KR20190063352A - Apparatus and method for clip connection of image contents by similarity analysis between clips - Google Patents

Abstract

An embodiment of the present invention relates to a clip connection apparatus for analyzing similarity with clips of other image content including subject or keyword distributions of similar clips based on clip subject or keyword distributions, and to a method thereof. According to the present invention, an image content clip connection method executed by an image content clip connection apparatus comprises the following steps: extracting one or more clips with a preset length from image content, and extracting a clip text corresponding to the clips from an image content text corresponding to the image content; using the image content text to generate an image content subject about the image content, and using the image content subject and the clip text to generate a clip subject about the extracted clip; and generating a reverse index about a word representing the clip subject, and calculating the distance between words when the same or similar word exists between the clip and other clips to analyze the similarity of the clip subject.

Description

[0001] APPARATUS AND METHOD FOR CLIP CONNECTION OF IMAGE CONTENTS BY SIMILARITY ANALYSIS BETWEEN CLIPS [0002]

More particularly, the present invention relates to an apparatus and a method for connecting clips of video contents by analyzing similarities between clips according to similarity of clip topics or similarity of clip keywords, .

In order to link image contents, conventionally, information such as a broadcast date, a production company, an actor information, a shooting place, and a background place for the image contents are manually tagged as metadata and utilized for searching or recommending image contents.

Such a technique may be difficult to determine how relevant the tagged metadata is to the image content. Therefore, research that can extract information about image contents automatically and confirm the association with image contents is needed.

According to an embodiment, a clip connection apparatus and method for generating a distribution of a clip subject and a distribution of a clip keyword based on a clip text, and expressing the meaning of the clip using a distribution of the clip subject and a distribution of the clip keyword .

According to one embodiment, it may be a clip connector and method for analyzing similarities with clips of other video content that include distributions of similar clip topics and distribution of clip keywords based on the distribution of clip topics or the distribution of clip keywords .

According to one embodiment, a clip connection apparatus and method for providing information that can link clips of other image contents having a similar distribution of clip subjects, or clips of other image contents having a similar distribution of clip keywords.

According to an aspect of the present invention, there is provided a clip connection method for video content performed by a clip connection device for video content, the method comprising: extracting one or more clips of a predetermined length from the video content; Extracting a corresponding clip text; Generating an image content subject for the image content using the image content text and generating a clip theme for the extracted clip using the image content theme and the clip text; And generating an inverse index related to a word representing the clip subject and analyzing the similarity of the clip subject by calculating distances between the words when there is the same or a significant word between the clip and the other clip And may be a clip connection method of contents.

Wherein the distribution of the clip topics comprises a set of identifiers of clip subjects and weights of clip subjects, the clip topics being a clip association method of video content comprising a set of words, a weight of words and a set of object names of words have.

The step of analyzing the degree of similarity of the clip subject may be a method of concatenating video contents that takes into consideration the identity of an entity name of the word when computing the distance between the words.

When analyzing the similarity of the clip subject by using the word and the object name of the word, considering the weight of the word and the weight of the clip subject, And may be a clip connection method of image contents for analyzing the similarity.

According to an aspect of the present invention, there is provided a clip connection method for video content performed by a clip connection device for video content, the method comprising: extracting one or more clips of a predetermined length from the video content; Extracting a corresponding clip text; Generating an image content subject for the image content using the image content text and generating a clip theme for the extracted clip using the image content theme and the clip text; Generating a clip keyword for the clip based on a weight of a word associated with the clip subject and a weight of a clip subject; And generating an inverse index related to the clip keyword and calculating a distance between the clip keywords if there is the same or a significant clip keyword between the clip and another clip to analyze the similarity of the clip keyword. Lt; / RTI >

Wherein the distribution of the clip keyword includes a clip keyword, a weight of a clip keyword, and a set of object names of a clip keyword, and the step of analyzing the similarity of the clip keyword includes: May be a method of linking video content clips that considers identity.

When the similarity degree of the clip keyword is analyzed using the clip keyword and the object name of the clip keyword, the weight of the clip keyword is taken into consideration, and thereby the same or a significant clip keyword or a clip keyword among the clips including the object name of the clip keyword And may be a clip connection method of image contents for analyzing the similarity.

According to an aspect of the present invention, there is provided a clip connection method for video content performed by a clip connection device for video content, the method comprising: extracting one or more clips of a predetermined length from the video content; Extracting a corresponding clip text; Generating a video content theme for the video content using the video content text, generating a clip theme for the extracted clip using the video content theme and the clip text, And generating a clip keyword for the clip based on the weight of the clip subject; And analyzing the similarity of the clip by calculating the distance between the words and the distance between the clips when the same keyword or the same word and the keyword of the clip are present using the clip subject and the clip keyword, Connection method.

Wherein the distribution of the clip topics includes a set of identifiers of clip subjects and weights of clip subjects, and the distribution of the clip keywords includes a clip keyword, a weight of clip keywords, and a clip connection Lt; / RTI >

According to an aspect of the present invention, there is provided a clip connection method of video content performed by a clip connection device of video content, the method comprising: identifying clip topics and clip keywords tagged in clips extracted from video content; Generating an inverse index related to a clip keyword and a reverse index of a word included in the clip topic using the clip topic and the clip keyword; Determining a degree of similarity between clips using an inverse index of a word generated in another image content and an inverse index of a clip keyword.

According to an aspect of the present invention, there is provided a clip connection device for video content, the clip connection device for video content including a processor, the processor extracting one or more clips of a preset length from the video content, Extracting a clip text corresponding to the clip from the image content text, generating an image content subject for the image content using the image content text, and extracting the extracted clip And generates a reverse index related to a word representing the clip topic. If there is an identical or significant word between the clip and another clip, the distance between the words is calculated to calculate the similarity of the clip topic And may be a clip connection device of the image contents to be analyzed.

Wherein the distribution of the clip topics comprises a set of identifiers of clip topics and weights of clip topics, the clip topics including a set of words, a weight of words and a set of object names of words, have.

The processor may be a clip connection device for video content that considers the identity of an entity name of the word when calculating the distance between the words to analyze the similarity of the clip subject.

The processor analyzes the similarity of the clip subject by using the word and the object name of the word, and by considering the weight of the word and the weight of the subject of the clip, And may be a clip connection device of image contents for analyzing the similarity of the subject of the clip between them.

A clip connection device for video content, the clip connection device for video content comprising a processor, the processor extracting one or more clips of a predetermined length from the video content, Extracting a clip text corresponding to the clip, generating an image content subject for the image content using the image content text, and extracting a clip theme for the extracted clip using the image content theme and the clip text Generating a clip keyword for the clip based on a weight of a word associated with the clip topic and a weight of the clip subject, generating a reverse index for the clip keyword, If there is one clip keyword, the clip keyword And may be a connection device of video content clips for analyzing the similarity of clip keywords.

Wherein the distribution of the clip keyword includes a clip keyword, a weight of a clip keyword, and a set of object names of a clip keyword, and the step of analyzing the similarity of the clip keyword includes: May be a linking device for video content clips that considers identity.

The processor analyzes the similarity of the clip keyword using the clip keyword and the object name of the clip keyword, and when considering the similarity of the clip keyword, A clip connection device for analyzing the similarity of the clip keywords among the clip contents.

According to an aspect of the present invention, there is provided a clip connection device for video content, the clip connection device for video content including a processor, the processor extracting one or more clips of a preset length from the video content, Extracting a clip text corresponding to the clip from the image content text, generating an image content subject for the image content using the image content text, and extracting the extracted clip Generating a clip keyword for the clip based on a weight of a word associated with the clip topic and a weight of the clip subject and generating a clip keyword for the clip using the same or a different If there is a word and a clip keyword, It may be Li, and clip connector of the video content, to analyze the degree of similarity of time by calculating the distance between the clip keyword.

Wherein the distribution of the clip topics includes a set of identifiers of clip subjects and weights of clip subjects, and the distribution of the clip keywords includes a clip keyword, a weight of clip keywords, and a clip connection Device.

According to an aspect of the present invention, there is provided a clip connection device for video content, the clip connection device for video content comprising a processor, the processor identifying clip topics and clip keywords tagged in clips extracted from video content, A reverse index of a word included in a clip subject and a clip keyword are generated using the clip subject and the clip keyword, and an inverse index of a word generated in another image content and a reverse index of a clip keyword are used And a clip connection device for video content that determines the similarity between the clips.

According to an embodiment, a clip connection apparatus and method for generating a distribution of a clip subject and a distribution of a clip keyword based on a clip text, and expressing the meaning of the clip using a distribution of the clip subject and a distribution of the clip keyword .

According to one embodiment, it may be a clip connector and method for analyzing similarities with clips of other video content that include distributions of similar clip topics and distribution of clip keywords based on the distribution of clip topics or the distribution of clip keywords .

According to one embodiment, a clip connection apparatus and method for providing information that can link clips of other image contents having a similar distribution of clip subjects, or clips of other image contents having a similar distribution of clip keywords.

1 is a diagram illustrating a theme and a keyword of a clip extracted from an image content according to an exemplary embodiment of the present invention.
2 is a diagram illustrating a process performed by a clip connection device for video content according to an exemplary embodiment of the present invention.
3 is a diagram illustrating a clip connection method of video contents performed by a clip connection device for video content according to an exemplary embodiment.
4 is a diagram illustrating a clip connection method of video contents performed by a clip connection device for video contents according to another embodiment.
5 is a diagram illustrating a clip connection method of video content performed by a clip connection apparatus for video content according to another embodiment.
FIG. 6 is a diagram illustrating a clip connection method of image contents performed by a clip connection device of image contents according to another embodiment.

Specific structural or functional descriptions of embodiments are set forth for illustration purposes only and may be embodied with various changes and modifications. Accordingly, the embodiments are not intended to be limited to the particular forms disclosed, and the scope of the present disclosure includes changes, equivalents, or alternatives included in the technical idea.

The terms first or second, etc. may be used to describe various elements, but such terms should be interpreted solely for the purpose of distinguishing one element from another. For example, the first component may be referred to as a second component, and similarly, the second component may also be referred to as a first component.

It is to be understood that when an element is referred to as being "connected" to another element, it may be directly connected or connected to the other element, although other elements may be present in between.

The singular expressions include plural expressions unless the context clearly dictates otherwise. In this specification, the terms "comprises ", or" having ", and the like, are used to specify one or more of the described features, numbers, steps, operations, elements, But do not preclude the presence or addition of steps, operations, elements, parts, or combinations thereof.

Unless otherwise defined, all terms used herein, including technical or scientific terms, have the same meaning as commonly understood by one of ordinary skill in the art. Terms such as those defined in commonly used dictionaries are to be interpreted as having a meaning consistent with the meaning of the context in the relevant art and, unless explicitly defined herein, are to be interpreted as ideal or overly formal Do not.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings.

1 is a diagram illustrating a theme and a keyword of a clip extracted from an image content according to an exemplary embodiment of the present invention.

According to one embodiment, the image content 110, 120- may include content that is broadcast / propagated through the medium. For example, the video content may include video content that includes video and audio that is broadcast on a TV.

The clip may be extracted from the image contents 110 and 120 - as a part included in the image contents. For example, clip 1 111 may be extracted from image content 110 and clip 2 112 may be extracted from image content 110. Clip 1 121 may be extracted from image content 120 and clip 2 122 may be extracted from image content 120. That is, the clip may be extracted from the same image content or extracted from other image content.

According to one embodiment, the clip may be extracted from the image content according to a predetermined length. Here, the predetermined length may be set automatically or manually. For example, the clip 1 111 can be extracted from the video content 110 according to the start time and end time corresponding to the playback time of the clip 1 111 in the entire time of the video content 110. Alternatively, the clip 2 112 may be extracted from the video content 110 according to the start time and end time corresponding to the playback time of the clip 2 112 in the entire time of the video content 110.

According to one embodiment, when a clip is extracted, the clip text corresponding to the clip can be extracted from the image content text corresponding to the image content. At this time, the image content text may include texts directly describing all related to the image content such as script, caption, and screen caption data. Alternatively, the video content text may include text that indirectly describes the video content, such as a blog, SNS, or news. Here, the indirectly describing text may include all the text including the content related to the image content after the image content is reproduced.

For example, the clip text may only contain text portions corresponding to clips extracted from texts that directly describe the video content, such as transcripts and subtitles. Alternatively, the clip text may include text describing the clip extracted from the text that indirectly describes the video content. Therefore, the clip text may include a script for a time corresponding to the playback time of the clip, a corresponding portion of the subtitle, or a portion described in the blog for a portion corresponding to the playback time of the clip.

As another example, a video content text broadcast through a TV may include a script describing the video content. At this time, the scenario includes time information synchronized with the audio of the video content, and can be displayed together with the playback of the video content in real time. At this time, not only the dialogue of the speaker but also the contents of the script including the information related to the effect sound, the OST, and the atmosphere can be displayed together.

Thus, the clip text may mean text during the length of the clip extracted from the video content text. For example, the clip text may mean a text including a description of clip 1 through a script, a blog, or the like during reproduction of clip 1 (111).

A clip subject corresponding to the extracted clip by analyzing the video content text and the clip text can be generated. Alternatively, the distribution of the image content subject may be generated using the image content text, and the clip subject may be generated using the distribution of the image content subject and the clip text. Here, the distribution of the image content subject may indicate that at least one image content subject is distributed.

For example, one or more video content topics may be generated using subtitles and / or scripts that are video content text. As more than one image content subject is created, the distribution of the image content subject may be created. One or more clip topics may be created using the distribution of the generated image content subject and clip text, and one or more clip topics may be distributed to the distribution of clip subjects. At this time, the clip subject may mean the subject of the corresponding clip.

According to one embodiment, a list of other clips having the same or similar theme, content, and meaning as the clip may be provided, reflecting the distribution of the clip theme of the video content and the distribution of the keyword.

2 is a diagram illustrating a process performed by a clip connection device for video content according to an exemplary embodiment of the present invention. The clip connector of the video content may include a processor, and the similarity between the clips may be determined by the processor.

The clip connection device 200 of the image content can receive the image content and the image content text. For example, the received video content may include a video content that includes a screen and audio played on the TV. In addition, the video content text may include text that directly explains all of the video content, such as script, caption, and screen caption data. Alternatively, the video content text may include text that indirectly describes the video content, such as a blog, SNS, or news. Here, the indirectly describing text may include all text associated with the image content after the image content is reproduced.

For example, the clip text may only contain text portions corresponding to clips extracted from texts that directly describe the video content, such as transcripts and subtitles. Alternatively, the clip text may include text describing the clip extracted from the text that indirectly describes the video content. Therefore, the clip text may include a script for a time corresponding to the playback time of the clip, a corresponding portion of the subtitle, or a portion described in the blog for a portion corresponding to the playback time of the clip.

As another example, a video content text broadcast through a TV may include a script describing the video content. At this time, the scenario includes time information synchronized with the audio of the video content, and can be displayed together with the playback of the video content in real time. At this time, not only the dialogue of the speaker but also the contents of the script including the information related to the effect sound, the OST, and the atmosphere can be displayed together.

The clip connector 200 of the image content can extract a clip of a predetermined length from the received image content. Here, the predetermined length may be set automatically or manually.

For example, the clip can be extracted from the image content in accordance with the start time and end time corresponding to the clip play time in the entire time of the video content. As another example, a clip may be extracted from a section including the same / similar topic by analyzing the connection relationship between the scenes included in the video content. As another example, the clip can be extracted according to the scene change of the image content. The clip is not limited to this example, and may include other examples.

The clip connector 200 of the image contents can extract the clip text corresponding to the extracted clip. The clip text may mean text during the length of the clip extracted from the video content text.

For example, the clip text may represent text corresponding to a predetermined time of start time and end time, text of a scene containing the same / similar subject, or text before the scene change. As another example, the clip text may include text describing the clip text, a description of the portion corresponding to the clip, such as a blog, SNS, etc., after being broadcast.

The clip connection device 200 of the image content may generate one or more image content topics using the image content text. The clip connection device 200 of the image contents can analyze the words included in the image content text. At this time, analysis methods can include methods such as co-occurrence and semantic distance.

For example, the clip connection device 200 for image contents learns words included in the image content text, and co-generates co-occurrences of words occurring on the same topic regardless of the position of the words in the image content text, occurrence can be measured. In another example, the clip connector 200 of the image contents can measure a word distance indicating how far two words have occurred at the same time when two words included in the image content text are generated at the same time have.

According to one embodiment, the clip connector 200 of the image content may generate weight values of the image content topic and the image content topic using words measured by the co-occurrence and semantic distance. At this time, the image content subject may be expressed as a set of {word, weight of word, object name of word} contained in the image content subject as an embodiment. Thus, the distribution of the image content subject to the image content can be generated.

Here, the word distribution representing the subject of the clip may include a set of {word, weight of clip subject, weight of word}. At this time, the weight of the word may be available after normalizing the word representing the image content subject. Here, the word distribution representing the clip subject can be determined while the distribution of the image content subject is generated, and the normalization can represent the process of setting the sum of the weights for the word distribution representing the subject of each clip to 1. [ At this time, the normalization may be performed selectively.

According to one embodiment, the distribution of the image content subject may comprise a set of {the identifier of the image content subject, the weight of the image content subject}. The distribution of the clip subject for each clip can be generated using the distribution of the image content subject and the clip text for the clip. Here, the distribution of the clip theme may indicate that one or more clip topics are distributed.

The distribution of the clip subject can be generated by distributing the distribution of the image content subject to the extracted clip of a certain length. Here, using the clip text, the distribution of the video content subject in which one or more video content subjects are distributed can be distributed to the clip. In one embodiment, the distribution of the image content subject may be normalized within the clip after being distributed to the clip. Likewise, normalization can be performed selectively.

According to one embodiment, when computing the similarity of a set of {word, word weight, word object name} sets, the similarity between words may be determined using a general algorithm such as cosine similarity, jacquard similarity, Can be used.

According to an embodiment, when a distribution of the image content subject is generated and redistributed to the corresponding clip, the unused word in the clip text may be used as the clip subject. That is, words that are not included in the clip text can be included in the clip theme.

Here, the distribution of the clip topics may include {a set of identifiers of clip subjects, a weight of clip subjects}. At this time, each clip theme is {word, weight of word. The object name of the word}, and the clip subject may be represented using one or more words.

At this time, the clip connection device of the image contents can distinguish words having homonyms by using the object names of words. In addition, the clip connection device of the image contents can be differentiated and utilized according to the object name of the word. Thus, by using the word ' s entity name, a word included in the distribution of the clip subject and / or the distribution of the keyword can be selectively used for the service. For example, when providing a service using a keyword for 'person' among a plurality of keywords used for providing a service, an object name of a word such as 'person, actor, character' among a plurality of object names is selected , And only keywords having an object name of such a word can be used for the service.

The clip connector 200 of the image contents can generate the distribution of clip keywords based on the distribution of the clip subjects. Here, the clip keyword may include a word indicating the clip theme, and the distribution of the clip keyword may be generated so as to correspond to the clip in which the distribution of the clip subject is generated. For example, the clip in which the distribution of the clip theme is generated and the clip in which the distribution of the clip keyword is generated may be the same clip.

According to one embodiment, when calculating the similarity between clips using a set of {clip keyword, weight of clip keyword, object name of clip keyword}, the similarity between clip keywords is calculated by a general algorithm such as cosine similarity and jacquard similarity Or use a plurality of algorithms.

According to one embodiment, the distribution of the clip keywords can be generated by analyzing the distribution of the subject of the clip and the word representing the subject of the clip, and considering the weight of the clip subject and the word weight in each clip subject. Here, the distribution of the clip keywords may include {set of clip keywords, weight of clip keywords, object names of clip keywords}.

Here, the clip keyword can be selected in consideration of the weight of the clip subject and the weight of the word. For example, when the value obtained by multiplying the weight of the clip subject by the weight of the word is equal to or greater than a preset threshold value, it can be selected as a clip keyword. Alternatively, the value obtained by multiplying the weight of the clip subject by the weight of the word may be sorted, and the word located at the upper level may be selected as the clip keyword. Alternatively, if a word in the clip theme is used more than a predetermined number of times, the word may be selected as a clip keyword.

In this case, when the clip keyword is selected, the weight of the clip keyword may be set to a value obtained by multiplying the weight of the clip subject by the word weight.

Further, the clip connection device of the image contents can distinguish clip keywords having homonyms by using the object names of the clip keywords. In addition, the clip connection device of the image contents can be differentiated and utilized according to the object name of the clip keyword. Therefore, by using the object name of the clip keyword, the clip keyword included in the distribution of the clip subject can be selectively used for the service. For example, when a service is provided by using a clip keyword for 'person' among a plurality of clip keywords used for providing a service, among the plurality of object names, objects of clip keywords such as 'person, actor, character' And a clip keyword having an object name of such a clip keyword can be used for the service.

The clip connector 200 of the image contents can analyze the similarity of the clip subjects between a plurality of clips included in the image contents. At this time, when analyzing the similarity of the clip subject, the weight of the clip subject and the weight of the word in the clip subject can be used.

Also, when analyzing the similarity of the clip subjects, it is possible to take into consideration the word and / or the word entity name of the clip subject. Thus, the similarity between clip subjects including one or more identical or similar words and / or entity names of words can be analyzed.

For example, when analyzing the similarity of a clip subject using a word and an object name of a word, the weight of the word and the weight of the subject of the clip are considered, The similarity of clip subjects can be analyzed.

The clip connector 200 of the image contents may analyze the similarity of clip keywords among a plurality of clips including the same or a significant clip keyword or an object name of the clip keyword. At this time, when analyzing the similarity degree of the clip keyword, the similarity of the clip keyword between the clips including the same or a significant clip keyword or the object name of the clip keyword can be analyzed by considering the clip keyword weight. Thus, the similarity between clip subjects including one or more of the same or similar clip keywords and / or object names of clip keywords can be analyzed.

According to one embodiment, the clip connector 200 of the image content may generate a ranking list in consideration of the similarity of the clip subject, based on a specific clip. Here, the higher ranking of the ranking list may mean that the similarity between the specific clip and the subject of the clip is high. At this time, the ranking list can be generated according to the similarity of the clip subjects between the specific clip and the rest of the entire clip. Further, when the clip connection device of the image contents receives new image contents, the ranking list may be updated.

Or, depending on the ranking list, you can link other clips with similar clips and clip themes. At this time, the ranking list can be tagged as metadata in the clip, and the clip connection device of the image contents can link the clip of the higher rank using the metadata.

According to another embodiment, the clip connector 200 of the image content may generate a ranking list in consideration of the similarity of clip keywords based on a specific clip. Here, the higher ranking of the ranking list may mean that the similarity between a specific clip and a clip keyword is high. At this time, the ranking list can be generated according to the similarity of clip keywords between a specific clip and the rest of the whole clip. Further, when the clip connection device of the image contents receives new image contents, the ranking list may be updated.

Or, depending on the ranking list, you can link other clips with similar clip and clip keywords. At this time, the ranking list can be tagged as metadata in the clip, and the clip connection device of the image contents can link the clip of the higher rank using the metadata.

3 is a diagram illustrating a clip connection method of video contents performed by a clip connection device for video content according to an exemplary embodiment.

In operation 310, the clip connection device of the image content extracts one or more clips having a predetermined length from the image content, and extracts the clip text corresponding to the clip from the image content text corresponding to the image content.

The clip connection device of the image contents can receive the image contents and the image contents text. For example, the received video content may include all video content broadcasted by the broadcasting station.

At this time, the image content text may include text describing the content of the image content. Here, the contents of the image contents may include all items included in the image contents such as script, OST, atmosphere, and background. In addition, the image content text may include not only direct text such as a script, but also indirect text such as a number of blogs, SNS, and news retrieved through image contents after the image contents are reproduced. For example, after the image content is broadcasted, the image content text may include a blog or an SNS in which the viewer of the image content describes his or her evaluation of a specific portion of the image content on his / her blog.

According to one embodiment, the clip connection device of the image content may extract a clip from the received image content. Here, the clip has a predetermined length set automatically or manually, and the predetermined length may be different depending on the image content.

For example, when the video content is a gag program, the video content can be generated by a number of small units of gag. Therefore, among the entire playback time of the video content, a clip corresponding to the playback time of the gag in a small unit can be extracted.

As another example, when the image content is a news program, the image content can be generated by a large number of small units of news. Therefore, among the entire playback time of the video content, the clip corresponding to the playback time of the news in a small unit can be extracted. However, the extraction of the clip is not limited to the above example, and the clip can be extracted by another method.

The clip connection device of the image contents can extract the clip text corresponding to the extracted clip. The clip text may mean text during the length of the clip extracted from the video content text. Accordingly, the clip text may include text corresponding to the start time and end time, and the clip text may include time information so that it can be synchronized with the reproduction of the clip in real time.

For example, the clip text may include a small unit of gag script or may include text posted on a blog for a small unit of gag after the video content has been broadcast.

As another example, the clip text may include a script of a small unit of news, or may include text posted to the SNS for news in small units since the video content was broadcast.

In step 320, the clip connection device of the image content may generate an image content subject for the image content using the image content text, and generate a clip theme for the extracted clip using the image content theme and clip text have.

The clip connection device of the image content can generate the image content theme using the image content text. The clip connection device of the image contents can analyze the words included in the image contents text. For example, the clip connection device of the image contents may generate a subject for the image contents by analyzing a script or SNS regarding the image contents.

At this time, analysis methods can include methods such as co-occurrence and semantic distance. However, the present invention is not limited to such an analysis method, and may include analysis of image content text by another analysis method.

For example, it is possible to learn the co-occurrences of co-occurring words occurring on the same subject regardless of the position of the word in the image content text by learning words contained in the image content text. As another example, if the two words contained in the image content text occur simultaneously, the word distance can be measured to indicate how far apart the two words occurred.

According to one embodiment, weights of image content subject and image content subject can be generated using words measured by coincidence and semantic distance. At this time, the image content subject can be expressed as a set of {word, weight of word, object name of word} contained in the image content subject as an embodiment. Here, one or more image content subjects may be generated, and the distribution of the image content subjects for the image content may be generated by using the weight of the image content subject.

According to one embodiment, the distribution of the image content subject may comprise a set of {the identifier of the image content subject, the weight of the image content subject}. A clip theme for each clip can be created using the distribution of the video content subject and the clip text for the clip, and the distribution of the clip theme can indicate that one or more clip topics are distributed.

The distribution of the clip subject can be generated by distributing the distribution of the image content subject to the extracted clip of a certain length. In one embodiment, the distribution of the image content subject may be normalized within the clip after being distributed to the clip.

According to an embodiment, when a distribution of the image content subject is generated and redistributed to the corresponding clip, the unused word in the clip text may be used as the clip subject. That is, words that are not included in the clip text can be included in the clip theme. For example, a clip theme can be expressed using a word that is not used in the clip text but included in the video content theme.

Here, the distribution of the clip topics may include {a set of identifiers of clip subjects, a weight of clip subjects}. At this time, each clip theme may include a set of {word, weight of word, object name of word}, and the clip theme may be expressed using one or more words. At this time, using the identifier of the clip subject, the clip connector can find the clip subject.

In operation 330, the clip connection device of the image content generates an inverse index related to a word representing a clip theme. If there is an identical or significant word between the clip and another clip, the distance between the words is calculated, The degree of similarity can be analyzed.

Therefore, it is possible to analyze the similarity of the clip theme and determine the similarity with other clips according to the order of similarity of the clip theme.

For example, when an index is displayed on data including a large number of words, if one word is set as a reverse index, the clip text including the word using the reverse index, the image content text, and the like can be linked. Thus, using the inverse index, a clip or image content to which a reverse index is connected among a large amount of data can be selected and used for analyzing the similarity of the image content subject and / or the similarity of the clip subject. Therefore, the data retrieval speed and the number of objects to be computed after retrieval can be reduced.

In addition, in order to distinguish homonyms, an inverted index can be generated using {word, word object name}. For example, {apple, fruit} and {apple, action} contain the same word apple, but can represent other reverse indices. Because the apple in front is a kind of fruit, and the apples in the back indicate appology.

The clip connection device of the image contents can analyze the similarity of the clip subjects among a plurality of clips included in the image contents. At this time, when analyzing the similarity of the clip subject, the weight of the clip subject and the weight of the word in the clip subject can be used.

Also, when analyzing the similarity of the clip subject, the clip connection device of the image contents may consider the word and / or the object name of the clip subject. Thus, the similarity between clip subjects including one or more identical or similar words and / or entity names of words can be analyzed.

For example, when analyzing the similarity of a clip subject by using word and word object names, the weight of the word and the weight of the clip subject are taken into account, so that the content of the clip subject between the clips containing the same or a meaningful word or word object name The similarity can be analyzed.

According to an exemplary embodiment, the clip connection device of the image contents may generate a ranking list in consideration of the similarity of the clip subject, based on a specific clip. Here, the higher ranking of the ranking list may mean that the similarity between the specific clip and the subject of the clip is high. At this time, the ranking list can be generated according to the similarity of the clip subjects between the specific clip and the rest of the entire clip. Further, when the clip connection device of the image contents receives new image contents, the ranking list may be updated.

For example, the clip connection device of the image content may display the similarity between the clip theme for a particular gag in the gag program or between gag clip topics in another gag program in the ranking list. In this case, the ranking of the ranking list reflects the ranking according to consideration of the same or a meaningful word or word entity name by considering the weight of the word representing the gag and the weight of the clip subject.

According to another embodiment, it is possible to link other clips with similar clip and clip theme according to the ranking list. At this time, the ranking list can be tagged as metadata in the clip, and the clip connection device of the image contents can link the clip of the higher rank using the metadata. For example, by using the metadata, the clip connection device of the image contents can continuously reproduce the gag having the similar clip theme.

The meta data includes not only the ranking list but also the identifier of the clip, the start time and end time of the clip in the image content, the identifier of the clip subject and the set of weights of the clip subject, the weight of words and words included in each clip theme, A clip keyword, a weight of a clip keyword, and a set of object names of a clip keyword.

4 is a diagram illustrating a clip connection method of video contents performed by a clip connection device for video contents according to another embodiment.

In operation 410, the clip connection device of the image content extracts one or more clips having a predetermined length from the image content, and extracts the clip text corresponding to the clip from the image content text corresponding to the image content.

The clip connection device of the image contents can receive the image contents and the image contents text. For example, the received video content may include all video content broadcasted by the broadcasting station.

At this time, the image content text may include text describing the content of the image content. Here, the contents of the image contents may include all items included in the image contents such as script, OST, atmosphere, and background. In addition, the image content text may include not only direct text such as a script, but also indirect text such as a number of blogs, SNS, and news retrieved through image contents after the image contents are reproduced. For example, after the image content is broadcasted, the image content text may include a blog or an SNS in which the viewer of the image content describes his or her evaluation of a specific portion of the image content on his / her blog.

According to one embodiment, the clip connection device of the image content may extract a clip from the received image content. Here, the clip has a predetermined length set automatically or manually, and the predetermined length may be different depending on the image content.

For example, when the video content is a gag program, the video content can be generated by a number of small units of gag. Therefore, among the entire playback time of the video content, a clip corresponding to the playback time of the gag in a small unit can be extracted.

As another example, when the image content is a news program, the image content can be generated by a large number of small units of news. Therefore, among the entire playback time of the video content, the clip corresponding to the playback time of the news in a small unit can be extracted. However, the extraction of the clip is not limited to the above example, and the clip can be extracted by another method.

The clip connection device of the image contents can extract the clip text corresponding to the extracted clip. The clip text may mean text during the length of the clip extracted from the video content text. Accordingly, the clip text may include text corresponding to the start time and end time, and the clip text may include time information so that it can be synchronized with the reproduction of the clip in real time.

For example, the clip text may include a small unit of gag script or may include text posted on a blog for a small unit of gag after the video content has been broadcast.

As another example, the clip text may include a script of a small unit of news, or may include text posted to the SNS for news in small units since the video content was broadcast.

In step 420, the clip connection device of the image content may generate the image content subject for the image content using the image content text and generate the clip subject for the extracted clip using the image content theme and clip text have.

The clip connection device of the image content can generate the image content theme using the image content text. The clip connection device of the image contents can analyze the words included in the image contents text. For example, the clip connection device of the image contents may generate a subject for the image contents by analyzing a script or SNS regarding the image contents.

At this time, analysis methods can include methods such as co-occurrence and semantic distance. However, the present invention is not limited to such an analysis method, and may include analysis of image content text by another analysis method.

For example, it is possible to learn the co-occurrences of co-occurring words occurring on the same subject regardless of the position of the word in the image content text by learning words contained in the image content text. As another example, if the two words contained in the image content text occur simultaneously, the word distance can be measured to indicate how far apart the two words occurred.

According to one embodiment, the clip connection device of the image contents can generate the weight of the image content subject and the image content subject by using the words measured by the co-occurrence and the semantic distance. At this time, the image content subject can be expressed as a set of {word, weight of word, object name of word} contained in the image content subject as an embodiment. Here, one or more image content subjects may be generated, and the distribution of the image content subjects for the image content may be generated by using the weight of the image content subject.

According to one embodiment, the distribution of the image content subject may comprise a set of {the identifier of the image content subject, the weight of the image content subject}. A clip theme for each clip can be created using the distribution of the video content subject and the clip text for the clip, and the distribution of the clip theme can indicate that one or more clip topics are distributed. At this time, the clip connection device of the image contents can find the image content subject by using the identifier of the image content subject.

The distribution of the clip subject can be generated by distributing the distribution of the image content subject to the extracted clip of a certain length. In one embodiment, the distribution of the image content subject may be normalized within the clip after being distributed to the clip.

According to an embodiment, when a distribution of the image content subject is generated and redistributed to the corresponding clip, the unused word in the clip text may be used as the clip subject. That is, words that are not included in the clip text can be included in the clip theme. For example, a clip theme can be expressed using a word that is not used in the clip text but included in the video content theme.

Here, the distribution of the clip topics may include {a set of identifiers of clip subjects, a weight of clip subjects}. At this time, each clip theme may include a set of {word, weight of word, object name of word}, and the clip theme may be expressed using one or more words.

At step 430, the clip connector of the image content may generate a clip keyword for the clip, based on the weight of the word associated with the clip theme and the weight of the clip theme.

The clip connection device of the image contents can generate the distribution of the clip keyword based on the distribution of the clip subject. Here, the distribution of the clip keyword may be generated so as to correspond to the clip in which the distribution of the clip subject is generated. For example, the clip in which the distribution of the clip theme is generated and the clip in which the distribution of the clip keyword is generated may be the same clip.

According to one embodiment, the distribution of clip keywords can be generated by analyzing the distribution of clip subjects and the words representing the clip subjects, and by considering the weight of the clip subject and the weight of the word in each clip subject. Here, the distribution of the clip keywords may include {set of clip keywords, weight of clip keywords, object names of clip keywords}.

Here, the clip keyword can be selected in consideration of the weight of the clip topic and the word weight. For example, if the value obtained by multiplying the weight of the clip subject by the weight of the word is equal to or greater than a preset threshold value, the keyword may be selected. Alternatively, a value obtained by multiplying the weight of the clip subject by the word weight can be sorted, and a word located in the upper part can be selected as a keyword. Alternatively, if a word in the clip theme is used more than a predetermined number of times, the word may be selected as a clip keyword.

Accordingly, among the words expressing the clip theme, when the predetermined condition is satisfied, the keyword can be selected as the clip keyword. In this case, when the clip keyword is selected, the weight of the clip keyword may be set to a value obtained by multiplying the weight of the clip subject by the word weight.

In step 440, the clip connection apparatus of the image contents generates an inverse index related to the clip keyword, and when there is the same or a significant word between the clip and the other clip, the distance between the words is calculated to analyze the similarity of the clip subject .

Therefore, the degree of similarity of the clip keywords can be analyzed, and the degree of similarity with other clips can be determined according to the order of similarity of the clip keywords.

The clip connection device of the image contents may analyze the similarity of the clip keywords among a plurality of clips including the same or a significant clip keyword or an object name of the clip keyword. At this time, when analyzing the similarity degree of the clip keyword, the similarity degree of the clip keyword between the clips including the same or a significant clip keyword or clip keyword object name can be analyzed by considering the weight of the clip keyword. Thus, the similarity between clip topics including one or more of the same or similar clip keywords and / or clip keyword entity names can be analyzed.

For example, if one clip keyword is set to a reverse index when displaying an index in the data including a large number of words, the clip text including the clip keyword, the image content text, and the like can be linked using the reverse index . Thus, using the inverse index, a clip or image content to which a reverse index is connected among a large amount of data can be selected and used for analyzing the similarity of the image content subject and / or the similarity of the clip keyword. Therefore, the data retrieval speed and the number of objects to be computed after retrieval can be reduced.

Also, in order to distinguish homonyms, an inverse index can be generated using {clip keyword, object name of clip keyword}. For example, {apple, fruit} and {apple, action} contain apple, which is the same clip keyword, but can represent another reverse index. Because the apple in front is a kind of fruit, and the apples in the back indicate appology.

According to one embodiment, the clip connection device of the image contents can generate a ranking list in consideration of the similarity of clip keywords based on a specific clip. Here, the higher ranking of the ranking list may mean that the similarity between a specific clip and a clip keyword is high. At this time, the ranking list can be generated according to the similarity of clip keywords between a specific clip and the rest of the whole clip. Further, when the clip connection device of the image contents receives new image contents, the ranking list may be updated.

For example, the clip connection device of the video content may display the similarity between the clip keywords for the specific gag in the gag program or between the gag clip keywords in other gag programs in the ranking list. At this time, the ranking of the ranking list reflects the ranking in consideration of the weight of the clip keyword representing the gag.

According to another embodiment, it is possible to link other clips having similar clip and clip keywords according to the ranking list. At this time, the ranking list can be tagged as metadata in the clip, and the clip connection device of the image contents can link the clip of the higher rank using the metadata. For example, by using the metadata, the clip connection device of the image content can continuously reproduce the gag similar to the clip keyword.

The meta data includes not only the ranking list but also the identifier of the clip, the start time and end time of the clip in the image content, the identifier of the clip subject and the set of weights of the clip subject, the weight of words and words included in each clip theme, A clip keyword, a weight of a clip keyword, and a set of object names of a clip keyword.

5 is a diagram illustrating a clip connection method of video content performed by a clip connection apparatus for video content according to another embodiment. A detailed description of FIG. 5 can be found in FIGS. 3 and 4. FIG.

In operation 510, the clip connection device of the image content may extract one or more clips of a predetermined length from the image content, and extract clip text corresponding to the clip from the image content text corresponding to the image content.

In step 520, the clip connection device of the image contents generates an image content theme for the image content using the image content text, generates a clip theme for the clip extracted using the image content theme and the clip text , A clip keyword for the clip may be generated based on the weight of the word associated with the clip subject and the weight of the clip subject.

In step 530, the clip connection device of the image contents uses the clip theme and the clip keyword to calculate the distance between the words and the distance between the clips if there is the same or a meaningful word and clip keyword to analyze the similarity of the clip .

Therefore, according to one embodiment, the similarity between clips can be analyzed by considering both the clip theme and the clip keyword.

FIG. 6 is a diagram illustrating a clip connection method of image contents performed by a clip connection device of image contents according to another embodiment.

In step 610, the clip connector of the image content may identify clip topics and clip keywords tagged in the clip extracted from the image content. The video content and clips can be pre-analyzed such that clip topics and clip keywords can be generated, and clip topics and clip keywords can be tagged to clips. Detailed descriptions of clip topics and clip keywords are omitted.

In step 620, the clip connection device of the image content may generate an inverse index of a word included in the clip subject and a reverse index of the clip keyword, using the clip subject and the clip keyword. The detailed description of the inverse index is omitted.

In step 630, the clip connector of the image content can determine the similarity between the clips using the inverse index of the word generated in the other image content and the inverse index of the clip keyword.

The inverse index can be generated using {word, object name of word}. At this time, similarity between different image contents or clips can be analyzed by using an inverse index including the same or similar {word, word object name}. Here, the inverse index may be generated by clip theme and / or image content theme. Or the inverse index may be generated for each clip keyword.

For example, the similarity between clips including the same or similar reverse index can be analyzed. Or the similarity between image contents including the same or similar reverse index can be analyzed. Or the similarity between clip keywords including the same or similar reverse index can be analyzed.

The embodiments described above may be implemented in hardware components, software components, and / or a combination of hardware components and software components. For example, the devices, methods, and components described in the embodiments may be implemented within a computer system, such as, for example, a processor, a controller, an arithmetic logic unit (ALU), a digital signal processor, such as an array, a programmable logic unit (PLU), a microprocessor, or any other device capable of executing and responding to instructions. The processing device may execute an operating system (OS) and one or more software applications running on the operating system. The processing device may also access, store, manipulate, process, and generate data in response to execution of the software. For ease of understanding, the processing apparatus may be described as being used singly, but those skilled in the art will recognize that the processing apparatus may have a plurality of processing elements and / As shown in FIG. For example, the processing unit may comprise a plurality of processors or one processor and one controller. Other processing configurations are also possible, such as a parallel processor.

The software may include a computer program, code, instructions, or a combination of one or more of the foregoing, and may be configured to configure the processing device to operate as desired or to process it collectively or collectively Device can be commanded. The software and / or data may be in the form of any type of machine, component, physical device, virtual equipment, computer storage media, or device , Or may be permanently or temporarily embodied in a transmitted signal wave. The software may be distributed over a networked computer system and stored or executed in a distributed manner. The software and data may be stored on one or more computer readable recording media.

The method according to an embodiment may be implemented in the form of a program command that can be executed through various computer means and recorded in a computer-readable medium. The computer-readable medium may include program instructions, data files, data structures, and the like, alone or in combination. The program instructions to be recorded on the medium may be those specially designed and configured for the embodiments or may be available to those skilled in the art of computer software. Examples of computer-readable media include magnetic media such as hard disks, floppy disks and magnetic tape; optical media such as CD-ROMs and DVDs; magnetic media such as floppy disks; Magneto-optical media, and hardware devices specifically configured to store and execute program instructions such as ROM, RAM, flash memory, and the like. Examples of program instructions include machine language code such as those produced by a compiler, as well as high-level language code that can be executed by a computer using an interpreter or the like. The hardware devices described above may be configured to operate as one or more software modules to perform the operations of the embodiments, and vice versa.

Although the embodiments have been described with reference to the drawings, various technical modifications and variations may be applied to those skilled in the art. For example, it is to be understood that the techniques described may be performed in a different order than the described methods, and / or that components of the described systems, structures, devices, circuits, Lt; RTI ID = 0.0 > and / or < / RTI > equivalents,

110, 120 ~: image content
111, 121 ~: Clip
200: clip connection device for image contents

Claims (20)

A clip connection method of video content performed by a clip connection device of video content,
Extracting at least one clip having a preset length from the image content and extracting clip text corresponding to the clip from the image content text corresponding to the image content;
Generating an image content subject for the image content using the image content text and generating a clip theme for the extracted clip using the image content theme and the clip text; And
A step of generating an inverse index related to a word representing the clip subject and analyzing the similarity of the clip subject by calculating the distance between the same and a different word between the clip and the other clip
And a clipping means for clipping the clips.
The method according to claim 1,
Wherein the distribution of the clip subject includes a set of weights of the clip subject and an identifier of the clip subject,
Wherein the clip subject comprises a set of words, weight of words, and object name of a word.
3. The method of claim 2,
Wherein analyzing the similarity of the clip subject comprises:
And calculating the distance between the words when considering the identity of the entity name of the word.
The method of claim 3,
When analyzing the similarity of the clip subject by using the word and the object name of the word, considering the weight of the word and the weight of the clip subject, Clipping Method of Image Contents to Analyze Similarity.
A clip connection method of video content performed by a clip connection device of video content,
Extracting at least one clip having a preset length from the image content and extracting clip text corresponding to the clip from the image content text corresponding to the image content;
Generating an image content subject for the image content using the image content text and generating a clip theme for the extracted clip using the image content theme and the clip text;
Generating a clip keyword for the clip based on a weight of a word associated with the clip subject and a weight of a clip subject; And
Generating an inverse index related to the clip keyword and analyzing the similarity degree of the clip keyword by calculating the distance between the clip keywords if there is the same or a significant clip keyword between the clip and the other clip
The method comprising the steps of:
6. The method of claim 5,
Wherein the distribution of the clip keyword includes a clip keyword, a weight of a clip keyword, and a set of object names of a clip keyword,
Wherein the step of analyzing the similarity of the clip keywords considers identity of the clip keywords when calculating distances between the clip keywords.
The method according to claim 6,
When the similarity degree of the clip keyword is analyzed using the clip keyword and the object name of the clip keyword, the weight of the clip keyword is taken into consideration, and thereby the same or a significant clip keyword or a clip keyword among the clips including the object name of the clip keyword Clipping Method of Image Contents to Analyze Similarity.
A clip connection method of video content performed by a clip connection device of video content,
Extracting at least one clip having a preset length from the image content and extracting clip text corresponding to the clip from the image content text corresponding to the image content;
Generating a video content theme for the video content using the video content text, generating a clip theme for the extracted clip using the video content theme and the clip text, And generating a clip keyword for the clip based on the weight of the clip subject; And
Analyzing the similarity degree of the clip by calculating the distance between the words and the distance between the clip keywords when there is the same or a meaningful word and clip keyword using the clip topic and the clip keyword;
And a clipping means for clipping the clips.
9. The method of claim 8,
Wherein the distribution of the clip topics includes a set of identifiers of clip subjects and weights of clip subjects, and the distribution of the clip keywords includes a clip keyword, a weight of clip keywords, and a clip connection Way.
A clip connection method of video content performed by a clip connection device of video content,
Identifying clip topics and clip keywords tagged in clips extracted from image content;
Generating an inverse index related to a clip keyword and a reverse index of a word included in the clip topic using the clip topic and the clip keyword;
Determining a degree of similarity between clips using an inverse index of a word generated in another image content and an inverse index of a clip keyword
And a clipping means for clipping the clips.
A clip connection device for video content,
Wherein the clip connection device of the image content comprises a processor,
The processor comprising:
Extracting at least one clip having a predetermined length from the image content, extracting clip text corresponding to the clip from the image content text corresponding to the image content,
Generating an image content theme for the image content using the image content text, generating a clip theme for the extracted clip using the image content theme and the clip text,
A clip linking unit for linking an image content to an image content to generate an inverse index related to a word representing the clip theme and analyzing the similarity of the clip theme by calculating a distance between the clip and another clip, .
12. The method of claim 11,
Wherein the distribution of the clip subject includes a set of weights of the clip subject and an identifier of the clip subject,
Wherein the clip subject comprises a set of words, word weights, and object names of words.
13. The method of claim 12,
The processor comprising:
And when the distance between the words is calculated in order to analyze the similarity of the clip subject, consideration is given to the identity of the entity name of the word.
14. The method of claim 13,
The processor comprising:
When analyzing the similarity of the clip subject by using the word and the object name of the word, considering the weight of the word and the weight of the clip subject, Clip connection device for video content analyzing similarity.
A clip connection device for video content,
Wherein the clip connection device of the image content comprises a processor,
The processor comprising:
Extracting at least one clip having a predetermined length from the image content, extracting clip text corresponding to the clip from the image content text corresponding to the image content,
Generating an image content theme for the image content using the image content text, generating a clip theme for the extracted clip using the image content theme and the clip text,
Generating a clip keyword for the clip based on a weight of a word associated with the clip subject and a weight of a clip subject,
And generating an inverse index related to the clip keyword and calculating a distance between the clip keywords if there is the same or a significant clip keyword between the clip and another clip to analyze the similarity of the clip keyword.
16. The method of claim 15,
Wherein the distribution of the clip keyword includes a clip keyword, a weight of a clip keyword, and a set of object names of a clip keyword,
Wherein the step of analyzing the degree of similarity of the clip keywords considers the identity of an object name of the clip keyword when calculating the distance between the clip keywords.
17. The method of claim 16,
The processor comprising:
When the similarity degree of the clip keyword is analyzed using the clip keyword and the object name of the clip keyword, the weight of the clip keyword is taken into consideration, and thereby the same or a significant clip keyword or a clip keyword among the clips including the object name of the clip keyword Clip connection device for video content analyzing similarity.
A clip connection device for video content,
Wherein the clip connection device of the image content comprises a processor,
The processor comprising:
Extracting at least one clip having a predetermined length from the image content, extracting clip text corresponding to the clip from the image content text corresponding to the image content,
Generating a video content theme for the video content using the video content text, generating a clip theme for the extracted clip using the video content theme and the clip text, And generating a clip keyword for the clip based on the weight of the clip subject,
Wherein the similarity of the clip is analyzed by calculating the distance between the words and the distance between the clips when the same keyword or the same word and the clip keyword exist using the clip subject and the clip keyword.
19. The method of claim 18,
Wherein the distribution of the clip topics includes a set of identifiers of clip subjects and weights of clip subjects, and the distribution of the clip keywords includes a clip keyword, a weight of clip keywords, and a clip connection Device.
A clip connection device for video content,
Wherein the clip connection device of the image content comprises a processor,
The processor comprising:
Identifying clip topics and clip keywords tagged in clips extracted from image content,
Generating an inverse index related to a clip keyword and a reverse index of a word included in the clip topic using the clip topic and the clip keyword,
Wherein the similarity between the clips is determined using an inverse index of a word generated in another image content and an inverse index related to the clip keyword.

Images