Expand description
Hybrid entity extraction: regex pre-filter + candle BERT NER (graceful degradation). Entity and URL extraction pipeline (NER + regex prefilter).
Runs named-entity recognition and regex heuristics to extract structured entities and hyperlinks from raw memory bodies before embedding.
Structs§
- Extracted
Entity - Extracted
Url - URL with source offset extracted from the memory body.
- Extraction
Result - Regex
Extractor
Traits§
Functions§
- extract_
graph_ auto - extract_
urls - Extracts URLs from a memory body, deduplicated by text.
URLs are stored in the
memory_urlstable separately from graph entities. v1.0.24: split of the URL block that polluted apply_regex_prefilter with entity_type=‘concept’.