def run_ner_linking()

in src/utils.py [0:0]

• 15 lines of code
• 5 McCabe index (conditional complexity)

def run_ner_linking(texts: typing.List[str], ner_model_path: str):
    """Loads and runs the Named Entity Recognition + Entity Linking model on all `texts`, 
    saving their named entity labels and Wikidata IDs if found.
    """
    nlp = spacy.load(ner_model_path)

    text_to_info = {}
    for text in tqdm(texts, desc="Running Named Entity Linking"):
        doc = nlp(text)
        datum = []
        for e in doc.ents:
            kb_id = (
                e.kb_id_ if "Q" == e.kb_id_[0] and e.kb_id_[1:].isnumeric() else None
            )
            datum.append(
                {"text": e.text, "label": e.label_, "id": kb_id,}
            )
        text_to_info[text] = datum
    return text_to_info