MCPcopy Create free account
hub / github.com/KnowledgeXLab/LeanRAG / entity_evaluate

Method entity_evaluate

CommonKG/llm_infer.py:287–306  ·  view source on GitHub ↗
(self, entities)

Source from the content-addressed store, hash-verified

285 logger.info(f"Maximum error tolerance reached, skipping text id: {id}")
286
287 def entity_evaluate(self, entities):
288 entities = "\n".join(entities)
289 user_prompt = (
290 "Analyze the entity list I provide and extract all entities (e.g., proper nouns, discipline terminologies, abstract and collective nouns, etc.). Follow these rules STRICTLY:\n"
291 "1. Extract as many nouns from the text as possible\n"
292 "2. Do not include any verbs or words without specific meanings such as time, location, number, measurement, etc.\n"
293 "3. Output EXACTLY one entity per line\n"
294 "4. Maintain ORIGINAL spelling/case from the input\n"
295 "5. Ensure that the entities are not repeated\n"
296
297 "Input entity list:\n"
298 f"{entities}\n"
299
300 "Now process this list. \n"
301 )
302
303 response = self.call_api(user_prompt)
304 verify_entities = response.split("\n")
305 verify_entities = list(set([item.strip() for item in verify_entities if len(item.strip()) > 0]))
306 return verify_entities

Callers 1

process_llm_batchFunction · 0.80

Calls 1

call_apiMethod · 0.95

Tested by

no test coverage detected