-
Notifications
You must be signed in to change notification settings - Fork 0
Expand file tree
/
Copy pathresearch.html
More file actions
447 lines (435 loc) · 16.6 KB
/
research.html
File metadata and controls
447 lines (435 loc) · 16.6 KB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
<!doctype html>
<html lang="en" data-theme="dark">
<head>
<meta charset="UTF-8" />
<meta name="viewport" content="width=device-width, initial-scale=1.0" />
<title>Research — ExplainGrade academic background</title>
<link rel="preconnect" href="https://fonts.googleapis.com" />
<link rel="preconnect" href="https://fonts.gstatic.com" crossorigin />
<link
href="https://fonts.googleapis.com/css2?family=Inter:wght@400;500;600;700;800;900&display=swap"
rel="stylesheet"
/>
<link rel="stylesheet" href="style.css" />
</head>
<body>
<div class="bg-blobs">
<div class="blob blob-1"></div>
<div class="blob blob-3"></div>
</div>
<nav class="navbar" id="navbar">
<div class="container">
<a href="index.html" class="nav-logo">🔬 ExplainGrade</a>
<ul class="nav-links">
<li><a href="index.html" class="nav-link">Home</a></li>
<li><a href="pipeline.html" class="nav-link">Pipeline</a></li>
<li><a href="research.html" class="nav-link active">Research</a></li>
<li><a href="usage.html" class="nav-link">How to Use</a></li>
<li><a href="chat.html" class="nav-link">Assistant 💬</a></li>
</ul>
</div>
</nav>
<main class="section container" style="margin-top: 100px">
<span class="badge">Research Background</span>
<h1 class="hero-title">Academic Foundation</h1>
<p class="hero-subtitle">
ExplainGrade is built on two published research works and benchmarked
against the Mohler ASAG Dataset.
</p>
<div
style="
display: grid;
grid-template-columns: 1fr 1fr;
gap: 2rem;
margin-top: 4rem;
"
>
<div class="glass-card">
<h3 style="margin-bottom: 1.5rem">🎓 Project Overview</h3>
<p style="color: var(--text-dim); margin-bottom: 1rem">
ExplainGrade addresses the "length-bias noise" (longer answers
getting artificially inflated scores) and "black-box scoring"
(students receive no actionable feedback) present in most ASAG
systems.
</p>
<p style="color: var(--text-dim)">
By anchoring every grade to specific, measurable NLP comparisons and
providing sentence-level attributions, every point is traceable.
</p>
</div>
<div class="glass-card">
<h3 style="margin-bottom: 1.5rem">📄 Key References</h3>
<ul style="list-style: none; display: grid; gap: 1.5rem">
<li style="display: flex; gap: 1rem">
<span style="font-weight: 800; color: var(--primary)">01</span>
<div>
<strong>Kulkarni et al. (2015)</strong><br />
<span style="font-size: 0.85rem; color: var(--text-muted)"
>Statistically Significant Detection of Linguistic Change.
Foundation for temporal semantic drift tracking and concept
evolution detection.</span
>
</div>
</li>
<li style="display: flex; gap: 1rem">
<span style="font-weight: 800; color: var(--primary)">02</span>
<div>
<strong>Hamilton et al. (2016)</strong><br />
<span style="font-size: 0.85rem; color: var(--text-muted)"
>Cultural shift or linguistic drift? Word embeddings and
diachronic semantics. Core methodology for temporal semantic
drift measurement.</span
>
</div>
</li>
<li style="display: flex; gap: 1rem">
<span style="font-weight: 800; color: var(--primary)">03</span>
<div>
<strong>Bamler & Mandt (2017)</strong><br />
<span style="font-size: 0.85rem; color: var(--text-muted)"
>Dynamic Word Embeddings. Framework for tracking semantic
changes over time sequences.</span
>
</div>
</li>
<li style="display: flex; gap: 1rem">
<span style="font-weight: 800; color: var(--primary)">04</span>
<div>
<strong>Gama et al. (2014)</strong><br />
<span style="font-size: 0.85rem; color: var(--text-muted)"
>A Survey on Concept Drift Adaptation. Classification of
concept drift detection techniques applied to student
understanding tracking.</span
>
</div>
</li>
<li style="display: flex; gap: 1rem">
<span style="font-weight: 800; color: var(--primary)">05</span>
<div>
<strong>Ahmad Ayaan (2024)</strong><br />
<span style="font-size: 0.85rem; color: var(--text-muted)"
>PMC12171532. Automated grading using NLP and semantic
analysis.</span
>
</div>
</li>
<li style="display: flex; gap: 1rem">
<span style="font-weight: 800; color: var(--primary)">06</span>
<div>
<strong>Filighera et al. (2023)</strong><br />
<span style="font-size: 0.85rem; color: var(--text-muted)"
>Our System for Short Answer Grading using Generative Models.
BEA Workshop, ACL 2023. Sentence-level attribution
framework.</span
>
</div>
</li>
<li style="display: flex; gap: 1rem">
<span style="font-weight: 800; color: var(--primary)">07</span>
<div>
<strong>Mohler et al. (2011)</strong><br />
<span style="font-size: 0.85rem; color: var(--text-muted)"
>Learning to grade short answer questions using semantic
similarity. ACL. Foundational ASAG dataset and
methodology.</span
>
</div>
</li>
</ul>
</div>
</div>
<div class="glass-card" style="margin-top: 4rem">
<h3 class="mb-4">🛠 Tech Stack & Tools</h3>
<div style="display: flex; flex-wrap: wrap; gap: 0.75rem">
<span class="badge">Python 3.12</span>
<span class="badge">KeyBERT</span>
<span class="badge">Sentence Transformers</span>
<span class="badge">all-MiniLM-L6-v2</span>
<span class="badge">SHAP</span>
<span class="badge">PapaParse</span>
<span class="badge">Canvas API</span>
<span class="badge">Mammoth .docx</span>
<span class="badge">XLSX.js</span>
</div>
<div
style="margin-top: 2rem; color: var(--text-muted); font-size: 0.88rem"
>
ExplainGrade leverages modern NLP models for semantic understanding
while remaining light enough to run client-side in the browser for
privacy and speed.
</div>
</div>
<!-- ═══ TEMPORAL SEMANTIC DRIFT ANALYSIS SECTION ═══ -->
<div
style="
margin-top: 6rem;
padding-top: 3rem;
border-top: 2px solid var(--border-glass);
"
>
<span class="badge">Innovation</span>
<h2 class="hero-title" style="margin-top: 1rem">
📈 Temporal Semantic Drift Analysis
</h2>
<p class="hero-subtitle">
Track how student understanding evolves over multiple submissions.
</p>
<div
style="
display: grid;
grid-template-columns: 1fr 1fr;
gap: 2rem;
margin-top: 3rem;
"
>
<div class="glass-card">
<h3 style="margin-bottom: 1.5rem">🔬 What It Measures</h3>
<ul
style="
list-style: none;
display: flex;
flex-direction: column;
gap: 1rem;
"
>
<li style="display: flex; gap: 1rem">
<span style="font-size: 1.5rem">📊</span>
<div>
<strong>Improvement Score (-1 to +1)</strong><br />
<span style="color: var(--text-dim); font-size: 0.9rem"
>Tracks whether student answers are getting better or worse
across submissions.</span
>
</div>
</li>
<li style="display: flex; gap: 1rem">
<span style="font-size: 1.5rem">⚡</span>
<div>
<strong>Consistency Score (0-1)</strong><br />
<span style="color: var(--text-dim); font-size: 0.9rem"
>Measures stability of understanding. High = consistent
understanding, Low = volatile responses.</span
>
</div>
</li>
<li style="display: flex; gap: 1rem">
<span style="font-size: 1.5rem">🎯</span>
<div>
<strong>Learning Trend</strong><br />
<span style="color: var(--text-dim); font-size: 0.9rem"
>Classification: Improving, Degrading, or Stable—shows
learning direction over time.</span
>
</div>
</li>
<li style="display: flex; gap: 1rem">
<span style="font-size: 1.5rem">🌊</span>
<div>
<strong>Volatility</strong><br />
<span style="color: var(--text-dim); font-size: 0.9rem"
>Unpredictability measure—identifies erratic or inconsistent
response patterns.</span
>
</div>
</li>
</ul>
</div>
<div class="glass-card">
<h3 style="margin-bottom: 1.5rem">🎓 How It Works</h3>
<p style="color: var(--text-dim); margin-bottom: 1rem">
When you submit multiple answer attempts in the demo:
</p>
<ol
style="
color: var(--text-dim);
display: flex;
flex-direction: column;
gap: 0.75rem;
margin-left: 1rem;
"
>
<li>
<strong>Submission 1</strong> — Answer is graded and stored
</li>
<li>
<strong>Submission 2+</strong> — System automatically computes
trajectory
</li>
<li>
<strong>Analysis</strong> — Temporal metrics are calculated and
visualized
</li>
<li>
<strong>Feedback</strong> — You see improvement/consistency
scores and learning trend
</li>
</ol>
<p
style="
color: var(--text-dim);
margin-top: 1.5rem;
font-size: 0.9rem;
"
>
<strong>Try it:</strong> Go to the
<a
href="index.html#demo"
style="color: var(--primary); text-decoration: underline"
>Live Demo</a
>, grade an answer, then modify it and grade again. The temporal
analysis will appear automatically.
</p>
</div>
</div>
<div
class="glass-card"
style="margin-top: 2rem; border-left: 4px solid var(--accent1)"
>
<h3 style="margin-bottom: 1rem">📚 Research Foundation</h3>
<p style="color: var(--text-dim); margin-bottom: 1.5rem">
Temporal Semantic Drift Analysis is built directly on four
peer-reviewed research papers (references 01-04):
</p>
<div style="display: grid; gap: 1rem">
<div
style="
padding: 1rem;
background: rgba(255, 255, 255, 0.03);
border-radius: var(--radius-sm);
border: 1px solid var(--border-glass);
"
>
<strong style="color: var(--primary)"
>Kulkarni et al. (2015)</strong
>
<p
style="
color: var(--text-dim);
font-size: 0.9rem;
margin-top: 0.5rem;
"
>
Provides statistical methodology for detecting significant
linguistic changes across time periods. Directly implemented for
improvement_score computation.
</p>
</div>
<div
style="
padding: 1rem;
background: rgba(255, 255, 255, 0.03);
border-radius: var(--radius-sm);
border: 1px solid var(--border-glass);
"
>
<strong style="color: var(--primary)"
>Hamilton et al. (2016)</strong
>
<p
style="
color: var(--text-dim);
font-size: 0.9rem;
margin-top: 0.5rem;
"
>
Word embedding drift measurement framework used to compute
semantic shift magnitude and direction across submissions.
</p>
</div>
<div
style="
padding: 1rem;
background: rgba(255, 255, 255, 0.03);
border-radius: var(--radius-sm);
border: 1px solid var(--border-glass);
"
>
<strong style="color: var(--primary)"
>Bamler & Mandt (2017)</strong
>
<p
style="
color: var(--text-dim);
font-size: 0.9rem;
margin-top: 0.5rem;
"
>
Dynamic embeddings framework validates temporal similarity
tracking and consistency scoring methodology.
</p>
</div>
<div
style="
padding: 1rem;
background: rgba(255, 255, 255, 0.03);
border-radius: var(--radius-sm);
border: 1px solid var(--border-glass);
"
>
<strong style="color: var(--primary)">Gama et al. (2014)</strong>
<p
style="
color: var(--text-dim);
font-size: 0.9rem;
margin-top: 0.5rem;
"
>
Concept drift detection taxonomy—identifies sudden understanding
changes, critical for education applications.
</p>
</div>
</div>
</div>
<div
class="glass-card"
style="margin-top: 2rem; border-left: 4px solid var(--accent2)"
>
<h3 style="margin-bottom: 1rem">💡 Why This Matters for Education</h3>
<ul
style="
list-style: none;
display: flex;
flex-direction: column;
gap: 1rem;
color: var(--text-dim);
"
>
<li>
<strong style="color: var(--text-primary)"
>📉 Detect confusion early:</strong
>
A degrading trend signals the student may be struggling, not just
different approaches.
</li>
<li>
<strong style="color: var(--text-primary)"
>✅ Recognize improvement patterns:</strong
>
Chart actual learning trajectories—improving score + improving
consistency = true learning.
</li>
<li>
<strong style="color: var(--text-primary)"
>🎯 Personalized feedback:</strong
>
System can suggest whether to clarify concepts vs. encourage
refinement.
</li>
<li>
<strong style="color: var(--text-primary)"
>📊 Research-backed insights:</strong
>
Directly implements four peer-reviewed papers on temporal
linguistics and concept drift.
</li>
</ul>
</div>
</div>
</main>
<footer class="footer">
<div class="container">
<p>© 2026 ExplainGrade · Research & Documentation</p>
</div>
</footer>
</body>
</html>