Semantic chunking
Semantic chunking: Chunking that creates segments based on content meaning and structure rather than fixed length alone. It can use paragraphs, similarity changes or logical units such as clauses and tables. It produces more coherent segments but requires heuristics and document-specific evaluation.
In italiano: Semantic ChunkingLegge I · Embedding, feature e spazi vettorialiIt can use paragraphs, similarity changes or logical units such as clauses and tables. It produces more coherent segments but requires heuristics and document-specific evaluation.
Examples
- Topic-based splitting
- Contract clauses
- Tables kept with their captions