0033] Figure 8 A schematic diagram of a video barrage generation and recommendation process provided in an embodiment of the present application;

[0034] Fig. 9 A schematic diagram of the structure of a video barrage generation device provided in an embodiment of the present application;

[0035] Fig.10 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0036] The technical solutions in the embodiments of the present application will be described below in conjunction with the drawings in the embodiments of the present application.

[0037] Figure 1 A flowchart of a method for generating video bullet screen provided in an embodiment of the present application is shown in FIG. Figure 1 As shown, the video barrage generation method may include: step 101, step 102, step 103, step 104, step 105 and step 106.

[0038] Step 101: Obtain the video plot text of the video file within a specified video playback time period.

[0039] The embodiments of the present application can be applied to a scenario in which a video plot text is analyzed to generate video comments related to the video plot.

[0040] In this example, the video file may be a variety show video file, or a film and television video file, etc. The type of the video file may be determined according to actual conditions, and this embodiment does not impose any limitation on this.

[0041] In a specific implementation, the playing time of a video file can be divided into several time periods for plot analysis, and the designated video playing time period is the video playing time period of the video file for generating video barrage. In this example, the designated video playing time period can be the time period from the 30th minute to the 35th minute of the playing time of the video file. For example, taking a film and television video as an example, the designated video playing time period can be the 1-minute playing time period at the beginning of the film and television video, or the 1-minute playing time period at the end, etc.

[0042] The video plot text may be used to indicate text related to the video plot in a specified video playback period of a video file. In this embodiment, the video plot text may include at least one of script text, video character dialogue text, video screen narration text, and the like.

[0043] When generating video bullet comments related to the video plot in the specified video playback period of a video file, the video plot text of the video file in the specified video playback period can be obtained. In a specific implementation, when performing plot analysis on the specified video playback period of a video file, the initial plot text of the video file in the specified video playback period can be collected, and the initial plot text can be preprocessed, such as word segmentation, part-of-speech tagging, syntactic analysis, etc., to obtain the video plot text of the video file in the specified video playback period. For this implementation process, it will be combined in the following implementation Figure 2 A detailed description is given below and this embodiment will not be described in detail here.

[0044] After the video plot text of the video file within the specified video playback time period is obtained, step 102 is executed.

[0045] Step 102: Obtain semantic feature information of video keywords in the video plot text.

[0046] After obtaining the video plot text of the video file within the specified video playback period, the video keywords in the video plot text can be extracted, and the extracted video keywords can be semantically analyzed to obtain semantic feature information of the video keywords, such as the meaning, part of speech, emotion, etc. of the video keywords. Specifically, the plot text can be processed based on a pre-trained semantic analysis model to obtain candidate semantic feature information of the video keywords of the video plot text, and the candidate semantic feature information can be post-processed (such as deduplication, part of speech restoration, etc.) to obtain semantic feature information. This implementation process will be described in the following embodiments in combination with Figure 3 A detailed description is given below and this embodiment will not be described in detail here.

[0047] After the semantic feature information of the video keywords in the video plot text is obtained, step 103 is executed.

[0048] Step 103: Encoding the video plot text and the semantic feature information respectively to obtain a text encoding vector of the video plot text and a semantic feature vector corresponding to the semantic feature information.

[0049] After obtaining the semantic feature information of the video keywords in the video plot text, the video plot text and the semantic feature information can be encoded separately to obtain a text encoding vector of the video plot text and a semantic feature vector corresponding to the semantic feature information, that is, the semantic feature information of the video plot text and the video keywords are vectorized respectively to obtain the corresponding text encoding vector and semantic feature vector.

[0050] In a specific implementation, the video plot text and video keywords can be vectorized based on a preset model (BERT-based large language model LLM, etc.) to obtain a text encoding vector and a semantic feature vector represented by a vectorization. Of course, it is not limited to this. In a specific implementation, other methods can be used to obtain text encoding vectors and semantic feature vectors. This embodiment does not limit the method for obtaining text encoding vectors and semantic feature vectors.

[0051] After encoding the video plot text and the semantic feature information to obtain a text encoding vector of the video plot text and a semantic feature vector corresponding to the semantic feature information, step 104 is performed.

[0052] Step 104: Determine, based on the semantic feature vector and the text encoding vector, target video keywords among the video keywords whose similarity index with the video plot text is greater than a threshold.

[0053] The threshold refers to a preset threshold of the similarity index for screening video keywords with high similarity. In this example, the specific value of the threshold can be determined according to business requirements, and this embodiment does not impose any limitation on this.

[0054] The target video keyword refers to a video keyword selected from the video keywords whose similarity index with the video plot text is greater than a threshold.

[0055] After encoding the video plot text and semantic feature information to obtain the text encoding vector of the video plot text and the semantic feature vector corresponding to the semantic feature information, the target video keywords in the video keywords whose similarity index with the video plot text is greater than a threshold value can be determined based on the semantic feature vector and the text encoding vector. Specifically, the semantic feature vector and the text encoding vector can be subjected to correlation analysis to obtain the similarity index between the video keywords and the video plot text, and the target video keywords can be screened out from the video keywords based on the similarity index and the threshold value. This implementation process will be described in conjunction with the following embodiments. Figure 4 A detailed description is given below and this embodiment will not be described in detail here.

[0056] After determining the target video keywords among the video keywords whose similarity index with the video plot text is greater than the threshold value according to the semantic feature vector and the text encoding vector, step 105 is executed.

[0057] Step 105: Generate video barrage text for the video file during the specified video playback period based on the target video keywords.

[0058] After obtaining the target video keywords, the video barrage text of the video file in the specified video playback period can be generated according to the target video keywords. Specifically, the target video keywords can be directly used as the video barrage text. The target video keywords can also be optimized to generate the video barrage text, etc.

[0059] In this embodiment, after obtaining the target video keyword, the barrage text generated by the target video keyword can also be updated in combination with the user's historical barrage data or the user's barrage requirements to obtain a video barrage text that meets the user's requirements. Figure 5 and Figure 6 A detailed description is given below and this embodiment will not be described in detail here.

[0060] After the video barrage text of the video file in the specified video playback period is generated according to the target video keywords, step 106 is executed.

[0061] Step 106: Store the video file identifier of the video file, the designated video play time period, and the video barrage text in an associated manner.

[0062] After generating the video barrage text of the video file in the specified video playback period according to the target video keywords, the video file identifier of the video file, the specified video playback period and the video barrage text can be associated and stored, that is, the storage format is <video id, playback time, vocabulary (containing one or more video barrage texts)>. When the user enters the barrage, the specific vocabulary can be found according to the video id and the current playback time, and the corresponding words can be recommended from the vocabulary, so that the user can quickly select and complete the barrage related to the video plot for sending.

[0063] The embodiment of the present application analyzes and understands the video plot text of the video file within a specified video playback period to generate a specific video barrage vocabulary library related to the video plot, so that users can express their viewing experience more accurately when inputting barrage.

[0064] Next, combine Figure 2 The process of obtaining the video plot text is described in detail.

[0065] Reference Figure 2 , shows a flowchart of the steps of a method for obtaining video plot text provided by an embodiment of the present application. Figure 2 As shown, the method for obtaining video plot text may include: step 201 and step 202.

[0066] Step 201: Obtain the initial plot text of the video file in the specified video playback period.

[0067] In this embodiment, when generating video barrage related to the video plot within the specified video playback period of the video file, the initial plot text of the video file in the specified video playback period can be obtained, such as script text, dialogue text, narration text, etc.

[0068] After the initial plot text of the video file in the specified video playback period is obtained, step 202 is executed.

[0069] Step 202: Process the initial plot text based on a preprocessing method to obtain the video plot text.

[0070] In this embodiment, the preprocessing method may include: at least one of word segmentation processing, part-of-speech tagging processing and syntactic analysis processing.

[0071] After the initial plot text of the video file in the specified video playback period is obtained, the initial plot text may be processed based on a preprocessing method to obtain a video plot text.

[0072] The embodiment of the present application can facilitate subsequent plot analysis by preprocessing the plot text in advance (such as word segmentation, part-of-speech tagging, syntactic analysis, etc.).

[0073] Next, combine Figure 3 The implementation process of obtaining the semantic feature information of video keywords is described in detail.

[0074] Reference Figure 3 , shows a flowchart of the steps of a method for obtaining semantic feature information provided by an embodiment of the present application. Figure 3 As shown, the method for acquiring semantic feature information may include: step 301 and step 302.

[0075] Step 301: Process the video plot text based on a pre-trained semantic analysis model to obtain candidate semantic feature information of video keywords in the video plot text.

[0076] In this embodiment, the pre-trained semantic analysis model refers to a model for extracting keywords and performing keyword semantic analysis on the video plot text. In this example, the semantic analysis model can be but is not limited to a BERT-based LLM (Large Language Model l) model.

[0077] After obtaining the video plot text of the video file, the video plot text can be processed based on the pre-trained semantic analysis model to obtain candidate semantic feature information of the video keywords of the video plot text. For example, the keyword semantic information extraction function of the LLM model can be used to extract the video keywords in the video plot text, and analyze the candidate semantic feature information of each video keyword. The candidate semantic feature information may include the meaning, part of speech, emotion, etc. of the video keyword.

[0078] Specifically, the model processing process can be as follows: 1. Input the video plot text into a semantic analysis model (such as an LLM model). 2. Use the context information extraction function in the semantic analysis model to obtain the context information of each word in the video plot text, wherein the context information may include words and phrases in the preceding and following texts. 3. Use the keyword semantic information extraction function in the semantic analysis model to obtain the semantic information of each keyword in the video plot text, wherein the semantic information may include the meaning, part of speech, emotion, etc. of the keyword. 4. Feature fusion: Fusion of context information and keyword semantic information to obtain feature representation of the video plot text, wherein the feature representation may include: the theme, emotion, key information, etc. of the video plot text. 5. Feature selection: According to task requirements, semantic feature information related to the video plot text can be selected according to the feature representation of the fused video plot text (such as by setting a threshold, sorting, etc.) as candidate semantic feature information.

[0079] After the video plot text is processed based on the pre-trained semantic analysis model to obtain candidate semantic feature information of the video keywords of the video plot text, step 302 is performed.

[0080] Step 302: Process the candidate semantic feature information based on a preset post-processing method to obtain the semantic feature information.

[0081] After processing the video plot text based on the pre-trained semantic analysis model to obtain the candidate semantic feature information of the video keywords of the video plot text, the candidate semantic feature information can be processed using a preset post-processing method to obtain the semantic feature information of the video keywords. In this example, the preset post-processing method may include at least one of: repeated feature removal, feature part-of-speech restoration processing, etc. Among them, repeated feature removal refers to eliminating features that appear repeatedly in the candidate semantic feature information to reduce redundancy and improve processing efficiency. Feature part-of-speech restoration processing refers to restoring words with different forms but the same meaning (such as different tenses of verbs, singular and plural forms of nouns, etc.) to a unified basic form to facilitate subsequent semantic analysis and processing.

[0082] The embodiment of the present application can improve the accuracy of plot analysis by using a pre-trained semantic analysis model to extract keywords and perform semantic analysis on the video plot text. At the same time, by processing the candidate semantic feature information in a preset post-processing manner, such as removing repeated keywords and performing part-of-speech restoration, repeated keywords can be avoided and subsequent similarity analysis can be facilitated.

[0083] Next, combine Figure 4 The screening process of target video keywords is described in detail.

[0084] Reference Figure 4 , shows a flowchart of the steps of a target video keyword screening method provided by an embodiment of the present application. Figure 4 As shown, the target video keyword screening method may include: step 401 and step 402.

[0085] Step 401: performing a correlation analysis on the semantic feature vector and the text encoding vector to obtain a similarity index between the video keyword and the video plot text.

[0086] In this embodiment, after obtaining the semantic feature vector corresponding to the semantic feature information of the video keyword and the text encoding vector of the video plot text, a correlation analysis can be performed on the semantic feature vector and the text encoding vector to obtain a similarity index between the video keyword and the video plot text.

[0087] In a specific implementation, the cosine distance between the semantic feature vector and the text encoding vector can be obtained, and the cosine distance can be used as the similarity index between the video keywords and the video plot text. The Euclidean distance between the semantic feature vector and the text encoding vector can also be obtained, and the Euclidean distance can be used as the similarity index between the video keywords and the video plot text.

[0088] Of course, this is not limited to this. In actual applications, other methods can be used to obtain the similarity index between video keywords and video plot text, such as calculating the Euclidean distance between the semantic feature vector and the text encoding vector as a similarity index, etc. Specifically, the method for obtaining the similarity index can be determined according to business needs, and this embodiment does not limit this.

[0089] After performing correlation analysis on the semantic feature vector and the text encoding vector to obtain a similarity index between the video keyword and the video plot text, step 402 is executed.

[0090] Step 402: Filter out target video keywords whose similarity index is greater than a threshold from the video keywords.

[0091] After performing correlation analysis on the semantic feature vector and the text encoding vector to obtain the similarity index between the video keywords and the video plot text, the target video keywords with a similarity index greater than a threshold can be screened out from the video keywords. For example, the video keywords include: keyword 1, keyword 2, keyword 3 and keyword 4, and the similarity indexes between keyword 1, keyword 2, keyword 3 and keyword 4 and the video plot text are: 0.8, 0.9, 0.4 and 0.6 respectively. When the threshold is 0.5, keyword 1, keyword 2 and keyword 4 can be used as target video keywords, etc.

[0092] It can be understood that the above examples are merely examples listed for a better understanding of the technical solutions of the embodiments of the present application, and are not intended to be the sole limitation to the embodiments.

[0093] The embodiment of the present application analyzes the similarity index between video keywords and video plot text to screen out target video keywords as the basis for subsequent generation of video barrage text, so that the generated video barrage text is relevant to the video plot, allowing users to express their viewing experience more accurately when inputting barrage text.

[0094] Next, combine Figure 5 The implementation process of generating video barrage text by combining historical barrage data is described in detail.

[0095] Reference Figure 5 , shows a flowchart of the steps of a method for obtaining video barrage text provided by an embodiment of the present application. Figure 5 As shown, the method for obtaining video barrage text may include: step 501 and step 502.

[0096] Step 501: Generate the first video bullet screen text of the video file in the specified video playback time period according to the target video keyword.

[0097] In this embodiment, after obtaining the target video keyword, the first video bullet text of the video file in the specified video playback period can be generated according to the target video keyword. Specifically, the target video keyword can be directly used as the first video bullet text, or the target video keyword can be optimized (such as colloquial processing of the target video keyword, etc.) to obtain the first video bullet text, etc.

[0098] After generating the first video barrage text of the video file in the specified video playback period according to the target video keyword, execute step 502.

[0099] Step 502: Update the first video barrage text according to the historical barrage data of the first user to obtain the video barrage text corresponding to the first user.

[0100] After the first video barrage text of the video file in the specified video playback period is generated according to the target video keyword, the first video barrage text can be updated according to the historical barrage data of the first user, thereby obtaining the video barrage text corresponding to the first user. For example, if it is determined according to the historical barrage data that the user likes to use emoticons, relevant emoticons can be added to the first video barrage text (such as adding corresponding emoticons according to the emotion of the first video barrage text, etc.) to generate video barrage text, etc.

[0101] It can be understood that the above examples are merely examples listed for a better understanding of the technical solutions of the embodiments of the present application, and are not intended to be the sole limitation to the embodiments.

[0102] The embodiment of the present application generates corresponding video barrage text by combining the user's historical barrage data, so that the generated video barrage text can be made to conform to the user's barrage input style on the basis of being relevant to the video plot, thereby improving the user's barrage interaction experience.

[0103] Next, combine Figure 6 The implementation process of generating video barrage text based on user needs is described in detail.

[0104] Reference Figure 6 , shows a flowchart of another method for obtaining video barrage text provided by an embodiment of the present application. Figure 6 As shown, the method for obtaining video barrage text may include: step 601 and step 602.

[0105] Step 601: Generate a second video bullet screen text for the video file in the specified video playback time period according to the target video keyword.

[0106] In this embodiment, after obtaining the target video keyword, the second video bullet text of the video file in the specified video playback period can be generated according to the target video keyword. Specifically, the target video keyword can be directly used as the second video bullet text, or the target video keyword can be optimized (such as colloquial processing of the target video keyword, etc.) to obtain the second video bullet text, etc.

[0107] After generating the second video bullet screen text of the video file in the specified video playback time period according to the target video keyword, execute step 602.

[0108] Step 602: Update the second video barrage text according to the barrage requirement information of the second user, and obtain the video barrage text corresponding to the second user.

[0109] After generating the second video barrage text of the video file in the specified video playback period according to the target video keywords, the second video barrage text can be updated according to the barrage requirement information of the second user to obtain the video barrage text corresponding to the second user. For example, when generating the video barrage text, a barrage requirement input box can be displayed on the screen of the electronic device used by the second user, and several requirement options can be given for the user to input the barrage requirement, and the video barrage text corresponding to the second user can be generated by combining the barrage requirement and the second video barrage text. Or, for example, when the barrage requirement input by the user is to add special marks, symbols, etc. at the end of the barrage text, special marks, symbols, etc. can be added at the end of the second video barrage text to obtain the video barrage text, etc.

[0110] It can be understood that the above examples are merely examples listed for a better understanding of the technical solutions of the embodiments of the present application, and are not intended to be the sole limitation to the embodiments.

[0111] The embodiment of the present application generates corresponding video barrage text by combining the user's barrage needs, so that the generated video barrage text can meet the user's personalized needs on the basis of being relevant to the video plot, thereby improving the user's barrage interaction experience.

[0112] Next, combine Figure 7 The recommendation process of video barrage is described in detail.

[0113] Reference Figure 7 , shows a flowchart of the steps of a method for recommending target video barrage text provided by an embodiment of the present application, such as Figure 7 As shown, the target video barrage text recommendation method may include: step 701, step 702, step 703 and step 704.

[0114] Step 701: During the playback of a target video file, when a bullet screen initiation operation triggered by a user is received, a target video playback time of the target video file corresponding to the bullet screen initiation operation is obtained.

[0115] In this embodiment, when a user-triggered bullet screen initiation operation is received during the playback of the target video file (such as the user pulling up the bullet screen publisher, etc.), the target video playback time of the target video file corresponding to the bullet screen initiation operation can be obtained. For example, when a user is watching a variety show video, at 00:55 of the video, the video playback time of the 55th minute of the variety show video can be used as the target video playback time, etc.

[0116] In this example, the video files mentioned in the above implementation process include the target video file.

[0117] Step 702: Obtain a target video file identifier of the target video file.

[0118] When the user plays the target video file, the target video file identifier of the target video file may be obtained.

[0119] Step 703: According to the pre-stored correspondence between the video file identifier, the designated video playback time period and the video bullet text, obtain the target video bullet text corresponding to the time period of the target video playback moment and the target video file identifier.

[0120] After obtaining the target video playback time of the target video file corresponding to the barrage initiation operation and the target video file identifier, the target video barrage text corresponding to the time period of the target video playback time and the target video file identifier can be obtained based on the correspondence between the pre-stored video file identifier, the specified video playback time period and the video barrage text.

[0121] In this example, the number of target video barrage texts may be one or more, and specifically, it may be determined according to actual conditions, and this embodiment does not impose any limitation on this.

[0122] After obtaining the target video barrage text, execute step 704.

[0123] Step 704: Recommend the target video barrage text to the user.

[0124] After obtaining the target video bullet text, the target video bullet text can be recommended to the user. In this embodiment, the target video bullet text can be displayed in a list form for bullet text recommendation, such as displaying a list in the middle area or other areas of the terminal screen, and the target video bullet text can be displayed in sequence in the list. The target video bullet text can also be displayed in a pop-up window form for bullet text recommendation. For example, a pop-up window can be displayed in the lower right corner area or the upper left corner area of ​​the terminal screen, and the target video bullet text can be displayed in the pop-up window. The target video bullet text can also be displayed in the form of a sidebar or bottom bar, for example, a fixed or expandable sidebar / bottom bar is set on one side or the bottom of the terminal screen to display the recommended target video bullet text.

[0125] It can be understood that the above examples are only examples listed for a better understanding of the technical solutions of the embodiments of the present application. In actual applications, other methods can also be used to display the target video barrage text, such as card-type display, etc. This embodiment does not limit this.

[0126] In an embodiment of the present application, when a user inputs a barrage, a specific vocabulary is found based on the video identifier and the current playback time, and corresponding words are recommended from the vocabulary, so that the user can quickly select and complete the barrage related to the video plot for sending.

[0127] The generation process of video bullet comments and the recommendation process of video bullet comments can be combined Figure 8 This is described in detail as follows.

[0128] Reference Figure 8 , which shows a schematic diagram of a video bullet screen generation and recommendation process provided by an embodiment of the present application. Figure 8 As shown, the video bullet screen generation and recommendation process can be: first, the video plot text can be collected. Specifically, for video files, its script, dialogue, narration and other information can be collected as the video plot text. Secondly, the plot keyword extraction function of the large language model of BERT can be used to obtain the semantic information of each keyword in the text, including the meaning, part of speech and emotion of the keyword. Then, a specific vocabulary can be generated. Specifically, the similarity index between the keyword and the video plot text can be calculated based on the semantic feature vector of the keyword and the text encoding vector of the video plot text, and the keywords with high similarity can be filtered out from the keywords according to the similarity index to generate a specific vocabulary (i.e., a database containing specific bullet screen text). The output data structure can be: <video id, playback time (minutes), vocabulary>. Finally, a bullet screen input prompt is performed: when the user enters the bullet screen, a specific vocabulary is found according to the video ID and the current playback time, and the corresponding words are recommended from the vocabulary, so that the user can quickly select and improve the bullet screen related to the video plot for sending.

[0129] This embodiment generates a specific word library closely related to the video plot by understanding and analyzing the plot text, so that users can more conveniently find expressions related to the video content when entering the bullet screen, thereby improving the experience and participation of the bullet screen interaction. In addition, the method can also dynamically generate corresponding specific word libraries according to different video contents, provide personalized bullet screen experience, and increase user stickiness.

[0130] The video bullet screen generation method provided in the embodiment of the present application is obtained by obtaining the video plot text of the video file within the specified video playback period. The semantic feature information of the video keywords in the video plot text is obtained. The video plot text and the semantic feature information are respectively encoded to obtain the text encoding vector of the video plot text and the semantic feature vector corresponding to the semantic feature information. According to the semantic feature vector and the text encoding vector, the target video keyword in the video keyword whose similarity index with the video plot text is greater than the threshold is determined. According to the target video keyword, the video bullet screen text of the video file in the specified video playback period is generated. The video file identifier, the specified video playback period and the video bullet screen text of the video file are stored in association. The embodiment of the present application analyzes and understands the video plot text of the video file within the specified video playback period to generate a specific video bullet screen word library related to the video plot, so that the user can express the viewing experience more accurately when inputting bullet screen, and improve the experience of bullet screen interaction.

[0131] Reference Fig. 9 , shows a schematic diagram of the structure of a video barrage generating device provided in an embodiment of the present application, such as Fig. 9 As shown, the video barrage generation device 900 may include the following modules:

[0132] The video plot text acquisition module 910 is used to acquire the video plot text of the video file within the specified video playback period;

[0133] A semantic feature information acquisition module 920 is used to acquire semantic feature information of video keywords in the video plot text;

[0134] A text semantic vector acquisition module 930 is used to encode the video plot text and the semantic feature information respectively to obtain a text encoding vector of the video plot text and a semantic feature vector corresponding to the semantic feature information;

[0135] A target keyword determination module 940 is used to determine, according to the semantic feature vector and the text encoding vector, a target video keyword whose similarity index with the video plot text is greater than a threshold value among the video keywords;

[0136] The video barrage text generating module 950 is used to generate the video barrage text of the video file in the specified video playback period according to the target video keyword;

[0137] The video barrage text storage module 960 is used to store the video file identifier of the video file, the specified video playback time period and the video barrage text in an associated manner.

[0138] Optionally, the target keyword determination module includes:

[0139] A similarity index obtaining unit, used for performing correlation analysis on the semantic feature vector and the text encoding vector to obtain a similarity index between the video keyword and the video plot text;

[0140] The target keyword screening unit is used to screen out the target video keywords whose similarity index is greater than a threshold from the video keywords.

[0141] Optionally, the similarity index obtaining unit includes:

[0142] A first index acquisition subunit is used to acquire a cosine distance between the semantic feature vector and the text encoding vector, and use the cosine distance as the similarity index;

[0143] The second index acquisition subunit is used to acquire the Euclidean distance between the semantic feature vector and the text encoding vector, and use the Euclidean distance as the similarity index.

[0144] Optionally, the video barrage text generation module includes:

[0145] A first bullet screen text generating unit, configured to generate a first video bullet screen text for the video file in the specified video playback period according to the target video keyword;

[0146] The first video barrage text acquisition unit is used to update the first video barrage text according to the historical barrage data of the first user, and obtain the video barrage text corresponding to the first user.

[0147] Optionally, the video barrage text generation module includes:

[0148] A second bullet screen text generating unit, configured to generate a second video bullet screen text for the video file in the specified video playback period according to the target video keyword;

[0149] The second video barrage text acquisition unit is used to update the second video barrage text according to the barrage demand information of the second user, and obtain the video barrage text corresponding to the second user.

[0150] Optionally, the device further comprises:

[0151] A target playback time acquisition module is used to obtain the target video playback time of the target video file corresponding to the barrage initiation operation when a barrage initiation operation triggered by a user is received during the playback of the target video file;

[0152] A target file identifier acquisition module, used to acquire a target video file identifier of the target video file;

[0153] A target barrage text acquisition module is used to acquire the target video barrage text corresponding to the time period of the target video playback moment and the target video file identifier according to the correspondence between the pre-stored video file identifier, the designated video playback time period and the video barrage text;

[0154] The target barrage text recommendation module is used to recommend the target video barrage text to the user.

[0155] Optionally, the video plot text acquisition module includes:

[0156] An initial plot text acquisition unit, used to acquire the initial plot text of the video file in the specified video playback period;

[0157] A video plot text acquisition unit, used for processing the initial plot text based on a preprocessing method to obtain the video plot text;

[0158] The preprocessing method includes: at least one of word segmentation processing, part-of-speech tagging processing and syntactic analysis processing.

[0159] Optionally, the semantic feature information acquisition module includes:

[0160] Processing the video plot text based on a pre-trained semantic analysis model to obtain candidate semantic feature information of video keywords of the video plot text;

[0161] Processing the candidate semantic feature information based on a preset post-processing method to obtain the semantic feature information of the video keyword;

[0162] Wherein, the preset post-processing method includes: at least one of repeated feature removal and feature part-of-speech restoration processing.

[0163] The video bullet screen generation device provided in the embodiment of the present application obtains the video plot text of the video file within the specified video playback period. Obtain the semantic feature information of the video keywords in the video plot text. Encode the video plot text and the semantic feature information respectively to obtain the text encoding vector of the video plot text and the semantic feature vector corresponding to the semantic feature information. According to the semantic feature vector and the text encoding vector, determine the target video keyword in the video keyword whose similarity index with the video plot text is greater than the threshold. According to the target video keyword, generate the video bullet screen text of the video file in the specified video playback period. The video file identifier, the specified video playback period and the video bullet screen text of the video file are stored in association. The embodiment of the present application analyzes and understands the video plot text of the video file within the specified video playback period to generate a specific video bullet screen word library related to the video plot, so that the user can express the viewing experience more accurately when inputting bullet screen, and improve the experience of bullet screen interaction.

[0164] The present application also provides an electronic device, such as Fig.10 As shown, it includes a processor 1001, a communication interface 1002, a memory 1003 and a communication bus 1004, wherein the processor 1001, the communication interface 1002, and the memory 1003 communicate with each other through the communication bus 1004.

[0165] Memory 1003, used for storing computer programs;

[0166] The processor 1001 is used to execute the program stored in the memory 1003, and implements the following steps:

[0167] Get the video plot text of the video file within the specified video playback period;

[0168] Obtaining semantic feature information of video keywords in the video plot text;

[0169] Encoding the video plot text and the semantic feature information respectively to obtain a text encoding vector of the video plot text and a semantic feature vector corresponding to the semantic feature information;

[0170] Determine, according to the semantic feature vector and the text encoding vector, a target video keyword whose similarity index with the video plot text is greater than a threshold value among the video keywords;

[0171] Generate video barrage text for the video file during the specified video playback period according to the target video keyword;

[0172] The video file identifier of the video file, the designated video play time period and the video barrage text are stored in association.

[0173] Optionally, the determining, according to the semantic feature vector and the text encoding vector, target video keywords among the video keywords whose similarity index with the video plot text is greater than a threshold value comprises:

[0174] Performing correlation analysis on the semantic feature vector and the text encoding vector to obtain a similarity index between the video keyword and the video plot text;

[0175] Target video keywords whose similarity index is greater than a threshold are selected from the video keywords.

[0176] Optionally, the performing correlation analysis on the semantic feature vector and the text encoding vector to obtain a similarity index between the video keyword and the video plot text includes:

[0177] Obtaining a cosine distance between the semantic feature vector and the text encoding vector, and using the cosine distance as the similarity index; or

[0178] The Euclidean distance between the semantic feature vector and the text encoding vector is obtained, and the Euclidean distance is used as the similarity index.

[0179] Optionally, generating the video barrage text of the video file in the specified video playback period according to the target video keyword includes:

[0180] Generate the first video barrage text of the video file in the specified video playback time period according to the target video keyword;

[0181] According to the historical barrage data of the first user, the first video barrage text is updated to obtain the video barrage text corresponding to the first user.

[0182] Optionally, generating the video barrage text of the video file in the specified video playback period according to the target video keyword includes:

[0183] Generate a second video bullet screen text of the video file in the specified video playback time period according to the target video keyword;

[0184] According to the barrage demand information of the second user, the barrage text of the second video is updated to obtain the video barrage text corresponding to the second user.

[0185] Optionally, after the video file identifier of the video file, the designated video play time period and the video barrage text are stored in association, the method further includes:

[0186] During the playback of the target video file, when a bullet screen initiation operation triggered by a user is received, obtaining a target video playback time of the target video file corresponding to the bullet screen initiation operation;

[0187] Obtain a target video file identifier of the target video file;

[0188] According to the correspondence between the pre-stored video file identifier, the designated video playback time period and the video barrage text, obtaining the target video barrage text corresponding to the time period of the target video playback moment and the target video file identifier;

[0189] The target video barrage text is recommended to the user.

[0190] Optionally, obtaining the video plot text of the video file within a specified video playback period includes:

[0191] Obtaining the initial plot text of the video file during the specified video playback period;

[0192] Processing the initial plot text based on a preprocessing method to obtain the video plot text;

[0193] The preprocessing method includes: at least one of word segmentation processing, part-of-speech tagging processing and syntactic analysis processing.

[0194] Optionally, the obtaining of semantic feature information of video keywords in the video plot text includes:

[0195] Processing the video plot text based on a pre-trained semantic analysis model to obtain candidate semantic feature information of video keywords of the video plot text;

[0196] Processing the candidate semantic feature information based on a preset post-processing method to obtain the semantic feature information of the video keyword;

[0197] Wherein, the preset post-processing method includes: at least one of repeated feature removal and feature part-of-speech restoration processing.

[0198] The communication bus mentioned in the above terminal can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. The communication bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, only one thick line is used in the figure, but it does not mean that there is only one bus or one type of bus.

[0199] The communication interface is used for communication between the above terminal and other devices.

[0200] The memory may include a random access memory (RAM) or a non-volatile memory, such as at least one disk memory. Optionally, the memory may also be at least one storage device located away from the aforementioned processor.

[0201] The above-mentioned processor can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components.

[0202] In another embodiment provided in the present application, a computer-readable storage medium is also provided, in which instructions are stored. When the computer-readable storage medium is run on a computer, the computer executes the video barrage generation method described in any of the above embodiments.

[0203] In another embodiment provided in the present application, a computer program product comprising instructions is also provided, on which a computer program is stored. When the computer program is run on a computer, the computer executes any of the above-mentioned methods for generating video barrage.

[0204] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium, for example, the computer instructions can be transmitted from a website site, computer, server or data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) mode to another website site, computer, server or data center. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server, data center, etc. that contains one or more available media integrated. The available medium may be a magnetic medium (eg, a floppy disk, a hard disk, a magnetic tape), an optical medium (eg, a DVD), or a semiconductor medium (eg, a solid state drive (SSD)).

[0205] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.

[0206] Each embodiment in this specification is described in a related manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0207] The above description is only a preferred embodiment of the present application and is not intended to limit the protection scope of the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application are included in the protection scope of the present application.

Claims

1. A video barrage generation method, characterized in that: include: Get the video plot text of the video file within the specified video playback period; Obtaining semantic feature information of video keywords in the video plot text; Encoding the video plot text and the semantic feature information respectively to obtain a text encoding vector of the video plot text and a semantic feature vector corresponding to the semantic feature information; Determine, according to the semantic feature vector and the text encoding vector, a target video keyword whose similarity index with the video plot text is greater than a threshold value among the video keywords; Generate video barrage text for the video file during the specified video playback period according to the target video keyword; The video file identifier of the video file, the designated video play time period and the video barrage text are stored in association.

2. The method according to claim 1, characterized in that The step of determining, according to the semantic feature vector and the text encoding vector, a target video keyword whose similarity index with the video plot text is greater than a threshold value among the video keywords comprises: Performing correlation analysis on the semantic feature vector and the text encoding vector to obtain a similarity index between the video keyword and the video plot text; Target video keywords whose similarity index is greater than a threshold are selected from the video keywords.

3. The method according to claim 2, characterized in that The performing correlation analysis on the semantic feature vector and the text encoding vector to obtain a similarity index between the video keyword and the video plot text includes: Obtaining a cosine distance between the semantic feature vector and the text encoding vector, and using the cosine distance as the similarity index; or The Euclidean distance between the semantic feature vector and the text encoding vector is obtained, and the Euclidean distance is used as the similarity index.

4. The method according to claim 1, characterized in that: The step of generating the video barrage text of the video file in the specified video playback period according to the target video keyword includes: Generate the first video barrage text of the video file in the specified video playback time period according to the target video keyword; According to the historical barrage data of the first user, the first video barrage text is updated to obtain the video barrage text corresponding to the first user.

5. The method according to claim 1, characterized in that The step of generating the video barrage text of the video file in the specified video playback period according to the target video keyword includes: Generate a second video bullet screen text of the video file in the specified video playback time period according to the target video keyword; According to the barrage demand information of the second user, the barrage text of the second video is updated to obtain the video barrage text corresponding to the second user.

6. The method according to claim 1, characterized in that After the video file identifier of the video file, the designated video play time period and the video barrage text are stored in association, the method further includes: During the playback of the target video file, when a bullet screen sending operation triggered by a user is received, obtaining a target video playback time of the target video file corresponding to the bullet screen sending operation; Obtain a target video file identifier of the target video file; According to the correspondence between the pre-stored video file identifier, the designated video playback time period and the video barrage text, obtaining the target video barrage text corresponding to the time period of the target video playback moment and the target video file identifier; The target video barrage text is recommended to the user.

7. The method according to claim 1, characterized in that The step of obtaining the video plot text of the video file within the specified video playback period includes: Obtaining the initial plot text of the video file during the specified video playback period; Processing the initial plot text based on a preprocessing method to obtain the video plot text; The preprocessing method includes: at least one of word segmentation processing, part-of-speech tagging processing and syntactic analysis processing.

8. The method according to claim 1, characterized in that The obtaining of semantic feature information of video keywords in the video plot text includes: Processing the video plot text based on a pre-trained semantic analysis model to obtain candidate semantic feature information of video keywords of the video plot text; Processing the candidate semantic feature information based on a preset post-processing method to obtain the semantic feature information of the video keyword; Wherein, the preset post-processing method includes: at least one of repeated feature removal and feature part-of-speech restoration processing.

9. A video barrage generation device, characterized in that: include: A video plot text acquisition module is used to acquire the video plot text of a video file within a specified video playback period; A semantic feature information acquisition module, used to acquire semantic feature information of video keywords in the video plot text; A text semantic vector acquisition module, used to encode the video plot text and the semantic feature information respectively, to obtain a text encoding vector of the video plot text and a semantic feature vector corresponding to the semantic feature information; A target keyword determination module, used to determine, according to the semantic feature vector and the text encoding vector, a target video keyword whose similarity index with the video plot text is greater than a threshold value among the video keywords; A video barrage text generation module, used to generate a video barrage text of the video file in the specified video playback period according to the target video keyword; The video barrage text storage module is used to associate and store the video file identifier of the video file, the specified video playback time period and the video barrage text.

10. An electronic device, characterized in that: It includes a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other through the communication bus; Memory, used to store computer programs; A processor, for implementing the method steps described in any one of claims 1 to 8 when executing a program stored in a memory.

11. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the method according to any one of claims 1 to 8 is implemented.

12. A computer program product comprising instructions, on which a computer program is stored, characterized in that When the computer program is executed on a computer, the computer is caused to execute the method according to any one of claims 1 to 8.

Images