Spaces:
Running
Running
File size: 9,510 Bytes
a977115 17b8d44 a977115 17b8d44 a977115 17b8d44 9e7bcc4 a977115 17b8d44 a977115 9e7bcc4 d0ab9c6 a977115 d0ab9c6 17b8d44 a977115 17b8d44 a977115 17b8d44 a977115 17b8d44 a977115 17b8d44 a977115 17b8d44 a977115 17b8d44 a977115 17b8d44 a977115 17b8d44 a977115 17b8d44 a977115 17b8d44 a977115 9e7bcc4 a977115 9e7bcc4 a977115 17b8d44 9e7bcc4 a977115 17b8d44 9e7bcc4 a977115 17b8d44 a977115 17b8d44 9e7bcc4 a977115 17b8d44 a977115 17b8d44 9e7bcc4 a977115 | 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 | """
Priority Engine — AI-Powered Complaint Priority Scoring
Formula: Score = (Urgency × 0.4) + (Impact × 0.3) + (Recurrence × 0.2) + (Sentiment × 0.1)
Score 0-100 → P0 (Critical), P1 (High), P2 (Medium), P3 (Routine)
"""
from typing import Dict
from services.nlp_service import nlp_service
from services.sentiment_service import sentiment_service
from services.qwen_priority_service import qwen_priority_service
PRIORITY_MAP = {
"P0": {"label": "CRITICAL", "response": "< 1 hour", "color": "#DC2626", "min_score": 85},
"P1": {"label": "HIGH", "response": "< 6 hours", "color": "#EA580C", "min_score": 65},
"P2": {"label": "MEDIUM", "response": "< 48 hours", "color": "#CA8A04", "min_score": 40},
"P3": {"label": "ROUTINE", "response": "< 2 weeks", "color": "#16A34A", "min_score": 0},
}
# Category base impact scores
CATEGORY_IMPACT = {
"Water Supply": 75,
"Roads & Potholes": 65,
"Drainage": 70,
"Electricity": 60,
"Garbage & Sanitation": 55,
"Safety & Security": 85,
"Public Health": 90,
"Education": 50,
"Public Transport": 45,
"Others": 35,
}
class PriorityEngine:
MODEL_VERSION = "priority-engine-v2.2-qwen"
def calculate_starvation_bonus(self, unresponded_hours: float = 0.0) -> float:
"""Escalate stale, unresponded complaints to prevent starvation in queues."""
hours = max(0.0, float(unresponded_hours or 0.0))
if hours >= 168:
return 35.0
if hours >= 96:
return 26.0
if hours >= 48:
return 18.0
if hours >= 24:
return 12.0
if hours >= 6:
return 6.0
return 0.0
def score_to_priority(self, score: float) -> str:
return self._score_to_priority(score)
def calculate_score(
self,
text: str,
category: str,
ward: str,
recurrence_count: int = 0,
local_cluster_count: int = 0,
social_mentions: int = 0,
unresponded_hours: float = 0.0,
enable_qwen: bool = True,
) -> Dict:
"""
Calculate AI priority score using the weighted formula.
Returns detailed scoring breakdown.
"""
qwen_result = {"used": False, "reason": "disabled_by_workflow"}
if enable_qwen:
qwen_result = qwen_priority_service.score_issue(
text=text,
category=category,
ward=ward,
)
# 1. Urgency Score (40% weight)
if qwen_result.get("used"):
urgency_score = float(qwen_result.get("urgency", 50.0))
if urgency_score >= 85:
urgency_level = "critical"
elif urgency_score >= 65:
urgency_level = "high"
elif urgency_score >= 40:
urgency_level = "medium"
else:
urgency_level = "low"
else:
urgency_level, urgency_score = nlp_service.assess_urgency(text)
# 2. Impact Score (30% weight)
if qwen_result.get("used"):
impact_score = float(qwen_result.get("impact", 50.0))
else:
base_impact = CATEGORY_IMPACT.get(category, 35)
# Boost impact if near schools, hospitals, main roads
impact_boost = 0
text_lower = text.lower()
if any(w in text_lower for w in ["school", "hospital", "college", "market", "temple"]):
impact_boost += 15
if any(w in text_lower for w in ["main road", "highway", "national", "state"]):
impact_boost += 10
if any(w in text_lower for w in ["children", "elderly", "patients", "pregnant"]):
impact_boost += 20
impact_score = min(base_impact + impact_boost, 100)
# 3. Recurrence/Local Cluster Score (20% weight)
recurrence_signal = max(0, int(recurrence_count or 0)) + max(0, int(local_cluster_count or 0))
if recurrence_signal >= 6:
recurrence_score = 95.0
elif recurrence_signal >= 4:
recurrence_score = 75.0
elif recurrence_signal >= 2:
recurrence_score = 55.0
elif recurrence_signal >= 1:
recurrence_score = 35.0
else:
recurrence_score = 10.0
# 4. Sentiment + public pressure signal (10% weight)
if qwen_result.get("used"):
sentiment_label = str(qwen_result.get("sentiment_label", "neutral"))
sentiment_confidence = float(qwen_result.get("confidence", 0.5) or 0.5)
text_sentiment_score = float(qwen_result.get("sentiment_score", 45.0) or 45.0)
else:
sentiment_payload = sentiment_service.analyze(text)
sentiment_label = sentiment_payload.get("sentiment", "neutral")
sentiment_confidence = float(sentiment_payload.get("confidence", 0.5) or 0.5)
if sentiment_label == "negative":
text_sentiment_score = 85.0 + (10.0 * sentiment_confidence)
elif sentiment_label == "positive":
text_sentiment_score = 20.0
else:
text_sentiment_score = 45.0
if social_mentions >= 50:
social_pressure_score = 90.0
elif social_mentions >= 20:
social_pressure_score = 70.0
elif social_mentions >= 5:
social_pressure_score = 50.0
else:
social_pressure_score = 20.0
sentiment_score = round(min(100.0, (text_sentiment_score * 0.75) + (social_pressure_score * 0.25)), 1)
# FINAL SCORE
base_score = (
urgency_score * 0.4
+ impact_score * 0.3
+ recurrence_score * 0.2
+ sentiment_score * 0.1
)
starvation_bonus = self.calculate_starvation_bonus(unresponded_hours)
final_score = round(min(100.0, base_score + starvation_bonus), 1)
# Map to priority level
priority = self._score_to_priority(final_score)
priority_info = PRIORITY_MAP[priority]
# Build explanation
explanation = self._build_explanation(
urgency_level, urgency_score, impact_score, recurrence_count,
recurrence_score, local_cluster_count, sentiment_label,
sentiment_confidence, social_mentions, sentiment_score, final_score, priority,
starvation_bonus, unresponded_hours,
)
return {
"score": final_score,
"priority": priority,
"urgency": urgency_score,
"impact": impact_score,
"recurrence": recurrence_score,
"sentiment": sentiment_score,
"sentiment_label": sentiment_label,
"sentiment_confidence": round(sentiment_confidence, 3),
"starvation_bonus": starvation_bonus,
"response_time": priority_info["response"],
"model_version": self.MODEL_VERSION,
"score_source": "qwen" if qwen_result.get("used") else "heuristic_fallback",
"weights": {
"urgency": 0.4,
"impact": 0.3,
"recurrence": 0.2,
"sentiment": 0.1,
},
"breakdown": {
"urgency": round(urgency_score, 1),
"impact": round(impact_score, 1),
"recurrence": round(recurrence_score, 1),
"sentiment": round(sentiment_score, 1),
"recurrence_count": recurrence_count,
"local_cluster_count": local_cluster_count,
"social_mentions": social_mentions,
"qwen_reasoning": qwen_result.get("reasoning") if qwen_result.get("used") else None,
"qwen_fallback_reason": qwen_result.get("reason") if not qwen_result.get("used") else None,
},
"explanation": explanation,
}
def _score_to_priority(self, score: float) -> str:
if score >= 85:
return "P0"
elif score >= 65:
return "P1"
elif score >= 40:
return "P2"
return "P3"
def _build_explanation(
self, urgency_level, urgency_score, impact_score, recurrence_count,
recurrence_score, local_cluster_count, sentiment_label,
sentiment_confidence, social_mentions, sentiment_score, final_score, priority,
starvation_bonus, unresponded_hours,
) -> str:
parts = []
recurrence_signal = max(0, int(recurrence_count or 0)) + max(0, int(local_cluster_count or 0))
parts.append(f"Urgency: {urgency_level.upper()} ({urgency_score}/100, weight 40%)")
parts.append(f"Impact: {impact_score}/100 (weight 30%)")
parts.append(
f"Recurrence: {recurrence_count} prior + {local_cluster_count} nearby -> {recurrence_signal} signal -> {recurrence_score}/100 (weight 20%)"
)
parts.append(
f"Sentiment: {sentiment_label.upper()} (conf {round(sentiment_confidence, 3)}) + {social_mentions} mentions -> {sentiment_score}/100 (weight 10%)"
)
if starvation_bonus > 0:
parts.append(
f"Starvation Guard: +{starvation_bonus} boost for {round(unresponded_hours, 1)}h unresponded"
)
parts.append(f"FINAL: {final_score} -> {priority} ({PRIORITY_MAP[priority]['label']})")
parts.append(f"Target Response: {PRIORITY_MAP[priority]['response']}")
return " | ".join(parts)
# Singleton
priority_engine = PriorityEngine()
|