Spaces:
Sleeping
Sleeping
File size: 52,823 Bytes
66b9109 102e87a 66b9109 5b0f561 66b9109 102e87a 66b9109 102e87a a8443d7 70de5e9 a8443d7 102e87a ee7ef03 102e87a 9b1ad5f 102e87a a8443d7 102e87a a8443d7 102e87a a8443d7 ac8959c 102e87a a8443d7 102e87a a8443d7 102e87a a8443d7 102e87a a8443d7 102e87a 66b9109 a8443d7 102e87a 66b9109 102e87a 66b9109 102e87a 66b9109 102e87a 66b9109 102e87a 66b9109 102e87a 66b9109 102e87a 66b9109 102e87a 66b9109 102e87a 66b9109 102e87a 66b9109 102e87a 66b9109 2b4639d 102e87a 66b9109 2b4639d 5b0f561 2e2ec5c 2b4639d 5b0f561 2e2ec5c 07164fa 95fbfc5 5b0f561 2e2ec5c 95fbfc5 2b4639d 5b0f561 2b4639d 102e87a 66b9109 102e87a 66b9109 2b4639d 102e87a 66b9109 2b4639d 102e87a 66b9109 102e87a 66b9109 102e87a 2b4639d 102e87a 2b4639d 102e87a 2b4639d 102e87a 2b4639d 102e87a 66b9109 102e87a 2b4639d 102e87a 2b4639d 102e87a 2b4639d 102e87a 2b4639d 102e87a 2b4639d 102e87a 2b4639d 102e87a 2eed23e 66b9109 b17ff40 66b9109 b17ff40 66b9109 102e87a 66b9109 102e87a 66b9109 102e87a 66b9109 102e87a 66b9109 102e87a 66b9109 102e87a 66b9109 102e87a 66b9109 102e87a 66b9109 102e87a 66b9109 102e87a 66b9109 102e87a 66b9109 102e87a 66b9109 102e87a 66b9109 102e87a 99cc145 66b9109 99cc145 66b9109 99cc145 66b9109 | 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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 | """
Enhanced data processing pipeline for advanced sentiment analysis application.
Handles translation, ABSA, intent classification, priority scoring, and co-occurrence analysis.
"""
import pandas as pd
import numpy as np
from typing import Dict, List, Tuple, Any, Optional
import re
from datetime import datetime, timedelta
import logging
from langdetect import detect
import streamlit as st
from collections import Counter, defaultdict
from itertools import combinations
import networkx as nx
import requests
import os
import time
# Suppress GitPython warnings for PyABSA
os.environ['GIT_PYTHON_REFRESH'] = 'quiet'
# Set up logging
logging.basicConfig(level=logging.INFO)
logger = logging.getLogger(__name__)
class DataValidator:
"""Validates uploaded CSV data format and content."""
REQUIRED_COLUMNS = ['id', 'reviews_title', 'review', 'date', 'user_id']
@staticmethod
def validate_csv(df: pd.DataFrame) -> Tuple[bool, List[str]]:
"""
Validate CSV format and content.
Args:
df: Uploaded dataframe
Returns:
Tuple of (is_valid, error_messages)
"""
errors = []
# Check required columns
missing_cols = set(DataValidator.REQUIRED_COLUMNS) - set(df.columns)
if missing_cols:
errors.append(f"Missing required columns: {missing_cols}")
if not errors:
# Check for empty values
if df['review'].isnull().any() or (df['review'] == '').any():
errors.append("Found empty review entries")
# Validate date format
try:
pd.to_datetime(df['date'], errors='coerce')
if df['date'].isnull().any():
errors.append("Invalid date format detected")
except Exception as e:
errors.append(f"Date validation error: {str(e)}")
return len(errors) == 0, errors
@staticmethod
def clean_data(df: pd.DataFrame) -> pd.DataFrame:
"""Clean and preprocess the data."""
df = df.copy()
# Convert date column
df['date'] = pd.to_datetime(df['date'], errors='coerce')
# Clean text data
df['review'] = df['review'].astype(str).str.strip()
df['reviews_title'] = df['reviews_title'].astype(str).str.strip()
# Remove rows with null reviews
df = df.dropna(subset=['review'])
# Remove duplicate reviews
df = df.drop_duplicates(subset=['review'], keep='first')
return df.reset_index(drop=True)
class TranslationService:
"""Handles translation using HF Inference API - much more reliable than local models."""
def __init__(self):
self.api_token = self._get_hf_token()
# Using AI4Bharat IndicTrans2 - specialized for Indian languages
self.translation_model = "ai4bharat/indictrans2-en-indic-1.3B"
self.base_url = "https://router.huggingface.co"
logger.info("Initialized HF Inference API for translation (IndicTrans2)")
def _get_hf_token(self) -> Optional[str]:
"""Get HF token from environment or Streamlit secrets."""
try:
return st.secrets["HF_TOKEN"]
except:
pass
token = os.getenv("HF_TOKEN")
if not token:
logger.warning("No HF_TOKEN found. Translation will be limited.")
else:
# Log first 10 chars to verify token is loaded (without exposing full token)
logger.info(f"HF_TOKEN loaded successfully (starts with: {token[:10]}...)")
return token
def _call_hf_translation_api(self, text: str, source_lang: str = "hi", target_lang: str = "en") -> str:
"""Call HF Translation API with fallback."""
if not self.api_token:
logger.debug("No API token, skipping translation")
return text
try:
headers = {"Authorization": f"Bearer {self.api_token}"}
# Try using serverless inference endpoint
url = f"https://api-inference.huggingface.co/models/{self.translation_model}"
# IndicTrans2 requires simple input format
payload = {"inputs": text}
response = requests.post(url, headers=headers, json=payload, timeout=10)
if response.status_code == 200:
result = response.json()
# IndicTrans2 returns: [{"translation_text": "..."}] or {"generated_text": "..."}
if isinstance(result, list) and len(result) > 0:
translated = result[0].get("translation_text", "") or result[0].get("generated_text", "")
if translated:
return translated
elif isinstance(result, dict):
translated = result.get("generated_text", "") or result.get("translation_text", "")
if translated:
return translated
# Silently fallback to original text (translation is optional)
logger.debug(f"Translation unavailable, using original text (status: {response.status_code})")
return text
except Exception as e:
logger.debug(f"Translation skipped: {str(e)}")
return text
def detect_language(self, text: str) -> str:
"""Detect language of the text."""
try:
lang = detect(text)
return lang
except:
return 'unknown'
def translate_to_english(self, text: str, source_lang: str = 'hi') -> str:
"""
Translate text to English using HF API.
Args:
text: Text to translate
source_lang: Source language code
Returns:
Translated text
"""
if source_lang == 'en' or source_lang == 'unknown':
return text
return self._call_hf_translation_api(text, source_lang, "en")
def process_reviews(self, reviews: List[str]) -> Tuple[List[str], List[str]]:
"""
Process list of reviews for translation.
Args:
reviews: List of review texts
Returns:
Tuple of (translated_reviews, detected_languages)
"""
translated_reviews = []
detected_languages = []
for i, review in enumerate(reviews):
if i % 20 == 0: # Progress logging
logger.info(f"Processing translation {i+1}/{len(reviews)}")
lang = self.detect_language(review)
detected_languages.append(lang)
if lang == 'hi': # Hindi detected
translated = self.translate_to_english(review, 'hi')
translated_reviews.append(translated)
else:
translated_reviews.append(review) # Keep original if not Hindi
return translated_reviews, detected_languages
class ABSAProcessor:
"""Enhanced ABSA using PyABSA for accurate aspect extraction and sentiment analysis."""
def __init__(self):
self.model = None
self.task_manager = None
self._load_pyabsa_model()
logger.info("Initialized PyABSA for ABSA processing")
def _load_pyabsa_model(self):
"""Load PyABSA multilingual model with caching."""
try:
# Suppress additional git warnings
import warnings
warnings.filterwarnings('ignore', category=DeprecationWarning)
# PyABSA v2.4.2 correct API - use AspectTermExtraction module
from pyabsa import AspectTermExtraction as ATEPC
# Check if git is available
try:
import subprocess
git_check = subprocess.run(['git', '--version'], capture_output=True, timeout=5)
if git_check.returncode == 0:
logger.info(f"Git available: {git_check.stdout.decode().strip()}")
else:
logger.warning("Git command failed, PyABSA may have issues downloading checkpoints")
except Exception as git_err:
logger.warning(f"Git not found or not executable: {str(git_err)}")
# Load aspect extractor using correct PyABSA v2.4.2 API
logger.info("Loading PyABSA ATEPC multilingual model...")
logger.info("This may take a few minutes on first run (downloading checkpoint)...")
# Use AspectExtractor class - the correct way in v2.4.2
self.model = ATEPC.AspectExtractor('multilingual')
logger.info("✅ PyABSA model loaded successfully")
except Exception as e:
import traceback
logger.error(f"❌ Failed to load PyABSA model: {str(e)}")
logger.error(f"Traceback: {traceback.format_exc()}")
logger.warning("⚠️ Using fallback method for aspect extraction")
self.model = None
def set_task_manager(self, task_manager):
"""Set task manager for cancellation support."""
self.task_manager = task_manager
def _get_hf_token(self) -> Optional[str]:
"""Get HF token from environment or Streamlit secrets."""
# Try Streamlit secrets first
try:
return st.secrets["HF_TOKEN"]
except:
pass
# Try environment variable
token = os.getenv("HF_TOKEN")
if not token:
logger.warning("No HF_TOKEN found. Some features may be limited.")
return token
def _call_hf_api(self, model_name: str, inputs: str, max_retries: int = 3) -> Dict:
"""Call HF Inference API with retry logic."""
headers = {}
if self.api_token:
headers["Authorization"] = f"Bearer {self.api_token}"
url = f"{self.base_url}/{model_name}"
payload = {"inputs": inputs}
for attempt in range(max_retries):
try:
response = requests.post(url, headers=headers, json=payload, timeout=30)
if response.status_code == 503:
# Model is loading, wait and retry
wait_time = 2 ** attempt # Exponential backoff
logger.info(f"Model loading, waiting {wait_time}s before retry...")
time.sleep(wait_time)
continue
response.raise_for_status()
return response.json()
except requests.exceptions.RequestException as e:
logger.error(f"API call failed (attempt {attempt + 1}): {str(e)}")
if attempt == max_retries - 1:
return {"error": str(e)}
time.sleep(1)
return {"error": "Max retries exceeded"}
def extract_aspects_and_sentiments(self, reviews: List[str], task_id: Optional[str] = None) -> List[Dict[str, Any]]:
"""
Extract aspects and sentiments using PyABSA with fallback and cancellation support.
Args:
reviews: List of review texts (preferably in English after translation)
task_id: Optional task ID for cancellation tracking
Returns:
List of dictionaries containing extracted information or cancellation dict
"""
import gc
logger.info(f"Processing {len(reviews)} reviews with PyABSA")
processed_results = []
batch_size = 5 # Process 5 reviews at a time for responsive cancellation
for batch_start in range(0, len(reviews), batch_size):
# Check cancellation before each batch
if task_id and self.task_manager and self.task_manager.is_cancelled(task_id):
logger.info(f"ABSA processing cancelled at review {batch_start}/{len(reviews)}")
del processed_results
gc.collect()
return {'status': 'cancelled', 'message': 'ABSA processing cancelled by user'}
batch_end = min(batch_start + batch_size, len(reviews))
batch_reviews = reviews[batch_start:batch_end]
# Process batch
for i, review in enumerate(batch_reviews):
review_idx = batch_start + i
if review_idx % 10 == 0: # Progress logging
logger.info(f"Processing review {review_idx+1}/{len(reviews)}")
# Try PyABSA first, fallback to rule-based if unavailable
if self.model is not None:
try:
result = self._extract_with_pyabsa(review)
except Exception as e:
logger.warning(f"PyABSA failed for review {review_idx}: {str(e)}, using fallback")
result = self._extract_with_fallback(review)
else:
result = self._extract_with_fallback(review)
processed_results.append(result)
# Update progress after each batch (50-90% range)
if task_id and self.task_manager:
progress = 50 + int((batch_end / len(reviews)) * 40)
self.task_manager.update_task(task_id, progress=progress)
logger.info(f"Successfully processed {len(processed_results)} reviews")
return processed_results
def _extract_with_pyabsa(self, review: str) -> Dict[str, Any]:
"""Extract aspects and sentiments using PyABSA model."""
result = self.model.predict(review, print_result=False, save_result=False)
# PyABSA returns: aspect, sentiment, confidence, position
aspects = result.get('aspect', [])
sentiments = result.get('sentiment', [])
positions = result.get('position', [])
confidence_scores = result.get('confidence', [])
# Handle single aspect case
if not isinstance(aspects, list):
aspects = [aspects] if aspects else []
sentiments = [sentiments] if sentiments else []
positions = [positions] if positions else []
confidence_scores = [confidence_scores] if confidence_scores else []
# If no aspects found, use fallback
if not aspects:
return self._extract_with_fallback(review)
return {
'sentence': review,
'aspects': aspects,
'sentiments': sentiments,
'positions': positions,
'confidence_scores': confidence_scores,
'tokens': review.split(),
'iob_tags': ['O'] * len(review.split())
}
def _extract_with_fallback(self, review: str) -> Dict[str, Any]:
"""Fallback rule-based extraction when PyABSA is unavailable."""
sentiment = self._get_rule_based_sentiment(review)
aspects = self._extract_simple_aspects(review)
return {
'sentence': review,
'aspects': aspects,
'sentiments': [sentiment] * len(aspects),
'positions': [[0, len(review)]] * len(aspects),
'confidence_scores': [0.7] * len(aspects), # Lower confidence for rule-based
'tokens': review.split(),
'iob_tags': ['O'] * len(review.split())
}
def _get_hf_sentiment(self, text: str) -> str:
"""Get sentiment from HF Inference API with fallback."""
if not self.api_token:
# Fallback to rule-based if no API token
return self._get_rule_based_sentiment(text)
try:
result = self._call_hf_api(self.sentiment_model, text)
if "error" in result:
logger.warning(f"API error, using rule-based fallback: {result['error']}")
return self._get_rule_based_sentiment(text)
# Parse HF sentiment result
if isinstance(result, list) and len(result) > 0:
predictions = result[0]
if isinstance(predictions, list) and len(predictions) > 0:
top_prediction = max(predictions, key=lambda x: x.get('score', 0))
label = top_prediction.get('label', 'NEUTRAL').upper()
# Map HF labels to our format
if 'POSITIVE' in label or 'POS' in label:
return 'Positive'
elif 'NEGATIVE' in label or 'NEG' in label:
return 'Negative'
else:
return 'Neutral'
# Fallback if parsing fails
return self._get_rule_based_sentiment(text)
except Exception as e:
logger.error(f"HF API error: {str(e)}, using rule-based fallback")
return self._get_rule_based_sentiment(text)
def _get_rule_based_sentiment(self, review: str) -> str:
"""Fallback rule-based sentiment analysis with enhanced negative detection."""
review_lower = review.lower()
# Enhanced sentiment words
positive_words = ['good', 'great', 'excellent', 'amazing', 'love', 'best', 'awesome',
'fantastic', 'wonderful', 'perfect', 'satisfied', 'happy', 'pleased',
'outstanding', 'brilliant', 'superb', 'delighted', 'impressed', 'working',
'अच्छा', 'बढ़िया', 'शानदार', 'बेहतरीन']
negative_words = ['bad', 'terrible', 'awful', 'hate', 'worst', 'horrible', 'poor',
'disappointing', 'frustrated', 'angry', 'broken', 'failed', 'useless',
'pathetic', 'disgusting', 'annoying', 'waste', 'regret', 'problem', 'issue',
'not working', 'doesn\'t work', 'never', 'delayed', 'late', 'slow', 'error',
'खराब', 'समस्या', 'देर', 'नहीं', 'बुरा']
# Strong negative phrases (count as 2 points)
negative_phrases = ['too late', 'never comes', 'not received', 'doesn\'t arrive',
'delayed', 'not working', 'बहुत देर', 'नहीं आता']
pos_count = sum(1 for word in positive_words if word in review_lower)
neg_count = sum(1 for word in negative_words if word in review_lower)
# Check for negative phrases (stronger signal)
for phrase in negative_phrases:
if phrase in review_lower:
neg_count += 2
if pos_count > neg_count:
return 'Positive'
elif neg_count > pos_count:
return 'Negative'
else:
return 'Neutral'
def _extract_simple_aspects(self, review: str) -> List[str]:
"""Extract aspects using enhanced keyword matching."""
review_lower = review.lower()
aspects = []
# Enhanced aspect keywords with Hindi/English variants
aspect_keywords = {
'OTP/Verification': ['otp', 'atp', 'verification', 'verify', 'code', 'pin', 'authentication', 'ओटीपी', 'कोड', 'सत्यापन'],
'Login/Account': ['login', 'sign in', 'signin', 'account', 'password', 'username', 'register', 'signup', 'लॉगिन', 'खाता'],
'App Performance': ['app', 'application', 'crash', 'freeze', 'hang', 'loading', 'lag', 'slow', 'एप', 'एप्लिकेशन'],
'Payment': ['payment', 'pay', 'transaction', 'refund', 'money', 'bank', 'upi', 'wallet', 'भुगतान', 'पैसा'],
'Quality': ['quality', 'build', 'material', 'construction', 'durability', 'solid', 'sturdy', 'cheap', 'flimsy', 'गुणवत्ता'],
'Price': ['price', 'cost', 'expensive', 'cheap', 'value', 'money', 'affordable', 'budget', 'worth', 'कीमत', 'दाम'],
'Service': ['service', 'support', 'help', 'staff', 'customer', 'response', 'team', 'representative', 'सेवा', 'सहायता'],
'Delivery': ['delivery', 'shipping', 'fast', 'quick', 'slow', 'delayed', 'arrive', 'package', 'डिलीवरी', 'शिपिंग'],
'Design': ['design', 'look', 'appearance', 'beautiful', 'ugly', 'style', 'color', 'aesthetic', 'डिज़ाइन', 'रूप'],
'Performance': ['performance', 'speed', 'fast', 'slow', 'efficiency', 'works', 'function', 'smooth', 'प्रदर्शन'],
'Usability': ['easy', 'difficult', 'user', 'interface', 'intuitive', 'complex', 'simple', 'confusing', 'उपयोग'],
'Features': ['feature', 'function', 'capability', 'option', 'setting', 'mode', 'tool', 'फीचर', 'सुविधा'],
'Size': ['size', 'big', 'small', 'large', 'compact', 'tiny', 'huge', 'dimension', 'आकार'],
'Battery': ['battery', 'charge', 'power', 'energy', 'last', 'drain', 'life', 'बैटरी', 'चार्ज']
}
for aspect, keywords in aspect_keywords.items():
if any(keyword in review_lower for keyword in keywords):
aspects.append(aspect)
# Default aspect if none found
if not aspects:
aspects = ['General']
return aspects
class IntentClassifier:
"""Enhanced intent classifier with severity scoring for complaints and type classification."""
INTENT_KEYWORDS = {
'complaint': {
'high_severity': ['terrible', 'worst', 'horrible', 'awful', 'hate', 'useless', 'trash', 'garbage', 'waste', 'pathetic'],
'medium_severity': ['bad', 'disappointed', 'frustrated', 'annoying', 'poor', 'defective', 'broken', 'failed'],
'low_severity': ['problem', 'issue', 'concern', 'slow', 'okay but', 'could be better']
},
'praise': {
'high_positive': ['excellent', 'amazing', 'fantastic', 'wonderful', 'perfect', 'outstanding', 'brilliant', 'superb'],
'medium_positive': ['good', 'great', 'love', 'best', 'awesome', 'nice', 'satisfied', 'happy'],
'low_positive': ['fine', 'decent', 'acceptable', 'adequate', 'reasonable']
},
'question': ['how', 'what', 'when', 'where', 'why', 'which', 'who', '?', 'can you', 'could you', 'is it possible'],
'suggestion': ['should', 'could', 'would', 'recommend', 'suggest', 'improve', 'better', 'enhancement', 'feature request'],
'comparison': ['better than', 'worse than', 'compared to', 'versus', 'vs', 'similar to', 'different from'],
'neutral': ['okay', 'fine', 'average', 'normal', 'standard', 'typical', 'nothing special']
}
@classmethod
def classify_intent_with_severity(cls, review: str) -> Tuple[str, str, float]:
"""
Classify intent with severity/type scoring.
Args:
review: Review text
Returns:
Tuple of (intent, severity/type, confidence_score)
"""
review_lower = review.lower()
# Check for complaints with severity
complaint_scores = {}
for severity, keywords in cls.INTENT_KEYWORDS['complaint'].items():
score = sum(1 for keyword in keywords if keyword in review_lower)
if score > 0:
complaint_scores[severity] = score
if complaint_scores:
severity = max(complaint_scores, key=complaint_scores.get)
confidence = min(complaint_scores[severity] / 3.0, 1.0) # Normalize confidence
return 'complaint', severity, confidence
# Check for praise with positivity level
praise_scores = {}
for positivity, keywords in cls.INTENT_KEYWORDS['praise'].items():
score = sum(1 for keyword in keywords if keyword in review_lower)
if score > 0:
praise_scores[positivity] = score
if praise_scores:
positivity = max(praise_scores, key=praise_scores.get)
confidence = min(praise_scores[positivity] / 3.0, 1.0)
return 'praise', positivity, confidence
# Check other intents
for intent, keywords in cls.INTENT_KEYWORDS.items():
if intent not in ['complaint', 'praise']:
score = sum(1 for keyword in keywords if keyword in review_lower)
if score > 0:
confidence = min(score / 2.0, 1.0)
return intent, 'standard', confidence
return 'neutral', 'standard', 0.1
@classmethod
def classify_batch_enhanced(cls, reviews: List[str]) -> List[Dict[str, Any]]:
"""Classify intents with enhanced information for a batch of reviews."""
results = []
for review in reviews:
intent, severity_type, confidence = cls.classify_intent_with_severity(review)
results.append({
'intent': intent,
'severity_type': severity_type,
'confidence': confidence
})
return results
class AspectAnalytics:
"""Advanced analytics for aspect analysis including priority scoring and co-occurrence."""
@staticmethod
def calculate_aspect_scores(df: pd.DataFrame) -> Tuple[pd.DataFrame, pd.DataFrame]:
"""
Calculate priority scores for negative aspects and strength scores for positive aspects.
Args:
df: Processed dataframe with aspects and sentiments
Returns:
Tuple of (areas_of_improvement_df, strength_anchors_df)
"""
aspect_data = []
# Extract aspect-sentiment pairs with intent information
for idx, row in df.iterrows():
aspects = row['aspects'] if isinstance(row['aspects'], list) else []
sentiments = row['aspect_sentiments'] if isinstance(row['aspect_sentiments'], list) else []
intent = row['intent']
intent_severity = row.get('intent_severity', 'standard')
date = row['date']
for aspect, sentiment in zip(aspects, sentiments):
aspect_data.append({
'aspect': aspect,
'sentiment': sentiment,
'intent': intent,
'intent_severity': intent_severity,
'date': date,
'review_id': row['id']
})
if not aspect_data:
empty_df = pd.DataFrame(columns=['aspect', 'score', 'frequency', 'sentiment_ratio'])
return empty_df, empty_df
aspect_df = pd.DataFrame(aspect_data)
# Calculate metrics for each aspect
aspect_metrics = {}
for aspect in aspect_df['aspect'].unique():
aspect_subset = aspect_df[aspect_df['aspect'] == aspect]
total_count = len(aspect_subset)
positive_count = len(aspect_subset[aspect_subset['sentiment'] == 'Positive'])
negative_count = len(aspect_subset[aspect_subset['sentiment'] == 'Negative'])
# Calculate sentiment ratios
positivity_ratio = positive_count / total_count if total_count > 0 else 0
negativity_ratio = negative_count / total_count if total_count > 0 else 0
# Calculate intent severity weighting
complaint_subset = aspect_subset[aspect_subset['intent'] == 'complaint']
severity_weight = 0
if len(complaint_subset) > 0:
severity_mapping = {'high_severity': 3, 'medium_severity': 2, 'low_severity': 1, 'standard': 1}
severity_weight = complaint_subset['intent_severity'].map(severity_mapping).mean()
# Calculate priority score for negative aspects
priority_score = negativity_ratio * total_count * (1 + severity_weight)
# Calculate strength score for positive aspects
strength_score = positivity_ratio * total_count * (1 + (positivity_ratio * 2))
aspect_metrics[aspect] = {
'frequency': total_count,
'positivity_ratio': positivity_ratio,
'negativity_ratio': negativity_ratio,
'priority_score': priority_score,
'strength_score': strength_score,
'intent_severity': severity_weight
}
# Create Areas of Improvement DataFrame (top negative aspects)
improvement_data = []
for aspect, metrics in aspect_metrics.items():
if metrics['negativity_ratio'] > 0.1: # Only include aspects with >10% negativity
improvement_data.append({
'aspect': aspect,
'negativity_pct': round(metrics['negativity_ratio'] * 100, 1),
'intent_severity': round(metrics['intent_severity'], 2),
'frequency': metrics['frequency'],
'priority_score': round(metrics['priority_score'], 2)
})
# Handle empty DataFrame case
if improvement_data:
areas_of_improvement = pd.DataFrame(improvement_data).sort_values('priority_score', ascending=False)
else:
# Return empty DataFrame with correct columns
areas_of_improvement = pd.DataFrame(columns=['aspect', 'negativity_pct', 'intent_severity', 'frequency', 'priority_score'])
# Create Strength Anchors DataFrame (top positive aspects)
strength_data = []
for aspect, metrics in aspect_metrics.items():
if metrics['positivity_ratio'] > 0.3: # Only include aspects with >30% positivity
strength_data.append({
'aspect': aspect,
'positivity_pct': round(metrics['positivity_ratio'] * 100, 1),
'intent_type': 'praise', # Simplified for now
'frequency': metrics['frequency'],
'strength_score': round(metrics['strength_score'], 2)
})
# Handle empty DataFrame case
if strength_data:
strength_anchors = pd.DataFrame(strength_data).sort_values('strength_score', ascending=False)
else:
# Return empty DataFrame with correct columns
strength_anchors = pd.DataFrame(columns=['aspect', 'positivity_pct', 'intent_type', 'frequency', 'strength_score'])
return areas_of_improvement, strength_anchors
@staticmethod
def calculate_aspect_cooccurrence(df: pd.DataFrame) -> nx.Graph:
"""
Calculate aspect co-occurrence for network analysis.
Args:
df: Processed dataframe with aspects
Returns:
NetworkX graph with aspect co-occurrence data
"""
G = nx.Graph()
# Calculate co-occurrence matrix
cooccurrence_counts = defaultdict(int)
aspect_sentiments = defaultdict(list)
aspect_frequencies = defaultdict(int)
for idx, row in df.iterrows():
aspects = row['aspects'] if isinstance(row['aspects'], list) else []
sentiments = row['aspect_sentiments'] if isinstance(row['aspect_sentiments'], list) else []
# Count individual aspects
for aspect, sentiment in zip(aspects, sentiments):
aspect_frequencies[aspect] += 1
aspect_sentiments[aspect].append(sentiment)
# Count co-occurrences
for aspect1, aspect2 in combinations(aspects, 2):
pair = tuple(sorted([aspect1, aspect2]))
cooccurrence_counts[pair] += 1
# Add nodes with attributes
for aspect, frequency in aspect_frequencies.items():
sentiments = aspect_sentiments[aspect]
positive_pct = sentiments.count('Positive') / len(sentiments) if sentiments else 0
negative_pct = sentiments.count('Negative') / len(sentiments) if sentiments else 0
# Determine overall sentiment color
if positive_pct > negative_pct:
color = 'green'
sentiment_score = positive_pct
elif negative_pct > positive_pct:
color = 'red'
sentiment_score = -negative_pct
else:
color = 'gray'
sentiment_score = 0
G.add_node(aspect,
frequency=frequency,
sentiment_score=sentiment_score,
color=color,
positive_pct=positive_pct,
negative_pct=negative_pct)
# Add edges with weights
for (aspect1, aspect2), count in cooccurrence_counts.items():
if count >= 2: # Only include co-occurrences that happen at least twice
G.add_edge(aspect1, aspect2, weight=count)
return G
@staticmethod
def detect_sentiment_spikes(df: pd.DataFrame, window_days: int = 7) -> List[Dict[str, Any]]:
"""
Detect week-over-week spikes in negative sentiment for aspects.
Args:
df: Processed dataframe with date and aspect information
window_days: Number of days for the rolling window
Returns:
List of alerts for aspects with significant negative spikes
"""
alerts = []
if len(df) < 2:
return alerts
# Extract aspect-sentiment data with dates
aspect_data = []
for idx, row in df.iterrows():
aspects = row['aspects'] if isinstance(row['aspects'], list) else []
sentiments = row['aspect_sentiments'] if isinstance(row['aspect_sentiments'], list) else []
date = row['date']
for aspect, sentiment in zip(aspects, sentiments):
aspect_data.append({
'aspect': aspect,
'sentiment': sentiment,
'date': date
})
if not aspect_data:
return alerts
aspect_df = pd.DataFrame(aspect_data)
aspect_df['date'] = pd.to_datetime(aspect_df['date'])
# Group by aspect and analyze trends
for aspect in aspect_df['aspect'].unique():
aspect_subset = aspect_df[aspect_df['aspect'] == aspect]
if len(aspect_subset) < 4: # Need minimum data points
continue
# Create daily negative sentiment counts
daily_negative = aspect_subset[aspect_subset['sentiment'] == 'Negative'].groupby(
aspect_subset['date'].dt.date
).size().reindex(
pd.date_range(aspect_subset['date'].min().date(),
aspect_subset['date'].max().date()).date,
fill_value=0
)
if len(daily_negative) >= window_days * 2:
# Calculate rolling averages
recent_avg = daily_negative.tail(window_days).mean()
previous_avg = daily_negative.iloc[-(window_days*2):-window_days].mean()
# Check for spike (>50% increase and at least 2 more complaints)
if recent_avg > previous_avg * 1.5 and (recent_avg - previous_avg) >= 2:
spike_magnitude = ((recent_avg - previous_avg) / previous_avg * 100) if previous_avg > 0 else 100
alerts.append({
'aspect': aspect,
'spike_magnitude': round(spike_magnitude, 1),
'recent_avg_negative': round(recent_avg, 1),
'previous_avg_negative': round(previous_avg, 1),
'alert_severity': 'high' if spike_magnitude > 100 else 'medium'
})
return sorted(alerts, key=lambda x: x['spike_magnitude'], reverse=True)
class SummaryGenerator:
"""Generates macro and micro-level summaries for aspects and overall sentiment."""
@staticmethod
def generate_macro_summary(df: pd.DataFrame, areas_of_improvement: pd.DataFrame,
strength_anchors: pd.DataFrame) -> Dict[str, str]:
"""
Generate high-level summary of sentiment analysis.
Args:
df: Processed dataframe
areas_of_improvement: Problem areas dataframe
strength_anchors: Strength areas dataframe
Returns:
Dictionary with macro-level insights
"""
total_reviews = len(df)
positive_pct = (df['overall_sentiment'] == 'Positive').mean() * 100
negative_pct = (df['overall_sentiment'] == 'Negative').mean() * 100
# Top issues and strengths
top_issue = areas_of_improvement.iloc[0]['aspect'] if len(areas_of_improvement) > 0 else "None identified"
top_strength = strength_anchors.iloc[0]['aspect'] if len(strength_anchors) > 0 else "None identified"
# Intent distribution
complaint_pct = (df['intent'] == 'complaint').mean() * 100
summary = {
'overall_sentiment': f"Out of {total_reviews} reviews, {positive_pct:.1f}% are positive and {negative_pct:.1f}% are negative.",
'top_issues': f"Primary concern: '{top_issue}' - requires immediate attention based on complaint frequency and severity.",
'top_strengths': f"Greatest strength: '{top_strength}' - leverage this positive aspect in marketing and product positioning.",
'intent_insights': f"{complaint_pct:.1f}% of reviews contain complaints, indicating specific areas for product improvement.",
'recommendation': "Focus on addressing top negative aspects while maintaining and promoting strengths."
}
return summary
@staticmethod
def generate_aspect_micro_summaries(df: pd.DataFrame, top_aspects: List[str],
max_aspects: int = 5) -> Dict[str, str]:
"""
Generate detailed summaries for specific aspects.
Args:
df: Processed dataframe
top_aspects: List of aspects to analyze
max_aspects: Maximum number of aspects to summarize
Returns:
Dictionary with aspect-specific insights
"""
summaries = {}
for aspect in top_aspects[:max_aspects]:
# Filter reviews mentioning this aspect
aspect_reviews = []
aspect_sentiments = []
for idx, row in df.iterrows():
aspects = row['aspects'] if isinstance(row['aspects'], list) else []
sentiments = row['aspect_sentiments'] if isinstance(row['aspect_sentiments'], list) else []
if aspect in aspects:
aspect_idx = aspects.index(aspect)
if aspect_idx < len(sentiments):
aspect_reviews.append(row['translated_review'])
aspect_sentiments.append(sentiments[aspect_idx])
if not aspect_reviews:
continue
positive_count = aspect_sentiments.count('Positive')
negative_count = aspect_sentiments.count('Negative')
total_count = len(aspect_sentiments)
# Generate contextual summary
if negative_count > positive_count:
sentiment_trend = "predominantly negative"
key_issues = "Issues include quality concerns, performance problems, and user dissatisfaction."
elif positive_count > negative_count:
sentiment_trend = "predominantly positive"
key_issues = "Users appreciate the quality, performance, and overall experience."
else:
sentiment_trend = "mixed"
key_issues = "Reviews show both satisfaction and areas for improvement."
summaries[aspect] = f"'{aspect}' mentioned in {total_count} reviews with {sentiment_trend} sentiment ({positive_count} positive, {negative_count} negative). {key_issues}"
return summaries
class DataProcessor:
"""Enhanced main data processing pipeline coordinator."""
def __init__(self):
self.translator = TranslationService()
self.absa_processor = ABSAProcessor()
self.validator = DataValidator()
self.analytics = AspectAnalytics()
self.summary_generator = SummaryGenerator()
self.task_manager = None # Will be set by API
def set_task_manager(self, task_manager):
"""Set task manager for cancellation support."""
self.task_manager = task_manager
# Also set it for ABSA processor
self.absa_processor.set_task_manager(task_manager)
def process_uploaded_data(self, df: pd.DataFrame, task_id: Optional[str] = None) -> Dict[str, Any]:
"""
Complete enhanced processing pipeline for uploaded data with cancellation support.
Args:
df: Uploaded dataframe
task_id: Optional task ID for cancellation tracking
Returns:
Dictionary containing all processed results with advanced analytics
"""
# Import gc for cleanup
import gc
# Check cancellation at start
if task_id and self.task_manager and self.task_manager.is_cancelled(task_id):
return {'status': 'cancelled', 'message': 'Task cancelled by user'}
# Validate data
if task_id and self.task_manager:
self.task_manager.update_task(task_id, stage='validation', progress=5, status='processing')
is_valid, errors = self.validator.validate_csv(df)
if not is_valid:
return {'error': errors}
# Clean data
df_clean = self.validator.clean_data(df)
reviews = df_clean['review'].tolist()
# Check cancellation before translation
if task_id and self.task_manager and self.task_manager.is_cancelled(task_id):
del df_clean, reviews
gc.collect()
return {'status': 'cancelled', 'message': 'Task cancelled during validation'}
# Translation with batch processing and cancellation checks
if task_id and self.task_manager:
self.task_manager.update_task(task_id, stage='translation', progress=10)
with st.spinner("Translating reviews..."):
translated_reviews = []
detected_languages = []
# Process in batches to allow cancellation
batch_size = 10
for i in range(0, len(reviews), batch_size):
# Check cancellation
if task_id and self.task_manager and self.task_manager.is_cancelled(task_id):
del df_clean, reviews, translated_reviews, detected_languages
gc.collect()
return {'status': 'cancelled', 'message': 'Task cancelled during translation'}
batch = reviews[i:i+batch_size]
batch_translated, batch_langs = self.translator.process_reviews(batch)
translated_reviews.extend(batch_translated)
detected_languages.extend(batch_langs)
# Update progress (10-40%)
if task_id and self.task_manager:
progress = 10 + int((i / len(reviews)) * 30)
self.task_manager.update_task(task_id, progress=progress)
# Check cancellation before intent classification
if task_id and self.task_manager and self.task_manager.is_cancelled(task_id):
del df_clean, reviews, translated_reviews, detected_languages
gc.collect()
return {'status': 'cancelled', 'message': 'Task cancelled after translation'}
# Enhanced intent classification
if task_id and self.task_manager:
self.task_manager.update_task(task_id, stage='intent_classification', progress=40)
with st.spinner("Classifying intents with severity analysis..."):
intent_results = IntentClassifier.classify_batch_enhanced(translated_reviews)
# Check cancellation before ABSA (most expensive step)
if task_id and self.task_manager and self.task_manager.is_cancelled(task_id):
del df_clean, reviews, translated_reviews, detected_languages, intent_results
gc.collect()
return {'status': 'cancelled', 'message': 'Task cancelled after intent classification'}
# ABSA processing with cancellation support
if task_id and self.task_manager:
self.task_manager.update_task(task_id, stage='aspect_extraction', progress=50)
with st.spinner("Extracting aspects and sentiments..."):
absa_results = self.absa_processor.extract_aspects_and_sentiments(
translated_reviews,
task_id=task_id
)
# Check if ABSA was cancelled
if isinstance(absa_results, dict) and absa_results.get('status') == 'cancelled':
del df_clean, reviews, translated_reviews, detected_languages, intent_results
gc.collect()
return absa_results
# Combine results with enhanced structure
if task_id and self.task_manager:
self.task_manager.update_task(task_id, stage='combining_results', progress=90)
df_processed = df_clean.copy()
df_processed['translated_review'] = translated_reviews
df_processed['detected_language'] = detected_languages
df_processed['intent'] = [r['intent'] for r in intent_results]
df_processed['intent_severity'] = [r['severity_type'] for r in intent_results]
df_processed['intent_confidence'] = [r['confidence'] for r in intent_results]
# Add ABSA results with enhanced structure
aspects_list = []
aspect_sentiments_list = []
overall_sentiment = []
for result in absa_results:
aspects_list.append(result['aspects'])
aspect_sentiments_list.append(result['sentiments'])
# Calculate overall sentiment
if result['sentiments']:
positive_count = result['sentiments'].count('Positive')
negative_count = result['sentiments'].count('Negative')
if positive_count > negative_count:
overall_sentiment.append('Positive')
elif negative_count > positive_count:
overall_sentiment.append('Negative')
else:
overall_sentiment.append('Neutral')
else:
overall_sentiment.append('Neutral')
df_processed['aspects'] = aspects_list
df_processed['aspect_sentiments'] = aspect_sentiments_list
df_processed['overall_sentiment'] = overall_sentiment
# Advanced analytics
if task_id and self.task_manager:
self.task_manager.update_task(task_id, stage='analytics', progress=95)
with st.spinner("Calculating aspect analytics and priority scores..."):
areas_of_improvement, strength_anchors = self.analytics.calculate_aspect_scores(df_processed)
aspect_network = self.analytics.calculate_aspect_cooccurrence(df_processed)
sentiment_alerts = self.analytics.detect_sentiment_spikes(df_processed)
# Generate summaries
with st.spinner("Generating AI-powered summaries..."):
macro_summary = self.summary_generator.generate_macro_summary(
df_processed, areas_of_improvement, strength_anchors
)
# Get top aspects for micro summaries
top_negative_aspects = areas_of_improvement['aspect'].head(3).tolist() if len(areas_of_improvement) > 0 else []
top_positive_aspects = strength_anchors['aspect'].head(3).tolist() if len(strength_anchors) > 0 else []
top_aspects = top_negative_aspects + top_positive_aspects
micro_summaries = self.summary_generator.generate_aspect_micro_summaries(
df_processed, top_aspects
)
# Mark task as complete
if task_id and self.task_manager:
self.task_manager.update_task(task_id, stage='completed', progress=100)
# ========== NEW: ASPECT-LEVEL DATA TRANSFORMATION ==========
aspect_level_data = []
mixed_sentiment_reviews = []
for idx, row in df_processed.iterrows():
aspects = row['aspects'] if isinstance(row['aspects'], list) else []
aspect_sentiments = row['aspect_sentiments'] if isinstance(row['aspect_sentiments'], list) else []
# Check for mixed sentiments (conflicting aspect sentiments)
unique_sentiments = set(aspect_sentiments)
is_mixed = ('Positive' in unique_sentiments and 'Negative' in unique_sentiments)
if is_mixed:
mixed_sentiment_reviews.append({
'review_id': row['id'],
'review': row['review'],
'aspects': aspects,
'aspect_sentiments': aspect_sentiments,
'intent': row['intent'],
'date': row['date']
})
# Create aspect-level records
for aspect, sentiment in zip(aspects, aspect_sentiments):
aspect_level_data.append({
'review_id': row['id'],
'review': row['review'],
'aspect': aspect,
'aspect_sentiment': sentiment,
'overall_sentiment': row['overall_sentiment'],
'intent': row['intent'],
'intent_severity': row['intent_severity'],
'date': row['date'],
'language': row['detected_language']
})
aspect_level_df = pd.DataFrame(aspect_level_data) if aspect_level_data else pd.DataFrame()
mixed_sentiment_df = pd.DataFrame(mixed_sentiment_reviews) if mixed_sentiment_reviews else pd.DataFrame()
return {
'processed_data': df_processed,
'aspect_level_data': aspect_level_df, # NEW: Aspect-level granular data
'mixed_sentiment_reviews': mixed_sentiment_df, # NEW: Mixed sentiment detection
'absa_details': absa_results,
'areas_of_improvement': areas_of_improvement,
'strength_anchors': strength_anchors,
'aspect_network': aspect_network,
'sentiment_alerts': sentiment_alerts,
'macro_summary': macro_summary,
'micro_summaries': micro_summaries,
'summary': {
'total_reviews': len(df_processed),
'total_aspects': len(aspect_level_df),
'mixed_sentiment_count': len(mixed_sentiment_df),
'mixed_sentiment_pct': round(len(mixed_sentiment_df) / len(df_processed) * 100, 1) if len(df_processed) > 0 else 0,
'languages_detected': list(set(detected_languages)),
'intents_distribution': pd.Series([r['intent'] for r in intent_results]).value_counts().to_dict(),
'sentiment_distribution': pd.Series(overall_sentiment).value_counts().to_dict(),
'top_problem_areas': len(areas_of_improvement),
'top_strength_anchors': len(strength_anchors),
'active_alerts': len(sentiment_alerts)
}
} |