Smart text enrichment algorithms for media retrieval applications
Steamer is a Flemish project during which we developed a software that automatically enriches texts and videos with metadata. The software makes articles and videos more searchable and makes connections between different content items.
The consortium partners developed different enrichment algorithms:
- IPTCcategorization: dividing articles or content into categories and topics
- A tagging algorithm that suggests the most important keywords
- Sentiment analysis
- Similarity detection, with eg. Event detection
- Onthe-fly video concept detection
Two demonstrators were built: a search engine with the enriched data of articles and video items, and a tool for journalists that suggests tags for the articles they wrote.