|
|
""" |
|
|
Functions operating on the verbalisation of an ontology, providing support for |
|
|
generating documentation, extracting competency questions, and preliminarly |
|
|
testing an ontology via competency questions. |
|
|
""" |
|
|
|
|
|
import config |
|
|
from openai import OpenAI |
|
|
|
|
|
cqe_prompt_a = "You are asked to provide a comprehensive list of competency "\ |
|
|
"questions describing all the possible requirements that can be "\ |
|
|
"addressed by the ontology described before." |
|
|
|
|
|
cqt_prompt_a = "You are asked to infer if the ontology described before can "\ |
|
|
"address the following competency question: \"{}\" "\ |
|
|
"Valid answers are: Yes, No." |
|
|
|
|
|
|
|
|
class ChatInterface: |
|
|
|
|
|
def __init__(self, |
|
|
api_key: str, |
|
|
model_name: str = config.DEFAULT_MODEL, |
|
|
sampling_seed: int = config.DEFAULT_SEED, |
|
|
temperature: int = config.DEFAULT_TEMPERATURE): |
|
|
|
|
|
self.client = OpenAI(api_key=api_key) |
|
|
self.model_name = model_name |
|
|
self.sampling_seed = sampling_seed |
|
|
self.temperature = temperature |
|
|
|
|
|
def chat_completion(self, messages, **kwargs): |
|
|
|
|
|
model = kwargs["model"] if "model" in kwargs else self.model_name |
|
|
temperature = kwargs["temperature"] if "temperature" in kwargs \ |
|
|
else self.temperature |
|
|
|
|
|
completion = self.client.chat.completions.create( |
|
|
model=model, |
|
|
messages=messages, |
|
|
seed=self.sampling_seed, |
|
|
temperature=temperature, |
|
|
) |
|
|
return completion.choices[0].message.content |
|
|
|
|
|
|
|
|
def extract_competency_questions(onto_verbalisation: str, |
|
|
chat_interface: ChatInterface, |
|
|
prompt: str = cqe_prompt_a): |
|
|
""" |
|
|
Extract competency questions from the verbalisation of an ontology. |
|
|
|
|
|
Parameters |
|
|
---------- |
|
|
onto_verbalisation : str |
|
|
A string expressing the ontology verbalisation as output from a |
|
|
supported method in the `verbaliser` module. |
|
|
chat_interface : ChatInterface |
|
|
An instance of a chat interface holding the API session. |
|
|
prompt : str, optional |
|
|
CQ extraction prompt, by default cqe_prompt_a |
|
|
|
|
|
Returns |
|
|
------- |
|
|
competency_questions : str |
|
|
A list of competency questions induced from the verbalisation. |
|
|
|
|
|
""" |
|
|
full_prompt = onto_verbalisation + "\n" + prompt |
|
|
conversation_history = [ |
|
|
{"role": "system", "content": "You are an ontology expert."}, |
|
|
{"role": "user", "content": full_prompt} |
|
|
] |
|
|
|
|
|
competency_questions = chat_interface.chat_completion( |
|
|
conversation_history, model="gpt-3.5-turbo-16k") |
|
|
|
|
|
return competency_questions |
|
|
|
|
|
|
|
|
def test_competency_questions(onto_verbalisation: str, |
|
|
competency_questions: list[str], |
|
|
chat_interface: ChatInterface, |
|
|
cq_prompt: str = cqt_prompt_a): |
|
|
""" |
|
|
Performs a preliminary test of the ontology to assess whether its |
|
|
verbalisation allows for addressing each competency questions given. |
|
|
|
|
|
Parameters |
|
|
---------- |
|
|
onto_verbalisation : str |
|
|
A string expressing the ontology verbalisation as output from a |
|
|
supported method in the `verbaliser` module. |
|
|
competency_questions: list[str] |
|
|
A list of competency questions to use for preliminary testing. |
|
|
chat_interface : ChatInterface |
|
|
An instance of a chat interface holding the API session. |
|
|
cq_prompt : str, optional |
|
|
CQ test prompt, by default cqt_prompt_a |
|
|
|
|
|
Returns |
|
|
------- |
|
|
cq_test_dict : dict |
|
|
A dictionary holding an outcome (yes/no) as a preliminary test of each |
|
|
competency question. Keys correspond to CQs. |
|
|
|
|
|
""" |
|
|
cq_test_dict = {} |
|
|
for cq in competency_questions: |
|
|
full_prompt = onto_verbalisation + "\n" + cq_prompt.format(cq) |
|
|
conversation_history = [ |
|
|
{"role": "system", "content": "You are an ontology engineer."}, |
|
|
{"role": "user", "content": full_prompt} |
|
|
] |
|
|
outcome = chat_interface.chat_completion( |
|
|
conversation_history, model="gpt-3.5-turbo-16k") |
|
|
cq_test_dict[cq] = outcome |
|
|
|
|
|
return cq_test_dict |
|
|
|
