UNDER CONSTRUCTION
Saga processes raw text into normalized tokens, entities and semantic tags. The output can be used for question-answering, full-text analysis (fact extraction), semantic search, content vectors and matching, and many other purposes.
- Handles the full range of text processing
- Tokens extraction & cleansing, entity extraction, syntactic analysis and semantic analysis
- Scalable to extremely large dictionaries and pattern databases (>10s of millions of patterns)
- Makes it possible to build patterns from machine learning algorithms
- Disambiguation is a first-class citizen
- Saves all interpretations all the time (nothing is thrown away)
- Multiple disambiguation methods
- Confidence is captured at every step
- Confidence builds up as patterns are matched
- Fast enough to process documents for full database scans
Use Cases:
- Query interpretation
- Question answering
- Chatbots
- Full document fact extraction
- Vector generation for statistical and machine learning