entity is not in entities list，but in relationship list?

[LLL1st]

Look at this code
`
elif startEntity_in_input_entities is None and endEntity_in_input_entities is None:
print(f"[INFO][INVENTED ENTITIES] Aie; the entities {startEntity} and {endEntity} are invented. Solving them ...")
startEntity.embed_Entity(embeddings_function=self.langchain_output_parser.calculate_embeddings,
entity_label_weight=entity_label_weight,
entity_name_weight=entity_name_weight)
endEntity.embed_Entity(embeddings_function=self.langchain_output_parser.calculate_embeddings,
entity_label_weight=entity_label_weight,
entity_name_weight=entity_name_weight)

            startEntity = self.matcher.find_match(obj1=startEntity, list_objects=entities, threshold=0.5)
            endEntity = self.matcher.find_match(obj1=endEntity, list_objects=entities, threshold=0.5)
            
            curated_relationships.append(Relationship(startEntity= startEntity, 
                                  endEntity = endEntity,
                                  name = relationship["name"]))

`
According to the ”find_match“, startEntity is returned if it does not find a similar entity among entities
At this point, startEntity is not in the entities list, but in the curated_relationships list. Is this correct?

[lairgiyassir]

Hello,

Yes, your understanding is correct. Based on the behavior of find_match, if it does not find a similar entity among the entities list, it simply returns the startEntity as-is.

I have lowered the value of threshold to 0.5 to maximise the chances to have a match because we assume that the extracted entities are exhaustive, so any invented entity might be just a rewriting of an original entity from the entities list.

However, to keep the code clean and handle all the possible cases, I will add a feature which consists of adding the invented entity (in case no match found under 0.5 threshold) to the input entities list.

Your question is very pertinent, thank you :)