System for producing promotional media content and method thereof

• US 10,311,479 B2
• Filed: 01/25/2016
• Issued: 06/04/2019
• Est. Priority Date: 10/20/2015
• Status: Active Grant

• Current Assignee: Korea Electronics Technology Institute (Government of South Korea)
• Original Assignee: Korea Electronics Technology Institute (Government of South Korea)
• Inventors: Park, Jong Bin, Lim, Tae beom, Kim, Kyung Won, Jung, Jong Jin, Moon, Jae won, Kum, Seung Woo
• Primary Examiner(s): Sittner, Matthew T
• Application Number:
  US15/006,043
• Publication Number:
  US 20170109786A1
• Time in Patent Office: 1,226 Days
• Field of Search: None
• CPC Class Codes: G06F 16/285 Clustering or classification
• CPC Class Codes: G06F 16/78 Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
• CPC Class Codes: G06F 16/901 Indexing; Data structures therefor; Storage structures for retrieval from the web G06F16/951
• CPC Class Codes: G06F 16/90335 Query processing
• CPC Class Codes: G06F 16/9535 Search customisation based on user profiles and personalisation
• CPC Class Codes: G06F 16/9538 Presentation of query results
• CPC Class Codes: G06Q 30/0256 User search

Abstract

A system for producing promotional media content includes a strategic meta information analysis processing unit configured to process various types of pieces of strategic meta information in various formats to an unstructured data batch having a certain weight value, a media content analysis processing unit configured to classify and analyze each meta information by similarity by using meta information of various types of media content or meta information actively extracted by analyzing each media content, an association analyzing unit configured to calculate association between strategic meta information processed through the strategic meta information analysis processing unit and the media content processed through the media content analysis processing unit, and a media content synthesizing unit configured to finally synthesize promotional media content by using the various types of strategic meta information, the media content, and the association information calculated by the association analyzing unit.

Claims

1. A system for producing promotional media content, the system comprising:

a strategic meta information analysis processing unit configured to process strategic meta information in a plurality of formats to an unstructured data batch having a predetermined weight value;

a semantic word extracting unit configured to extract a semantic word regarding N number of pieces of meta information regarding a plurality of media contents;

a semantic word dictionary unit configured to create a dictionary by comparing semantic words extracted by the semantic word extracting unit and words in a predefined dictionary;

a semantic word-based association analyzing unit configured to calculate similarity value between the pieces of meta information regarding the plurality of media contents on the basis of semantic words by using the dictionary generated by the dictionary unit, and classify the pieces of meta information by using the similarity value;

an association analyzing unit configured to calculate association between the strategic meta information processed through the strategic meta information analysis processing unit and the plurality of media contents processed through the semantic word-based association analyzing unit; and

a media content synthesizing unit configured to synthesize promotional media content by using the strategic meta information, the plurality of media contents, and the association information calculated by the association analyzing unit,

wherein the strategic meta information is expressed with N number of pieces of partial strategic meta information as expressed in Equation (1) below;

q=q₁+q₂+ . . . +q_N 

 

Equation (1)

wherein each of the pieces of partial strategic meta information has the predetermined weight value as expressed in Equation (2) below, and the sum of all predetermined weight values of the N number of pieces of partial strategic meta information is 1;

{w₁,w₂, . . . w_N}, w₁+w₂+ . . . +w_N=1 

 

Equation (2)

where q is strategic meta information, and w is a weight value.

2. The system of claim 1, wherein the strategic meta information in the plurality of formats include at least one of marketing strategy information, publicity strategy information, user information, user intention information, statistical service information, and feedback information.

3. The system of claim 1, wherein the plurality of media contents is classified in system into

a common resource item representing media resources such as a picture, video, music, and text information commonly used in a service system or a program;

a personal resource item representing media content that each user has directly uploaded or purchased; and

a story container representing an outcome designed by using a gain filter, a change effect, and an overlay effect in order to specifically express an abstract concept called story.

4. The system of claim 3, wherein the plurality of media contents includes at least one of a picture, video, music, text, and font.

5. The system of claim 3, wherein the common resource item, the personal resource item, and the story container include different meta information.

6. The system of claim 1, wherein the semantic word extracting unit extracts the semantic word by using

a first method of extracting the semantic word by analyzing similarity between a text string input by using a dynamic program and words in the predefined dictionary,

a second method of separating the input text string to a word set by using a regular expression, removing a meaningless word from the separated word set, and extracting the semantic word, or

a combination of the first method and the second method.

7. The system of claim 6, wherein when the semantic word extracted by the semantic word extracting unit is not included in the predefined dictionary, the semantic word dictionary unit provides an index to the semantic word and add the semantic word to the dictionary, configures meta information with words present in the created dictionary, converts the meta information into a set of dictionary indices, and stores the converted set in a memory or a file, and thereafter, whenever new media content is added, the semantic word dictionary unit updates the semantic word to create a new dictionary by repeating the process.

8. The system of claim 7, wherein the semantic word-based association analyzing unit uses at least one of topic modeling techniques including at least one of a term frequency-inverse document frequency (TF-IDF) technique, a latent semantic analysis/indexing (LSA) method, and a latent dirichlet allocation (LDA) technique.

9. The system of claim 1, wherein the media content synthesizing unit includes:

a project conversion unit configured to convert a project file as a set of meta information regarding the plurality of media contents to fit media synthesizing; and

a rendering unit configured to synthesize the plurality of media contents by using the pieces of meta information converted through the project conversion unit.

10. The system of claim 7, wherein the association analyzing unit creates a query statement by using each strategic meta information such as marketing, publicity strategy, and publicity intention, expresses the created query statement as an index by using the dictionary created by the dictionary unit, expresses the query statement expressed as the dictionary index in a vector space based on topic modeling, calculates similarity value between pieces of meta information describing the media content expressed in the topic-modeled vector space and the input query statement, sorts the pieces of meta information according to the calculated similarity value, and outputs media content having high similarity to the media content synthesizing unit.

11. A method for producing promotional media content, the method performed by a server computer, the method comprising:

processing strategic meta information in a plurality of formats to an unstructured data batch having a predetermined weight value;

extracting a semantic word regarding N number of pieces of meta information regarding a plurality of media contents;

creating a dictionary by comparing the extracted semantic word and words in a predefined dictionary;

calculating similarity value between the pieces of meta information regarding the plurality of media contents on the basis of semantic words by using the generated dictionary and classifying the pieces of meta information by using the similarity value;

calculating association between the strategic meta information and the plurality of media contents; and

synthesizing promotional media content by using the strategic meta information, the plurality of media contents, and the calculated association information,

wherein the strategic meta information is expressed with N number of pieces of partial strategic meta information as expressed in Equation (1) below;

q=q₁+q₂+ . . . +q_N 

 

Equation (1)

wherein each of the pieces of partial strategic meta information has the predetermined weight value as expressed in Equation (2) below, and the sum of all predetermined weight values of the N number of pieces of partial strategic meta information is 1;

{w₁,w₂, . . . w_N}, w₁+w₂+ . . . +w_N=1 

 

Equation (2)

where q is strategic meta information, and w is a weight value.

12. The method of claim 11, wherein the strategic meta information in the plurality of formats include at least one of marketing strategy information, publicity strategy information, user information, user intention information, statistical service information, and feedback information.

13. The method of claim 11, wherein the plurality of media contents is classified in system into

a common resource item representing media resources such as a picture, video, music, and text information commonly used in a service system or a program;

a personal resource item representing media content that each user has directly uploaded or purchased; and

a story container representing an outcome designed by using a gain filter, a change effect, and an overlay effect in order to specifically express an abstract concept called story.

14. The method of claim 13, wherein the plurality of media contents includes at least one of a picture, video, music, text, and font.

15. The method of claim 13, wherein the common resource item, the personal resource item, and the story container include different meta information.

16. The method of claim 11, wherein the extracting of a semantic word comprise extracting a semantic word by using a first method of extracting the semantic word by analyzing similarity between a text string input by using a dynamic program and words in the predefined dictionary, by using a second method of separating the input text string to a word set by using a regular expression, removing a meaningless word from the separated word set, and extracting the semantic word, or by combining the first method and the second method.

17. The method of claim 16, wherein the creating of a dictionary comprises:

when the extracted semantic word is not included in the predefined dictionary, providing an index to the semantic word and adding the semantic word to the dictionary;

configuring meta information with words present in the created dictionary, converting the meta information into a set of dictionary indices, and storing the converted set in a memory or a file; and

repeating the process to perform updating whenever new media content is added.

18. The method of claim 17, wherein the classifying of the pieces of meta information comprises classifying the pieces of meta information by using at least one of topic modeling techniques including at least one of a term frequency-inverse document frequency (TF-IDF) technique, a latent semantic analysis/indexing (LSA) method, and a latent dirichlet allocation (LDA) technique.

19. The method of claim 11, wherein the synthesizing of media content comprises:

converting a project file as a set of meta information regarding the plurality of media content to fit media synthesizing; and

rendering the plurality of media contents by using the converted pieces of meta information.

20. The method of claim 17, wherein the calculating of association comprises:

creating a query statement by using each strategic meta information such as marketing, publicity strategy, and publicity intention;

expressing the created query statement as an index by using the dictionary created by the dictionary unit;

expressing the query statement expressed as the dictionary index in a vector space based on topic modeling;

calculating similarity value between pieces of meta information describing the media content expressed in the topic-modeled vector space and the input query statement; and

sorting the pieces of meta information according to the calculated similarity value and subsequently outputting media content having high similarity.