swarmauri_parser_entityrecognition 0.8.4.dev2

Entity Recognition Parser for Swarmauri.

Swarmauri Parser Entityrecognition

Named-entity recognition (NER) parser for Swarmauri built on spaCy. Extracts entities (PERSON, ORG, GPE, etc.) from unstructured text and returns Document objects with entity metadata.

Features

• Uses spaCy's en_core_web_sm model by default (downloads automatically if missing).
• Falls back to a blank English pipeline with minimal regex-based tagging when the full model is unavailable (best-effort mode).
• Emits Document instances containing the entity text and metadata (entity_type, entity_id).

Prerequisites

• Python 3.10 or newer.
• spaCy and its English model. The parser attempts to download en_core_web_sm if missing; set SPACY_HOME or pre-install the model in production deployments.
• If running without internet access, install the model ahead of time: python -m spacy download en_core_web_sm.

Installation

# pip
pip install swarmauri_parser_entityrecognition

# poetry
poetry add swarmauri_parser_entityrecognition

# uv (pyproject-based projects)
uv add swarmauri_parser_entityrecognition

Quickstart

from swarmauri_parser_entityrecognition import EntityRecognitionParser

text = "Barack Obama was born in Hawaii and served as President of the United States."
parser = EntityRecognitionParser()
entities = parser.parse(text)

for entity_doc in entities:
    print(entity_doc.content, entity_doc.metadata["entity_type"])

Batch Processing

texts = [
    "Apple Inc. unveiled new MacBooks in California.",
    "Tim Cook met investors in New York City.",
]

parser = EntityRecognitionParser()
results = [parser.parse(t) for t in texts]

for doc_set in results:
    for doc in doc_set:
        print(doc.content, doc.metadata["entity_type"])

Handling Fallback Mode

When spaCy's English model is unavailable, the parser performs best-effort matching using a blank pipeline and simple regex patterns. Check for entity_type values and the entity_id metadata to understand which mode produced the result.

parser = EntityRecognitionParser()
entities = parser.parse("Tim Cook announced new products in New York City for Apple Inc.")
print([d.metadata for d in entities])

Install spaCy models before production use to avoid fallback accuracy losses.

Tips

• For languages beyond English, load a different spaCy model by changing the initialization logic (e.g., subclass the parser and load es_core_news_sm).
• Preprocess text to remove noise (HTML tags, markup) before parsing to improve NER accuracy.
• Combine with Swarmauri middleware or pipelines to fuse entity data with downstream tasks (e.g., knowledge graph enrichment, anonymization).

Want to help?

If you want to contribute to swarmauri-sdk, read up on our guidelines for contributing that will help you get started.