Spaces:
Running
Running
Update app.py
Browse files
app.py
CHANGED
|
@@ -7,13 +7,13 @@ import random
|
|
| 7 |
import re
|
| 8 |
warnings.filterwarnings('ignore')
|
| 9 |
|
| 10 |
-
#
|
| 11 |
STANDARD_ENTITIES = [
|
| 12 |
'DATE', 'EVENT', 'FAC', 'GPE', 'LANG', 'LOC',
|
| 13 |
-
'MISC', 'NORP', 'ORG', 'PER', 'PRODUCT', '
|
| 14 |
]
|
| 15 |
|
| 16 |
-
#
|
| 17 |
STANDARD_COLORS = {
|
| 18 |
'DATE': '#FF6B6B', # Red
|
| 19 |
'EVENT': '#4ECDC4', # Teal
|
|
@@ -26,10 +26,10 @@ STANDARD_COLORS = {
|
|
| 26 |
'ORG': '#55A3FF', # Light Blue
|
| 27 |
'PER': '#00B894', # Green
|
| 28 |
'PRODUCT': '#E17055', # Orange-Red
|
| 29 |
-
'
|
| 30 |
}
|
| 31 |
|
| 32 |
-
# Additional
|
| 33 |
CUSTOM_COLOR_PALETTE = [
|
| 34 |
'#FF9F43', '#10AC84', '#EE5A24', '#0FBC89', '#5F27CD',
|
| 35 |
'#FF3838', '#2F3640', '#3742FA', '#2ED573', '#FFA502',
|
|
@@ -44,28 +44,16 @@ class HybridNERManager:
|
|
| 44 |
self.flair_models = {}
|
| 45 |
self.all_entity_colors = {}
|
| 46 |
self.model_names = [
|
| 47 |
-
'
|
| 48 |
-
'
|
| 49 |
-
'
|
| 50 |
-
'
|
| 51 |
]
|
| 52 |
|
| 53 |
-
def load_gliner_model(self):
|
| 54 |
-
"""Load GLiNER model for custom entities"""
|
| 55 |
-
if self.gliner_model is None:
|
| 56 |
-
try:
|
| 57 |
-
# Use a more stable model for HF Spaces
|
| 58 |
-
self.gliner_model = GLiNER.from_pretrained("urchade/gliner_medium-v2.1")
|
| 59 |
-
print("β GLiNER model loaded successfully")
|
| 60 |
-
except Exception as e:
|
| 61 |
-
print(f"Error loading GLiNER model: {str(e)}")
|
| 62 |
-
return None
|
| 63 |
-
return self.gliner_model
|
| 64 |
-
|
| 65 |
def load_model(self, model_name):
|
| 66 |
"""Load the specified model"""
|
| 67 |
try:
|
| 68 |
-
if
|
| 69 |
return self.load_spacy_model()
|
| 70 |
elif 'flair' in model_name:
|
| 71 |
return self.load_flair_model(model_name)
|
|
@@ -76,16 +64,21 @@ class HybridNERManager:
|
|
| 76 |
return None
|
| 77 |
|
| 78 |
def load_spacy_model(self):
|
| 79 |
-
"""Load spaCy model for
|
| 80 |
if self.spacy_model is None:
|
| 81 |
try:
|
| 82 |
import spacy
|
| 83 |
try:
|
| 84 |
-
|
| 85 |
-
|
|
|
|
| 86 |
except OSError:
|
| 87 |
-
|
| 88 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
| 89 |
except Exception as e:
|
| 90 |
print(f"Error loading spaCy model: {str(e)}")
|
| 91 |
return None
|
|
@@ -98,15 +91,66 @@ class HybridNERManager:
|
|
| 98 |
from flair.models import SequenceTagger
|
| 99 |
if 'ontonotes' in model_name:
|
| 100 |
model = SequenceTagger.load("flair/ner-english-ontonotes-large")
|
|
|
|
| 101 |
else:
|
| 102 |
-
model = SequenceTagger.load("flair/ner-english-large")
|
|
|
|
| 103 |
self.flair_models[model_name] = model
|
| 104 |
-
print(f"β {model_name} loaded successfully")
|
| 105 |
except Exception as e:
|
| 106 |
print(f"Error loading {model_name}: {str(e)}")
|
| 107 |
-
|
|
|
|
| 108 |
return self.flair_models[model_name]
|
| 109 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 110 |
def extract_spacy_entities(self, text, entity_types):
|
| 111 |
"""Extract entities using spaCy"""
|
| 112 |
model = self.load_spacy_model()
|
|
@@ -131,35 +175,6 @@ class HybridNERManager:
|
|
| 131 |
print(f"Error with spaCy extraction: {str(e)}")
|
| 132 |
return []
|
| 133 |
|
| 134 |
-
def assign_colors(self, standard_entities, custom_entities):
|
| 135 |
-
"""Assign colors to all entity types"""
|
| 136 |
-
self.all_entity_colors = {}
|
| 137 |
-
|
| 138 |
-
# Assign standard colors
|
| 139 |
-
for entity in standard_entities:
|
| 140 |
-
self.all_entity_colors[entity.upper()] = STANDARD_COLORS.get(entity, '#CCCCCC')
|
| 141 |
-
|
| 142 |
-
# Assign custom colors
|
| 143 |
-
for i, entity in enumerate(custom_entities):
|
| 144 |
-
if i < len(CUSTOM_COLOR_PALETTE):
|
| 145 |
-
self.all_entity_colors[entity.upper()] = CUSTOM_COLOR_PALETTE[i]
|
| 146 |
-
else:
|
| 147 |
-
# Generate random color if we run out
|
| 148 |
-
self.all_entity_colors[entity.upper()] = f"#{random.randint(0, 0xFFFFFF):06x}"
|
| 149 |
-
|
| 150 |
-
return self.all_entity_colors
|
| 151 |
-
|
| 152 |
-
def extract_entities_by_model(self, text, entity_types, model_name, threshold=0.3):
|
| 153 |
-
"""Extract entities using the specified model"""
|
| 154 |
-
if model_name == 'spacy_en_core_web_sm':
|
| 155 |
-
return self.extract_spacy_entities(text, entity_types)
|
| 156 |
-
elif 'flair' in model_name:
|
| 157 |
-
return self.extract_flair_entities(text, entity_types, model_name)
|
| 158 |
-
elif 'gliner' in model_name:
|
| 159 |
-
return self.extract_gliner_entities(text, entity_types, threshold, is_custom=False)
|
| 160 |
-
else:
|
| 161 |
-
return []
|
| 162 |
-
|
| 163 |
def extract_flair_entities(self, text, entity_types, model_name):
|
| 164 |
"""Extract entities using Flair"""
|
| 165 |
model = self.load_flair_model(model_name)
|
|
@@ -172,7 +187,7 @@ class HybridNERManager:
|
|
| 172 |
model.predict(sentence)
|
| 173 |
entities = []
|
| 174 |
for entity in sentence.get_spans('ner'):
|
| 175 |
-
# Map Flair labels to our
|
| 176 |
label = entity.tag
|
| 177 |
if label == 'PERSON':
|
| 178 |
label = 'PER'
|
|
@@ -213,7 +228,7 @@ class HybridNERManager:
|
|
| 213 |
'start': entity['start'],
|
| 214 |
'end': entity['end'],
|
| 215 |
'confidence': entity.get('score', 0.0),
|
| 216 |
-
'source': 'GLiNER-Custom' if is_custom else 'GLiNER-
|
| 217 |
})
|
| 218 |
return result
|
| 219 |
except Exception as e:
|
|
@@ -241,7 +256,7 @@ def find_overlapping_entities(entities):
|
|
| 241 |
|
| 242 |
# Check if entities overlap
|
| 243 |
if (current_entity['start'] <= next_entity['start'] < current_entity['end'] or
|
| 244 |
-
next_entity['start'] <= current_entity['start'] <
|
| 245 |
current_entity['text'].lower() == next_entity['text'].lower()):
|
| 246 |
overlapping_entities.append(next_entity)
|
| 247 |
sorted_entities.pop(j)
|
|
@@ -303,7 +318,7 @@ def create_highlighted_html(text, entities, entity_colors):
|
|
| 303 |
html_parts.append(text[last_end:entity['start']])
|
| 304 |
|
| 305 |
if entity.get('is_shared', False):
|
| 306 |
-
# Handle shared entity with multiple
|
| 307 |
html_parts.append(create_shared_entity_html(entity, entity_colors))
|
| 308 |
else:
|
| 309 |
# Handle single entity
|
|
@@ -326,39 +341,39 @@ def create_highlighted_html(text, entities, entity_colors):
|
|
| 326 |
def create_single_entity_html(entity, entity_colors):
|
| 327 |
"""Create HTML for a single entity"""
|
| 328 |
label = entity['label']
|
| 329 |
-
|
| 330 |
confidence = entity.get('confidence', 0.0)
|
| 331 |
source = entity.get('source', 'Unknown')
|
| 332 |
|
| 333 |
-
return (f'<span style="background-color: {
|
| 334 |
f'border-radius: 3px; margin: 0 1px; '
|
| 335 |
-
f'border: 1px solid {
|
| 336 |
f'title="{label} ({source}) - confidence: {confidence:.2f}">'
|
| 337 |
f'{entity["text"]}</span>')
|
| 338 |
|
| 339 |
def create_shared_entity_html(entity, entity_colors):
|
| 340 |
-
"""Create HTML for a shared entity with multiple
|
| 341 |
labels = entity['labels']
|
| 342 |
sources = entity['sources']
|
| 343 |
confidences = entity['confidences']
|
| 344 |
|
| 345 |
-
# Get
|
| 346 |
-
|
| 347 |
for label in labels:
|
| 348 |
-
|
| 349 |
-
|
| 350 |
|
| 351 |
# Create gradient background
|
| 352 |
-
if len(
|
| 353 |
-
gradient = f"linear-gradient(to right, {
|
| 354 |
else:
|
| 355 |
-
# For more
|
| 356 |
-
segment_size = 100 / len(
|
| 357 |
gradient_parts = []
|
| 358 |
-
for i,
|
| 359 |
start = i * segment_size
|
| 360 |
end = (i + 1) * segment_size
|
| 361 |
-
gradient_parts.append(f"{
|
| 362 |
gradient = f"linear-gradient(to right, {', '.join(gradient_parts)})"
|
| 363 |
|
| 364 |
# Create tooltip
|
|
@@ -371,12 +386,12 @@ def create_shared_entity_html(entity, entity_colors):
|
|
| 371 |
f'border-radius: 3px; margin: 0 1px; '
|
| 372 |
f'border: 2px solid #333; color: white; font-weight: bold;" '
|
| 373 |
f'title="SHARED: {tooltip}">'
|
| 374 |
-
f'{entity["text"]}
|
| 375 |
|
| 376 |
-
def
|
| 377 |
-
"""Create
|
| 378 |
if not entities:
|
| 379 |
-
return "
|
| 380 |
|
| 381 |
# Share overlapping entities
|
| 382 |
shared_entities = find_overlapping_entities(entities)
|
|
@@ -394,80 +409,37 @@ def create_entity_table_html(entities, entity_colors):
|
|
| 394 |
entity_groups[key].append(entity)
|
| 395 |
|
| 396 |
if not entity_groups:
|
| 397 |
-
return "
|
| 398 |
-
|
| 399 |
-
# Create tabbed interface
|
| 400 |
-
tab_html = "<div style='margin: 20px 0;'>"
|
| 401 |
-
|
| 402 |
-
# Tab headers
|
| 403 |
-
tab_html += "<div style='border-bottom: 2px solid #ddd; margin-bottom: 20px;'>"
|
| 404 |
-
tab_headers = []
|
| 405 |
-
|
| 406 |
-
for i, entity_type in enumerate(sorted(entity_groups.keys())):
|
| 407 |
-
count = len(entity_groups[entity_type])
|
| 408 |
-
|
| 409 |
-
if entity_type == 'SHARED_ENTITIES':
|
| 410 |
-
color = '#666666'
|
| 411 |
-
icon = "π"
|
| 412 |
-
display_name = "SHARED"
|
| 413 |
-
else:
|
| 414 |
-
color = entity_colors.get(entity_type.upper(), '#f0f0f0')
|
| 415 |
-
# Determine if it's standard or custom
|
| 416 |
-
is_standard = entity_type in STANDARD_ENTITIES
|
| 417 |
-
icon = "π―" if is_standard else "β¨"
|
| 418 |
-
display_name = entity_type
|
| 419 |
-
|
| 420 |
-
active_style = f"background-color: #f8f9fa; border-bottom: 3px solid {color};" if i == 0 else "background-color: #fff;"
|
| 421 |
-
tab_headers.append(f"""
|
| 422 |
-
<button onclick="showTab('{entity_type}')" id="tab-{entity_type}"
|
| 423 |
-
style="padding: 12px 24px; margin-right: 5px; border: 1px solid #ddd;
|
| 424 |
-
border-bottom: none; cursor: pointer; font-weight: bold; {active_style}">
|
| 425 |
-
{icon} {display_name} ({count})
|
| 426 |
-
</button>
|
| 427 |
-
""")
|
| 428 |
-
|
| 429 |
-
tab_html += ''.join(tab_headers)
|
| 430 |
-
tab_html += "</div>"
|
| 431 |
-
|
| 432 |
-
# Tab content
|
| 433 |
-
for i, entity_type in enumerate(sorted(entity_groups.keys())):
|
| 434 |
-
entities_of_type = entity_groups[entity_type]
|
| 435 |
-
display_style = "display: block;" if i == 0 else "display: none;"
|
| 436 |
|
|
|
|
|
|
|
|
|
|
|
|
|
| 437 |
if entity_type == 'SHARED_ENTITIES':
|
| 438 |
-
|
| 439 |
-
|
| 440 |
-
|
| 441 |
-
|
| 442 |
-
|
| 443 |
-
|
| 444 |
-
|
| 445 |
-
|
| 446 |
-
|
| 447 |
-
|
| 448 |
-
|
| 449 |
-
|
| 450 |
-
|
| 451 |
-
|
| 452 |
-
|
| 453 |
-
|
| 454 |
-
|
| 455 |
-
tab_html += f"""
|
| 456 |
-
<tr style="background-color: {color}; color: white;">
|
| 457 |
-
<th style="padding: 12px; text-align: left; border: 1px solid #ddd;">Entity Text</th>
|
| 458 |
-
<th style="padding: 12px; text-align: left; border: 1px solid #ddd;">All Labels</th>
|
| 459 |
-
<th style="padding: 12px; text-align: left; border: 1px solid #ddd;">Sources</th>
|
| 460 |
-
<th style="padding: 12px; text-align: left; border: 1px solid #ddd;">Count</th>
|
| 461 |
-
</tr>
|
| 462 |
-
</thead>
|
| 463 |
-
<tbody>
|
| 464 |
"""
|
| 465 |
-
|
| 466 |
for entity in entities_of_type:
|
| 467 |
labels_text = " | ".join(entity['labels'])
|
| 468 |
sources_text = " | ".join(entity['sources'])
|
| 469 |
-
|
| 470 |
-
|
| 471 |
<tr style="background-color: #fff;">
|
| 472 |
<td style="padding: 10px; border: 1px solid #ddd; font-weight: bold;">{entity['text']}</td>
|
| 473 |
<td style="padding: 10px; border: 1px solid #ddd;">{labels_text}</td>
|
|
@@ -479,32 +451,48 @@ def create_entity_table_html(entities, entity_colors):
|
|
| 479 |
</td>
|
| 480 |
</tr>
|
| 481 |
"""
|
|
|
|
|
|
|
|
|
|
|
|
|
| 482 |
else:
|
| 483 |
-
|
| 484 |
-
|
| 485 |
-
|
| 486 |
-
|
| 487 |
-
|
| 488 |
-
|
| 489 |
-
|
| 490 |
-
|
| 491 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 492 |
"""
|
| 493 |
-
|
| 494 |
# Sort by confidence score
|
| 495 |
entities_of_type.sort(key=lambda x: x.get('confidence', 0), reverse=True)
|
| 496 |
-
|
| 497 |
for entity in entities_of_type:
|
| 498 |
confidence = entity.get('confidence', 0.0)
|
| 499 |
-
|
| 500 |
source = entity.get('source', 'Unknown')
|
| 501 |
source_badge = f"<span style='background-color: #007bff; color: white; padding: 2px 6px; border-radius: 10px; font-size: 11px;'>{source}</span>"
|
| 502 |
-
|
| 503 |
-
|
| 504 |
<tr style="background-color: #fff;">
|
| 505 |
<td style="padding: 10px; border: 1px solid #ddd; font-weight: bold;">{entity['text']}</td>
|
| 506 |
<td style="padding: 10px; border: 1px solid #ddd;">
|
| 507 |
-
<span style="color: {
|
| 508 |
{confidence:.3f}
|
| 509 |
</span>
|
| 510 |
</td>
|
|
@@ -512,46 +500,15 @@ def create_entity_table_html(entities, entity_colors):
|
|
| 512 |
<td style="padding: 10px; border: 1px solid #ddd;">{source_badge}</td>
|
| 513 |
</tr>
|
| 514 |
"""
|
| 515 |
-
|
| 516 |
-
|
| 517 |
-
|
| 518 |
-
|
| 519 |
-
|
| 520 |
-
|
| 521 |
-
|
| 522 |
-
# JavaScript for tab switching
|
| 523 |
-
tab_html += """
|
| 524 |
-
<script>
|
| 525 |
-
function showTab(entityType) {
|
| 526 |
-
// Hide all content
|
| 527 |
-
var contents = document.querySelectorAll('[id^="content-"]');
|
| 528 |
-
contents.forEach(function(content) {
|
| 529 |
-
content.style.display = 'none';
|
| 530 |
-
});
|
| 531 |
-
|
| 532 |
-
// Reset all tab styles
|
| 533 |
-
var tabs = document.querySelectorAll('[id^="tab-"]');
|
| 534 |
-
tabs.forEach(function(tab) {
|
| 535 |
-
tab.style.backgroundColor = '#fff';
|
| 536 |
-
tab.style.borderBottom = 'none';
|
| 537 |
-
});
|
| 538 |
-
|
| 539 |
-
// Show selected content
|
| 540 |
-
document.getElementById('content-' + entityType).style.display = 'block';
|
| 541 |
-
|
| 542 |
-
// Highlight selected tab
|
| 543 |
-
var activeTab = document.getElementById('tab-' + entityType);
|
| 544 |
-
activeTab.style.backgroundColor = '#f8f9fa';
|
| 545 |
-
activeTab.style.borderBottom = '3px solid #4ECDC4';
|
| 546 |
-
}
|
| 547 |
-
</script>
|
| 548 |
-
"""
|
| 549 |
-
|
| 550 |
-
tab_html += "</div>"
|
| 551 |
-
return tab_html
|
| 552 |
|
| 553 |
def create_legend_html(entity_colors, standard_entities, custom_entities):
|
| 554 |
-
"""Create a legend showing entity
|
| 555 |
if not entity_colors:
|
| 556 |
return ""
|
| 557 |
|
|
@@ -560,11 +517,11 @@ def create_legend_html(entity_colors, standard_entities, custom_entities):
|
|
| 560 |
|
| 561 |
if standard_entities:
|
| 562 |
html += "<div style='margin-bottom: 15px;'>"
|
| 563 |
-
html += "<h5 style='margin: 0 0 8px 0;'>π―
|
| 564 |
html += "<div style='display: flex; flex-wrap: wrap; gap: 8px;'>"
|
| 565 |
for entity_type in standard_entities:
|
| 566 |
-
|
| 567 |
-
html += f"<span style='background-color: {
|
| 568 |
html += "</div></div>"
|
| 569 |
|
| 570 |
if custom_entities:
|
|
@@ -572,8 +529,8 @@ def create_legend_html(entity_colors, standard_entities, custom_entities):
|
|
| 572 |
html += "<h5 style='margin: 0 0 8px 0;'>β¨ Custom Entities:</h5>"
|
| 573 |
html += "<div style='display: flex; flex-wrap: wrap; gap: 8px;'>"
|
| 574 |
for entity_type in custom_entities:
|
| 575 |
-
|
| 576 |
-
html += f"<span style='background-color: {
|
| 577 |
html += "</div></div>"
|
| 578 |
|
| 579 |
html += "</div>"
|
|
@@ -582,45 +539,55 @@ def create_legend_html(entity_colors, standard_entities, custom_entities):
|
|
| 582 |
# Initialize the NER manager
|
| 583 |
ner_manager = HybridNERManager()
|
| 584 |
|
| 585 |
-
def process_text(text, standard_entities, custom_entities_str, confidence_threshold, selected_model):
|
| 586 |
-
"""Main processing function for Gradio interface"""
|
| 587 |
if not text.strip():
|
| 588 |
-
return "β Please enter some text to
|
| 589 |
|
|
|
|
|
|
|
| 590 |
# Parse custom entities
|
| 591 |
custom_entities = []
|
| 592 |
if custom_entities_str.strip():
|
| 593 |
custom_entities = [entity.strip() for entity in custom_entities_str.split(',') if entity.strip()]
|
| 594 |
|
| 595 |
-
# Parse
|
| 596 |
selected_standard = [entity for entity in standard_entities if entity]
|
| 597 |
|
| 598 |
if not selected_standard and not custom_entities:
|
| 599 |
-
return "β Please select at least one
|
| 600 |
|
|
|
|
|
|
|
| 601 |
all_entities = []
|
| 602 |
|
| 603 |
-
# Extract
|
| 604 |
if selected_standard and selected_model:
|
|
|
|
| 605 |
standard_entities_results = ner_manager.extract_entities_by_model(text, selected_standard, selected_model, confidence_threshold)
|
| 606 |
all_entities.extend(standard_entities_results)
|
| 607 |
|
| 608 |
# Extract custom entities using GLiNER
|
| 609 |
if custom_entities:
|
|
|
|
| 610 |
custom_entity_results = ner_manager.extract_gliner_entities(text, custom_entities, confidence_threshold, is_custom=True)
|
| 611 |
all_entities.extend(custom_entity_results)
|
| 612 |
|
| 613 |
if not all_entities:
|
| 614 |
-
return "β No entities found. Try lowering the confidence threshold or using different entity types.", "",
|
| 615 |
|
| 616 |
-
|
| 617 |
-
|
|
|
|
|
|
|
| 618 |
|
| 619 |
# Create outputs
|
| 620 |
legend_html = create_legend_html(entity_colors, selected_standard, custom_entities)
|
| 621 |
highlighted_html = create_highlighted_html(text, all_entities, entity_colors)
|
| 622 |
-
|
| 623 |
|
|
|
|
|
|
|
| 624 |
# Create summary with shared entities terminology
|
| 625 |
total_entities = len(all_entities)
|
| 626 |
shared_entities = find_overlapping_entities(all_entities)
|
|
@@ -635,7 +602,9 @@ def process_text(text, standard_entities, custom_entities_str, confidence_thresh
|
|
| 635 |
- **Average confidence:** {sum(e.get('confidence', 0) for e in all_entities) / total_entities:.3f}
|
| 636 |
"""
|
| 637 |
|
| 638 |
-
|
|
|
|
|
|
|
| 639 |
|
| 640 |
# Create Gradio interface
|
| 641 |
def create_interface():
|
|
@@ -643,23 +612,23 @@ def create_interface():
|
|
| 643 |
gr.Markdown("""
|
| 644 |
# π― Hybrid NER + Custom GLiNER Entity Recognition Tool
|
| 645 |
|
| 646 |
-
Combine
|
| 647 |
|
| 648 |
-
##
|
| 649 |
|
| 650 |
### How to use:
|
| 651 |
1. **π Enter your text** in the text area below
|
| 652 |
-
2. **π― Select a model** from the dropdown for
|
| 653 |
-
3. **βοΈ Select
|
| 654 |
4. **β¨ Add custom entities** (comma-separated) like "relationships, occupations, skills"
|
| 655 |
5. **βοΈ Adjust confidence threshold**
|
| 656 |
-
6. **π Click "
|
| 657 |
""")
|
| 658 |
|
| 659 |
with gr.Row():
|
| 660 |
with gr.Column(scale=2):
|
| 661 |
text_input = gr.Textbox(
|
| 662 |
-
label="π Text to
|
| 663 |
placeholder="Enter your text here...",
|
| 664 |
lines=6,
|
| 665 |
max_lines=10
|
|
@@ -676,21 +645,21 @@ def create_interface():
|
|
| 676 |
|
| 677 |
with gr.Row():
|
| 678 |
with gr.Column():
|
| 679 |
-
gr.Markdown("### π―
|
| 680 |
|
| 681 |
# Model selector
|
| 682 |
model_dropdown = gr.Dropdown(
|
| 683 |
choices=ner_manager.model_names,
|
| 684 |
value=ner_manager.model_names[0],
|
| 685 |
-
label="Select Model for
|
| 686 |
-
info="Choose which model to use for
|
| 687 |
)
|
| 688 |
|
| 689 |
-
#
|
| 690 |
standard_entities = gr.CheckboxGroup(
|
| 691 |
choices=STANDARD_ENTITIES,
|
| 692 |
value=['PER', 'ORG', 'LOC', 'MISC'], # Default selection
|
| 693 |
-
label="Select
|
| 694 |
)
|
| 695 |
|
| 696 |
# Select/Deselect All button
|
|
@@ -727,7 +696,7 @@ def create_interface():
|
|
| 727 |
- financial terms, business roles
|
| 728 |
""")
|
| 729 |
|
| 730 |
-
|
| 731 |
|
| 732 |
# Output sections
|
| 733 |
with gr.Row():
|
|
@@ -736,12 +705,104 @@ def create_interface():
|
|
| 736 |
with gr.Row():
|
| 737 |
highlighted_output = gr.HTML(label="Highlighted Text")
|
| 738 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 739 |
with gr.Row():
|
| 740 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 741 |
|
| 742 |
# Connect the button to the processing function
|
| 743 |
-
|
| 744 |
-
fn=
|
| 745 |
inputs=[
|
| 746 |
text_input,
|
| 747 |
standard_entities,
|
|
@@ -749,32 +810,32 @@ def create_interface():
|
|
| 749 |
confidence_threshold,
|
| 750 |
model_dropdown
|
| 751 |
],
|
| 752 |
-
outputs=[summary_output, highlighted_output,
|
| 753 |
)
|
| 754 |
|
| 755 |
# Add examples
|
| 756 |
gr.Examples(
|
| 757 |
examples=[
|
| 758 |
[
|
| 759 |
-
"John Smith works at Google in New York. He graduated from Stanford University in 2015 and
|
| 760 |
["PER", "ORG", "LOC", "DATE"],
|
| 761 |
"relationships, occupations, educational background",
|
| 762 |
0.3,
|
| 763 |
-
"
|
| 764 |
],
|
| 765 |
[
|
| 766 |
"The meeting between CEO Jane Doe and the board of directors at Microsoft headquarters in Seattle discussed the Q4 financial results and the new AI strategy for 2024.",
|
| 767 |
["PER", "ORG", "LOC", "DATE"],
|
| 768 |
"corporate roles, business events, financial terms",
|
| 769 |
0.4,
|
| 770 |
-
"
|
| 771 |
],
|
| 772 |
[
|
| 773 |
"Dr. Emily Watson published a research paper on machine learning algorithms at MIT. She collaborates with her colleague Prof. David Chen on natural language processing projects.",
|
| 774 |
-
["PER", "ORG", "
|
| 775 |
"academic titles, research topics, collaborations",
|
| 776 |
0.3,
|
| 777 |
-
"
|
| 778 |
]
|
| 779 |
],
|
| 780 |
inputs=[
|
|
|
|
| 7 |
import re
|
| 8 |
warnings.filterwarnings('ignore')
|
| 9 |
|
| 10 |
+
# Common NER entity types
|
| 11 |
STANDARD_ENTITIES = [
|
| 12 |
'DATE', 'EVENT', 'FAC', 'GPE', 'LANG', 'LOC',
|
| 13 |
+
'MISC', 'NORP', 'ORG', 'PER', 'PRODUCT', 'Work of Art'
|
| 14 |
]
|
| 15 |
|
| 16 |
+
# Colour schemes
|
| 17 |
STANDARD_COLORS = {
|
| 18 |
'DATE': '#FF6B6B', # Red
|
| 19 |
'EVENT': '#4ECDC4', # Teal
|
|
|
|
| 26 |
'ORG': '#55A3FF', # Light Blue
|
| 27 |
'PER': '#00B894', # Green
|
| 28 |
'PRODUCT': '#E17055', # Orange-Red
|
| 29 |
+
'WORK OF ART': '#DDA0DD' # Plum
|
| 30 |
}
|
| 31 |
|
| 32 |
+
# Additional colours for custom entities
|
| 33 |
CUSTOM_COLOR_PALETTE = [
|
| 34 |
'#FF9F43', '#10AC84', '#EE5A24', '#0FBC89', '#5F27CD',
|
| 35 |
'#FF3838', '#2F3640', '#3742FA', '#2ED573', '#FFA502',
|
|
|
|
| 44 |
self.flair_models = {}
|
| 45 |
self.all_entity_colors = {}
|
| 46 |
self.model_names = [
|
| 47 |
+
'entities_flair_ner-large',
|
| 48 |
+
'entities_spacy_en_core_web_trf',
|
| 49 |
+
'entities_flair_ner-ontonotes-large',
|
| 50 |
+
'entities_gliner_knowledgator/modern-gliner-bi-large-v1.0'
|
| 51 |
]
|
| 52 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 53 |
def load_model(self, model_name):
|
| 54 |
"""Load the specified model"""
|
| 55 |
try:
|
| 56 |
+
if 'spacy' in model_name:
|
| 57 |
return self.load_spacy_model()
|
| 58 |
elif 'flair' in model_name:
|
| 59 |
return self.load_flair_model(model_name)
|
|
|
|
| 64 |
return None
|
| 65 |
|
| 66 |
def load_spacy_model(self):
|
| 67 |
+
"""Load spaCy model for common NER"""
|
| 68 |
if self.spacy_model is None:
|
| 69 |
try:
|
| 70 |
import spacy
|
| 71 |
try:
|
| 72 |
+
# Try transformer model first, fallback to small model
|
| 73 |
+
self.spacy_model = spacy.load("en_core_web_trf")
|
| 74 |
+
print("β spaCy transformer model loaded successfully")
|
| 75 |
except OSError:
|
| 76 |
+
try:
|
| 77 |
+
self.spacy_model = spacy.load("en_core_web_sm")
|
| 78 |
+
print("β spaCy common model loaded successfully")
|
| 79 |
+
except OSError:
|
| 80 |
+
print("spaCy model not found. Using GLiNER for all entity types.")
|
| 81 |
+
return None
|
| 82 |
except Exception as e:
|
| 83 |
print(f"Error loading spaCy model: {str(e)}")
|
| 84 |
return None
|
|
|
|
| 91 |
from flair.models import SequenceTagger
|
| 92 |
if 'ontonotes' in model_name:
|
| 93 |
model = SequenceTagger.load("flair/ner-english-ontonotes-large")
|
| 94 |
+
print("β Flair OntoNotes model loaded successfully")
|
| 95 |
else:
|
| 96 |
+
model = SequenceTagger.load("flair/ner-english-large")
|
| 97 |
+
print("β Flair large model loaded successfully")
|
| 98 |
self.flair_models[model_name] = model
|
|
|
|
| 99 |
except Exception as e:
|
| 100 |
print(f"Error loading {model_name}: {str(e)}")
|
| 101 |
+
# Fallback to GLiNER
|
| 102 |
+
return self.load_gliner_model()
|
| 103 |
return self.flair_models[model_name]
|
| 104 |
|
| 105 |
+
def load_gliner_model(self):
|
| 106 |
+
"""Load GLiNER model for custom entities"""
|
| 107 |
+
if self.gliner_model is None:
|
| 108 |
+
try:
|
| 109 |
+
# Try the modern GLiNER model first, fallback to stable model
|
| 110 |
+
self.gliner_model = GLiNER.from_pretrained("knowledgator/gliner-bi-large-v1.0")
|
| 111 |
+
print("β GLiNER knowledgator model loaded successfully")
|
| 112 |
+
except Exception as e:
|
| 113 |
+
print(f"Primary GLiNER model failed: {str(e)}")
|
| 114 |
+
try:
|
| 115 |
+
# Fallback to stable model
|
| 116 |
+
self.gliner_model = GLiNER.from_pretrained("urchade/gliner_medium-v2.1")
|
| 117 |
+
print("β GLiNER fallback model loaded successfully")
|
| 118 |
+
except Exception as e2:
|
| 119 |
+
print(f"Error loading GLiNER model: {str(e2)}")
|
| 120 |
+
return None
|
| 121 |
+
return self.gliner_model
|
| 122 |
+
|
| 123 |
+
def assign_colours(self, standard_entities, custom_entities):
|
| 124 |
+
"""Assign colours to all entity types"""
|
| 125 |
+
self.all_entity_colors = {}
|
| 126 |
+
|
| 127 |
+
# Assign common colours
|
| 128 |
+
for entity in standard_entities:
|
| 129 |
+
# Handle the special case of "Work of Art"
|
| 130 |
+
colour_key = "WORK OF ART" if entity == "Work of Art" else entity.upper()
|
| 131 |
+
self.all_entity_colors[entity.upper()] = STANDARD_COLORS.get(colour_key, '#CCCCCC')
|
| 132 |
+
|
| 133 |
+
# Assign custom colours
|
| 134 |
+
for i, entity in enumerate(custom_entities):
|
| 135 |
+
if i < len(CUSTOM_COLOR_PALETTE):
|
| 136 |
+
self.all_entity_colors[entity.upper()] = CUSTOM_COLOR_PALETTE[i]
|
| 137 |
+
else:
|
| 138 |
+
# Generate random colour if we run out
|
| 139 |
+
self.all_entity_colors[entity.upper()] = f"#{random.randint(0, 0xFFFFFF):06x}"
|
| 140 |
+
|
| 141 |
+
return self.all_entity_colors
|
| 142 |
+
|
| 143 |
+
def extract_entities_by_model(self, text, entity_types, model_name, threshold=0.3):
|
| 144 |
+
"""Extract entities using the specified model"""
|
| 145 |
+
if 'spacy' in model_name:
|
| 146 |
+
return self.extract_spacy_entities(text, entity_types)
|
| 147 |
+
elif 'flair' in model_name:
|
| 148 |
+
return self.extract_flair_entities(text, entity_types, model_name)
|
| 149 |
+
elif 'gliner' in model_name:
|
| 150 |
+
return self.extract_gliner_entities(text, entity_types, threshold, is_custom=False)
|
| 151 |
+
else:
|
| 152 |
+
return []
|
| 153 |
+
|
| 154 |
def extract_spacy_entities(self, text, entity_types):
|
| 155 |
"""Extract entities using spaCy"""
|
| 156 |
model = self.load_spacy_model()
|
|
|
|
| 175 |
print(f"Error with spaCy extraction: {str(e)}")
|
| 176 |
return []
|
| 177 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 178 |
def extract_flair_entities(self, text, entity_types, model_name):
|
| 179 |
"""Extract entities using Flair"""
|
| 180 |
model = self.load_flair_model(model_name)
|
|
|
|
| 187 |
model.predict(sentence)
|
| 188 |
entities = []
|
| 189 |
for entity in sentence.get_spans('ner'):
|
| 190 |
+
# Map Flair labels to our common set
|
| 191 |
label = entity.tag
|
| 192 |
if label == 'PERSON':
|
| 193 |
label = 'PER'
|
|
|
|
| 228 |
'start': entity['start'],
|
| 229 |
'end': entity['end'],
|
| 230 |
'confidence': entity.get('score', 0.0),
|
| 231 |
+
'source': 'GLiNER-Custom' if is_custom else 'GLiNER-Common'
|
| 232 |
})
|
| 233 |
return result
|
| 234 |
except Exception as e:
|
|
|
|
| 256 |
|
| 257 |
# Check if entities overlap
|
| 258 |
if (current_entity['start'] <= next_entity['start'] < current_entity['end'] or
|
| 259 |
+
next_entity['start'] <= current_entity['start'] < current_entity['end'] or
|
| 260 |
current_entity['text'].lower() == next_entity['text'].lower()):
|
| 261 |
overlapping_entities.append(next_entity)
|
| 262 |
sorted_entities.pop(j)
|
|
|
|
| 318 |
html_parts.append(text[last_end:entity['start']])
|
| 319 |
|
| 320 |
if entity.get('is_shared', False):
|
| 321 |
+
# Handle shared entity with multiple colours
|
| 322 |
html_parts.append(create_shared_entity_html(entity, entity_colors))
|
| 323 |
else:
|
| 324 |
# Handle single entity
|
|
|
|
| 341 |
def create_single_entity_html(entity, entity_colors):
|
| 342 |
"""Create HTML for a single entity"""
|
| 343 |
label = entity['label']
|
| 344 |
+
colour = entity_colors.get(label.upper(), '#CCCCCC')
|
| 345 |
confidence = entity.get('confidence', 0.0)
|
| 346 |
source = entity.get('source', 'Unknown')
|
| 347 |
|
| 348 |
+
return (f'<span style="background-color: {colour}; padding: 2px 4px; '
|
| 349 |
f'border-radius: 3px; margin: 0 1px; '
|
| 350 |
+
f'border: 1px solid {colour}; color: white; font-weight: bold;" '
|
| 351 |
f'title="{label} ({source}) - confidence: {confidence:.2f}">'
|
| 352 |
f'{entity["text"]}</span>')
|
| 353 |
|
| 354 |
def create_shared_entity_html(entity, entity_colors):
|
| 355 |
+
"""Create HTML for a shared entity with multiple colours"""
|
| 356 |
labels = entity['labels']
|
| 357 |
sources = entity['sources']
|
| 358 |
confidences = entity['confidences']
|
| 359 |
|
| 360 |
+
# Get colours for each label
|
| 361 |
+
colours = []
|
| 362 |
for label in labels:
|
| 363 |
+
colour = entity_colors.get(label.upper(), '#CCCCCC')
|
| 364 |
+
colours.append(colour)
|
| 365 |
|
| 366 |
# Create gradient background
|
| 367 |
+
if len(colours) == 2:
|
| 368 |
+
gradient = f"linear-gradient(to right, {colours[0]} 50%, {colours[1]} 50%)"
|
| 369 |
else:
|
| 370 |
+
# For more colours, create equal segments
|
| 371 |
+
segment_size = 100 / len(colours)
|
| 372 |
gradient_parts = []
|
| 373 |
+
for i, colour in enumerate(colours):
|
| 374 |
start = i * segment_size
|
| 375 |
end = (i + 1) * segment_size
|
| 376 |
+
gradient_parts.append(f"{colour} {start}%, {colour} {end}%")
|
| 377 |
gradient = f"linear-gradient(to right, {', '.join(gradient_parts)})"
|
| 378 |
|
| 379 |
# Create tooltip
|
|
|
|
| 386 |
f'border-radius: 3px; margin: 0 1px; '
|
| 387 |
f'border: 2px solid #333; color: white; font-weight: bold;" '
|
| 388 |
f'title="SHARED: {tooltip}">'
|
| 389 |
+
f'{entity["text"]} π§©π§©</span>')
|
| 390 |
|
| 391 |
+
def create_entity_table_gradio_tabs(entities, entity_colors):
|
| 392 |
+
"""Create Gradio tabs for entity results"""
|
| 393 |
if not entities:
|
| 394 |
+
return "No entities found."
|
| 395 |
|
| 396 |
# Share overlapping entities
|
| 397 |
shared_entities = find_overlapping_entities(entities)
|
|
|
|
| 409 |
entity_groups[key].append(entity)
|
| 410 |
|
| 411 |
if not entity_groups:
|
| 412 |
+
return "No entities found."
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 413 |
|
| 414 |
+
# Create content for each tab
|
| 415 |
+
tab_contents = {}
|
| 416 |
+
|
| 417 |
+
for entity_type, entities_of_type in entity_groups.items():
|
| 418 |
if entity_type == 'SHARED_ENTITIES':
|
| 419 |
+
colour = '#666666'
|
| 420 |
+
header = f"π§©π§© Shared Entities ({len(entities_of_type)} found)"
|
| 421 |
+
|
| 422 |
+
# Create table for shared entities
|
| 423 |
+
table_html = f"""
|
| 424 |
+
<div style="margin: 15px 0;">
|
| 425 |
+
<h4 style="color: {colour}; margin-bottom: 15px;">{header}</h4>
|
| 426 |
+
<table style="width: 100%; border-collapse: collapse; border: 1px solid #ddd;">
|
| 427 |
+
<thead>
|
| 428 |
+
<tr style="background-color: {colour}; color: white;">
|
| 429 |
+
<th style="padding: 12px; text-align: left; border: 1px solid #ddd;">Entity Text</th>
|
| 430 |
+
<th style="padding: 12px; text-align: left; border: 1px solid #ddd;">All Labels</th>
|
| 431 |
+
<th style="padding: 12px; text-align: left; border: 1px solid #ddd;">Sources</th>
|
| 432 |
+
<th style="padding: 12px; text-align: left; border: 1px solid #ddd;">Count</th>
|
| 433 |
+
</tr>
|
| 434 |
+
</thead>
|
| 435 |
+
<tbody>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 436 |
"""
|
| 437 |
+
|
| 438 |
for entity in entities_of_type:
|
| 439 |
labels_text = " | ".join(entity['labels'])
|
| 440 |
sources_text = " | ".join(entity['sources'])
|
| 441 |
+
|
| 442 |
+
table_html += f"""
|
| 443 |
<tr style="background-color: #fff;">
|
| 444 |
<td style="padding: 10px; border: 1px solid #ddd; font-weight: bold;">{entity['text']}</td>
|
| 445 |
<td style="padding: 10px; border: 1px solid #ddd;">{labels_text}</td>
|
|
|
|
| 451 |
</td>
|
| 452 |
</tr>
|
| 453 |
"""
|
| 454 |
+
|
| 455 |
+
table_html += "</tbody></table></div>"
|
| 456 |
+
tab_contents[f"π§©π§© SHARED ({len(entities_of_type)})"] = table_html
|
| 457 |
+
|
| 458 |
else:
|
| 459 |
+
colour = entity_colors.get(entity_type.upper(), '#f0f0f0')
|
| 460 |
+
# Determine if it's common or custom
|
| 461 |
+
is_standard = entity_type in STANDARD_ENTITIES
|
| 462 |
+
icon = "π―" if is_standard else "β¨"
|
| 463 |
+
source_text = "Common NER" if is_standard else "Custom GLiNER"
|
| 464 |
+
header = f"{icon} {source_text} - {entity_type} ({len(entities_of_type)} found)"
|
| 465 |
+
|
| 466 |
+
# Create table for this entity type
|
| 467 |
+
table_html = f"""
|
| 468 |
+
<div style="margin: 15px 0;">
|
| 469 |
+
<h4 style="color: {colour}; margin-bottom: 15px;">{header}</h4>
|
| 470 |
+
<table style="width: 100%; border-collapse: collapse; border: 1px solid #ddd;">
|
| 471 |
+
<thead>
|
| 472 |
+
<tr style="background-color: {colour}; color: white;">
|
| 473 |
+
<th style="padding: 12px; text-align: left; border: 1px solid #ddd;">Entity Text</th>
|
| 474 |
+
<th style="padding: 12px; text-align: left; border: 1px solid #ddd;">Confidence</th>
|
| 475 |
+
<th style="padding: 12px; text-align: left; border: 1px solid #ddd;">Type</th>
|
| 476 |
+
<th style="padding: 12px; text-align: left; border: 1px solid #ddd;">Source</th>
|
| 477 |
+
</tr>
|
| 478 |
+
</thead>
|
| 479 |
+
<tbody>
|
| 480 |
"""
|
| 481 |
+
|
| 482 |
# Sort by confidence score
|
| 483 |
entities_of_type.sort(key=lambda x: x.get('confidence', 0), reverse=True)
|
| 484 |
+
|
| 485 |
for entity in entities_of_type:
|
| 486 |
confidence = entity.get('confidence', 0.0)
|
| 487 |
+
confidence_colour = "#28a745" if confidence > 0.7 else "#ffc107" if confidence > 0.4 else "#dc3545"
|
| 488 |
source = entity.get('source', 'Unknown')
|
| 489 |
source_badge = f"<span style='background-color: #007bff; color: white; padding: 2px 6px; border-radius: 10px; font-size: 11px;'>{source}</span>"
|
| 490 |
+
|
| 491 |
+
table_html += f"""
|
| 492 |
<tr style="background-color: #fff;">
|
| 493 |
<td style="padding: 10px; border: 1px solid #ddd; font-weight: bold;">{entity['text']}</td>
|
| 494 |
<td style="padding: 10px; border: 1px solid #ddd;">
|
| 495 |
+
<span style="color: {confidence_colour}; font-weight: bold;">
|
| 496 |
{confidence:.3f}
|
| 497 |
</span>
|
| 498 |
</td>
|
|
|
|
| 500 |
<td style="padding: 10px; border: 1px solid #ddd;">{source_badge}</td>
|
| 501 |
</tr>
|
| 502 |
"""
|
| 503 |
+
|
| 504 |
+
table_html += "</tbody></table></div>"
|
| 505 |
+
tab_label = f"{icon} {entity_type} ({len(entities_of_type)})"
|
| 506 |
+
tab_contents[tab_label] = table_html
|
| 507 |
+
|
| 508 |
+
return tab_contents
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 509 |
|
| 510 |
def create_legend_html(entity_colors, standard_entities, custom_entities):
|
| 511 |
+
"""Create a legend showing entity colours"""
|
| 512 |
if not entity_colors:
|
| 513 |
return ""
|
| 514 |
|
|
|
|
| 517 |
|
| 518 |
if standard_entities:
|
| 519 |
html += "<div style='margin-bottom: 15px;'>"
|
| 520 |
+
html += "<h5 style='margin: 0 0 8px 0;'>π― Common Entities:</h5>"
|
| 521 |
html += "<div style='display: flex; flex-wrap: wrap; gap: 8px;'>"
|
| 522 |
for entity_type in standard_entities:
|
| 523 |
+
colour = entity_colors.get(entity_type.upper(), '#ccc')
|
| 524 |
+
html += f"<span style='background-color: {colour}; padding: 4px 8px; border-radius: 15px; color: white; font-weight: bold; font-size: 12px;'>{entity_type}</span>"
|
| 525 |
html += "</div></div>"
|
| 526 |
|
| 527 |
if custom_entities:
|
|
|
|
| 529 |
html += "<h5 style='margin: 0 0 8px 0;'>β¨ Custom Entities:</h5>"
|
| 530 |
html += "<div style='display: flex; flex-wrap: wrap; gap: 8px;'>"
|
| 531 |
for entity_type in custom_entities:
|
| 532 |
+
colour = entity_colors.get(entity_type.upper(), '#ccc')
|
| 533 |
+
html += f"<span style='background-color: {colour}; padding: 4px 8px; border-radius: 15px; color: white; font-weight: bold; font-size: 12px;'>{entity_type}</span>"
|
| 534 |
html += "</div></div>"
|
| 535 |
|
| 536 |
html += "</div>"
|
|
|
|
| 539 |
# Initialize the NER manager
|
| 540 |
ner_manager = HybridNERManager()
|
| 541 |
|
| 542 |
+
def process_text(text, standard_entities, custom_entities_str, confidence_threshold, selected_model, progress=gr.Progress()):
|
| 543 |
+
"""Main processing function for Gradio interface with progress tracking"""
|
| 544 |
if not text.strip():
|
| 545 |
+
return "β Please enter some text to analyse", "", {}
|
| 546 |
|
| 547 |
+
progress(0.1, desc="Initialising...")
|
| 548 |
+
|
| 549 |
# Parse custom entities
|
| 550 |
custom_entities = []
|
| 551 |
if custom_entities_str.strip():
|
| 552 |
custom_entities = [entity.strip() for entity in custom_entities_str.split(',') if entity.strip()]
|
| 553 |
|
| 554 |
+
# Parse common entities
|
| 555 |
selected_standard = [entity for entity in standard_entities if entity]
|
| 556 |
|
| 557 |
if not selected_standard and not custom_entities:
|
| 558 |
+
return "β Please select at least one common entity type OR enter custom entity types", "", {}
|
| 559 |
|
| 560 |
+
progress(0.2, desc="Loading models...")
|
| 561 |
+
|
| 562 |
all_entities = []
|
| 563 |
|
| 564 |
+
# Extract common entities using selected model
|
| 565 |
if selected_standard and selected_model:
|
| 566 |
+
progress(0.4, desc="Extracting common entities...")
|
| 567 |
standard_entities_results = ner_manager.extract_entities_by_model(text, selected_standard, selected_model, confidence_threshold)
|
| 568 |
all_entities.extend(standard_entities_results)
|
| 569 |
|
| 570 |
# Extract custom entities using GLiNER
|
| 571 |
if custom_entities:
|
| 572 |
+
progress(0.6, desc="Extracting custom entities...")
|
| 573 |
custom_entity_results = ner_manager.extract_gliner_entities(text, custom_entities, confidence_threshold, is_custom=True)
|
| 574 |
all_entities.extend(custom_entity_results)
|
| 575 |
|
| 576 |
if not all_entities:
|
| 577 |
+
return "β No entities found. Try lowering the confidence threshold or using different entity types.", "", {}
|
| 578 |
|
| 579 |
+
progress(0.8, desc="Processing results...")
|
| 580 |
+
|
| 581 |
+
# Assign colours
|
| 582 |
+
entity_colors = ner_manager.assign_colours(selected_standard, custom_entities)
|
| 583 |
|
| 584 |
# Create outputs
|
| 585 |
legend_html = create_legend_html(entity_colors, selected_standard, custom_entities)
|
| 586 |
highlighted_html = create_highlighted_html(text, all_entities, entity_colors)
|
| 587 |
+
tab_contents = create_entity_table_gradio_tabs(all_entities, entity_colors)
|
| 588 |
|
| 589 |
+
progress(0.9, desc="Creating summary...")
|
| 590 |
+
|
| 591 |
# Create summary with shared entities terminology
|
| 592 |
total_entities = len(all_entities)
|
| 593 |
shared_entities = find_overlapping_entities(all_entities)
|
|
|
|
| 602 |
- **Average confidence:** {sum(e.get('confidence', 0) for e in all_entities) / total_entities:.3f}
|
| 603 |
"""
|
| 604 |
|
| 605 |
+
progress(1.0, desc="Complete!")
|
| 606 |
+
|
| 607 |
+
return summary, legend_html + highlighted_html, tab_contents
|
| 608 |
|
| 609 |
# Create Gradio interface
|
| 610 |
def create_interface():
|
|
|
|
| 612 |
gr.Markdown("""
|
| 613 |
# π― Hybrid NER + Custom GLiNER Entity Recognition Tool
|
| 614 |
|
| 615 |
+
Combine common NER categories with your own custom entity types! This tool uses both traditional NER models and GLiNER for comprehensive entity extraction.
|
| 616 |
|
| 617 |
+
## π§©π§© NEW: Overlapping entities are automatically shared with split-colour highlighting!
|
| 618 |
|
| 619 |
### How to use:
|
| 620 |
1. **π Enter your text** in the text area below
|
| 621 |
+
2. **π― Select a model** from the dropdown for common entities
|
| 622 |
+
3. **βοΈ Select common entities** you want to find (PER, ORG, LOC, etc.)
|
| 623 |
4. **β¨ Add custom entities** (comma-separated) like "relationships, occupations, skills"
|
| 624 |
5. **βοΈ Adjust confidence threshold**
|
| 625 |
+
6. **π Click "Analyse Text"** to see results with tabbed output
|
| 626 |
""")
|
| 627 |
|
| 628 |
with gr.Row():
|
| 629 |
with gr.Column(scale=2):
|
| 630 |
text_input = gr.Textbox(
|
| 631 |
+
label="π Text to Analyse",
|
| 632 |
placeholder="Enter your text here...",
|
| 633 |
lines=6,
|
| 634 |
max_lines=10
|
|
|
|
| 645 |
|
| 646 |
with gr.Row():
|
| 647 |
with gr.Column():
|
| 648 |
+
gr.Markdown("### π― Common Entity Types")
|
| 649 |
|
| 650 |
# Model selector
|
| 651 |
model_dropdown = gr.Dropdown(
|
| 652 |
choices=ner_manager.model_names,
|
| 653 |
value=ner_manager.model_names[0],
|
| 654 |
+
label="Select Model for Common Entities",
|
| 655 |
+
info="Choose which model to use for common NER"
|
| 656 |
)
|
| 657 |
|
| 658 |
+
# Common entities with select all functionality
|
| 659 |
standard_entities = gr.CheckboxGroup(
|
| 660 |
choices=STANDARD_ENTITIES,
|
| 661 |
value=['PER', 'ORG', 'LOC', 'MISC'], # Default selection
|
| 662 |
+
label="Select Common Entities"
|
| 663 |
)
|
| 664 |
|
| 665 |
# Select/Deselect All button
|
|
|
|
| 696 |
- financial terms, business roles
|
| 697 |
""")
|
| 698 |
|
| 699 |
+
analyse_btn = gr.Button("π Analyse Text", variant="primary", size="lg")
|
| 700 |
|
| 701 |
# Output sections
|
| 702 |
with gr.Row():
|
|
|
|
| 705 |
with gr.Row():
|
| 706 |
highlighted_output = gr.HTML(label="Highlighted Text")
|
| 707 |
|
| 708 |
+
# Create dynamic tabs for results
|
| 709 |
+
results_tabs = gr.State({})
|
| 710 |
+
|
| 711 |
+
def update_tabs(tab_contents):
|
| 712 |
+
"""Update the results tabs based on the analysis"""
|
| 713 |
+
if not tab_contents or not isinstance(tab_contents, dict):
|
| 714 |
+
return {gr.HTML("No results to display"): gr.update(visible=True)}
|
| 715 |
+
|
| 716 |
+
# Create tabs dynamically
|
| 717 |
+
tab_components = {}
|
| 718 |
+
for tab_name, content in tab_contents.items():
|
| 719 |
+
tab_components[tab_name] = gr.HTML(content)
|
| 720 |
+
|
| 721 |
+
return tab_components
|
| 722 |
+
|
| 723 |
+
# Results section with tabs
|
| 724 |
with gr.Row():
|
| 725 |
+
with gr.Column():
|
| 726 |
+
gr.Markdown("### π Detailed Results")
|
| 727 |
+
|
| 728 |
+
# We'll update this section dynamically
|
| 729 |
+
results_container = gr.HTML(label="Results")
|
| 730 |
+
|
| 731 |
+
# Function to process and display results
|
| 732 |
+
def process_and_display(text, standard_entities, custom_entities, confidence_threshold, selected_model):
|
| 733 |
+
# Get results from main processing function
|
| 734 |
+
summary, highlighted, tab_contents = process_text(
|
| 735 |
+
text, standard_entities, custom_entities, confidence_threshold, selected_model
|
| 736 |
+
)
|
| 737 |
+
|
| 738 |
+
# Create tabs HTML manually since Gradio dynamic tabs are complex
|
| 739 |
+
if isinstance(tab_contents, dict) and tab_contents:
|
| 740 |
+
tabs_html = """
|
| 741 |
+
<div style="margin: 20px 0;">
|
| 742 |
+
<div style="border-bottom: 2px solid #ddd; margin-bottom: 20px;">
|
| 743 |
+
"""
|
| 744 |
+
|
| 745 |
+
# Create tab buttons
|
| 746 |
+
tab_names = list(tab_contents.keys())
|
| 747 |
+
for i, tab_name in enumerate(tab_names):
|
| 748 |
+
active_style = "background-color: #f8f9fa; border-bottom: 3px solid #4ECDC4;" if i == 0 else "background-color: #fff;"
|
| 749 |
+
tabs_html += f"""
|
| 750 |
+
<button onclick="showResultTab('{i}')" id="result-tab-{i}"
|
| 751 |
+
style="padding: 12px 24px; margin-right: 5px; border: 1px solid #ddd;
|
| 752 |
+
border-bottom: none; cursor: pointer; font-weight: bold; {active_style}">
|
| 753 |
+
{tab_name}
|
| 754 |
+
</button>
|
| 755 |
+
"""
|
| 756 |
+
|
| 757 |
+
tabs_html += "</div>"
|
| 758 |
+
|
| 759 |
+
# Create tab content
|
| 760 |
+
for i, (tab_name, content) in enumerate(tab_contents.items()):
|
| 761 |
+
display_style = "display: block;" if i == 0 else "display: none;"
|
| 762 |
+
tabs_html += f"""
|
| 763 |
+
<div id="result-content-{i}" style="{display_style}">
|
| 764 |
+
{content}
|
| 765 |
+
</div>
|
| 766 |
+
"""
|
| 767 |
+
|
| 768 |
+
# Add JavaScript for tab switching
|
| 769 |
+
tabs_html += """
|
| 770 |
+
<script>
|
| 771 |
+
function showResultTab(tabIndex) {
|
| 772 |
+
// Hide all content
|
| 773 |
+
var contents = document.querySelectorAll('[id^="result-content-"]');
|
| 774 |
+
contents.forEach(function(content) {
|
| 775 |
+
content.style.display = 'none';
|
| 776 |
+
});
|
| 777 |
+
|
| 778 |
+
// Reset all tab styles
|
| 779 |
+
var tabs = document.querySelectorAll('[id^="result-tab-"]');
|
| 780 |
+
tabs.forEach(function(tab) {
|
| 781 |
+
tab.style.backgroundColor = '#fff';
|
| 782 |
+
tab.style.borderBottom = 'none';
|
| 783 |
+
});
|
| 784 |
+
|
| 785 |
+
// Show selected content
|
| 786 |
+
document.getElementById('result-content-' + tabIndex).style.display = 'block';
|
| 787 |
+
|
| 788 |
+
// Highlight selected tab
|
| 789 |
+
var activeTab = document.getElementById('result-tab-' + tabIndex);
|
| 790 |
+
activeTab.style.backgroundColor = '#f8f9fa';
|
| 791 |
+
activeTab.style.borderBottom = '3px solid #4ECDC4';
|
| 792 |
+
}
|
| 793 |
+
</script>
|
| 794 |
+
</div>
|
| 795 |
+
"""
|
| 796 |
+
|
| 797 |
+
results_display = tabs_html
|
| 798 |
+
else:
|
| 799 |
+
results_display = str(tab_contents) if tab_contents else "No results to display"
|
| 800 |
+
|
| 801 |
+
return summary, highlighted, results_display
|
| 802 |
|
| 803 |
# Connect the button to the processing function
|
| 804 |
+
analyse_btn.click(
|
| 805 |
+
fn=process_and_display,
|
| 806 |
inputs=[
|
| 807 |
text_input,
|
| 808 |
standard_entities,
|
|
|
|
| 810 |
confidence_threshold,
|
| 811 |
model_dropdown
|
| 812 |
],
|
| 813 |
+
outputs=[summary_output, highlighted_output, results_container]
|
| 814 |
)
|
| 815 |
|
| 816 |
# Add examples
|
| 817 |
gr.Examples(
|
| 818 |
examples=[
|
| 819 |
[
|
| 820 |
+
"John Smith works at Google in New York. He graduated from Stanford University in 2015 and specialises in artificial intelligence research. His wife Sarah is a doctor at Mount Sinai Hospital.",
|
| 821 |
["PER", "ORG", "LOC", "DATE"],
|
| 822 |
"relationships, occupations, educational background",
|
| 823 |
0.3,
|
| 824 |
+
"entities_spacy_en_core_web_trf"
|
| 825 |
],
|
| 826 |
[
|
| 827 |
"The meeting between CEO Jane Doe and the board of directors at Microsoft headquarters in Seattle discussed the Q4 financial results and the new AI strategy for 2024.",
|
| 828 |
["PER", "ORG", "LOC", "DATE"],
|
| 829 |
"corporate roles, business events, financial terms",
|
| 830 |
0.4,
|
| 831 |
+
"entities_flair_ner-ontonotes-large"
|
| 832 |
],
|
| 833 |
[
|
| 834 |
"Dr. Emily Watson published a research paper on machine learning algorithms at MIT. She collaborates with her colleague Prof. David Chen on natural language processing projects.",
|
| 835 |
+
["PER", "ORG", "Work of Art"],
|
| 836 |
"academic titles, research topics, collaborations",
|
| 837 |
0.3,
|
| 838 |
+
"entities_gliner_knowledgator/modern-gliner-bi-large-v1.0"
|
| 839 |
]
|
| 840 |
],
|
| 841 |
inputs=[
|