CN109408672B - Article generation method, article generation device, server and storage medium - Google Patents
Article generation method, article generation device, server and storage medium Download PDFInfo
- Publication number
- CN109408672B CN109408672B CN201811535231.7A CN201811535231A CN109408672B CN 109408672 B CN109408672 B CN 109408672B CN 201811535231 A CN201811535231 A CN 201811535231A CN 109408672 B CN109408672 B CN 109408672B
- Authority
- CN
- China
- Prior art keywords
- video
- image
- information
- news
- target
- 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.)
- Active
Links
Images
Abstract
The embodiment of the invention discloses an article generation method, an article generation device, a server and a storage medium, wherein the method comprises the following steps: acquiring a video image set from a target video; identifying at least one piece of news related to the target video and the video images thereof; and combining the video knowledge information in the pre-established knowledge graph information base and the at least one piece of news to generate abstract information related to each video image, and generating an article related to the target video according to the abstract information and the video image set. The embodiment of the invention realizes the fast and efficient generation of the articles with high quality and rich content related to the video, can meet the appeal of users to the articles of high-quality popular videos, and improves the commercial value of related products.
Description
Technical Field
The embodiment of the invention relates to the technical field of internet, in particular to an article generation method, an article generation device, a server and a storage medium.
Background
In the era of internet information explosion, trending applications and products spread around network data are receiving wide attention. For various types of videos, such as a movie in a movie theater, that is, a high-definition album of a highlight in the movie theater, through the album article, the movie theater can be helped to attract a user to watch, and the user can confirm whether the movie theater has a watching interest while watching the movie theater. Therefore, the gallery article associated with videos is the most direct bridge between communicating users and these videos.
At present, most of the video album articles seen by people, such as movie and movie drama articles, are small-sized articles edited manually, for example, some popular dramas are selected manually, then highlight pictures are selected as drama pictures, finally drama introduction, character introduction and the like are performed on the dramas, and finally the movie and movie drama album articles are generated. Obviously, the whole article generation process needs manual participation, manual intervention is excessive, procedures are complicated, and the article generation is long in time consumption and low in yield. That is, besides the disadvantages of tedious data collection, insufficient data resources, poor information timeliness, etc., the method is far from reaching the mass production and real-time update.
Disclosure of Invention
Embodiments of the present invention provide an article generation method, an article generation device, a server, and a storage medium, which enable articles with high quality and rich content related to a video to be generated quickly and efficiently, and can meet the user's appeal for articles of high-quality popular videos, and improve the commercial value of related products.
In a first aspect, an embodiment of the present invention provides an article generation method, including:
acquiring a video image set from a target video;
identifying at least one piece of news related to the target video and the video images thereof;
and combining the video knowledge information in the pre-established knowledge graph information base and the at least one piece of news to generate abstract information related to each video image, and generating an article related to the target video according to the abstract information and the video image set.
In a second aspect, an embodiment of the present invention further provides an article generating apparatus, including:
the image acquisition module is used for acquiring a video image set from a target video;
the news acquisition module is used for identifying at least one piece of news related to the target video and the video image thereof;
and the article generation module is used for generating abstract information related to each video image by combining the video knowledge information in the pre-established knowledge graph information base and the at least one piece of news, and generating an article related to the target video according to the abstract information and the video image set.
In a third aspect, an embodiment of the present invention further provides a server, where the server includes:
one or more processors;
a storage device for storing one or more programs,
when executed by the one or more processors, cause the one or more processors to implement a method of article generation as described in any embodiment of the invention.
In a fourth aspect, embodiments of the present invention also provide a storage medium containing computer-executable instructions, which when executed by a computer processor, are configured to perform a method of article generation according to any of the embodiments of the present invention.
The embodiment of the invention obtains a video image set from a target video; identifying at least one piece of news related to the target video and the video images thereof; and combining the video knowledge information in the pre-established knowledge graph information base and the at least one piece of news to generate abstract information related to each video image, and generating an article related to the target video according to the abstract information and the video image set, so that the article related to the video is generated quickly and efficiently. Meanwhile, the materials of the articles are acquired from news and knowledge map information bases by combining with the video images, so that the articles are high in quality and rich in content, the demands of users on high-quality popular video articles can be met, and the commercial value of related products is improved.
Drawings
FIG. 1 is a flowchart of an article generation method according to a first embodiment of the present invention;
FIG. 2 is a flowchart of an article generation method according to a second embodiment of the present invention;
fig. 3 is a flowchart of a method for identifying news in an article generation method according to a second embodiment of the present invention;
fig. 4 is a schematic structural diagram of an article generation apparatus according to a third embodiment of the present invention;
fig. 5 is a schematic structural diagram of a server in the fourth embodiment of the present invention.
Detailed Description
The present invention will be described in further detail with reference to the accompanying drawings and examples. It is to be understood that the specific embodiments described herein are merely illustrative of the invention and are not limiting of the invention. It should be further noted that, for the convenience of description, only some of the structures related to the present invention are shown in the drawings, not all of the structures.
Example one
Fig. 1 is a flowchart of an article generation method according to an embodiment of the present invention, where this embodiment is applicable to the generation situation of an article, and the method may be executed by an article generation apparatus, where the apparatus may be implemented by hardware and/or software and may be configured in a server, and the method specifically includes:
and S110, acquiring a video image set from the target video.
The target video may be a video related to an article to be produced or generated, and may include a short video, a micro movie, a television series, a movie, and the like, and the video image set may be a video image composition extracted from the target video. If a drama article is to be produced, the target video may be a movie associated with the drama article to be produced or generated. Illustratively, to produce a movie article for movie B, movie B is the target video (target movie). When a movie and television play article is produced, a video image related to a target movie and television play needs to be selected first, and specifically, the video image can be directly identified from a movie and television play library, for example, a corresponding movie can be analyzed from a video of the movie and television play. The series library may be an existing video resource library, which contains a large amount of series related information, such as the name of the series or movie, the shooting time, the related series and definition, etc. The video image collection may be a candidate collection formed for one or more related series of the target movie. Here, the video image set may also be obtained from other existing information knowledge databases, such as an encyclopedia database, and the source of the video image set is not limited in this embodiment.
And S120, identifying at least one piece of news related to the target video and the video image thereof.
The video image may be one or more video-related images in a set of video images. When making related album articles, other descriptive information is needed besides images, for example, when making the material of a drama article, information of a textual expression or description related to the drama article is needed besides the drama.
In this embodiment, at least one piece of news related to an image of a video, such as a movie or television series, may be identified from a news library and utilized to generate text related to the image in the video. The news library includes a large amount of news information, so that one or more news items related to the video and the video images thereof can be selected from the news library. Taking a video as a movie, for example, a video image collection of a tv series C includes 5 movies, and the 5 movies are selected as video images, and further, news information related to the tv series C and the 5 movies can be identified from a news library. Of course, the rules for identifying news may be pre-established identification rules or policies, and may be, for example, identifying news information related to a character role or a play role in a video image; or the related news can be identified by using the identification model trained by corresponding sample data. Here, the news may also be obtained from other existing information knowledge databases, and the source of the related news is not limited in this embodiment.
S130, combining video knowledge information in a pre-established knowledge graph information base and the at least one piece of news to generate abstract information related to each video image, and generating an article related to the target video according to the abstract information and the video image set.
The pre-established knowledge map information base may be a pre-established database including more comprehensive video related information, may include video related knowledge information, and the knowledge information may be production information and content information of the video. For example, the knowledge information of videos such as movie plays, micro-movies, etc. may include at least the title, the scenario, the actors, the characters, the broadcasting time, the production time, etc. Summary information may be summary or primary information for video information. Furthermore, the articles related to the video can be formed by combining the video images and the corresponding abstract information. Illustratively, corresponding abstract information can be correspondingly added below each video image, and finally an atlas article is formed.
Taking the generation of the movie and television play article as an example, the knowledge information of the movie and television play and at least one piece of news can be used to generate the abstract information related to the movie and television play, and the abstract information can be the information related to the movie and television play, such as the description introduction information of the movie and television play, extracted or analyzed from the knowledge information of the movie and television play and the news. Of course, a plurality of news items may be used to generate summary information corresponding to the drama. And then generating a movie article of the movie play by using each movie and the corresponding abstract information.
Optionally, before acquiring the video image set from the target video, the method further includes:
and mining a hot video from a video resource library, and taking the hot video as the target video.
The popular video can be the video which is most concerned by the public or net friends within a certain time and a certain range. For example, if the audience rating of the series a is higher than that of other series at the same time, or the topic discussion volume of the series a is very large, the series a may be regarded as a popular series, and accordingly, the series a may be regarded as a target series. The popular video is taken as the article related to the video, so that the reading requirement of the public can be better met, and the eyeball of the audience can be better attracted.
Optionally, the mining of the popular video from the video resource library includes:
acquiring description information of each video in the video resource library, wherein the description information is used for introducing the videos from different dimensions;
capturing selection characteristics of each video from the Internet according to the description information, wherein the selection characteristics are used for evaluating video heat from different angles;
and determining the total heat score of each video by the selected characteristics through a weighted summation method, and mining the hot video according to the total heat score.
Specifically, the description information may be specific information for describing or explaining the video, wherein the description information includes at least a name, a shooting time, and a definition. For example, the description information for mining the popular video may further include: the personas of each video and the corresponding decorators. The selected features of each video are captured from the internet or other databases according to the description information, and capturing of the selected features can be performed by using a preset capture model or existing capture tools, software and the like.
The selection features are used to evaluate video popularity from different angles, and may include at least dimensions of movie popularity, actor popularity, and play popularity, for example. Illustratively, if a movie is adapted according to a novel, the reading heat of the corresponding novel can also be used as the selection feature. And determining the total heat score of each video by a method of weighting and summing different selected features, wherein the weight values of different features can be preset according to business requirements or different types and requirements of the videos. For example, for some even-image dramas, the weighting value corresponding to the heat of the actor may be set to be larger. And finally, determining the hot video according to the total heat score of each video, wherein the higher the total heat score is, the higher the heat of the video is, the hot video can be regarded as the hot video.
According to the technical scheme of the embodiment of the invention, the selected video images and the related news are combined, further, the abstract information of the video images is generated by combining the video knowledge information and the news, and the articles are finally generated by utilizing the video images and the corresponding abstract information, so that the articles related to the videos are generated quickly and efficiently. Meanwhile, the materials of the articles are obtained from news and knowledge map information bases, so that the content is high in quality and rich, the demands of users on the atlas articles of high-quality popular videos can be met, and the commercial value of related products is improved.
Example two
Fig. 2 is a flowchart of an article generation method provided in the second embodiment of the present invention, and further optimization is performed on the basis of the second embodiment, as shown in fig. 2, the method specifically includes:
s210, analyzing a video source of the target video, and randomly acquiring a preset number of video frame images in seconds from all video frame images obtained through analysis to obtain a plurality of candidate video frame images of the target video.
In order to obtain a video image, all video frame images can be obtained by analyzing a video source of a target video, and then a plurality of candidate video frame images are selected according to a certain rule. For example, for a video having 24 video frames per second, a plurality of images in the video source are acquired in such a manner that 3 images are randomly extracted per second, thereby obtaining a preset number of video frame images.
S220, obtaining identification features in each candidate video frame image, screening the candidate video frame images by using a high-quality image identification model obtained through pre-training and combining the identification features to obtain a video image candidate set, wherein the identification features comprise at least one feature used for measuring the quality of the video frame images.
After a plurality of video frame images are acquired, quality identification or filtering needs to be performed on the images. Specifically, images can be identified through a high-quality drama identification model obtained through pre-training, and for a drama article, it is necessary to identify which high-quality dramas can be used. The high-quality drama recognition model can be obtained by performing two-classification learning on the dramas by combining the image classification algorithm based on deep learning RESNET (content) and the prior knowledge of the dramas, such as whether the recognition characteristics of actors, the number of the actors, whether the actors are closed eyes or not, whether the pictures are fuzzy or not and the like are taken as sample data. In addition, a priori knowledge about the screenshots may be identified and provided by other third party services.
When the high-quality photos are identified, different photos and corresponding prior knowledge can be input into the identification model to obtain the quality value of each photo, and when the quality value is higher than a preset threshold value, the photos can be regarded as high-quality photos and can be used for the photo articles.
And S230, carrying out duplicate removal on the video frame images in the video image candidate set according to the image similarity to obtain the video image set.
This may reduce the reading experience of the reader, as there may be duplicates or similarities in the filtered high quality image. Therefore, the screened photos can be subjected to the duplicate removal operation again. Illustratively, the SITF algorithm may be used for filtering similar images.
And S240, capturing attribute information related to the target movie and television play from the Internet, wherein the attribute information at least comprises a video name, a director name, an actor name, roles played by actors and a plot.
The attribute information may be related information for describing a movie, especially more detailed and specific information for the movie itself, and may include, in addition to a video title, a director name, an actor name, a role played by the actor, and a scenario, other related information, such as, if the movie is a main play, introduction information such as a historical background corresponding to the scenario, and the like. Specifically, for example, the related attribute information of the target movie may be crawled from the internet by using an existing crawling tool or software.
And S250, combining the names of actors involved in the video images, comparing the attribute information with news in a news library, and identifying the news related to the target video and the video images thereof according to the number of hits of the attribute information types, the hit frequency of all types of attribute information and the news word number threshold in the comparison process.
In the process of identifying news related to the target video and the video image, the news in the news library may be compared with the actor names related to the video image and the attribute information of the target video through different dimensions or features, for example, the situation of comparing features such as the actor names, the actors, the roles, and the like. Specifically, the characteristics of number of hits of the title of the drama, number of hits of the actors, number of hits of the roles, hit frequency, number of news words, whether the title includes movie and television drama name information, whether the title includes the names of the actors, and the like can be selected as the filtering conditions. The number of hits may be the number of features that hit in a common way among the features, and the hit frequency may be the number of times that the features appear in a common way in news.
For example, fig. 3 is a flowchart of a method for identifying news provided by an embodiment of the present invention, as shown in fig. 3, first, some low-quality news, such as incomplete news content or unhealthy news, may be filtered from a news library, and then statistics is performed on the above characteristics in the news, if statistics shows that in the above characteristics, the number of hits is greater than 3, the number of hits is greater than 5, and the number of news words is less than 300, the news may be considered as meeting the requirements, and may be used as final news. Of course, a specific rule for identifying news may be preset, or a news identification model trained in advance may be used for identification, so that news information related to a movie and television series may be screened from a large amount of news data.
S260, randomly selecting a preset number of video images from the video image set to generate a video image set.
Since the video image set includes one or more images, a certain number of images can be randomly selected from the video image set. For example, for the production of a drama article, 5 dramas may be required according to a preset number of dramas or a drama article template, and 5 dramas may be randomly selected from a video image set for use in the generation of a subsequent drama article.
S270, acquiring video knowledge information related to the identification characteristics and the attribute information from the knowledge graph information base according to the identification characteristics of each image in the image graph set and the attribute information related to the target video.
The knowledge-graph information base may be created in advance in which a large amount of useful information is recorded. In this embodiment, the text-text combined with the video image needs to be generated, so that the video knowledge information related to the identification features of the video image and the attribute information of the target video can be acquired from the knowledge graph information base for the text matching of the atlas text. The information in the knowledge-graph information base generally has certain accuracy and is relatively comprehensive in coverage, so that the accuracy of the article can be improved and the readability of the article can be improved by constructing the atlas article by using the information in the knowledge-graph information base. For example, for a single star drama, the related knowledge information of the star and the related knowledge information of the star about the current target movie and television drama can be obtained from the knowledge-graph information base.
And S280, generating abstract information related to each image in the video map set from the at least one piece of news according to the identification characteristics of each image in the video map set and the attribute information related to the target video and by combining the video knowledge information.
The summary information is used as a collocation text with images in the atlas article, and can be extracted from the identified news, and further, the number of the news can be set to be not more than 3, namely, the news is extracted from 1-2 news, so that the finally generated summary information has good consistency and the readability is increased. For a movie article, the extraction needs to generate summary information related to each movie according to the identification features of the movie, for example, the summary information related to the actors in the movie and the current target movie and television play. In addition, movie and television drama knowledge information acquired from the knowledge map information base can be combined to be used as a character supplement for news. As for the method for extracting the abstract from the known text, in the embodiment of the present invention, any abstraction-type or generation-type abstract generation algorithm in the prior art may be adopted, and therefore, the details are not described herein.
And S290, combining each image in the video image set with the corresponding abstract information to obtain an article related to the target video.
Illustratively, if 8-10 high-quality pictures are randomly selected from the video image set as the screenplay album candidates, then a summary corresponding to each screenplay is used as the introduction of the screenplay album, and the final screenplay articles are obtained through combination. Of course, if the data amount of the summary information is insufficient, the description of the movie and television scenario or the description of the actor of encyclopedia can be used as the supplement of the description of the atlas.
According to the embodiment of the invention, the preset knowledge map information base is utilized, the advanced deep learning technology is utilized, the high-quality movie and television drama is obtained from a large amount of disordered video data, the abstract information for describing and commenting the movie and television drama is generated for the movie by utilizing the advantage of natural news information, and finally the movie and abstract information is combined into the video album article.
EXAMPLE III
Fig. 4 is a schematic structural diagram of an article generating apparatus according to a third embodiment of the present invention, and as shown in fig. 4, the apparatus includes:
an image obtaining module 410, configured to obtain a video image set from a target video. The news retrieval module 420 is configured to identify at least one piece of news related to the target video and the video images thereof.
An article generating module 430, configured to combine video knowledge information in a pre-established knowledge-graph information base with the at least one piece of news to generate summary information related to each video image, and generate an article related to the target video according to the summary information and the video image set.
Optionally, the apparatus further comprises:
and the target video mining module is used for mining the hot video from the video resource library before the image acquisition module 410 acquires the video image set from the target video, and taking the hot video as the target video.
Optionally, the target video mining module includes:
the description information acquisition unit is used for acquiring description information of each video in the video resource library, wherein the description information is used for introducing the videos from different dimensions;
the selection feature capturing unit is used for capturing selection features of all videos from the Internet according to the description information, wherein the selection features are used for evaluating video heat from different angles;
and the heat total score calculating unit is used for determining the heat total score of each video by a weighted summation method through the selection characteristics, and mining the hot video according to the heat total score.
Optionally, the image obtaining module 410 includes:
the video source analysis unit is used for analyzing the video source of the target video, and randomly acquiring a preset number of video frame images in seconds from all the video frame images obtained through analysis to obtain a plurality of candidate video frame images of the target video;
the video candidate set screening unit is used for obtaining identification features in each candidate video frame image, screening the candidate video frame images by using a high-quality image identification model obtained through pre-training and combining the identification features to obtain a video image candidate set, wherein the identification features comprise at least one feature used for measuring the image quality of the video frame;
and the duplication removing unit is used for carrying out duplication removal on the video frame images in the video image candidate set according to the image similarity to obtain the video image set.
Optionally, the news acquisition module 420 includes:
the attribute information capturing unit is used for capturing attribute information related to the target video from the Internet, wherein the attribute information at least comprises a movie and television play name, a director name, an actor name, and roles and scenarios played by actors;
and the news identification unit is used for comparing the actor names related to the video images with news in a news library according to the attribute information, and identifying the news related to the target video and the video images thereof according to the number of hits of the attribute information types, the hit frequency of all types of attribute information and the news word number threshold in the comparison process.
Optionally, the article generating module 430 includes:
the video atlas selection unit is used for randomly selecting a preset number of video images from the video image set to generate a video atlas;
a knowledge information acquisition unit, configured to acquire video knowledge information related to the identification features and attribute information from the knowledge map information base according to the identification features of each image in the video map set and the attribute information related to a target video;
the summary information generating unit is used for generating summary information related to each image in the video image set from the at least one piece of news according to the identification characteristics of each image in the video image set and the attribute information related to the target video and combining the video knowledge information;
and the article combination unit is used for combining each image in the video image set with the corresponding abstract information to obtain an article related to the target video.
The article generation device provided by the embodiment of the invention can execute the article generation method provided by any embodiment of the invention, and has corresponding functional modules and beneficial effects of the execution method. For technical details that are not described in detail in this embodiment, reference may be made to an article generation method provided in any embodiment of the present invention.
Example four
Referring to fig. 5, the present embodiment provides a server 500, which includes: one or more processors 520; the storage 510 is configured to store one or more programs, and when the one or more programs are executed by the one or more processors 520, the one or more processors 520 implement an article generation method provided in an embodiment of the present invention, including:
acquiring a video image set from a target video;
identifying at least one piece of news related to the target video and the video images thereof;
and combining the video knowledge information in the pre-established knowledge graph information base and the at least one piece of news to generate abstract information related to each video image, and generating an article related to the target video according to the abstract information and the video image set.
Of course, those skilled in the art will appreciate that processor 520 may also implement aspects of a method for article generation provided in any of the embodiments of the present invention.
The server 500 shown in fig. 5 is only an example, and should not bring any limitation to the function and the scope of use of the embodiments of the present invention.
As shown in FIG. 5, server 500 is in the form of a general purpose computing device. The components of server 500 may include, but are not limited to: one or more processors 520, a memory device 510, and a bus 550 that couples the various system components (including the memory device 510 and the processors 520).
A program/utility 514 having a set (at least one) of program modules 515 may be stored, for instance, in storage 510, such program modules 515 including, but not limited to, an operating system, one or more application programs, other program modules, and program data, each of which examples or some combination thereof may comprise an implementation of a network environment. Program modules 515 generally perform the functions and/or methods of any of the embodiments described herein.
The server 500 may also communicate with one or more external devices 560 (e.g., keyboard, pointing device, display 570, etc.), with one or more devices that enable a user to interact with the server 500, and/or with any devices (e.g., network card, modem, etc.) that enable the server 500 to communicate with one or more other computing devices. Such communication may occur via input/output (I/O) interfaces 530. Also, server 500 may communicate with one or more networks (e.g., a Local Area Network (LAN), a Wide Area Network (WAN) and/or a public network, such as the Internet) through network adapter 540. As shown in FIG. 5, the network adapter 540 communicates with the other modules of the server 500 via a bus 550. It should be appreciated that although not shown in the figures, other hardware and/or software modules may be used in conjunction with the server 500, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems, among others.
The processor 520 executes various functional applications and data processing by executing programs stored in the storage device 510, for example, implementing an article generation method provided by an embodiment of the present invention.
EXAMPLE five
An embodiment of the present invention provides a storage medium containing computer-executable instructions, which when executed by a computer processor, perform a method for article generation, the method including:
acquiring a video image set from a target video;
identifying at least one piece of news related to the target video and the video images thereof;
and combining the video knowledge information in the pre-established knowledge graph information base and the at least one piece of news to generate abstract information related to each video image, and generating an article related to the target video according to the abstract information and the video image set.
Of course, the storage medium provided by the embodiment of the present invention contains computer-executable instructions, and the computer-executable instructions are not limited to the method operations described above, and may also perform related operations in an article generation method provided by any embodiment of the present invention.
Computer storage media for embodiments of the invention may employ any combination of one or more computer-readable media. The computer readable medium may be a computer readable signal medium or a computer readable storage medium. A computer readable storage medium may be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the foregoing. More specific examples (a non-exhaustive list) of the computer readable storage medium would include the following: an electrical connection having one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. In the context of this document, a computer readable storage medium may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device.
A computer readable signal medium may include a propagated data signal with computer readable program code embodied therein, for example, in baseband or as part of a carrier wave. Such a propagated data signal may take many forms, including, but not limited to, electro-magnetic, optical, or any suitable combination thereof. A computer readable signal medium may also be any computer readable medium that is not a computer readable storage medium and that can communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device.
Program code embodied on a computer readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.
Computer program code for carrying out operations for aspects of the present invention may be written in any combination of one or more programming languages, including an object oriented programming language such as Java, Smalltalk, C + + or the like and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a Local Area Network (LAN) or a Wide Area Network (WAN), or the connection may be made to an external computer (for example, through the Internet using an Internet service provider).
It is to be noted that the foregoing is only illustrative of the preferred embodiments of the present invention and the technical principles employed. It will be understood by those skilled in the art that the present invention is not limited to the particular embodiments described herein, but is capable of various obvious changes, rearrangements and substitutions as will now become apparent to those skilled in the art without departing from the scope of the invention. Therefore, although the present invention has been described in greater detail by the above embodiments, the present invention is not limited to the above embodiments, and may include other equivalent embodiments without departing from the spirit of the present invention, and the scope of the present invention is determined by the scope of the appended claims.
Claims (14)
1. A method for article generation, the method comprising:
acquiring a video image set from a target video;
identifying at least one piece of news related to the target video and the video images thereof;
and combining the video knowledge information in the pre-established knowledge graph information base and the at least one piece of news to generate abstract information related to each video image, and combining the abstract information according to the video images to form an article related to the target video.
2. The method of claim 1, wherein prior to obtaining the set of video images from the target video, the method further comprises:
and mining a hot video from a video resource library, and taking the hot video as the target video.
3. The method of claim 2, wherein mining the trending video from the video repository comprises:
acquiring description information of each video in the video resource library, wherein the description information is used for introducing the videos from different dimensions;
capturing selection characteristics of each video from the Internet according to the description information, wherein the selection characteristics are used for evaluating video heat from different angles;
and determining the total heat score of each video by the selected characteristics through a weighted summation method, and mining the hot video according to the total heat score.
4. The method of claim 1, wherein said obtaining a set of video images from a target video comprises:
analyzing a video source of a target video, and randomly acquiring a preset number of video frame images in a second unit from all video frame images obtained by analyzing to obtain a plurality of candidate video frame images of the target video;
acquiring identification features in each candidate video frame image, and screening the candidate video frame images by using a high-quality image identification model obtained by pre-training and combining the identification features to obtain a video image candidate set, wherein the identification features comprise at least one feature for measuring the image quality of the video frame;
and carrying out duplicate removal on the video frame images in the video image candidate set according to the image similarity to obtain the video image set.
5. The method of claim 4, wherein identifying at least one news item associated with the target video and the video images thereof comprises:
capturing attribute information related to a target video from the Internet, wherein the attribute information at least comprises a video name, a director name, an actor name, roles played by actors and a plot;
and comparing the attribute information with news in a news library by combining the names of actors involved in the video images, and identifying the news related to the target video and the video images thereof according to the number of hits of the attribute information types, the hit frequency of all types of attribute information and the news word number threshold in the comparison process.
6. The method of claim 5, wherein combining the video knowledge information in the pre-established knowledge-graph information base and the at least one piece of news to generate summary information associated with each video image, and generating an article associated with the target video according to the summary information and the video image set comprises:
randomly selecting a preset number of video images from the video image set to generate a video image set;
acquiring video knowledge information related to the identification characteristics and the attribute information from the knowledge graph information base according to the identification characteristics of each image in the video graph set and the attribute information related to a target video;
generating summary information related to each image in the video map set from the at least one piece of news according to the identification characteristics of each image in the video map set and the attribute information related to the target video and combining the video knowledge information;
and combining each image in the video image set with the corresponding abstract information to obtain an article related to the target video.
7. An article generation apparatus, the apparatus comprising:
the image acquisition module is used for acquiring a video image set from a target video;
the news acquisition module is used for identifying at least one piece of news related to the target video and the video image thereof;
and the article generation module is used for combining the video knowledge information in the pre-established knowledge graph information base and the at least one piece of news to generate abstract information related to each video image, and forming an article related to the target video according to the video images and corresponding to the abstract information.
8. The apparatus of claim 7, further comprising:
and the target video mining module is used for mining the hot video from the video resource library before the image acquisition module acquires the video image set from the target video, and taking the hot video as the target video.
9. The apparatus of claim 8, wherein the target video mining module comprises:
the description information acquisition unit is used for acquiring description information of each video in the video resource library, wherein the description information is used for introducing the videos from different dimensions;
the selection feature capturing unit is used for capturing selection features of all videos from the Internet according to the description information, wherein the selection features are used for evaluating video heat from different angles;
and the heat total score calculating unit is used for determining the heat total score of each video by a weighted summation method through the selection characteristics, and mining the hot video according to the heat total score.
10. The apparatus of claim 7, wherein the image acquisition module comprises:
the video source analysis unit is used for analyzing the video source of the target video, and randomly acquiring a preset number of video frame images in seconds from all the video frame images obtained through analysis to obtain a plurality of candidate video frame images of the target video;
the video candidate set screening unit is used for obtaining identification features in each candidate video frame image, screening the candidate video frame images by using a high-quality image identification model obtained through pre-training and combining the identification features to obtain a video image candidate set, wherein the identification features comprise at least one feature used for measuring the image quality of the video frame;
and the duplication removing unit is used for carrying out duplication removal on the video frame images in the video image candidate set according to the image similarity to obtain the video image set.
11. The apparatus of claim 10, wherein the news gathering module comprises:
the attribute information capturing unit is used for capturing attribute information related to the target video from the Internet, wherein the attribute information at least comprises a movie and television play name, a director name, an actor name, and roles and scenarios played by actors;
and the news identification unit is used for comparing the actor names related to the video images with news in a news library according to the attribute information, and identifying the news related to the target video and the video images thereof according to the number of hits of the attribute information types, the hit frequency of all types of attribute information and the news word number threshold in the comparison process.
12. The apparatus of claim 11, wherein the article generation module comprises:
the video atlas selection unit is used for randomly selecting a preset number of video images from the video image set to generate a video atlas;
a knowledge information acquisition unit, configured to acquire video knowledge information related to the identification features and attribute information from the knowledge map information base according to the identification features of each image in the video map set and the attribute information related to a target video;
the summary information generating unit is used for generating summary information related to each image in the video image set from the at least one piece of news according to the identification characteristics of each image in the video image set and the attribute information related to the target video and combining the video knowledge information;
and the article combination unit is used for combining each image in the video image set with the corresponding abstract information to obtain an article related to the target video.
13. A server, characterized in that the server comprises:
one or more processors;
a storage device for storing one or more programs,
the one or more programs, when executed by the one or more processors, cause the one or more processors to implement a method of article generation as recited in any of claims 1-6.
14. A storage medium containing computer-executable instructions for performing a method of article generation as claimed in any one of claims 1-6 when executed by a computer processor.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811535231.7A CN109408672B (en) | 2018-12-14 | 2018-12-14 | Article generation method, article generation device, server and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811535231.7A CN109408672B (en) | 2018-12-14 | 2018-12-14 | Article generation method, article generation device, server and storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109408672A CN109408672A (en) | 2019-03-01 |
| CN109408672B true CN109408672B (en) | 2020-09-29 |
Family
ID=65459369
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811535231.7A Active CN109408672B (en) | 2018-12-14 | 2018-12-14 | Article generation method, article generation device, server and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109408672B (en) |
Families Citing this family (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110377891B (en) * | 2019-06-19 | 2023-01-06 | 北京百度网讯科技有限公司 | Method, device and equipment for generating event analysis article and computer readable storage medium |
| WO2021058116A1 (en) * | 2019-09-27 | 2021-04-01 | Huawei Technologies Co., Ltd. | Mood based multimedia content summarization |
| CN113660541B (en) * | 2021-07-16 | 2023-10-13 | 北京百度网讯科技有限公司 | Method and device for generating abstract of news video |
| CN113569065A (en) * | 2021-07-28 | 2021-10-29 | 北京达佳互联信息技术有限公司 | Topic list display method, device, system, equipment and storage medium |
| CN114417808B (en) * | 2022-02-25 | 2023-04-07 | 北京百度网讯科技有限公司 | Article generation method and device, electronic equipment and storage medium |
| CN115334367B (en) * | 2022-07-11 | 2023-10-17 | 北京达佳互联信息技术有限公司 | Method, device, server and storage medium for generating abstract information of video |
Family Cites Families (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103945234A (en) * | 2014-03-27 | 2014-07-23 | 百度在线网络技术（北京）有限公司 | Video-related information providing method and device |
| US10255361B2 (en) * | 2015-08-19 | 2019-04-09 | International Business Machines Corporation | Video clips generation system |
| CN106844322A (en) * | 2017-01-22 | 2017-06-13 | 百度在线网络技术（北京）有限公司 | Intelligent article generation method and device |
| CN106997387B (en) * | 2017-03-28 | 2019-08-09 | 中国科学院自动化研究所 | Based on the multi-modal automaticabstracting of text-images match |
| CN107316248A (en) * | 2017-06-26 | 2017-11-03 | 高岚 | A kind of system and method for increasing explanatory note for picture and generating blog article |
| CN107239801B (en) * | 2017-06-28 | 2020-07-28 | 安徽大学 | Video attribute representation learning method and video character description automatic generation method |
| CN108062364A (en) * | 2017-12-05 | 2018-05-22 | 优酷网络技术（北京）有限公司 | Information displaying method and device |
-
2018
- 2018-12-14 CN CN201811535231.7A patent/CN109408672B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN109408672A (en) | 2019-03-01 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109408672B (en) | Article generation method, article generation device, server and storage medium | |
| US10277946B2 (en) | Methods and systems for aggregation and organization of multimedia data acquired from a plurality of sources | |
| US10165307B2 (en) | Automatic recognition of entities in media-captured events | |
| CN106354861B (en) | Film label automatic indexing method and automatic indexing system | |
| CN108694223B (en) | User portrait database construction method and device | |
| US20200134456A1 (en) | Video data processing method and apparatus, and readable storage medium | |
| US10108709B1 (en) | Systems and methods for queryable graph representations of videos | |
| KR101816113B1 (en) | Estimating and displaying social interest in time-based media | |
| US8750681B2 (en) | Electronic apparatus, content recommendation method, and program therefor | |
| CN109862397B (en) | Video analysis method, device, equipment and storage medium | |
| US8909617B2 (en) | Semantic matching by content analysis | |
| US20150296228A1 (en) | Systems and Methods for Performing Multi-Modal Video Datastream Segmentation | |
| CN108520046B (en) | Method and device for searching chat records | |
| US9652534B1 (en) | Video-based search engine | |
| WO2023011094A1 (en) | Video editing method and apparatus, electronic device, and storage medium | |
| CN113010703B (en) | Information recommendation method and device, electronic equipment and storage medium | |
| CN111368141B (en) | Video tag expansion method, device, computer equipment and storage medium | |
| CN111444357A (en) | Content information determination method and device, computer equipment and storage medium | |
| US10897658B1 (en) | Techniques for annotating media content | |
| CN111783712A (en) | Video processing method, device, equipment and medium | |
| CN111314732A (en) | Method for determining video label, server and storage medium | |
| CN112287168A (en) | Method and apparatus for generating video | |
| JP2014153977A (en) | Content analysis device, content analysis method, content analysis program, and content reproduction system | |
| Husa et al. | HOST-ATS: automatic thumbnail selection with dashboard-controlled ML pipeline and dynamic user survey | |
| CN113301382B (en) | Video processing method, device, medium, and program product |
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 |