CN116958971A - Video tag determining method, sheet information displaying method and related devices - Google Patents

Video tag determining method, sheet information displaying method and related devices Download PDF

Info

Publication number
CN116958971A
CN116958971A CN202310909433.8A CN202310909433A CN116958971A CN 116958971 A CN116958971 A CN 116958971A CN 202310909433 A CN202310909433 A CN 202310909433A CN 116958971 A CN116958971 A CN 116958971A
Authority
CN
China
Prior art keywords
interactive
text
video
interactive text
clustering
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202310909433.8A
Other languages
Chinese (zh)
Inventor
柳政
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Beijing QIYI Century Science and Technology Co Ltd
Original Assignee
Beijing QIYI Century Science and Technology Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Beijing QIYI Century Science and Technology Co Ltd filed Critical Beijing QIYI Century Science and Technology Co Ltd
Priority to CN202310909433.8A priority Critical patent/CN116958971A/en
Publication of CN116958971A publication Critical patent/CN116958971A/en
Pending legal-status Critical Current

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/70—Labelling scene content, e.g. deriving syntactic or semantic representations
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
    • G06F16/35—Clustering; Classification
    • G06F16/353—Clustering; Classification into predefined classes
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/70—Information retrieval; Database structures therefor; File system structures therefor of video data
    • G06F16/73—Querying
    • G06F16/738—Presentation of query results
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/70—Information retrieval; Database structures therefor; File system structures therefor of video data
    • G06F16/78—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
    • G06F16/783—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using metadata automatically derived from the content
    • G06F16/7844—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using metadata automatically derived from the content using original textual content or text extracted from visual content or transcript of audio data
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/70—Information retrieval; Database structures therefor; File system structures therefor of video data
    • G06F16/78—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
    • G06F16/7867—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using information manually generated, e.g. tags, keywords, comments, title and artist information, manually generated time, location and usage information, user ratings
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/22—Matching criteria, e.g. proximity measures
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/23—Clustering techniques
    • G06F18/232—Non-hierarchical techniques
    • G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
    • G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/40—Scenes; Scene-specific elements in video content
    • G06V20/41—Higher-level, semantic clustering, classification or understanding of video scenes, e.g. detection, labelling or Markovian modelling of sport events or news items

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • General Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • Multimedia (AREA)
  • Computational Linguistics (AREA)
  • Databases & Information Systems (AREA)
  • Evolutionary Computation (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Library & Information Science (AREA)
  • Software Systems (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Evolutionary Biology (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Biomedical Technology (AREA)
  • Health & Medical Sciences (AREA)
  • Probability & Statistics with Applications (AREA)
  • Biophysics (AREA)
  • General Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • Mathematical Physics (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)

Abstract

The embodiment of the invention provides a video tag determining method, a sheet information displaying method and a related device, wherein the method comprises the following steps: acquiring an interactive text issued aiming at a target video; clustering the interactive text based on the text feature similarity of the interactive text to obtain one or more interactive text clusters and clustering labels of the interactive text clusters; and determining topic labels of the target video based on the cluster labels of the interactive text clusters. More intelligent, refined and multiple video tags can be determined.

Description

Video tag determining method, sheet information displaying method and related devices
Technical Field
The invention relates to the technical field of computers, in particular to a video tag determining method, a sheet information displaying method and a related device.
Background
Currently, video tags are widely used in tasks such as classification, recommendation and retrieval of videos. Therefore, the accuracy and the diversity of the video tags are important for the management and popularization of the video.
However, the current video tag is mainly generated based on actors, themes and the like, personalized information transmitted by users is difficult to utilize, related topics of episodes spontaneously formed by the users cannot be automatically tracked on a large scale, and the diversity of the video tag is insufficient.
Disclosure of Invention
The embodiment of the invention aims to provide a video tag determining method, a sheet information displaying method and a related device so as to improve the diversity of video tags. The specific technical scheme is as follows:
in a first aspect of the present invention, there is provided a video tag determining method, including:
acquiring an interactive text issued aiming at a target video;
clustering the interactive texts based on the text feature similarity of the interactive texts to obtain one or more interactive text clusters and clustering labels of the interactive text clusters;
and determining the topic label of the target video based on the cluster label of the interactive text cluster.
Optionally, the method further comprises:
determining text similarity among the topic labels determined for a plurality of target videos, and aggregating target videos corresponding to topic labels with the text similarity not smaller than a similarity threshold value to obtain a video recommendation list.
Optionally, the step of clustering the interactive text based on the text feature similarity of the interactive text includes:
extracting text feature vectors of the interactive texts, determining text feature similarities among the interactive texts based on the similarity among the text feature vectors of the interactive texts and/or based on time differences among corresponding time stamps of the interactive texts in the target video, and clustering the interactive texts based on the text feature similarities.
Optionally, before the clustering of the interactive text based on the text feature similarity of the interactive text, the clustering method further includes:
classifying the interactive texts based on a pre-trained classification model, and eliminating the interactive texts of a preset category; the preset categories include: a specific behavior category and/or a negative rating category.
Optionally, before the clustering of the interactive text based on the text feature similarity of the interactive text, the clustering method further includes:
and eliminating the non-ideographic characters in the interactive text.
Optionally, the step of determining the topic label of the target video based on the cluster label of the interactive text cluster includes:
determining the clustering label of each interactive text cluster as the topic label of the target video;
or alternatively, the first and second heat exchangers may be,
determining the quantity of the interactive texts contained in each interactive text cluster, and determining the cluster labels of the interactive text clusters, the quantity of which is not smaller than a preset value, as the topic labels of the target video.
In a second aspect of the present invention, a method for displaying sheet information is provided, applied to a client, and includes:
acquiring sheet search information, and sending the sheet search information to a server so that the server determines a recommended sheet matched with the sheet search information based on the sheet search information; each video in the recommended sheet contains topic labels with text feature similarity not smaller than preset similarity with the search information, and the topic label corresponding to each video is determined based on clustering labels of an obtained interactive text cluster after the interactive text issued for the video is clustered;
And receiving the recommended sheet returned by the server and displaying the recommended sheet.
In a third aspect of the present invention, there is also provided a video tag determining apparatus, including:
the acquisition module is used for acquiring the interactive text issued aiming at the target video;
the clustering module is used for clustering the interactive text based on the text feature similarity of the interactive text to obtain one or more interactive text clusters and clustering labels of the interactive text clusters;
and the determining module is used for determining the topic label of the target video based on the cluster label of the interactive text cluster.
In a fourth aspect of the present invention, there is provided a display device for sheet information, applied to a client, including:
the acquisition module is used for acquiring the sheet searching information and sending the sheet searching information to a server so that the server can determine a recommended sheet matched with the sheet searching information based on the sheet searching information; each video in the recommended sheet contains topic labels with text feature similarity not smaller than preset similarity with the search information, and the topic label corresponding to each video is determined based on clustering labels of an obtained interactive text cluster after the interactive text issued for the video is clustered;
And the display module is used for receiving the recommended sheet returned by the server and displaying the recommended sheet.
In a fifth aspect of the present invention, there is also provided an electronic device, including a processor, a communication interface, a memory, and a communication bus, where the processor, the communication interface, and the memory complete communication with each other through the communication bus;
a memory for storing a computer program;
and the processor is used for realizing the video tag determining method or the sheet information displaying method when executing the program stored in the memory.
In still another aspect of the present invention, there is further provided a computer readable storage medium, in which a computer program is stored, the computer program implementing any one of the above video tag determination methods or the presentation method of sheet information when executed by a processor.
In yet another aspect of the present invention, there is also provided a computer program product containing instructions which, when run on a computer, cause the computer to perform any of the video tag determination methods or the presentation method of sheet information described above.
According to the video tag determining method, the sheet information displaying method and the related device, the topic tag of the target video is determined based on the clustering tag of the interactive text cluster obtained by clustering by acquiring the interactive text issued for the target video and clustering the interactive text. Related topics spontaneously formed by a user aiming at the video can be tracked through the interactive text, and clustering labels which can be used for representing the related topics are obtained through clustering the interactive text, so that topics which are interested in the target video by the user can be effectively represented based on the topic labels determined by the clustering labels, and compared with the traditional video labels generated based on actors or themes, the topic labels of the video are more intelligent, finer and more numerous.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below.
Fig. 1 is a schematic flow chart of a video tag determining method according to an embodiment of the present invention;
fig. 2 is another flow chart of a video tag determining method according to an embodiment of the present invention;
fig. 3 is a schematic diagram of a video tag determining method according to an embodiment of the present invention;
fig. 4 is a flow chart of a method for displaying sheet information according to an embodiment of the present invention;
fig. 5 is a schematic structural diagram of a video tag determining apparatus according to an embodiment of the present invention;
fig. 6 is a schematic structural diagram of a display device for sheet information according to an embodiment of the present invention;
fig. 7 is a schematic structural diagram of an electronic device according to an embodiment of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be described below with reference to the accompanying drawings in the embodiments of the present invention.
In order to solve the problem of insufficient pluripotency of a conventional video tag, an embodiment of the present invention provides a video tag determining method, and fig. 1 is a schematic flow diagram of the video tag determining method provided by the embodiment of the present invention, referring to fig. 1, the video tag determining method provided by the embodiment of the present invention specifically includes the following steps:
Step S101: and acquiring an interactive text issued aiming at the target video.
Step S102: based on the text feature similarity of the interactive text, clustering the interactive text to obtain one or more interactive text clusters and clustering labels of the interactive text clusters.
Step S103: and determining topic labels of the target video based on the cluster labels of the interactive text clusters.
According to the video tag determining method provided by the embodiment of the invention, the topic tag of the target video is determined based on the clustering tag of the interactive text cluster obtained by clustering by acquiring the interactive text issued for the target video and clustering the interactive text. Related topics spontaneously formed by a user aiming at the video can be tracked through the interactive text, and clustering labels which can be used for representing the related topics are obtained through clustering the interactive text, so that topics which are interested in the target video by the user can be effectively represented based on the topic labels determined by the clustering labels, and compared with the traditional video labels generated based on actors or themes, the topic labels of the video are more intelligent, finer and more numerous.
The following describes the steps of the embodiments of the present invention, respectively:
In step S101, the target video is a video that needs to be video-tagged.
Specifically, a user usually publishes various text information for sharing his own evaluation or emotion and realizing interaction with other users during or after watching a video, and such text information is the interactive text in the embodiment of the present invention. For example, a user posts a bullet screen and comments, i.e., interactive text, for a video.
In the embodiment of the invention, the interactive text aiming at the target video release can be obtained through platforms such as a video website, a video evaluation website and the like.
In step S102, the interactive texts are clustered, specifically, the similarity between any two interactive texts on the text features may be calculated, and the interactive text with the similarity between the text features meeting the preset condition, i.e., the interactive text with the higher similarity is used as an interactive text cluster, so as to obtain one or more interactive text clusters.
As an example, in the process of clustering the interactive text based on the text feature similarity of the interactive text, the interactive text may be first converted into text feature vectors, and the interactive text may be clustered based on the similarity between the text feature vectors of the interactive text.
In the embodiment of the invention, the clustering algorithm can be adopted to realize the clustering of the interactive text. The specific clustering algorithm may be selected based on actual requirements, and as an example, a k-means (a clustering algorithm) algorithm or a knn (a clustering algorithm) algorithm may be used to cluster the interactive text.
Specifically, after the clustering is completed, the interactive texts in each interactive text cluster have higher text feature similarity, so that the interactive texts in the same interactive text cluster tend to express similar semantics, and the interactive texts in the same interactive text cluster can be considered as discussions on the same topic.
On this basis, for each interactive text cluster determined based on step S102, a cluster tag can be determined to characterize the topic discussed by the interactive text in the interactive text cluster. In the embodiment of the invention, the label is specifically text, for example, the label can be a word or a phrase. As one example, the tag may specifically characterize content such as a person in the video or a video episode.
The specific method for determining the cluster labels of the interactive text clusters can refer to the content in the related art, and the embodiment of the invention is not limited to this. As one example, the tags of the interactive text clusters may be determined by a keyword generation algorithm or a keyword extraction algorithm.
Specifically, for each interactive text cluster determined in step S102, a cluster label of the interactive text cluster may be determined.
In step S103, one or more cluster labels of the interactive text clusters may be selected as topic labels of the target video according to actual requirements, and the embodiment of the present invention does not limit the specific manner and range of selecting the interactive text clusters.
As an example, the cluster labels of all the interactive text clusters obtained by clustering may be determined as the topic labels of the target video, and as another example, the cluster labels of only one of the interactive text clusters may be determined as the topic labels of the target video.
According to the video tag determining method provided by the embodiment of the invention, the topic tag of the target video is determined based on the clustering tag of the interactive text cluster obtained by clustering by acquiring the interactive text issued for the target video and clustering the interactive text. Related topics spontaneously formed by a user aiming at the video can be tracked through the interactive text, and clustering labels which can be used for representing the related topics are obtained through clustering the interactive text, so that topics which are interested in the target video by the user can be effectively represented based on the topic labels determined by the clustering labels, and compared with the traditional video labels generated based on actors or themes, the topic labels of the video are more intelligent, finer and more numerous.
In one embodiment of the present invention, the step of clustering the interactive text based on the text feature similarity of the interactive text includes:
extracting text feature vectors of the interactive texts, determining the text feature similarities among the interactive texts based on the similarity among the text feature vectors of the interactive texts and/or based on the time difference between corresponding time stamps of the interactive texts in the target video, and clustering the interactive texts based on the text feature similarities.
Specifically, the text feature vectors can represent text features of the interactive texts in a vectorized form, so that similarity among the interactive texts on the text feature vectors can be used for accurately representing the similarity among the text features of the interactive texts. The text feature vectors of the interactive text and the method for calculating the similarity between the text feature vectors may refer to the content in the related art, which is not limited in the embodiment of the present invention, and as an example, the text feature vectors of the interactive text may be extracted by TF-IDF (Term Frequency/Inverse Document Frequency, a text feature extraction algorithm) or word2vec (a text feature extraction algorithm), and the similarity between the text feature vectors may be determined based on cosine distances or euclidean distances between the text feature vectors.
In addition, in practical application, part of the interactive text has a corresponding timestamp in the video. Taking an interactive text as an example of a barrage, the barrage can be displayed when a video is played to a specific image frame, so that the time point of the image frame displayed in the video by the barrage can be understood as the corresponding time stamp of the barrage in a target video.
In practical application, a user can issue interactive text based on the content actually played by the video. Thus, the bullet screen corresponding to the same or temporally similar image frames in the target video, whose corresponding timestamps are relatively close, tends to be a discussion of the episodes corresponding to those image frames, and thus generally characterizes the same topic. Thus, if the time difference between the corresponding time stamps in the target video is less than the threshold, i.e., the corresponding time stamps in the target video are closer, it may be considered that the interactive texts tend to have a higher similarity in text characteristics.
Therefore, in the embodiment of the invention, if the time difference of the corresponding time stamp of the interactive text in the target video is smaller, the interactive text can be divided into the same interactive text cluster.
In the embodiment of the invention, the text feature vector of the interactive text and/or the corresponding timestamp of the interactive text in the target video can be used as the basis for determining the text feature similarity of the interactive text according to actual requirements, and the implementation of the invention is not limited to the determination. As an example, the text feature similarity between the interactive texts may be determined based on the text feature vectors of the interactive texts or the corresponding time stamps of the interactive texts in the target video only, or different weights may be determined for the text feature vectors of the interactive texts and the corresponding time stamps thereof in the video, and the text feature similarity of the interactive texts may be determined based on combining the text feature vectors and the corresponding time stamps.
In the embodiment of the invention, the text feature similarity between the interactive texts is determined based on the similarity between the text feature vectors of the interactive texts and/or the time difference between the corresponding time stamps of the interactive texts in the target video, and compared with the traditional text similarity algorithm which only considers the text feature vectors, the time difference between the corresponding time stamps of the interactive texts in the target video is used as the basis for determining the text feature similarity based on the actual characteristics of the interactive texts, so that the method has higher comprehensiveness when determining the text feature similarity between the interactive texts, and is helpful for improving the clustering effect of the interactive texts on the basis, thereby improving the accuracy of the determined topic labels of the target video.
In one embodiment of the present invention, the step of determining the topic label of the target video based on the cluster label of the interactive text cluster includes:
determining a clustering label of each interactive text cluster as a topic label of the target video;
or alternatively, the first and second heat exchangers may be,
determining the quantity of the interactive texts contained in each interactive text cluster, and determining the cluster labels of the interactive text clusters with the quantity of the contained interactive texts not smaller than a preset value as topic labels of the target video.
In the embodiment of the invention, the clustering label of each interactive text cluster obtained by clustering can be used as the topic label of the target video, and in this case, each topic label can more comprehensively represent topics related to the target video.
In the embodiment of the invention, the number of the interactive texts contained in each interactive text cluster can be determined first, and only the number of the contained interactive texts is greater than a preset value, namely, the cluster labels of the interactive text clusters containing more interactive texts are determined as the topic labels of the target video.
Specifically, if the number of interactive texts included in one interactive text cluster is small, there may be a deviation between the meaning expressed by the interactive texts and the actual topic of the target video, so that the cluster label of the interactive text cluster cannot be used to characterize the topic related to the target video. If the number of the interactive texts contained in one interactive text cluster is large, more users are characterized to discuss the target video based on the topics corresponding to the interactive text cluster, so that the cluster label of the interactive text cluster can be used for representing the topics of the target video.
Therefore, only the cluster labels of the interactive text clusters containing more interactive texts can be used as topic labels of the target video, in which case, the cluster labels of the selected interactive texts can be regarded as topics with a certain discussion heat about the target video, so that the cluster labels are used as topic labels of the target video with higher representativeness.
It should be understood that, based on the number of interactive texts included in each interactive text cluster, the interactive text clusters may be ordered in order from more to less, and the cluster labels of the interactive text clusters with a preset proportion or a preset number are used as topic labels of the target video, which is substantially the same as the above example of the embodiment of the present invention.
It has been mentioned above that video tags can be applied to video recommendation tasks. The current video recommendation method mainly recommends videos based on actors and themes, and personalized information transmitted by users is difficult to use. And, many episodes suddenly appear with some stems after a long time of play. The conventional video recommendation method cannot carry out large-scale tracking on the video related stems, and cannot carry out video recommendation based on the stems spontaneously formed by the users, so that the content of the sheet which can be recommended to the users on the basis is single.
In view of this problem, after determining the topic labels of the videos based on the above embodiments of the present invention, a sheet may also be generated based on the video topics, so that the videos can be recommended to the user in a more personalized and diversified manner. Fig. 2 is another flow chart of the video tag determining method provided by the embodiment of the present invention, referring to fig. 2, in one embodiment of the present invention, the video tag determining method provided by the embodiment of the present invention further includes the following steps:
step S104: determining text similarity among topic labels determined for a plurality of target videos, and aggregating target videos corresponding to topic labels with text similarity not smaller than a similarity threshold value to obtain a video recommendation list.
Through the foregoing steps S101-S103, one or more video tags may be determined for each target video.
In the step, topic labels determined for a plurality of target videos can be obtained, the text similarity of the topic labels is calculated, and then the target videos corresponding to the topic labels with the text similarity not smaller than a similarity threshold are listed in the same video recommendation list or sheet.
In the embodiment of the invention, the text similarity between the topic labels can be calculated through a text similarity algorithm. Specifically, the topic labels can be converted into text feature vectors, and then the similarity between the text feature vectors is calculated, so that the text similarity between the topic labels is obtained. The text similarity algorithm used in this process may be selected based on actual requirements, which is not limited by the embodiments of the present invention.
As an example, a SimHash (a text similarity algorithm) algorithm or a BERT model may be used to calculate text similarity between topic tags.
The similarity threshold may be configured according to actual requirements, for example, the similarity threshold may be set to 1, and then target videos with the same topic label may be listed in a video recommendation list, or the similarity threshold may be set to a value smaller than 1, and then target videos with the similar topic label may be listed in a video recommendation list.
As one example, if topic tags a and B are determined for video 1, topic tags a and C are determined for video 2, and topic tags D and E are determined for video 3. With the similarity threshold set to 1, video 1 and video 2 both have the topic tag A and are therefore listed in a video recommendation list. In the case where the similarity threshold is determined to be a value smaller than 1, if the similarity between the topic tag C and the topic tag D is not smaller than the similarity threshold, the video 2 and the video 3 may be listed in one video recommendation list.
After videos with the same or similar topic labels are listed in one video recommendation list, the videos in the video recommendation list can be put together on a video website for users to watch, and the video recommendation list can be pushed to the users, so that users interested in the corresponding topics can find the interested videos more conveniently through the video recommendation list for watching.
In the embodiment of the invention, on the basis that the determined topic labels are more diversified and more personalized, the video of the same or similar topic labels is listed in a video recommendation list, and when the video recommendation method is applied to video recommendation on a video website, personalized recommendation can be performed for users with different interests, so that the use time of the users is prolonged, and the number of reserved users of the users is increased.
In one embodiment of the invention, the obtained interactive text can be screened first, and only the screened interactive text is clustered later. In an embodiment of the present invention, before the step S102, the method further includes:
classifying the interactive texts based on a pre-trained classification model, and eliminating the interactive texts of a preset category; the preset categories include: a specific behavior category and/or a negative rating category.
In practical application, part of interactive texts released aiming at a target video are not substantially related to the target video, if clustering is carried out based on the interactive texts, the obtained clustering labels of the interactive text clusters cannot reflect topics related to the target video, and the determined topic labels on the basis have no reference value and are not sufficient in accuracy.
In order to avoid interference of the interactive text on the process of determining the topic label of the target video, in the embodiment of the invention, the interactive text of a specific behavior type can be determined according to actual requirements and can not be used for representing the topic of the target video, and the interactive text is removed.
It should be appreciated that a particular behavior category may specifically include multiple subcategories, and that a particular subcategory may be determined based on actual circumstances. As one example, the interactive text for a particular behavior category may specifically include: interactive text with end-of-head punch, such as "XXX me get" or "XXX me go"; interactive text released on a specific date, such as "X month X day; nonsensical interactive text, i.e. interactive text consisting of letters randomly spelled or scrambled, etc. It should be appreciated that the specific behavior categories listed herein are by way of example only, and that the interactive text of the specific behavior categories that may be culled in actual application is not limited thereto.
In addition, the embodiment of the invention can also reject the interactive text of the negative evaluation category, wherein the interactive text of the negative evaluation category refers to the interactive text for carrying out negative evaluation on the video, for example, the interactive text containing abuse or sensitive vocabulary can be the interactive text of the negative evaluation category.
Specifically, since the video tags are generally applied to video retrieval and recommendation, if the clusters are performed based on the interactive text of the negative evaluation category, the tags of the obtained interactive text clusters cannot objectively reflect the video content, and cannot play a role in guiding the users interested in the topic to find the video.
Therefore, the interactive texts of the specific behavior and the negative evaluation category cannot have a good effect when being used for determining the topic labels, so that the interactive texts of the specific behavior and the negative evaluation category can be rejected as noise interactive texts, the other interactive texts of the other category are normal interactive texts, and the topic labels of the target video can be determined by clustering the normal interactive texts.
Specific sub-categories which can be contained in specific behavior categories to be eliminated can be predetermined, an interactive text corpus used for training is obtained, means calibration is carried out on the categories of the interactive texts in the corpus, the interactive texts marked as one or more specific behavior categories and the interactive texts marked as negative evaluation categories are obtained, and training is carried out on the classification models based on the marked interactive texts, so that the classifier used for identifying the interactive texts of the specific behavior categories and the negative evaluation categories in the embodiment of the invention is obtained. The specific classification algorithm adopted by the classifier and the training method of the classifier can be selected based on actual requirements, and the embodiment of the invention is not limited to the specific classification algorithm.
As an example, interactive text may be classified using a BERT (Bidirectional Encoder Representations from Transformers, a language model) model or a GPT (generated Pre-Training) model as a classifier.
In the embodiment of the invention, the interactive text is classified by the classification model, and the interactive text with the specific behavior category and/or the negative evaluation category is removed, so that the follow-up interactive text for determining the topic label of the video can be ensured to effectively represent the topic related to the video content, thereby improving the effectiveness of the determined topic label.
In one embodiment of the present invention, the obtained interactive text may be preprocessed, and then the processed interactive text may be clustered. In an embodiment of the present invention, before the step S102, the method further includes:
and eliminating non-ideographic characters in the interactive text.
In the embodiment of the invention, each interactive text can be specifically composed of characters, wherein part of the characters do not work on personalized contents to be expressed by a user, and the part of the characters can be regarded as non-ideographic characters. The character types specifically included in the non-ideographic characters can be selected according to actual requirements, and the embodiment of the present invention is not limited to this, and as an example, the non-ideographic characters may include: topics in the video platform used for interactive text, such as "# XXX decade #"; expression bag; a foreign language; an invalid symbol, such as "@ # is%".
As an example, if the interactive text before preprocessing is "#xxx ten years #xxx is too general [ smile ]", the interactive text obtained after removing the non-ideographic characters therein is specifically "XXX is too general".
In practical application, specific rules for eliminating non-ideographic characters can be determined according to practical requirements, filters are written according to the rules, and the interactive text is filtered based on the filters to obtain the interactive text after the non-ideographic characters are eliminated. As one example, a filter may be written for a particular rule with words that reject a particular part of speech, a particular symbol, or a string in a particular format.
According to the embodiment of the invention, the non-ideographic characters in the interactive text are removed by preprocessing the interactive text, so that the interference of the non-ideographic characters in the interactive text when the topic label of the video is determined based on the interactive text can be avoided, the effectiveness of the determined topic label is improved, and the calculation efficiency in the process of determining the topic label is improved.
Fig. 3 is a schematic diagram of a video tag determining method according to an embodiment of the present invention, and an exemplary description of the video tag determining method according to the embodiment of the present invention is provided below with reference to fig. 3.
First, an original interactive text published by a user is obtained, where the original interactive text may be specifically an original barrage/comment published by the user, and an example of a portion of the original interactive text is shown in part (a) of fig. 3.
After the original interactive text issued by the user is obtained, text preprocessing is needed to be carried out, non-ideographic characters in the original interactive text are removed, and invalid symbols, foreign languages, topics and expressions can be specifically removed for each original interactive text. Referring to fig. 3 (b), which is a preprocessed interactive text, it can be seen that information such as invalid symbols, foreign languages, expression packages, etc. in the original interactive text is removed.
After the text preprocessing of the interactive text is completed, the interactive text needs to be classified, and only normal text in the interactive text is reserved. Taking the interactive text as a barrage as an example, the noise barrage without information can be specifically removed, and a normal barrage is reserved, wherein the specific category of the noise barrage can refer to the content in the embodiment. Referring to part (c) of fig. 3, the first bullet screen is identified as a normal bullet screen, and the second bullet screen is identified as a noise bullet screen, so that the second bullet screen is rejected.
After the text preprocessing of the interactive text is completed and the normal text in the interactive text is reserved, clustering integration can be performed on the reserved interactive text, and for clustered interactive text clusters, the clustering topics of the interactive text clusters, namely the labels of the interactive text clusters, can be determined. Taking the interactive text as a barrage, referring to the part (d) in fig. 3, the left side is a clustering topic of the interactive text cluster, and the right side is a specific barrage contained in the interactive text cluster, so that the clustering topic can be seen to effectively represent the topic concerned by the interactive text in the interactive text cluster.
After the clustering of the interactive text is completed, the clustering topics of the interactive text cluster can be selected to be determined as the topic labels of the videos, and then the videos with the same or similar topic labels are listed in a video recommendation list, so that personalized recommendation of the videos can be performed for users with different interests.
Based on the same inventive concept, the embodiment of the invention also provides a method for displaying sheet information, which is specifically applied to a client, and fig. 4 is a flow diagram of the method for displaying sheet information provided by the embodiment of the invention, referring to fig. 4, and the method specifically includes the following steps:
step S401: acquiring sheet search information, and sending the sheet search information to a server so that the server determines a recommended sheet matched with the sheet search information based on the sheet search information; each video in the recommended sheet contains topic labels with the text feature similarity between the video and the search information not smaller than the preset similarity, and the topic label corresponding to each video is determined based on the clustering label of the obtained interactive text cluster after the interactive text issued by the video is clustered.
Step S402: and receiving the recommended sheet returned by the server and displaying the recommended sheet.
In the embodiment of the application, the sheet searching information specifically may refer to text input by a user in a client, and the text may be used for searching a sheet or searching a video. For the sheet search information, text feature similarity between the sheet search information and topic labels of a plurality of videos can be calculated, and for the videos containing topic labels with the text feature similarity not smaller than preset similarity with the sheet search information, the videos can be used as the videos in the recommended sheet recommended for the sheet search information. In the embodiment of the application, the magnitude of the preset similarity can be set based on actual requirements.
The topic labels of each video may refer to the description in any of the foregoing embodiments of the present application, and are not described herein.
On the basis that the topic label of the video can represent personalized information related to the video in a multiple and comprehensive manner, a recommended sheet matched with the sheet search information is determined through the text similarity between the sheet search information and the topic label, the video related to the sheet search information can be displayed in a more comprehensive and multiple manner, a user can find the interested video to watch based on the sheet search information, and therefore the method is beneficial to improving the using time of the user to a client and the number of reserved users.
Based on the same inventive concept, the embodiment of the present invention further provides a video tag determining apparatus, referring to fig. 5, including:
the acquisition module 501 is configured to acquire an interactive text published for a target video;
the clustering module 502 is configured to cluster the interactive text based on the text feature similarity of the interactive text, and obtain one or more interactive text clusters and a cluster label of each interactive text cluster;
the determining module 503 determines the topic label of the target video based on the cluster label of the interactive text cluster.
According to the video tag determining device provided by the embodiment of the invention, the topic tag of the target video is determined based on the clustering tag of the interactive text cluster obtained by clustering by acquiring the interactive text issued for the target video and clustering the interactive text. Related topics spontaneously formed by a user aiming at the video can be tracked through the interactive text, and clustering labels which can be used for representing the related topics are obtained through clustering the interactive text, so that topics which are interested in the target video by the user can be effectively represented based on the topic labels determined by the clustering labels, and compared with the traditional video labels generated based on actors or themes, the topic labels of the video are more intelligent, finer and more numerous.
In one embodiment of the present invention, the video tag determination apparatus further includes:
the aggregation module is used for determining text similarity among topic labels determined for a plurality of target videos, and aggregating target videos corresponding to topic labels with text similarity not smaller than a similarity threshold value to obtain a video recommendation list.
In one embodiment of the present invention, the clustering module 502 is specifically configured to:
extracting text feature vectors of the interactive texts, determining the text feature similarities among the interactive texts based on the similarity among the text feature vectors of the interactive texts and/or based on the time difference between corresponding time stamps of the interactive texts in the target video, and clustering the interactive texts based on the text feature similarities.
In one embodiment of the present invention, the video tag determination apparatus further includes:
the first rejecting module is used for classifying the interactive texts based on a pre-trained classifying model and rejecting the interactive texts with preset categories; the preset categories include: a specific behavior category and/or a negative rating category.
In one embodiment of the present invention, the video tag determination apparatus further includes:
and the second rejecting module is used for rejecting the non-ideographic characters in the interactive text.
In one embodiment of the present invention, the determining module 503 is specifically configured to:
determining a clustering label of each interactive text cluster as a topic label of the target video;
or alternatively, the first and second heat exchangers may be,
determining the quantity of the interactive texts contained in each interactive text cluster, and determining the cluster labels of the interactive text clusters with the quantity of the contained interactive texts not smaller than a preset value as topic labels of the target video.
Based on the same inventive concept, the embodiment of the invention also provides a method for displaying sheet information, which is applied to a client, and referring to fig. 6, the device comprises:
the acquiring module 601 is configured to acquire the sheet search information, and send the sheet search information to the server, so that the server determines a recommended sheet matching the sheet search information based on the sheet search information; each video in the recommended sheet contains topic labels with text feature similarity not smaller than preset similarity with search information, and the topic label corresponding to each video is determined based on a clustering label of an obtained interactive text cluster after the interactive text issued for the video is clustered;
and the display module 602 is configured to receive the recommended sheet returned by the server and display the recommended sheet.
On the basis that the topic label of the video can represent personalized information related to the video in a multiple and comprehensive manner, a recommended sheet matched with the sheet search information is determined through the text similarity between the sheet search information and the topic label, the video related to the sheet search information can be displayed in a more comprehensive and multiple manner, a user can find the interested video to watch based on the sheet search information, and therefore the method is beneficial to improving the using time of the user to a client and the number of reserved users.
The embodiment of the present invention further provides an electronic device, as shown in fig. 7, including a processor 701, a communication interface 702, a memory 703 and a communication bus 704, where the processor 701, the communication interface 702, and the memory 703 perform communication with each other through the communication bus 704,
a memory 703 for storing a computer program;
the processor 701 is configured to execute the program stored in the memory 703, and implement the following steps:
acquiring an interactive text issued aiming at a target video;
based on the text feature similarity of the interactive text, clustering the interactive text to obtain one or more interactive text clusters and clustering labels of the interactive text clusters.
And determining topic labels of the target video based on the cluster labels of the interactive text clusters.
The communication bus mentioned by the above terminal may be a peripheral component interconnect standard (Peripheral Component Interconnect, abbreviated as PCI) bus or an extended industry standard architecture (Extended Industry Standard Architecture, abbreviated as EISA) bus, etc. The communication bus may be classified as an address bus, a data bus, a control bus, or the like. For ease of illustration, the figures are shown with only one bold line, but not with only one bus or one type of bus.
The communication interface is used for communication between the terminal and other devices.
The memory may include random access memory (Random Access Memory, RAM) or non-volatile memory (non-volatile memory), such as at least one disk memory. Optionally, the memory may also be at least one memory device located remotely from the aforementioned processor.
The processor may be a general-purpose processor, including a central processing unit (Central Processing Unit, CPU for short), a network processor (Network Processor, NP for short), etc.; but also digital signal processors (Digital Signal Processor, DSP for short), application specific integrated circuits (Application Specific Integrated Circuit, ASIC for short), field-programmable gate arrays (Field-Programmable Gate Array, FPGA for short) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components.
In yet another embodiment of the present invention, a computer readable storage medium is provided, where a computer program is stored, the computer program, when executed by a processor, implements the video tag determination method according to any one of the above embodiments.
In yet another embodiment of the present invention, a computer program product containing instructions that, when run on a computer, cause the computer to perform the video tag determination method of any of the above embodiments is also provided.
In the above embodiments, it may be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented in software, may 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 loaded and executed on a computer, produces a flow or function in accordance with embodiments of the present invention, in whole or in part. The computer may be a general purpose computer, a special purpose computer, a computer network, or other programmable apparatus. The computer instructions may be stored in or transmitted from one computer-readable storage medium to another, for example, by wired (e.g., coaxial cable, optical fiber, digital Subscriber Line (DSL)), or wireless (e.g., infrared, wireless, microwave, etc.). The computer readable storage medium may be any available medium that can be accessed by a computer or a data storage device such as a server, data center, etc. that contains an integration of one or more available media. The usable medium may be a magnetic medium (e.g., floppy Disk, hard Disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., solid State Disk (SSD)), etc.
It is noted that relational terms such as first and second, and the like are used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Moreover, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one … …" does not exclude the presence of other like elements in a process, method, article, or apparatus that comprises the element.
In this specification, each embodiment is described in a related manner, and identical and similar parts of each embodiment are all referred to each other, and each embodiment mainly describes differences from other embodiments. In particular, for the video tag determination apparatus, the electronic device, and the readable storage medium embodiments, since they are substantially similar to the method embodiments, the description is relatively simple, and the relevant points are referred to in the partial description of the method embodiments.
The foregoing description is only of the preferred embodiments of the present invention and is not intended to limit the scope of the present invention. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention are included in the protection scope of the present invention.

Claims (11)

1. A method for determining a video tag, comprising:
acquiring an interactive text issued aiming at a target video;
clustering the interactive texts based on the text feature similarity of the interactive texts to obtain one or more interactive text clusters and clustering labels of the interactive text clusters;
and determining the topic label of the target video based on the cluster label of the interactive text cluster.
2. The method according to claim 1, wherein the method further comprises:
determining text similarity among the topic labels determined for a plurality of target videos, and aggregating target videos corresponding to topic labels with the text similarity not smaller than a similarity threshold value to obtain a video recommendation list.
3. The method of claim 1, wherein the step of clustering the interactive text based on the text feature similarity of the interactive text comprises:
Extracting text feature vectors of the interactive texts, determining text feature similarities among the interactive texts based on the similarity among the text feature vectors of the interactive texts and/or based on time differences among corresponding time stamps of the interactive texts in the target video, and clustering the interactive texts based on the text feature similarities.
4. The method of any of claims 1-3, wherein prior to clustering the interactive text based on the text feature similarity of the interactive text, further comprising:
classifying the interactive texts based on a pre-trained classification model, and eliminating the interactive texts of a preset category; the preset categories include: a specific behavior category and/or a negative rating category.
5. The method of any of claims 1-3, wherein prior to clustering the interactive text based on the text feature similarity of the interactive text, further comprising:
and eliminating the non-ideographic characters in the interactive text.
6. The method of any of claims 1-3, wherein the step of determining the topic label of the target video based on the cluster labels of the interactive text clusters comprises:
Determining the clustering label of each interactive text cluster as the topic label of the target video;
or alternatively, the first and second heat exchangers may be,
determining the quantity of the interactive texts contained in each interactive text cluster, and determining the cluster labels of the interactive text clusters, the quantity of which is not smaller than a preset value, as the topic labels of the target video.
7. A method for displaying sheet information, which is applied to a client, the method comprising:
acquiring sheet search information, and sending the sheet search information to a server so that the server determines a recommended sheet matched with the sheet search information based on the sheet search information; each video in the recommended sheet contains topic labels with text feature similarity not smaller than preset similarity with the search information, and the topic label corresponding to each video is determined based on clustering labels of an obtained interactive text cluster after the interactive text issued for the video is clustered;
and receiving the recommended sheet returned by the server and displaying the recommended sheet.
8. A video tag determination apparatus, comprising:
The acquisition module is used for acquiring the interactive text issued aiming at the target video;
the clustering module is used for clustering the interactive text based on the text feature similarity of the interactive text to obtain one or more interactive text clusters and clustering labels of the interactive text clusters;
and the determining module is used for determining the topic label of the target video based on the cluster label of the interactive text cluster.
9. The utility model provides a display device of sheet information, which is characterized in that is applied to the customer end, includes:
the acquisition module is used for acquiring the sheet searching information and sending the sheet searching information to a server so that the server can determine a recommended sheet matched with the sheet searching information based on the sheet searching information; each video in the recommended sheet contains topic labels with text feature similarity not smaller than preset similarity with the search information, and the topic label corresponding to each video is determined based on clustering labels of an obtained interactive text cluster after the interactive text issued for the video is clustered;
and the display module is used for receiving the recommended sheet returned by the server and displaying the recommended sheet.
10. The electronic equipment is characterized by comprising a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory are communicated with each other through the communication bus;
a memory for storing a computer program;
a processor for carrying out the method steps of any one of claims 1-6 or 7 when executing a program stored on a memory.
11. A computer-readable storage medium, characterized in that the computer-readable storage medium has stored therein a computer program which, when executed by a processor, implements the method steps of any of claims 1-6 or 7.
CN202310909433.8A 2023-07-24 2023-07-24 Video tag determining method, sheet information displaying method and related devices Pending CN116958971A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202310909433.8A CN116958971A (en) 2023-07-24 2023-07-24 Video tag determining method, sheet information displaying method and related devices

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202310909433.8A CN116958971A (en) 2023-07-24 2023-07-24 Video tag determining method, sheet information displaying method and related devices

Publications (1)

Publication Number Publication Date
CN116958971A true CN116958971A (en) 2023-10-27

Family

ID=88452415

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202310909433.8A Pending CN116958971A (en) 2023-07-24 2023-07-24 Video tag determining method, sheet information displaying method and related devices

Country Status (1)

Country Link
CN (1) CN116958971A (en)

Similar Documents

Publication Publication Date Title
CN107766371B (en) Text information classification method and device
CN107463605B (en) Method and device for identifying low-quality news resource, computer equipment and readable medium
US20180232362A1 (en) Method and system relating to sentiment analysis of electronic content
CN112148889A (en) Recommendation list generation method and device
CN110888990A (en) Text recommendation method, device, equipment and medium
CN110263248A (en) A kind of information-pushing method, device, storage medium and server
CN107832338B (en) Method and system for recognizing core product words
JP2013517563A (en) User communication analysis system and method
CN112507167A (en) Method and device for identifying video collection, electronic equipment and storage medium
US10216989B1 (en) Providing additional information for text in an image
CN111782793A (en) Intelligent customer service processing method, system and equipment
CN111291551A (en) Text processing method and device, electronic equipment and computer readable storage medium
CN111737607B (en) Data processing method, device, electronic equipment and storage medium
CN112487181B (en) Keyword determination method and related equipment
CN110489740B (en) Semantic analysis method and related product
CN111783424A (en) Text clause dividing method and device
US20230090601A1 (en) System and method for polarity analysis
CN116108181A (en) Client information processing method and device and electronic equipment
CN106294292B (en) Chapter catalog screening method and device
CN116958971A (en) Video tag determining method, sheet information displaying method and related devices
CN114996571A (en) Information pushing method and device, storage medium and electronic equipment
CN114741501A (en) Public opinion early warning method and device, readable storage medium and electronic equipment
CN114911936A (en) Model training and comment recognition method and device, electronic equipment and medium
CN114255067A (en) Data pricing method and device, electronic equipment and storage medium
CN110825847B (en) Method and device for identifying intimacy between target people, electronic equipment and storage medium

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination