Semantic segmentation and tagging engine
In accordance with the embodiments of the present invention, a method and engine for assigning semantic tags to segments within media. The invention receives media and extracts textual information related to the media's content. It processes the textual information and creates a list of topics related to the content. The invention segments the media and intelligently assigns topical tags to the segments. The semantically segmented media data is outputted for storage or analysis.
1. A system for creating a database for semantically segmented media, comprising:
a media receiving means for receiving media;
a text extraction and preparation means that extracts textual content from the media and divides the textual content into overlapping text segments and non-overlapping text segments;
a topic detection means for generating a set of topical tags for said non-overlapping text segments by inferring the set of topical tags in said non-overlapping text segments in accordance with a probabilistic model;
a semantic segment tagging means for assigning a topical tag to each overlapping text segment, and for generating a weight for each topical tag assignment;
a tagging processing means for combining said overlapping text segments and non-overlapping text segments based on the topical tag assignment and weight for each overlapping text segment and non-overlapping segment; and
a storage means for storing records that map the topical tag assignments to the overlapping text segments and non-overlapping text segments of the media.
2. The system according to claim 1 , wherein the media is received from a media source.
3. The system according to claim 2 , wherein the media comprises video, audio, audio-visual content, text, images, slideshows, presentations, signals, data, or combinations thereof.
4. The system according to claim 2 , wherein the media source comprises digital files, local databases, online archives, pre-recorded media streams, live media streams, or combinations thereof.
5. The system according to claim 1 , wherein the text extraction and preparation means determines the time boundaries of the textual content to exclude and converts the textual content using natural language processing methods.
6. A method for creating a database for semantically segmented media, comprising:
receiving timed media;
extracting textual content from the media;
preparing the extracted textual content by determining the time boundaries of the textual content to exclude, processing the textual content using natural language processing methods, and dividing the textual content into overlapping text segments and non-overlapping text segments;
detecting at least one topic in prepared textual content;
segmenting prepared textual content;
assigning at least one topic to at least one segment of the prepared textual content; and;
storing said prepared textual content and said at least one topic in a database.
7. The method according to claim 6 , wherein media is received from a media source.
8. The method according to claim 7 , wherein the media comprises video, audio, audio-visual content, text, images, slideshows, presentations, signals, data, or combinations thereof.
9. The method according to claim 7 , wherein the media source comprises digital files, local databases, online archives, pre-recorded media streams, live media streams, or combinations thereof.
10. The method according to claim 6 , further comprising generating semantic tags by detecting topics contained in said non-overlapping text segments.
11. The method according to claim 6 , further comprising assigning the semantic tags to said overlapping text segments.
12. The method according to claim 6 , further comprising merging the overlapping text segments and non-overlapping text segments and expanding said semantic tags.
13. The method according to claim 6 , further comprising storing in a database records that map the semantic tag assignments to the overlapping text segments and non-overlapping text segments.