Exploration via TREC tracks
- Spanish (porting to non-English)
- Confusion (working with OCR input)
- Filtering (new metrics/scenario for routing task)
- Chinese (lack of word boundaries)
- Interactive (cross-group comparison using controls)
- Spoken Document Retrieval
- Cross-Language Information Retrieval (English, French, German)
- Very Large Corpus (100 GB of data)
- Small Web (ad hoc topics against 2 GB web)
- Q & A (200 questions; answers as 50 byte strings)