Segmentation of Video According to Narrative Theme

• US 20130036124A1
• Filed: 08/02/2011
• Published: 02/07/2013
• Est. Priority Date: 08/02/2011
• Status: Active Grant

• Current Assignee: Comcast Cable Communications LLC (Comcast Corporation)
• Original Assignee: Comcast Cable Communications LLC (Comcast Corporation)
• Inventors: Ambwani, Geetu, Davis, Anthony R.
• Granted Patent: US 10,467,289 B2
• Time in Patent Office: Days
• US Class Current: 707/749
• CPC Class Codes: G06F 16/7844 using original textual content or text extracted from visual content or transcript of audio data

Abstract

Content items may be segmented and labeled by a theme to provide information, such as audio or video, with greater functionality, accuracy and speed. The segments and other related information may be stored and made accessible to users through, for example, a search service and/or an on-demand service. Also provided is automatic segmentation that may include any one of receiving a textual description of a content item, determining themes within the textual description, determining relevance intervals corresponding to terms of the theme, filtering and/or merging the relevance intervals, and determining a total interval for each theme that represents one or more contextually-coherent segments of the content item.

Claims

1. A method comprising:

receiving, at a computing device, a textual description of a content item;

identifying one or more terms based on the textual description;

determining one or more relevance intervals based on each term of the one or more terms, resulting in a plurality of relevance intervals for the one or more terms;

determining a total interval for the one or more terms from the plurality of relevance intervals; and

preparing data that identifies the total interval.

2. The method of claim 1, further comprising:

receiving a transcript of the content item;

calculating relevance intervals for terms included in the transcript, wherein each relevance interval identifies one or more sections of the content item that are relevant to a term occurring in the transcript of the content item; and

storing the relevance intervals into a database;

wherein determining one or more relevance intervals based on each term of the one or more terms includes retrieving the plurality of relevance intervals from the database.

3. The method of claim 1, wherein the one or more terms represent a first theme and the method further comprises:

determining a second theme based on the textual description;

determining one or more relevance intervals for each term in the second theme, resulting in a plurality of relevance intervals for the second theme;

determining a second theme total interval from the plurality of relevance intervals for the second theme; and

storing data that identifies the second theme total interval and the second theme.

4. The method of claim 3, wherein the total interval is a first theme total interval and the method further comprises:

filtering the first theme total interval based on the second theme total interval by a process that includes determining that a sub-interval of the first theme total interval overlaps with a sub-interval of the second theme total interval, and removing at least one sub-interval from the first theme total interval.

5. The method of claim 3, wherein the plurality of relevance intervals for the one or more terms is a plurality of relevance intervals for the first theme and the total interval is a first theme total interval, and wherein the method further comprising:

filtering the plurality of relevance intervals for the first theme and the plurality of relevance intervals for the second theme;

merging the plurality of relevance intervals for the first theme;

merging the plurality of relevance intervals for the second theme;

filtering the first theme total interval and the second theme total interval;

merging the first theme total interval; and

merging the second theme total interval.

6. The method of claim 1, further comprises filtering the plurality of relevance intervals, wherein filtering the plurality of relevance intervals includes removing an interval from the plurality of relevance intervals, and further includes at least one of the following:

filtering intervals in the plurality of relevance intervals based on size of a relevance interval, filtering intervals in the plurality of relevance intervals based on overlap between relevance intervals, filtering intervals in the plurality of relevance intervals based on adjacency between relevance intervals, and filtering intervals in the plurality of relevance intervals based on content item location of a relevance interval.

7. The method of claim 1, further comprising merging the plurality of relevance intervals, wherein merging the plurality of relevance intervals includes at least one of merging intervals in the plurality of relevance intervals based on overlap between intervals, and merging intervals in the plurality of relevance intervals based on adjacency between intervals.

8. An apparatus comprising:

at least one processor; and

memory storing computer readable instructions configured to, when executed by the at least one processor, cause the apparatus to;

receive a textual description of a content item;

determine one or more terms based on the textual description;

determine one or more relevance intervals based on each term of the one or more terms, resulting in a plurality of relevance intervals for the one or more terms;

determine a total interval for the one or more terms from the plurality of relevance intervals; and

prepare data that identifies the total interval.

9. The apparatus of claim 8, wherein the memory further stores computer readable instructions configured to, when executed by the at least one processor, cause the apparatus to:

receive a transcript of the content item,

calculate relevance intervals for terms included in the transcript, wherein each relevance interval identifies one or more sections of the content item that are relevant to a term occurring in the transcript of the content item, and

store the relevance intervals into a database;

wherein determining one or more relevance intervals based on each term of the one or more terms includes retrieving the plurality of relevance intervals from the database.

10. The apparatus of claim 8, wherein the one or more terms represent a first theme and wherein the memory further stores computer readable instructions configured to, when executed by the at least one processor, cause the apparatus to:

determine a second theme based on the textual description,

determine one or more relevance intervals for each term in the second theme, resulting in a plurality of relevance intervals for the second theme,

determine a second theme total interval from the plurality of relevance intervals for the second theme, and

prepare data that identifies the second theme total interval and the second theme.

11. The apparatus of claim 10, wherein the total interval is a first theme total interval and wherein the memory further stores computer readable instructions configured to, when executed by the at least one processor, cause the apparatus to:

filter the first theme total interval based on the second theme total interval by a process that includes determining that a sub-interval of the first theme total interval overlaps with a sub-interval of the second theme total interval, and removing at least one sub-interval from the first theme total interval.

12. The apparatus of claim 10, wherein the plurality of relevance intervals for the one or more terms is a plurality of relevance intervals for the first theme and the total interval is a first theme total interval, and wherein the memory further stores computer readable instructions configured to, when executed by the at least one processor, cause the apparatus to:

filter the plurality of relevance intervals for the first theme and the plurality of relevance intervals for the second theme,

merge the plurality of relevance intervals for the first theme,

merge the plurality of relevance intervals for the second theme,

filter the first theme total interval and the second theme total interval,

merge the first theme total interval, and

merge the second theme total interval.

13. The apparatus of claim 8, wherein the memory further stores computer readable instructions configured to, when executed by the at least one processor, cause the apparatus to:

filter the plurality of relevance intervals, wherein filtering the plurality of relevance intervals includes removing an interval from the plurality of relevance intervals, and further includes at least one of the following;

filtering intervals in the plurality of relevance intervals based on size of a relevance interval, filtering intervals in the plurality of relevance intervals based on overlap between relevance intervals, filtering intervals in the plurality of relevance intervals based on adjacency between relevance intervals, and filtering intervals in the plurality of relevance intervals based on content item location of a relevance interval.

14. The apparatus of claim 8, wherein the memory further stores computer readable instructions configured to, when executed by the at least one processor, cause the apparatus to:

merge the plurality of relevance intervals, wherein merging the plurality of relevance intervals includes at least one of the following;

merging intervals in the plurality of relevance intervals based on overlap between intervals, and merging intervals in the plurality of relevance intervals based on adjacency between intervals.

15. A method comprising:

determining, at a computing device, a textual description of a content item;

identifying one or more terms based on the textual description;

determining one or more relevance intervals based on each term of the one or more terms, resulting in a plurality of relevance intervals for the one or more terms;

determining a total interval for the one or more terms from the plurality of relevance intervals; and

preparing data that identifies the total interval.

16. The method of claim 15, wherein the textual description includes one or more sentences, each sentence of the one or more sentences representing a theme, and wherein the one or more terms are from a first sentence of the one or more sentences, wherein the total interval is for the theme represented by the first sentence, and wherein preparing data that identifies the total interval includes preparing a data structure that includes data identifying the total interval and data identifying the theme represented by the first sentence.

17. The method of claim 16, wherein the theme represented by the first sentence is a first theme, and the method further comprises:

determining a second theme based on a second sentence of the textual description;

determining one or more relevance intervals for each term in the second theme, resulting in a plurality of relevance intervals for the second theme;

determining a second theme total interval from the plurality of relevance intervals for the second theme; and

preparing data that identifies the second theme total interval and the second theme.

18. The method of claim 17, wherein the total interval is a first theme total interval, and the method further comprises:

filtering the first theme total interval based on the second theme total interval by a process that includes determining that a sub-interval of the first theme total interval overlaps with a sub-interval of the second theme total interval, and removing at least one sub-interval from the first theme total interval.

19. The method of claim 16, further comprising:

filtering the plurality of relevance intervals, wherein filtering the plurality of relevance intervals includes removing an interval from the plurality of relevance intervals, and further includes at least one of the following;

filtering intervals in the plurality of relevance intervals based on size of a relevance interval, filtering intervals in the plurality of relevance intervals based on overlap between relevance intervals, filtering intervals in the plurality of relevance intervals based on adjacency between relevance intervals, and filtering intervals in the plurality of relevance intervals based on content item location of a relevance interval.

20. The method of claim 16, further comprising:

merging the plurality of relevance intervals, wherein merging the plurality of relevance intervals includes at least one of the following;

merging intervals in the plurality of relevance intervals based on overlap between intervals, and merging intervals in the plurality of relevance intervals based on adjacency between intervals.