CN112989118A - Video recall method and device - Google Patents
Video recall method and device Download PDFInfo
- Publication number
- CN112989118A CN112989118A CN202110156410.5A CN202110156410A CN112989118A CN 112989118 A CN112989118 A CN 112989118A CN 202110156410 A CN202110156410 A CN 202110156410A CN 112989118 A CN112989118 A CN 112989118A
- Authority
- CN
- China
- Prior art keywords
- recalled
- target
- video
- label
- vector
- 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.)
- Granted
Links
Images
Classifications
-
- 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
- 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
- G06F16/735—Filtering based on additional data, e.g. user or group profiles
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Multimedia (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Library & Information Science (AREA)
- Computational Linguistics (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The embodiment of the invention provides a video recall method and a video recall device, wherein the method comprises the following steps: acquiring a target label set of a target video and a to-be-recalled label set of a to-be-recalled video; acquiring a target label vector corresponding to a target label in the target label set, and acquiring a to-be-recalled label vector corresponding to a to-be-recalled label in the to-be-recalled label set; summing the target label vectors to obtain target label set vectors, and summing the to-be-recalled label vectors to obtain to-be-recalled label set vectors; calculating the similarity between the target label set vector and the to-be-recalled label set vector; and recalling videos from the videos to be recalled corresponding to the label set vectors to be recalled, of which the similarity meets the preset condition. The embodiment of the invention takes the label set as a retrieval unit to recall the video, can improve the correlation between the recalled video and the target video, and further improves the quality of the recalled video.
Description
Technical Field
The present invention relates to the field of computer technologies, and in particular, to a video recall method, a video recall apparatus, an electronic device, and a computer-readable storage medium.
Background
With the rapid development of informatization, various internet contents come up endlessly, and browsing videos becomes an important leisure mode for people, but generally, users are interested in only part of videos, and videos are recommended to the users in a personalized mode in order to save time of the users. Specifically, the video recommending process includes firstly recalling similar videos based on target videos browsed by the user, and then selecting videos from the recalled similar videos to recommend to the user.
However, when similar videos are recalled at present, it is sometimes difficult to acquire the similar videos with the target video, which causes the quality of the recalled videos to be reduced.
Disclosure of Invention
The embodiment of the invention aims to provide a video recall method, which is used for recalling videos based on a video label set of a target video and improving the quality of the recalled videos. The specific technical scheme is as follows:
in a first aspect of the present invention, there is provided a video recall method, including:
acquiring a target label set of a target video and a to-be-recalled label set of a to-be-recalled video;
acquiring a target label vector corresponding to a target label in the target label set, and acquiring a to-be-recalled label vector corresponding to a to-be-recalled label in the to-be-recalled label set; the method comprises the steps that a label vector is obtained from a vector set, and the vector set is generated according to video labels which have a co-occurrence relation and replace synonyms with designated synonyms corresponding to the synonyms;
summing the target label vectors to obtain target label set vectors, and summing the to-be-recalled label vectors to obtain to-be-recalled label set vectors;
calculating the similarity between the target label set vector and the to-be-recalled label set vector;
and recalling videos from the videos to be recalled corresponding to the label set vectors to be recalled, of which the similarity meets the preset condition.
Optionally, the obtaining a target tag vector corresponding to a target tag in the target tag set and obtaining a to-be-recalled tag vector corresponding to a to-be-recalled tag in the to-be-recalled tag set include:
obtaining a set of orientation quantities; the vector set comprises tag vectors corresponding to video tags;
and acquiring a target label vector corresponding to a target label in the target label set from the vector set, and acquiring a to-be-recalled label vector corresponding to a to-be-recalled label in the to-be-recalled label set from the vector set.
Optionally, before the obtaining the set of orientation quantities, the method further comprises:
acquiring video tags with a co-occurrence relation, wherein the co-occurrence relation is used for representing the probability of the simultaneous occurrence of the video tags;
when synonyms exist in the video tags, replacing the synonyms in the video tags with the designated synonyms corresponding to the synonyms;
and inputting a preset word vector model by adopting the video tag after replacing the synonym, obtaining a corresponding tag vector, and storing the tag vector into a vector set.
Optionally, when there is a synonym in the video tag, replacing the synonym in the video tag with a designated synonym corresponding to the synonym includes:
when the matched synonym exists in a preset synonym table in the video label, replacing the synonym in the video label with the designated synonym corresponding to the synonym.
Optionally, the obtaining, from the vector set, a target tag vector corresponding to a target tag in the target tag set, and obtaining, from the vector set, a to-be-recalled tag vector corresponding to a to-be-recalled tag in the to-be-recalled tag set include:
when the vector set does not have a target label vector corresponding to a target label in the target label set, performing word segmentation on the target label to obtain a sub-word, obtaining a target sub-label vector corresponding to the sub-word of the target label from the vector set, and summing the target sub-label vectors to obtain a target label vector, an
When the vector set does not have the to-be-recalled label vector corresponding to the to-be-recalled label in the to-be-recalled label set, performing word segmentation on the to-be-recalled label to obtain a sub-word, obtaining the to-be-recalled sub-label vector corresponding to the sub-word of the to-be-recalled label from the vector set, and summing the to-be-recalled sub-label vectors to obtain the to-be-recalled label vector.
Optionally, the recalling a video from the to-be-recalled videos corresponding to the to-be-recalled label set vector whose similarity satisfies a preset condition includes:
acquiring a target to-be-recalled label set corresponding to the to-be-recalled label set vector with the similarity larger than a preset similarity threshold and sorted in the top N, wherein N is a positive integer; wherein the target to-be-recalled labelset comprises a target labelset of the target video;
creating a similar video index list according to the target to-be-recalled label set;
acquiring a related video to be recalled as a target video to be recalled according to the similar video index list;
determining similar videos from the target video to be recalled based on preset posterior indexes; the posterior indicator includes click through rate.
Optionally, the target video and the to-be-recalled video have channel identifiers, and the acquiring the to-be-recalled video associated with the target to-be-recalled label set as the target to-be-recalled video includes:
and acquiring the video to be recalled which is associated with the target video to be recalled label set and has the same channel identification as that of the target video as the target video to be recalled.
In a second aspect implemented by the present invention, there is also provided a video recall apparatus, comprising:
the system comprises a tag set acquisition module, a target video retrieval module and a target video retrieval module, wherein the tag set acquisition module is used for acquiring a target tag set of a target video and a to-be-recalled tag set of a to-be-recalled video;
the tag vector acquisition module is used for acquiring a target tag vector corresponding to a target tag in the target tag set and acquiring a to-be-recalled tag vector corresponding to a to-be-recalled tag in the to-be-recalled tag set; the method comprises the steps that a label vector is obtained from a vector set, and the vector set is generated according to video labels which have a co-occurrence relation and replace synonyms with designated synonyms corresponding to the synonyms;
the vector summation module is used for summing the target label vectors to obtain target label set vectors and summing the to-be-recalled label vectors to obtain to-be-recalled label set vectors;
the similarity calculation module is used for calculating the similarity between the target label set vector and the to-be-recalled label set vector;
and the video recall module is used for recalling videos from the videos to be recalled corresponding to the label set vectors to be recalled, wherein the similarity of the videos to be recalled meets the preset condition.
In yet another aspect of the present invention, there is also provided a computer-readable storage medium having stored therein instructions, which when run on a computer, cause the computer to perform any of the video recall methods described above.
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 recall methods described above.
The video recall method provided by the embodiment of the invention obtains the target label set of the target video and the to-be-recalled label set of the to-be-recalled video, and obtaining a target label vector corresponding to a target label in the target label set and obtaining a to-be-recalled label vector corresponding to a to-be-recalled label in the to-be-recalled label set, then summing the target tag vectors to obtain target tag set vectors and summing the to-be-recalled tag vectors to obtain to-be-recalled tag set vectors, calculating the similarity between the target tag set vectors and the to-be-recalled tag set vectors, recalling videos from the to-be-recalled videos corresponding to the to-be-recalled tag set vectors with the similarity meeting preset conditions, and recalling the videos by taking the tag sets as retrieval units, the correlation between the recalled video and the target video can be improved, and therefore the quality of the recalled video is improved.
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 flowchart illustrating steps of a video recall method according to an embodiment of the present invention;
fig. 2 is a flowchart of a step of generating a tag vector according to an embodiment of the present invention;
FIG. 3 is a flowchart illustrating steps of another video recall method according to an embodiment of the present invention;
FIG. 4 is a block diagram of a video recall device according to an embodiment of the present invention;
fig. 5 is a block diagram of an electronic device provided in 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 drawings in the embodiments of the present invention.
Referring to fig. 1, which is a flowchart illustrating steps of a video recall method according to an embodiment of the present invention, as shown in fig. 1, the method may specifically include the following steps:
In an embodiment of the present invention, the target video may refer to a video historically viewed or currently viewed on the user video website. The video to be recalled can refer to all videos on a video website, and can also be videos screened out through preset screening conditions, wherein the screening conditions can be time, on-demand volume, browsing duration and the like, and can also be video types which are not interested by a user browsing a target video, for example, if the user is not interested in a horror video, the horror video can be removed in advance, so that the video to be recalled does not contain the horror video. For example, if the user is browsing video a on a video website, video a is the target video.
The tag set comprises a set of one or more video tags, the video tags are obtained by decomposing keywords of video content in different dimensions, so that description information of the video is formed, for example, the video tags can be categories, partitions, titles, brief descriptions and the like, and the video can be efficiently searched and accurately positioned through the video tags. For example, for a children's cartoon, the video tags in its tag set may include: cartoon, kid, daily, familiarity, and friendship, for a detective movie, the video tags in its tag set may include: movies, crime, brainstorming, love and comedy.
In embodiments of the present invention, each video has a corresponding set of tags, each video is represented by a unique set of tags, and the same set of tags may represent multiple videos. For example, in the embodiment of the present invention, the target video and the video to be recalled respectively have their corresponding unique tag sets, which are the target tag set and the video to be recalled respectively.
In the embodiment of the present invention, the video tags in the tag set may be represented by tag vectors, specifically, the tag vectors are a set of data that maps the video tags to real numbers, and the video tags are converted into a set of data that can be recognized by a machine, which is beneficial to subsequent analysis processing. Specifically, a text matching algorithm, such as word2vec, TFIDF, LDA, and the like, may be used to obtain a target label vector corresponding to a target label in the target label set, and a to-be-recalled label vector corresponding to a to-be-recalled label in the to-be-recalled label set.
The target label vector and the label vector to be recalled are label vectors, the label vectors are obtained from a vector set, and the vector set is generated according to video labels which have a co-occurrence relation and replace synonyms with designated synonyms corresponding to the synonyms. Specifically, the co-occurrence relationship is a probability that characterizes the simultaneous occurrence of the video tags, for example, two video tags of students and teachers have a high probability of occurring simultaneously, so that the two video tags can be considered as having the co-occurrence relationship, and in addition, in order to avoid processing too many similar words, the embodiment of the present invention uniformly determines the video tags that are synonyms as designated synonyms, for example, the video tags "company" and "enterprise" are synonyms, and uniformly determines the two video tags that are synonyms as designated synonyms "company".
In the embodiment of the present invention, after the tag vectors corresponding to the video tags in the tag set are obtained, the tag set vectors corresponding to the video may be generated based on the sum of the tag vectors, so that one video may be represented by one tag set vector.
Specifically, a target tag set vector of the target video can be obtained by summing target tag vectors in the target tag set, and a target tag set vector of the video to be recalled can be obtained by summing the target tag vectors in the tag set to be recalled.
And 104, calculating the similarity between the target label set vector and the to-be-recalled label set vector.
And 105, recalling videos from the videos to be recalled corresponding to the label set vectors to be recalled, of which the similarity meets the preset condition.
In the embodiment of the present invention, the similarity between the target tag set and the to-be-recalled tag set may be determined according to the similarity between the target tag set vector and the to-be-recalled tag set vector, and when the similarity is greater than a preset similarity threshold, it indicates that there is a certain similarity between the target tag set and the to-be-recalled tag set, so that a video is recalled from the to-be-recalled video based on the to-be-recalled tag set whose similarity is greater than the preset similarity threshold, and a video with a higher similarity to the target video may be recalled.
In the video recall method, a target label set of a target video and a to-be-recalled label set of a to-be-recalled video are obtained, a target label vector corresponding to a target label in the target label set and a to-be-recalled label vector corresponding to a to-be-recalled label in the to-be-recalled label set are obtained, then the target label vectors are summed to obtain a target label set vector and the to-be-recalled label vector are summed to obtain a to-be-recalled label set vector, the similarity between the target label set vector and the to-be-recalled label set vector is calculated, a video is recalled from the to-be-recalled video corresponding to the to-be-recalled label set vector with the similarity meeting a preset condition, and the label set is taken as a retrieval unit for video recall, so that the correlation between the recalled video and the target video can be improved, and the quality of the recalled video is improved.
In an exemplary embodiment of the present invention, the step 102, acquiring a target tag vector corresponding to a target tag in the target tag set, and acquiring a to-be-recalled tag vector corresponding to a to-be-recalled tag in the to-be-recalled tag set, includes:
obtaining a set of orientation quantities; the vector set comprises tag vectors corresponding to video tags;
and acquiring a target label vector corresponding to a target label in the target label set from the vector set, and acquiring a to-be-recalled label vector corresponding to a to-be-recalled label in the to-be-recalled label set from the vector set.
The word vector model can be word2vec, TFIDF or LDA algorithm. In the embodiment of the invention, the video tags are input into the word vector model in advance for training, so that the tag vectors corresponding to the video tags are generated, and then the tag vectors are stored into the vector set. Then, the label vector corresponding to the video label can be obtained from the vector set.
Specifically, after a target tag set of a target video and a to-be-recalled tag set of a to-be-recalled video are obtained, a vector set is obtained, then a target tag vector corresponding to a target tag in the target tag set can be obtained from the vector set, and a to-be-recalled tag vector corresponding to a to-be-recalled tag in the to-be-recalled tag set is obtained from the vector set.
In the above exemplary embodiment, the vector set is obtained by training the word vector model through the video tags, so that the target tag vector corresponding to the target tag in the target tag set and the to-be-recalled tag vector corresponding to the to-be-recalled tag in the to-be-recalled tag set are obtained from the vector set to calculate the similarity for recalling the video, and the relevance and accuracy of the recalled video can be improved.
In an exemplary embodiment of the invention, before the obtaining the set of orientation quantities, the method further comprises:
The co-occurrence relationship (tag sequence) is a probability that video tags appear simultaneously, for example, the video tags "kid" and "cartoon" often appear simultaneously, and the "kid" and the "cartoon" can be regarded as having a co-occurrence relationship, and for example, the video tags "game character" and "blood volume value" can be regarded as having a co-occurrence relationship between the "game character" and the "blood volume value". In the embodiment of the invention, the existing video tags are obtained, and the video tags which do not contain the co-occurrence relationship are removed from the existing video tags.
In a specific implementation, a large number of synonyms may exist in a video tag, and the synonyms refer to the same meaning, and if there is no difference in constructing a tag vector of the video tag, it cannot be guaranteed that an optimal video will be retained as a similar video of a target video, which causes a problem of quality degradation of a recalled similar video, so that synonyms existing in the video tag will be replaced in the embodiment of the present invention.
In an exemplary embodiment of the present invention, the step 202, when there is a synonym in the video tag, replacing the synonym in the video tag with a designated synonym corresponding to the synonym, includes: when the matched synonym exists in a preset synonym table in the video label, replacing the synonym in the video label with the designated synonym corresponding to the synonym.
Specifically, the embodiment of the present invention may determine whether the video tag includes a synonym according to a preset synonym table. For example, assuming that there are video tags including "prize and reward", "company encounter", "reward", which are determined to be synonyms according to the preset synonym table, and a designated synonym of these synonyms is set as "reward" in the preset synonym table, the "company encounter" and "reward" may be replaced with the "company reward" and "reward". In the above exemplary embodiment, synonyms in the video tag are normalized to be in the same representation form, thereby improving the accuracy of tag vector expression.
And 203, inputting a preset word vector model by adopting the video tags after replacing the synonyms to obtain corresponding tag vectors, and storing the tag vectors into a vector set.
In the embodiment of the invention, the video tags which do not contain the co-occurrence relationship in the video tags are removed, synonyms existing in the video tags are normalized to be in the same representation form, and then the video tags are trained in a word vector model, such as word2vec, to obtain tag vectors, and the tag vectors are stored in a vector set for later use.
In the above exemplary embodiment, the label vectors obtained by training input into the word vector model in the same expression form through normalization of the co-occurrence relationship and the synonyms can obtain the label vectors with strong correlation and high expression accuracy.
In an exemplary embodiment of the present invention, the obtaining, from the vector set, a target tag vector corresponding to a target tag in the target tag set, and obtaining, from the vector set, a to-be-recalled tag vector corresponding to a to-be-recalled tag in the to-be-recalled tag set includes:
when the vector set does not have a target label vector corresponding to a target label in the target label set, performing word segmentation on the target label to obtain a sub-word, obtaining a target sub-label vector corresponding to the sub-word of the target label from the vector set, and summing the target sub-label vectors to obtain a target label vector, an
When the vector set does not have the to-be-recalled label vector corresponding to the to-be-recalled label in the to-be-recalled label set, performing word segmentation on the to-be-recalled label to obtain a sub-word, obtaining the to-be-recalled sub-label vector corresponding to the sub-word of the to-be-recalled label from the vector set, and summing the to-be-recalled sub-label vectors to obtain the to-be-recalled label vector.
The video tags may include long-tail tags. Specifically, the long-tail tag refers to a video tag containing a keyword, and usually consists of 2-3 words or is a phrase, and usually further explanation and definition of the keyword. Such as: the keyword 'red bean' can be used as a video label, and the 'nutritional value of red bean' can also be used as a long-tail label containing the keyword. In practice, an accurate video can be located by the long-tail tag, but at the same time, there is a problem that any video cannot be located by the long-tail tag.
In the embodiment of the invention, when the video tags are trained based on the word vector model to obtain the tag vectors, a reasonable minimum word frequency is set for the word vectors, so that some video tags with a small occurrence number, such as long-tail tags, can be filtered, and therefore, the vector set may not contain or only contain a small number of vector tags with long-tail tags.
In view of the above problems, in the embodiment of the present invention, if a tag vector corresponding to a video tag exists in a vector set, the tag vector is directly obtained, and if a tag vector corresponding to a video tag does not exist in the vector set, it indicates that the video tag may be a long-tail tag, a word segmentation may be performed on the video tag to obtain a sub-word, then sub-tag vectors corresponding to sub-words of the video tag are obtained from the vector set, and the sub-tag vectors are summed to obtain a tag vector corresponding to the long-tail tag.
In the above exemplary embodiment, the video tags which are long-tail tags are segmented, so that sub-tag vectors corresponding to the long-tail tags are obtained and summed to obtain tag vectors corresponding to the long-tail tags, and the video tags can obtain corresponding tag vectors, so that the overall coverage of the recalled video can be improved.
In an exemplary embodiment of the present invention, in step 105, recalling a video from the to-be-recalled videos corresponding to the to-be-recalled tag set vector whose similarity satisfies a preset condition includes:
acquiring a target to-be-recalled label set corresponding to the to-be-recalled label set vector with the similarity larger than a preset similarity threshold and sorted in the top N, wherein N is a positive integer; wherein the target to-be-recalled labelset comprises a target labelset of the target video;
creating a similar video index list according to the target to-be-recalled label set;
acquiring a related video to be recalled according to the similar video index list to be used as a target video to be recalled;
determining similar videos from the target video to be recalled based on preset posterior indexes; the posterior indicator includes click through rate.
In the embodiment of the present invention, the similarity may be a cosine similarity, the preset similarity threshold may be set to 0.8, and N may be set to 5. In the embodiment of the invention, the cosine similarity between the target tag set vector and the to-be-recalled tag set vector is calculated, the cosine similarity is kept to be more than 0.8 according to a descending order, the to-be-recalled tag set corresponding to the to-be-recalled tag set vector in the top 5 is taken as the target to-be-recalled tag set, then the to-be-recalled video associated with the target to-be-recalled tag set is taken as the target to-be-recalled video, a similar video is further determined from the target to-be-recalled video based on a posterior index, and the video recall is completed.
The posterior index may be a Click-through Rate (CTR), and specifically, the Click-through Rate refers to a ratio of the number of times a certain video on a video website is clicked to the number of times the certain video is displayed, i.e., clicks/views, which is a percentage and reflects a degree of interest of the certain video on the video website. For example, assuming that a video window of video B is displayed on a certain web page of a video website, if one of 10 people clicks the video window of video B on an open web page, the click rate of video B is 10%. Of course, in addition to the click rate, a collection rate, a comment rate, a sharing rate, or a forwarding rate may also be selected as the posterior index, which is not limited in the embodiment of the present invention.
It should be noted that the video to be recalled may include the target video, and therefore, the target set of tags in the target set of tags to be recalled may include the target tag.
In the embodiment of the present invention, the target to-be-recalled tab set may be one or more tab sets, and may be combined into one target to-be-recalled tab set list, and the target to-be-recalled tab set list may include the target tab set of the target video, and then a similar video index list may be reversely created according to the target to-be-recalled tab set list, so as to determine the finally recalled similar video based on the similar video index list. Specifically, the similar video index list may include a video identifier, and the video identifier may be a name or a number of the video to be recalled, so that the video to be recalled may be acquired through the similar video index list.
Specifically, first, a video set V0 (target to-be-recalled video) corresponding to a tag set T0 in a target to-be-recalled tag set list is obtained, where T0 is the target tag set of the target video, V _ i represents a similar video index list (target video), and X videos with the highest posterior index (e.g., CTR) in V0 are taken as target videos to be written into V _ i; if the number of videos in V _ i is less than Y, sequentially taking out the video set V1 corresponding to the tag set T1 after T0 in the target to-be-recalled tag set list, taking the X videos with the highest CTR, writing the X videos into V _ i, and repeating the steps until the number of videos in V _ i meets Y.
In the above exemplary embodiment, the target to-be-recalled tag set corresponding to the to-be-recalled tag set vector with the similarity greater than the preset similarity threshold and ranked in the top N is obtained, then the associated target to-be-recalled video is determined based on the target to-be-recalled tag set, finally the similar video recalled for the target video is determined based on the posterior index, and the video is recalled through a plurality of combined screening conditions, so that the relevance of the recalled video is improved.
In an exemplary embodiment of the present invention, the acquiring the to-be-recalled video associated with the target to-be-recalled label set as the target to-be-recalled video includes:
and acquiring the video to be recalled which is associated with the target video to be recalled label set and has the same channel identification as that of the target video as the target video to be recalled.
Where each video may have a corresponding channel identification, such as "sports", "entertainment", "current events", etc. In the embodiment of the present invention, when determining a target video to be recalled, the target video to be recalled needs to be associated with a target set of tags to be recalled, and needs to have the same channel identifier as that of the target video. For example, assuming that the channel of the target video is identified as "entertainment", only the video to be recalled whose channel is identified as "entertainment" can be taken as the target video to be recalled.
In the above exemplary embodiment, the video to be recalled with the same channel identification as that of the target video is taken as the target video to be recalled, thereby ensuring the relevance of the final recalled similar video to the target video before.
Referring to fig. 3, which is a flowchart illustrating steps of another video recall method provided in an embodiment of the present invention, as shown in fig. 3, the method may specifically include the following steps:
301, acquiring video tags with a co-occurrence relationship, wherein the co-occurrence relationship is used for representing the probability of the video tags appearing simultaneously;
In the above video recall method, video tags having a co-occurrence relationship are acquired and synonyms existing in the video tags are normalized into a same representation form, then the video tags are used as training data, word2vec or other algorithms are used to obtain tag vectors and store the tag vectors into a vector set, then when similar videos of a target video are recalled, target tag vectors corresponding to target tags in the target tag set can be acquired from the vector set and summed to obtain target tag set vectors, and to-be-recalled tag vectors corresponding to-be-recalled tags in the to-be-recalled tag set are acquired from the vector set and summed to obtain to-be-recalled tag set vectors, then the similarity between the target tag set vectors and the to-be-recalled tag set vectors is calculated, and finally, based on the similarity, the to-be-recalled tag set vectors meeting preset conditions, are corresponding to the to-be-recalled video, and determining similar videos of the target video. Therefore, the embodiment of the invention improves the overall coverage of the label vectors of the video labels, thereby improving the accuracy of the label vector recall result, simultaneously proposes synonym replacement in consideration of the phenomenon that a large number of videos have video labels with a plurality of synonyms, takes the label set as the similarity index list of the retrieval unit, reversely creates similar videos with video granularity according to the service strategies such as posterior indexes and the like, improves the posterior indexes of the recalled similar videos on the basis of ensuring the correlation, and achieves the purpose of comprehensively improving the quality of the recalled videos.
It should be noted that, for simplicity of description, the method embodiments are described as a series of acts or combination of acts, but those skilled in the art will recognize that the present invention is not limited by the illustrated order of acts, as some steps may occur in other orders or concurrently in accordance with the embodiments of the present invention. Further, those skilled in the art will appreciate that the embodiments described in the specification are presently preferred and that no particular act is required to implement the invention.
Referring to fig. 4, which is a block diagram of a video recall apparatus provided in an embodiment of the present invention, as shown in fig. 4, the apparatus 40 may specifically include the following modules:
a tag set obtaining module 401, configured to obtain a target tag set of a target video and a to-be-recalled tag set of a to-be-recalled video;
a tag vector obtaining module 402, configured to obtain a target tag vector corresponding to a target tag in the target tag set, and obtain a to-be-recalled tag vector corresponding to a to-be-recalled tag in the to-be-recalled tag set; the method comprises the steps that a label vector is obtained from a vector set, and the vector set is generated according to video labels which have a co-occurrence relation and replace synonyms with designated synonyms corresponding to the synonyms;
a vector summing module 403, configured to sum the target tag vectors to obtain target tag set vectors, and sum the to-be-recalled tag vectors to obtain to-be-recalled tag set vectors;
a similarity calculation module 404, configured to calculate a similarity between the target labelset vector and the to-be-recalled labelset vector;
a video recall module 405, configured to recall a video from the to-be-recalled videos corresponding to the to-be-recalled tag set vectors whose similarity satisfies a preset condition.
In an exemplary embodiment of the present invention, the tag vector obtaining module 402 is configured to obtain a set of orientation vectors; the vector set comprises tag vectors corresponding to video tags; and acquiring a target label vector corresponding to a target label in the target label set from the vector set, and acquiring a to-be-recalled label vector corresponding to a to-be-recalled label in the to-be-recalled label set from the vector set.
In an exemplary embodiment of the invention, the apparatus further comprises: a vector set generation module; the vector set generation module is used for acquiring video tags with a co-occurrence relationship, and the co-occurrence relationship is used for representing the probability of the video tags appearing at the same time; when synonyms exist in the video tags, replacing the synonyms in the video tags with the designated synonyms corresponding to the synonyms; and inputting a preset word vector model by adopting the video tag after replacing the synonym, obtaining a corresponding tag vector, and storing the tag vector into a vector set.
In an exemplary embodiment of the present invention, the vector set generating module is configured to replace a synonym in the video tag with a designated synonym corresponding to the synonym when there is a matching synonym in a preset synonym table in the video tag.
In an exemplary embodiment of the present invention, the tag vector obtaining module 402 is configured to, when a target tag vector corresponding to a target tag in the target tag set does not exist in the vector set, perform word segmentation on the target tag to obtain a sub-word, obtain a target sub-tag vector corresponding to the sub-word of the target tag from the vector set, sum the target sub-tag vectors to obtain a target tag vector, when a to-be-recalled tag vector corresponding to the to-be-recalled tag in the to-be-recalled tag set does not exist in the vector set, perform word segmentation on the to-be-recalled tag to obtain a sub-word, obtain a to-be-recalled sub-tag vector corresponding to the sub-word of the to-be-recalled tag from the vector set, and sum the to-be-recalled sub-tag vector to obtain a to-be-recalled tag vector.
In an exemplary embodiment of the present invention, the video recall module 405 is configured to acquire a target to-be-recalled tag set corresponding to the to-be-recalled tag set vector whose similarity is greater than a preset similarity threshold and which is sorted in top N, where N is a positive integer; wherein the target to-be-recalled labelset comprises a target labelset of the target video; creating a similar video index list according to the target to-be-recalled label set; acquiring a related video to be recalled according to the similar video index list to be used as a target video to be recalled; determining similar videos from the target video to be recalled based on preset posterior indexes; the posterior indicator includes click through rate.
In an exemplary embodiment of the present invention, the video recall module 405 is configured to obtain, as the target video to be recalled, the video to be recalled, which is associated with the target set of tags to be recalled and has a channel identifier that is the same as a channel identifier of the target video.
In the embodiment of the invention, by acquiring a target label set of a target video and a to-be-recalled label set of a to-be-recalled video, acquiring a target label vector corresponding to a target label in the target label set and a to-be-recalled label vector corresponding to a to-be-recalled label in the to-be-recalled label set, summing the target label vectors to obtain a target label set vector and the to-be-recalled label vectors to obtain a to-be-recalled label set vector, calculating the similarity between the target label set vector and the to-be-recalled label set vector, and recalling the video from the to-be-recalled video corresponding to the to-be-recalled label set vector with the similarity meeting a preset condition, the video is recalled by taking the label set as a retrieval unit, so that the correlation between the recalled video and the target video can be improved, and the quality of the recalled video is improved.
For the above device embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and for the relevant points, refer to the partial description of the method embodiment.
An embodiment of the present invention further provides an electronic device, as shown in fig. 5, which includes a processor 501, a communication interface 502, a memory 503 and a communication bus 504, where the processor 501, the communication interface 502 and the memory 503 complete mutual communication through the communication bus 504,
a memory 503 for storing a computer program;
the processor 501, when executing the program stored in the memory 503, implements the following steps:
acquiring a target label set of a target video and a to-be-recalled label set of a to-be-recalled video;
acquiring a target label vector corresponding to a target label in the target label set, and acquiring a to-be-recalled label vector corresponding to a to-be-recalled label in the to-be-recalled label set; the method comprises the steps that a label vector is obtained from a vector set, and the vector set is generated according to video labels which have a co-occurrence relation and replace synonyms with designated synonyms corresponding to the synonyms;
summing the target label vectors to obtain target label set vectors, and summing the to-be-recalled label vectors to obtain to-be-recalled label set vectors;
calculating the similarity between the target label set vector and the to-be-recalled label set vector;
and recalling videos from the videos to be recalled corresponding to the label set vectors to be recalled, of which the similarity meets the preset condition.
Optionally, the obtaining a target tag vector corresponding to a target tag in the target tag set and obtaining a to-be-recalled tag vector corresponding to a to-be-recalled tag in the to-be-recalled tag set include:
obtaining a set of orientation quantities; the vector set comprises tag vectors corresponding to video tags;
and acquiring a target label vector corresponding to a target label in the target label set from the vector set, and acquiring a to-be-recalled label vector corresponding to a to-be-recalled label in the to-be-recalled label set from the vector set.
Optionally, before the obtaining the set of orientation quantities, the method further comprises:
acquiring video tags with a co-occurrence relation, wherein the co-occurrence relation is used for representing the probability of the simultaneous occurrence of the video tags;
when synonyms exist in the video tags, replacing the synonyms in the video tags with the designated synonyms corresponding to the synonyms;
and inputting a preset word vector model by adopting the video tag after replacing the synonym, obtaining a corresponding tag vector, and storing the tag vector into a vector set.
Optionally, when there is a synonym in the video tag, replacing the synonym in the video tag with a designated synonym corresponding to the synonym includes:
when the matched synonym exists in a preset synonym table in the video label, replacing the synonym in the video label with the designated synonym corresponding to the synonym.
Optionally, the obtaining, from the vector set, a target tag vector corresponding to a target tag in the target tag set, and obtaining, from the vector set, a to-be-recalled tag vector corresponding to a to-be-recalled tag in the to-be-recalled tag set include:
when the vector set does not have a target label vector corresponding to a target label in the target label set, performing word segmentation on the target label to obtain a sub-word, obtaining a target sub-label vector corresponding to the sub-word of the target label from the vector set, and summing the target sub-label vectors to obtain a target label vector, an
When the vector set does not have the to-be-recalled label vector corresponding to the to-be-recalled label in the to-be-recalled label set, performing word segmentation on the to-be-recalled label to obtain a sub-word, obtaining the to-be-recalled sub-label vector corresponding to the sub-word of the to-be-recalled label from the vector set, and summing the to-be-recalled sub-label vectors to obtain the to-be-recalled label vector.
Optionally, the recalling a video from the to-be-recalled videos corresponding to the to-be-recalled label set vector whose similarity satisfies a preset condition includes:
acquiring a target to-be-recalled label set corresponding to the to-be-recalled label set vector with the similarity larger than a preset similarity threshold and sorted in the top N, wherein N is a positive integer; wherein the target to-be-recalled labelset comprises a target labelset of the target video;
creating a similar video index list according to the target to-be-recalled label set;
acquiring a related video to be recalled according to the similar video index list to be used as a target video to be recalled;
determining similar videos from the target video to be recalled based on preset posterior indexes; the posterior indicator includes click through rate.
Optionally, the target video and the to-be-recalled video have channel identifiers, and the acquiring the to-be-recalled video associated with the target to-be-recalled label set as the target to-be-recalled video includes:
and acquiring the video to be recalled which is associated with the target video to be recalled label set and has the same channel identification as that of the target video as the target video to be recalled.
The communication bus mentioned in the above terminal may be a Peripheral Component Interconnect (PCI) bus, an Extended Industry Standard Architecture (EISA) bus, or the like. The communication bus may be divided into an address bus, a data bus, a control bus, etc. For ease of illustration, only one thick line is shown, but this does not mean that there is only one bus or one type of bus.
The communication interface is used for communication between the terminal and other equipment.
The Memory may include a Random Access Memory (RAM) or a 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 processor.
The processor may be a general-purpose processor, and includes a Central Processing Unit (CPU), a Network Processor (NP), and the like; the Integrated Circuit may also be a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field Programmable Gate Array (FPGA) or other Programmable logic device, a discrete Gate or transistor logic device, or a discrete hardware component.
In yet another embodiment of the present invention, there is also provided a computer-readable storage medium having stored therein instructions, which when run on a computer, cause the computer to perform the video recall method of any of the above embodiments.
In yet another embodiment provided by the present invention, there is also provided a computer program product containing instructions that, when run on a computer, cause the computer to perform the video recall method of any of the above embodiments.
In the above embodiments, the implementation may be wholly or partially realized 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, cause the processes or functions described in accordance with the embodiments of the invention to occur, in whole or in part. The computer may be a general purpose computer, a special purpose computer, a network of computers, or other programmable device. The computer instructions may be stored in a computer readable storage medium or transmitted from one computer readable storage medium to another, for example, from one website site, computer, server, or data center to another website site, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, Digital Subscriber Line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device, such as a server, a data center, etc., that incorporates one or more of the 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)), among others.
It is noted that, herein, relational terms such as first and second, and the like may be 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. Also, 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 an … …" does not exclude the presence of other identical elements in a process, method, article, or apparatus that comprises the element.
All the embodiments in the present specification are described in a related manner, and the same and similar parts among the embodiments may be referred to each other, and each embodiment focuses on the differences from the other embodiments. In particular, for the system embodiment, since it is substantially similar to the method embodiment, the description is simple, and for the relevant points, reference may be made to the partial description of the method embodiment.
The above description is only for the preferred embodiment of the present invention, and is not intended to limit the scope of the present invention. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention shall fall within the protection scope of the present invention.
Claims (10)
1. A method for video recall, the method comprising:
acquiring a target label set of a target video and a to-be-recalled label set of a to-be-recalled video;
acquiring a target label vector corresponding to a target label in the target label set, and acquiring a to-be-recalled label vector corresponding to a to-be-recalled label in the to-be-recalled label set; the method comprises the steps that a label vector is obtained from a vector set, and the vector set is generated according to video labels which have a co-occurrence relation and replace synonyms with designated synonyms corresponding to the synonyms;
summing the target label vectors to obtain target label set vectors, and summing the to-be-recalled label vectors to obtain to-be-recalled label set vectors;
calculating the similarity between the target label set vector and the to-be-recalled label set vector;
and recalling videos from the videos to be recalled corresponding to the label set vectors to be recalled, of which the similarity meets the preset condition.
2. The method of claim 1, wherein the obtaining a target tag vector corresponding to a target tag in the target tag set and obtaining a to-be-recalled tag vector corresponding to a to-be-recalled tag in the to-be-recalled tag set comprise:
obtaining a set of orientation quantities; the vector set comprises tag vectors corresponding to video tags;
and acquiring a target label vector corresponding to a target label in the target label set from the vector set, and acquiring a to-be-recalled label vector corresponding to a to-be-recalled label in the to-be-recalled label set from the vector set.
3. The method of claim 2, wherein prior to obtaining the set of orientation metrics, the method further comprises:
acquiring video tags with a co-occurrence relation, wherein the co-occurrence relation is used for representing the probability of the simultaneous occurrence of the video tags;
when synonyms exist in the video tags, replacing the synonyms in the video tags with the designated synonyms corresponding to the synonyms;
and inputting a preset word vector model by adopting the video tag after replacing the synonym, obtaining a corresponding tag vector, and storing the tag vector into a vector set.
4. The method according to claim 3, wherein the replacing the synonym in the video tag with the designated synonym corresponding to the synonym when the synonym exists in the video tag comprises:
when the matched synonym exists in a preset synonym table in the video label, replacing the synonym in the video label with the designated synonym corresponding to the synonym.
5. The method of claim 2, wherein the obtaining a target tag vector corresponding to a target tag in the target tag set from the vector set and obtaining a to-be-recalled tag vector corresponding to a to-be-recalled tag in the to-be-recalled tag set from the vector set comprises:
when the vector set does not have a target label vector corresponding to a target label in the target label set, performing word segmentation on the target label to obtain a sub-word, obtaining a target sub-label vector corresponding to the sub-word of the target label from the vector set, and summing the target sub-label vectors to obtain a target label vector, an
When the vector set does not have the to-be-recalled label vector corresponding to the to-be-recalled label in the to-be-recalled label set, performing word segmentation on the to-be-recalled label to obtain a sub-word, obtaining the to-be-recalled sub-label vector corresponding to the sub-word of the to-be-recalled label from the vector set, and summing the to-be-recalled sub-label vectors to obtain the to-be-recalled label vector.
6. The method according to claim 2, wherein the recalling a video from the videos to be recalled corresponding to the tag set vector to be recalled whose similarity satisfies a preset condition includes:
acquiring a target to-be-recalled label set corresponding to the to-be-recalled label set vector with the similarity larger than a preset similarity threshold and sorted in the top N, wherein N is a positive integer; wherein the target to-be-recalled labelset comprises a target labelset of the target video;
creating a similar video index list according to the target to-be-recalled label set;
acquiring a related video to be recalled according to the similar video index list to be used as a target video to be recalled;
determining similar videos from the target video to be recalled based on preset posterior indexes; the posterior indicator includes click through rate.
7. The method of claim 6, wherein the target video and the to-be-recalled video have channel identifications, and wherein the obtaining the to-be-recalled video associated with the target to-be-recalled label set as the target to-be-recalled video comprises:
and acquiring the video to be recalled which is associated with the target video to be recalled label set and has the same channel identification as that of the target video as the target video to be recalled.
8. A video recall apparatus, the apparatus comprising:
the system comprises a tag set acquisition module, a target video retrieval module and a target video retrieval module, wherein the tag set acquisition module is used for acquiring a target tag set of a target video and a to-be-recalled tag set of a to-be-recalled video;
the tag vector acquisition module is used for acquiring a target tag vector corresponding to a target tag in the target tag set and acquiring a to-be-recalled tag vector corresponding to a to-be-recalled tag in the to-be-recalled tag set; the method comprises the steps that a label vector is obtained from a vector set, and the vector set is generated according to video labels which have a co-occurrence relation and replace synonyms with designated synonyms corresponding to the synonyms;
the vector summation module is used for summing the target label vectors to obtain target label set vectors and summing the to-be-recalled label vectors to obtain to-be-recalled label set vectors;
the similarity calculation module is used for calculating the similarity between the target label set vector and the to-be-recalled label set vector;
and the video recall module is used for recalling videos from the videos to be recalled corresponding to the label set vectors to be recalled, wherein the similarity of the videos to be recalled meets the preset condition.
9. An electronic device is characterized by comprising a processor, a communication interface, a memory and a communication bus, wherein the processor and the communication interface are used for realizing mutual communication by the memory through the communication bus;
a memory for storing a computer program;
a processor for implementing the method steps of any of claims 1 to 7 when executing a program stored in the memory.
10. A computer-readable storage medium, on which a computer program is stored which, when being executed by a processor, carries out the method according to any one of claims 1-7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110156410.5A CN112989118B (en) | 2021-02-04 | 2021-02-04 | Video recall method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110156410.5A CN112989118B (en) | 2021-02-04 | 2021-02-04 | Video recall method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112989118A true CN112989118A (en) | 2021-06-18 |
| CN112989118B CN112989118B (en) | 2023-08-18 |
Family
ID=76347134
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110156410.5A Active CN112989118B (en) | 2021-02-04 | 2021-02-04 | Video recall method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112989118B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113392266A (en) * | 2021-08-17 | 2021-09-14 | 北京达佳互联信息技术有限公司 | Training and sorting method and device of sorting model, electronic equipment and storage medium |
| CN116915925A (en) * | 2023-06-20 | 2023-10-20 | 天翼爱音乐文化科技有限公司 | Video generation method, system, electronic equipment and medium based on video template |
Citations (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2006139484A (en) * | 2004-11-11 | 2006-06-01 | Nippon Telegr & Teleph Corp <Ntt> | Information retrieval method, system therefor and computer program |
| US20120016885A1 (en) * | 2010-07-16 | 2012-01-19 | Ibm Corporation | Adaptive and personalized tag recommendation |
| US20120323968A1 (en) * | 2011-06-14 | 2012-12-20 | Microsoft Corporation | Learning Discriminative Projections for Text Similarity Measures |
| US20130036061A1 (en) * | 2011-08-02 | 2013-02-07 | International Business Machines Corporation | Product recall information management |
| CN103279513A (en) * | 2013-05-22 | 2013-09-04 | 百度在线网络技术（北京）有限公司 | Method for generating content label and method and device for providing multi-media content information |
| CN103488787A (en) * | 2013-09-30 | 2014-01-01 | 北京奇虎科技有限公司 | Method and device for pushing online playing entry objects based on video retrieval |
| US20170140261A1 (en) * | 2015-11-17 | 2017-05-18 | Spotify Ab | Systems, methods and computer products for determining an activity |
| CN107122413A (en) * | 2017-03-31 | 2017-09-01 | 北京奇艺世纪科技有限公司 | A kind of keyword extracting method and device based on graph model |
| CN108228665A (en) * | 2016-12-22 | 2018-06-29 | 阿里巴巴集团控股有限公司 | Determine object tag, the method and device for establishing tab indexes, object search |
| US20180295394A1 (en) * | 2015-07-28 | 2018-10-11 | Google Llc | Methods, systems, and media for presenting media content items belonging to a media content group |
| CN109587554A (en) * | 2018-10-29 | 2019-04-05 | 百度在线网络技术（北京）有限公司 | Processing method, device and the readable storage medium storing program for executing of video data |
| CN109635157A (en) * | 2018-10-30 | 2019-04-16 | 北京奇艺世纪科技有限公司 | Model generating method, video searching method, device, terminal and storage medium |
| CN110457581A (en) * | 2019-08-02 | 2019-11-15 | 达而观信息科技（上海）有限公司 | A kind of information recommended method, device, electronic equipment and storage medium |
-
2021
- 2021-02-04 CN CN202110156410.5A patent/CN112989118B/en active Active
Patent Citations (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2006139484A (en) * | 2004-11-11 | 2006-06-01 | Nippon Telegr & Teleph Corp <Ntt> | Information retrieval method, system therefor and computer program |
| US20120016885A1 (en) * | 2010-07-16 | 2012-01-19 | Ibm Corporation | Adaptive and personalized tag recommendation |
| US20120323968A1 (en) * | 2011-06-14 | 2012-12-20 | Microsoft Corporation | Learning Discriminative Projections for Text Similarity Measures |
| US20130036061A1 (en) * | 2011-08-02 | 2013-02-07 | International Business Machines Corporation | Product recall information management |
| CN103279513A (en) * | 2013-05-22 | 2013-09-04 | 百度在线网络技术（北京）有限公司 | Method for generating content label and method and device for providing multi-media content information |
| CN103488787A (en) * | 2013-09-30 | 2014-01-01 | 北京奇虎科技有限公司 | Method and device for pushing online playing entry objects based on video retrieval |
| US20180295394A1 (en) * | 2015-07-28 | 2018-10-11 | Google Llc | Methods, systems, and media for presenting media content items belonging to a media content group |
| US20170140261A1 (en) * | 2015-11-17 | 2017-05-18 | Spotify Ab | Systems, methods and computer products for determining an activity |
| CN108228665A (en) * | 2016-12-22 | 2018-06-29 | 阿里巴巴集团控股有限公司 | Determine object tag, the method and device for establishing tab indexes, object search |
| CN107122413A (en) * | 2017-03-31 | 2017-09-01 | 北京奇艺世纪科技有限公司 | A kind of keyword extracting method and device based on graph model |
| CN109587554A (en) * | 2018-10-29 | 2019-04-05 | 百度在线网络技术（北京）有限公司 | Processing method, device and the readable storage medium storing program for executing of video data |
| CN109635157A (en) * | 2018-10-30 | 2019-04-16 | 北京奇艺世纪科技有限公司 | Model generating method, video searching method, device, terminal and storage medium |
| CN110457581A (en) * | 2019-08-02 | 2019-11-15 | 达而观信息科技（上海）有限公司 | A kind of information recommended method, device, electronic equipment and storage medium |
Non-Patent Citations (1)
| Title |
|---|
| 许良武: "基于向量化标签的视频推荐算法研究与实现", 《无线互联科技》, pages 132 - 134 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113392266A (en) * | 2021-08-17 | 2021-09-14 | 北京达佳互联信息技术有限公司 | Training and sorting method and device of sorting model, electronic equipment and storage medium |
| CN116915925A (en) * | 2023-06-20 | 2023-10-20 | 天翼爱音乐文化科技有限公司 | Video generation method, system, electronic equipment and medium based on video template |
| CN116915925B (en) * | 2023-06-20 | 2024-02-23 | 天翼爱音乐文化科技有限公司 | Video generation method, system, electronic equipment and medium based on video template |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112989118B (en) | 2023-08-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Li et al. | A bibliometric analysis of topic modelling studies (2000–2017) | |
| CN105224699B (en) | News recommendation method and device | |
| Zhu et al. | Ranking user authority with relevant knowledge categories for expert finding | |
| US8346701B2 (en) | Answer ranking in community question-answering sites | |
| CN108363821A (en) | A kind of information-pushing method, device, terminal device and storage medium | |
| US20170249389A1 (en) | Sentiment rating system and method | |
| CN112434151A (en) | Patent recommendation method and device, computer equipment and storage medium | |
| CN109189990B (en) | Search word generation method and device and electronic equipment | |
| Jiang et al. | Recommending new features from mobile app descriptions | |
| CN108228745B (en) | Recommendation algorithm and device based on collaborative filtering optimization | |
| CN111666448B (en) | Search method, search device, electronic equipment and computer readable storage medium | |
| CN112347778A (en) | Keyword extraction method and device, terminal equipment and storage medium | |
| Lin et al. | Mitigating sentiment bias for recommender systems | |
| CN111104590A (en) | Information recommendation method, device, medium and electronic equipment | |
| CN111984867B (en) | Network resource determining method and device | |
| CN112989118B (en) | Video recall method and device | |
| Yang et al. | Characteristics of Chinese online movie reviews and opinion leadership identification | |
| CN112836126A (en) | Recommendation method and device based on knowledge graph, electronic equipment and storage medium | |
| Liu et al. | Question popularity analysis and prediction in community question answering services | |
| Faisal et al. | A novel framework for social web forums’ thread ranking based on semantics and post quality features | |
| CN113220974A (en) | Click rate prediction model training and search recall method, device, equipment and medium | |
| CN112328881A (en) | Article recommendation method and device, terminal device and storage medium | |
| CN111353052B (en) | Multimedia object recommendation method and device, electronic equipment and storage medium | |
| Kim et al. | TrendsSummary: a platform for retrieving and summarizing trendy multimedia contents | |
| CN107368533B (en) | Content item recommendation method and device and electronic equipment |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |