Create app.py
Browse files
app.py
ADDED
@@ -0,0 +1,69 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import os; import json; import requests
|
2 |
+
import streamlit as st
|
3 |
+
ES_URL = os.environ.get("ES_URL")
|
4 |
+
|
5 |
+
question = 'What is the capital of Netherlands?'
|
6 |
+
query_text = 'Query used for keyword search (you can also edit, and experiment with the responses)'
|
7 |
+
written_question = st.text_input(query_text, question)
|
8 |
+
if written_question:
|
9 |
+
question = written_question
|
10 |
+
if st.button('Run keyword search'):
|
11 |
+
if question:
|
12 |
+
try:
|
13 |
+
# qa_result = pipe_exqa(question=question, context=paragraph)
|
14 |
+
url = f"{ES_URL}/document/_search?pretty"
|
15 |
+
# payload = json.dumps({"query":{"match":{"content":"moldova"}}})
|
16 |
+
payload = json.dumps({"query": {
|
17 |
+
"more_like_this": { "like": question, # "What is the capital city of Netherlands?"
|
18 |
+
"fields": ["content"], "min_term_freq": 1.9, "min_doc_freq": 4, "max_query_terms": 50
|
19 |
+
}}})
|
20 |
+
headers = {'Content-Type': 'application/json'}
|
21 |
+
response = requests.request("GET", url, headers=headers, data=payload)
|
22 |
+
qa_result = response.json() # print(response.text)
|
23 |
+
|
24 |
+
except Exception as e:
|
25 |
+
qa_result = str(e)
|
26 |
+
|
27 |
+
# if "answer" in qa_result.keys():
|
28 |
+
# answer_span, answer_score = qa_result["answer"], qa_result["score"]
|
29 |
+
# st.write(f'Answer: **{answer_span}**')
|
30 |
+
# start_par, stop_para = max(0, qa_result["start"]-86), min(qa_result["end"]+90, len(paragraph))
|
31 |
+
# answer_context = paragraph[start_par:stop_para].replace(answer_span, f'**{answer_span}**')
|
32 |
+
# st.write(f'Answer context (and score): ... _{answer_context}_ ... (score: {format(answer_score, ".3f")})')
|
33 |
+
|
34 |
+
st.write(f'Answer JSON: '); st.write(qa_result)
|
35 |
+
else:
|
36 |
+
st.write('Write a query to submit your keyword search'); st.stop()
|
37 |
+
|
38 |
+
"""
|
39 |
+
result_first_two_hits = result['hits']['hits'][:2] # print("First 2 results:")
|
40 |
+
question_similarity = [ (hit['_score'], hit['_source']['content'][:200])
|
41 |
+
for hit in result_first_two_hits ] # print(question_similarity)
|
42 |
+
|
43 |
+
top_hit = result['hits']['hits'][0]
|
44 |
+
context = top_hit['_source']['content']
|
45 |
+
# context = r" Extractive Question Answering is the task of extracting
|
46 |
+
# an answer from a text given a question. An example of a question
|
47 |
+
# answering dataset is the SQuAD dataset, which is entirely based
|
48 |
+
# on that task. If you would like to fine-tune a model on a SQuAD task,
|
49 |
+
# you may leverage the `examples/pytorch/question-answering/run_squad.py` script."
|
50 |
+
question = input # "What is extractive question answering?"
|
51 |
+
# "What is a good example of a question answering dataset?"
|
52 |
+
print(question)
|
53 |
+
context = context[:5000]
|
54 |
+
print(context)
|
55 |
+
try:
|
56 |
+
qa_result = pipe_exqa(question=question, context=context)
|
57 |
+
except Exception as e:
|
58 |
+
return {"output": str(e)}
|
59 |
+
|
60 |
+
return {"output": str(qa_result)}
|
61 |
+
|
62 |
+
answer = qa_result['answer']
|
63 |
+
score = round(qa_result['score'], 4)
|
64 |
+
span = f"start: {qa_result['start']}, end: {qa_result['end']}"
|
65 |
+
# st.write(answer); st.write(f"score: {score}"); st.write(f"span: {span}")
|
66 |
+
output = f"{str(answer)} \n {str(score)} \n {str(span)}"
|
67 |
+
|
68 |
+
return {"output": output} or {"output": str(question_similarity)} or result or {"Hello": "World!"}
|
69 |
+
"""
|