Upload 2 files
Browse files- rubrics.py +112 -0
- subjective_test_evaluation.py +252 -0
rubrics.py
ADDED
@@ -0,0 +1,112 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import streamlit as st
|
2 |
+
from pymongo import MongoClient
|
3 |
+
from openai import OpenAI
|
4 |
+
from bson import ObjectId
|
5 |
+
import json
|
6 |
+
from dotenv import load_dotenv
|
7 |
+
import os
|
8 |
+
|
9 |
+
load_dotenv()
|
10 |
+
MONGO_URI = os.getenv('MONGO_URI')
|
11 |
+
OPENAI_API_KEY = os.getenv('OPENAI_KEY')
|
12 |
+
|
13 |
+
client = MongoClient(MONGO_URI)
|
14 |
+
db = client['novascholar_db']
|
15 |
+
# db.create_collection("rubrics")
|
16 |
+
rubrics_collection = db['rubrics']
|
17 |
+
resources_collection = db['resources']
|
18 |
+
courses_collection = db['courses']
|
19 |
+
|
20 |
+
def generate_rubrics(api_key, session_title, outcome_description, taxonomy, pre_class_material):
|
21 |
+
prompt = f"""
|
22 |
+
You are an expert educational AI assistant specializing in instructional design. Generate a detailed rubric for the session titled "{session_title}". The rubric should be aligned with Bloom's Taxonomy level "{taxonomy}" and use numerical scoring levels (4,3,2,1) instead of descriptive levels. Use the following context:
|
23 |
+
|
24 |
+
Session Outcome Description:
|
25 |
+
{outcome_description}
|
26 |
+
|
27 |
+
Pre-class Material:
|
28 |
+
{pre_class_material}
|
29 |
+
|
30 |
+
Please generate the rubric in JSON format with these specifications:
|
31 |
+
1. Use numerical levels (4=Highest, 1=Lowest) instead of descriptive levels
|
32 |
+
2. Include 4-5 relevant criteria based on the session outcome
|
33 |
+
3. Each criterion should have clear descriptors for each numerical level
|
34 |
+
4. Focus on objectively measurable aspects for evaluation
|
35 |
+
5. Structure should be suitable for evaluating assignments and test answers
|
36 |
+
|
37 |
+
***IMPORTANT: DO NOT INCLUDE THE WORD JSON IN THE OUTPUT STRING, DO NOT INCLUDE BACKTICKS (```) IN THE OUTPUT, AND DO NOT INCLUDE ANY OTHER TEXT, OTHER THAN THE ACTUAL JSON RESPONSE. START THE RESPONSE STRING WITH AN OPEN CURLY BRACE {{ AND END WITH A CLOSING CURLY BRACE }}.***
|
38 |
+
"""
|
39 |
+
|
40 |
+
messages = [
|
41 |
+
{
|
42 |
+
"role": "system",
|
43 |
+
"content": "You are an expert educational AI assistant specializing in instructional design.",
|
44 |
+
},
|
45 |
+
{
|
46 |
+
"role": "user",
|
47 |
+
"content": prompt
|
48 |
+
},
|
49 |
+
]
|
50 |
+
|
51 |
+
try:
|
52 |
+
client = OpenAI(api_key=api_key)
|
53 |
+
response = client.chat.completions.create(
|
54 |
+
model="gpt-4-0125-preview",
|
55 |
+
messages=messages
|
56 |
+
)
|
57 |
+
return response.choices[0].message.content
|
58 |
+
except Exception as e:
|
59 |
+
st.error(f"Failed to generate rubrics: {e}")
|
60 |
+
return None
|
61 |
+
|
62 |
+
def display_rubrics_tab(session, course_id):
|
63 |
+
st.subheader("Generated Rubrics")
|
64 |
+
|
65 |
+
# Fetch session details from the courses collection
|
66 |
+
course_data = courses_collection.find_one(
|
67 |
+
{"course_id": course_id, "sessions.session_id": session['session_id']},
|
68 |
+
{"sessions.$": 1}
|
69 |
+
)
|
70 |
+
|
71 |
+
if course_data and 'sessions' in course_data and len(course_data['sessions']) > 0:
|
72 |
+
session_data = course_data['sessions'][0]
|
73 |
+
|
74 |
+
# Extract session learning outcomes
|
75 |
+
if 'session_learning_outcomes' in session_data and len(session_data['session_learning_outcomes']) > 0:
|
76 |
+
outcome = session_data['session_learning_outcomes'][0]
|
77 |
+
outcome_description = outcome.get('outcome_description', '')
|
78 |
+
taxonomy_level = outcome.get('bloom_taxonomy_level', '')
|
79 |
+
|
80 |
+
# Display fetched information
|
81 |
+
st.markdown("### Session Information")
|
82 |
+
st.markdown(f"**Session Title:** {session['title']}")
|
83 |
+
st.markdown(f"**Learning Outcome:** {outcome_description}")
|
84 |
+
st.markdown(f"**Taxonomy Level:** {taxonomy_level}")
|
85 |
+
|
86 |
+
# Fetch pre-class material
|
87 |
+
pre_class_material_docs = resources_collection.find({"session_id": session['session_id']})
|
88 |
+
pre_class_material = "\n".join([f"{doc.get('title', 'No Title')}: {doc.get('url', 'No URL')}" for doc in pre_class_material_docs])
|
89 |
+
|
90 |
+
if st.button("Generate Rubric"):
|
91 |
+
rubric = generate_rubrics(
|
92 |
+
OPENAI_API_KEY,
|
93 |
+
session['title'],
|
94 |
+
outcome_description,
|
95 |
+
taxonomy_level,
|
96 |
+
pre_class_material
|
97 |
+
)
|
98 |
+
|
99 |
+
if rubric:
|
100 |
+
st.json(rubric)
|
101 |
+
if st.button("Save Rubric"):
|
102 |
+
rubric_data = {
|
103 |
+
"course_id": course_id,
|
104 |
+
"session_id": session['session_id'],
|
105 |
+
"rubric": json.loads(rubric)
|
106 |
+
}
|
107 |
+
rubrics_collection.insert_one(rubric_data)
|
108 |
+
st.success("Rubric saved successfully!")
|
109 |
+
else:
|
110 |
+
st.error("No learning outcomes found for this session")
|
111 |
+
else:
|
112 |
+
st.error("Session data not found")
|
subjective_test_evaluation.py
ADDED
@@ -0,0 +1,252 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import openai
|
2 |
+
from pymongo import MongoClient
|
3 |
+
from datetime import datetime
|
4 |
+
import os
|
5 |
+
from dotenv import load_dotenv
|
6 |
+
import re
|
7 |
+
import streamlit as st
|
8 |
+
from bson import ObjectId
|
9 |
+
|
10 |
+
load_dotenv()
|
11 |
+
MONGO_URI = os.getenv('MONGO_URI')
|
12 |
+
OPENAI_API_KEY = os.getenv('OPENAI_KEY')
|
13 |
+
|
14 |
+
client = MongoClient(MONGO_URI)
|
15 |
+
db = client['novascholar_db']
|
16 |
+
rubrics_collection = db['rubrics']
|
17 |
+
resources_collection = db['resources']
|
18 |
+
subjective_tests_collection = db['subjective_tests']
|
19 |
+
subjective_test_analysis_collection = db['subjective_test_analysis']
|
20 |
+
|
21 |
+
openai.api_key = OPENAI_API_KEY
|
22 |
+
|
23 |
+
def evaluate_subjective_answers(test_id, student_id, course_id):
|
24 |
+
"""Evaluate subjective test answers using OpenAI."""
|
25 |
+
try:
|
26 |
+
# Get test and submission details
|
27 |
+
test_doc = subjective_tests_collection.find_one({
|
28 |
+
"_id": ObjectId(test_id),
|
29 |
+
"course_id": course_id
|
30 |
+
})
|
31 |
+
if not test_doc:
|
32 |
+
return {
|
33 |
+
"content_analysis": "Error: Test not found",
|
34 |
+
"analyzed_at": datetime.utcnow(),
|
35 |
+
"correctness_score": 0
|
36 |
+
}
|
37 |
+
|
38 |
+
submission = next(
|
39 |
+
(sub for sub in test_doc.get('submissions', []) if sub['student_id'] == student_id),
|
40 |
+
None
|
41 |
+
)
|
42 |
+
|
43 |
+
if not submission:
|
44 |
+
return {
|
45 |
+
"content_analysis": "Error: Submission not found",
|
46 |
+
"analyzed_at": datetime.utcnow(),
|
47 |
+
"correctness_score": 0
|
48 |
+
}
|
49 |
+
|
50 |
+
# Rest of the evaluation logic remains the same
|
51 |
+
questions = test_doc.get('questions', [])
|
52 |
+
student_answers = submission.get('answers', [])
|
53 |
+
|
54 |
+
if not questions or not student_answers:
|
55 |
+
return {
|
56 |
+
"content_analysis": "Error: No questions or answers found",
|
57 |
+
"analyzed_at": datetime.utcnow(),
|
58 |
+
"correctness_score": 0
|
59 |
+
}
|
60 |
+
|
61 |
+
# Retrieve rubrics for the session
|
62 |
+
rubric_doc = rubrics_collection.find_one({
|
63 |
+
"session_id": test_doc['session_id'],
|
64 |
+
"course_id": course_id
|
65 |
+
})
|
66 |
+
|
67 |
+
if not rubric_doc:
|
68 |
+
return {
|
69 |
+
"content_analysis": "Error: Rubric not found",
|
70 |
+
"analyzed_at": datetime.utcnow(),
|
71 |
+
"correctness_score": 0
|
72 |
+
}
|
73 |
+
|
74 |
+
rubric = rubric_doc.get('rubric', {})
|
75 |
+
|
76 |
+
# Retrieve pre-class materials
|
77 |
+
pre_class_materials = resources_collection.find({
|
78 |
+
"session_id": test_doc['session_id'],
|
79 |
+
"course_id": course_id
|
80 |
+
})
|
81 |
+
pre_class_content = "\n".join([material.get('text_content', '') for material in pre_class_materials])
|
82 |
+
|
83 |
+
# Analyze each question
|
84 |
+
all_analyses = []
|
85 |
+
total_score = 0
|
86 |
+
|
87 |
+
for i, (question, answer) in enumerate(zip(questions, student_answers), 1):
|
88 |
+
analysis_content = f"Question {i}: {question['question']}\nAnswer: {answer}\n\nRubric: {rubric}\n\nPre-class Materials: {pre_class_content}\n\n"
|
89 |
+
|
90 |
+
prompt_template = f"""As an educational assessor, evaluate this student's answer based on the provided rubric criteria and pre-class materials. Follow these assessment guidelines:
|
91 |
+
|
92 |
+
1. Evaluation Process:
|
93 |
+
- Use each rubric criterion (scored 1-4) for internal assessment
|
94 |
+
- Compare response with pre-class materials
|
95 |
+
- Check alignment with all rubric requirements
|
96 |
+
- Calculate final score: sum of criteria scores converted to 10-point scale
|
97 |
+
|
98 |
+
Pre-class Materials:
|
99 |
+
{pre_class_content}
|
100 |
+
|
101 |
+
Rubric Criteria:
|
102 |
+
{rubric}
|
103 |
+
|
104 |
+
Question and Answer:
|
105 |
+
{analysis_content}
|
106 |
+
|
107 |
+
Provide your assessment in the following format:
|
108 |
+
|
109 |
+
**Score and Evidence**
|
110 |
+
- Score: [X]/10
|
111 |
+
- Evidence for deduction: [One-line reference to most significant gap or inaccuracy]
|
112 |
+
**Key Areas for Improvement**
|
113 |
+
- [Concise improvement point 1]
|
114 |
+
- [Concise improvement point 2]
|
115 |
+
- [Concise improvement point 3]
|
116 |
+
"""
|
117 |
+
|
118 |
+
response = openai.Completion.create(
|
119 |
+
model="text-davinci-003",
|
120 |
+
prompt=prompt_template,
|
121 |
+
max_tokens=500,
|
122 |
+
temperature=0.7
|
123 |
+
)
|
124 |
+
|
125 |
+
individual_analysis = response.choices[0].text.strip()
|
126 |
+
|
127 |
+
try:
|
128 |
+
score_match = re.search(r'Score: (\d+)', individual_analysis)
|
129 |
+
question_score = int(score_match.group(1)) if score_match else 0
|
130 |
+
total_score += question_score
|
131 |
+
except:
|
132 |
+
question_score = 0
|
133 |
+
|
134 |
+
formatted_analysis = f"\n\n## Question {i} Analysis\n\n{individual_analysis}"
|
135 |
+
all_analyses.append(formatted_analysis)
|
136 |
+
|
137 |
+
average_score = round(total_score / len(questions)) if questions else 0
|
138 |
+
combined_analysis = "\n".join(all_analyses)
|
139 |
+
|
140 |
+
return {
|
141 |
+
"content_analysis": combined_analysis,
|
142 |
+
"analyzed_at": datetime.utcnow(),
|
143 |
+
"correctness_score": average_score
|
144 |
+
}
|
145 |
+
|
146 |
+
except Exception as e:
|
147 |
+
return {
|
148 |
+
"content_analysis": f"Error evaluating answers: {str(e)}",
|
149 |
+
"analyzed_at": datetime.utcnow(),
|
150 |
+
"correctness_score": 0
|
151 |
+
}
|
152 |
+
|
153 |
+
def display_evaluation_to_faculty(session_id, student_id, course_id):
|
154 |
+
"""Display submitted tests with improved error handling and debugging"""
|
155 |
+
st.subheader("Evaluate Subjective Tests")
|
156 |
+
|
157 |
+
try:
|
158 |
+
# Convert all IDs to strings for consistent comparison
|
159 |
+
session_id = str(session_id)
|
160 |
+
student_id = str(student_id)
|
161 |
+
course_id = str(course_id)
|
162 |
+
|
163 |
+
print(f"Searching for tests with session_id: {session_id}, student_id: {student_id}, course_id: {course_id}")
|
164 |
+
|
165 |
+
# Query for tests
|
166 |
+
query = {
|
167 |
+
"session_id": session_id,
|
168 |
+
"course_id": course_id,
|
169 |
+
"submissions": {
|
170 |
+
"$elemMatch": {
|
171 |
+
"student_id": student_id
|
172 |
+
}
|
173 |
+
}
|
174 |
+
}
|
175 |
+
|
176 |
+
# Log the query for debugging
|
177 |
+
print(f"MongoDB Query: {query}")
|
178 |
+
|
179 |
+
# Fetch tests
|
180 |
+
tests = list(subjective_tests_collection.find(query))
|
181 |
+
print(f"Found {len(tests)} tests matching query")
|
182 |
+
|
183 |
+
if not tests:
|
184 |
+
# Check if any tests exist for this session
|
185 |
+
all_session_tests = list(subjective_tests_collection.find({
|
186 |
+
"session_id": session_id,
|
187 |
+
"course_id": course_id
|
188 |
+
}))
|
189 |
+
|
190 |
+
if all_session_tests:
|
191 |
+
print(f"Found {len(all_session_tests)} tests for this session, but no submissions from student {student_id}")
|
192 |
+
st.warning("No submitted tests found for this student, but tests exist for this session.")
|
193 |
+
else:
|
194 |
+
print("No tests found for this session at all")
|
195 |
+
st.info("No tests have been created for this session yet.")
|
196 |
+
return
|
197 |
+
|
198 |
+
# Display tests and handle evaluation
|
199 |
+
for test in tests:
|
200 |
+
with st.expander(f"Test: {test.get('title', 'Untitled Test')}", expanded=True):
|
201 |
+
# Find student submission
|
202 |
+
submission = next(
|
203 |
+
(sub for sub in test.get('submissions', [])
|
204 |
+
if sub['student_id'] == student_id),
|
205 |
+
None
|
206 |
+
)
|
207 |
+
|
208 |
+
if submission:
|
209 |
+
st.write("### Student's Answers")
|
210 |
+
for i, (question, answer) in enumerate(zip(test['questions'], submission['answers'])):
|
211 |
+
st.markdown(f"**Q{i+1}:** {question['question']}")
|
212 |
+
st.markdown(f"**A{i+1}:** {answer}")
|
213 |
+
st.markdown("---")
|
214 |
+
|
215 |
+
# Generate/display analysis
|
216 |
+
if st.button(f"Generate Analysis for {test.get('title')}"):
|
217 |
+
with st.spinner("Analyzing responses..."):
|
218 |
+
analysis = evaluate_subjective_answers(
|
219 |
+
str(test['_id']),
|
220 |
+
student_id,
|
221 |
+
course_id
|
222 |
+
)
|
223 |
+
|
224 |
+
if analysis:
|
225 |
+
st.markdown("### Analysis")
|
226 |
+
st.markdown(analysis['content_analysis'])
|
227 |
+
st.metric("Score", f"{analysis['correctness_score']}/10")
|
228 |
+
else:
|
229 |
+
st.error("Submission data not found for this student")
|
230 |
+
|
231 |
+
except Exception as e:
|
232 |
+
st.error("An error occurred while loading the tests")
|
233 |
+
with st.expander("Error Details"):
|
234 |
+
st.write(f"Error: {str(e)}")
|
235 |
+
st.write(f"Session ID: {session_id}")
|
236 |
+
st.write(f"Student ID: {student_id}")
|
237 |
+
st.write(f"Course ID: {course_id}")
|
238 |
+
|
239 |
+
def check_test_submission(session_id, student_id, course_id):
|
240 |
+
"""Utility function to check test submission status"""
|
241 |
+
try:
|
242 |
+
query = {
|
243 |
+
"session_id": str(session_id),
|
244 |
+
"course_id": str(course_id),
|
245 |
+
"submissions.student_id": str(student_id)
|
246 |
+
}
|
247 |
+
|
248 |
+
test = subjective_tests_collection.find_one(query)
|
249 |
+
return bool(test)
|
250 |
+
except Exception as e:
|
251 |
+
print(f"Error checking submission: {e}")
|
252 |
+
return False
|