Spaces:
Sleeping
Sleeping
File size: 10,791 Bytes
feeaf83 | 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 | import re
from typing import Dict, List, Any
from datetime import datetime
class StudentClassifier:
"""
Lightweight, fast NLP classifier for Student News that avoids heavy ML/LLM models.
Uses keyword heuristics and regex to categorize, summarize, and extract metadata.
"""
CATEGORIES = {
"Scholarships & Internships": ["scholarship", "internship", "fellowship", "stipend", "student grant", "student funding", "financial assistance for students", "startup grant", "startup funding", "training program", "summer internship"],
"Exams & Results": ["exam date", "results declared", "scorecard", "admit card", "nta", "jee main", "neet ug", "upsc prelims", "cut-off marks", "answer key", "date sheet", "counseling", "mock test", "merit list", "board exams", "entrance exam", "examination", "test result", "passing marks", "results out"],
"Admissions & Courses": ["admission open", "apply now", "enrollment", "course curriculum", "degree program", "admission-notice", "university admission", "college admission", "enroll now", "last date to apply"],
"Career & Jobs": ["campus placement", "fresher hiring", "graduate recruitment", "university placement", "off-campus drive", "fresher vacancy", "job for students", "career fair", "recruitment drive", "employment news", "hiring", "job alert"],
"Education & Policy": ["education policy", "syllabus", "ugc", "cbse", "aicte", "nep 2020", "ministry of education", "school board", "educational reform", "government school", "higher education", "campus life", "university news", "school news", "student union", "academic research", "research paper", "phd thesis"]
}
PROFILES = {
"School Student (10th/12th)": ["cbse", "ncert", "class 10", "class 12", "board exam", "icse", "state board", "school student", "secondary education"],
"Engineering Aspirant": ["jee", "b.tech", "engineering entrance", "gate exam", "iit", "nit", "bitsat", "josaa", "engineering student", "iitian"],
"Medical Aspirant": ["neet", "mbbs admission", "aiims", "bds", "pharmacy entrance", "ayush counseling", "jipmer", "pgimer", "medical student"],
"Govt Job Aspirant": ["upsc notification", "ssc cgl", "bank po entrance", "rrb ntpc", "ibps po", "nda exam", "cds notification", "civil services prelims", "sbi po", "state psc", "government job", "employment news"],
"Graduate/Techie": ["fresher", "placement", "startup", "coding", "software engineer", "developer", "internship", "graduation"]
}
AUTHORITIES = ["UGC", "CBSE", "AICTE", "NTA", "UPSC", "SSC", "Ministry of Education", "State Board", "University Grants Commission"]
URGENCY_KEYWORDS = {
"High": ["deadline", "tomorrow", "today", "urgent", "breaking", "last date", "closing", "alert"],
"Medium": ["upcoming", "soon", "next week", "announced", "scheduled", "notification"],
"Low": ["proposed", "planned", "expected", "future", "report", "study"]
}
STRICT_KEYWORDS = [
"university admission", "college admission", "campus placement", "scholarship",
"exam result", "admit card", "student visa", "study abroad", "cut-off marks",
"board exam", "jee main", "neet ug", "education policy", "fellowship", "internship",
"syllabus", "ncert", "ugc", "cbse", "nta", "upsc notification", "ssc cgl",
"startup grant", "startup funding", "hackathon", "coding competition",
"educational", "learning", "academic", "placement drive", "hiring fresher",
"recruitment drive", "campus hiring", "student achievement", "campus life",
"exam update", "university news", "school news", "career guide",
"intern", "stipend", "fresher", "placement", "graduation", "degree", "diploma",
"research paper", "higher education", "student protest", "tuition fees"
]
def _extract_specific_exam(self, text: str) -> str | None:
# Match standard Indian exams as whole words
exams = ["NEET", "JEE", "UPSC", "SSC", "CBSE", "CAT", "CLAT", "GATE", "ICSE", "CUET", "NDA", "CDS", "IBPS"]
for exam in exams:
if re.search(rf'\b{exam}\b', text, re.IGNORECASE):
return exam
return None
def process_article(self, title: str, content: str) -> Dict[str, Any] | None:
"""
Main pipeline to process a raw article into a structured student news item.
Returns None if it does not strictly cover Student News keywords.
"""
combined_text = f"{title} {content}".lower()
# Strict context guard
strict_matches = 0
for kw in self.STRICT_KEYWORDS:
if re.search(rf'\b{kw}\b', combined_text, re.IGNORECASE):
strict_matches += 1
specific_exam = self._extract_specific_exam(combined_text)
# Relaxation: if the category is Education or it hits at least one strict keyword, it's a match
if strict_matches == 0 and not specific_exam:
# Check if it's generally about education
if "education" in combined_text or "student" in combined_text:
# Allow it but maybe with lower trend score
pass
else:
return None
category = self._assign_category(combined_text)
tags = self._generate_tags(combined_text, category)
profiles = self._assign_profiles(combined_text)
direct_links = self._extract_links(text=content) # Extract links only from content
# Inject exact matching exam if found
if specific_exam:
tags.insert(0, f"#{specific_exam}")
dates = self._extract_dates(combined_text)
authority = self._extract_authority(combined_text)
urgency = self._determine_urgency(combined_text)
summary = self._generate_summary(content)
trend_score = self._calculate_trend_score(combined_text, urgency)
return {
"title": title,
"summary": summary,
"category": category,
"tags": tags,
"profiles": profiles,
"direct_links": direct_links,
"important_dates": dates,
"authority": authority,
"urgency": urgency,
"trend_score": trend_score
}
def _assign_category(self, text: str) -> str:
best_match = "General Student News"
max_score = 0
for cat, keywords in self.CATEGORIES.items():
score = 0
for kw in keywords:
if re.search(rf'\b{kw}\b', text, re.IGNORECASE):
score += 1
if score > max_score:
max_score = score
best_match = cat
return best_match
def _generate_tags(self, text: str, category: str) -> List[str]:
tags = set()
# Base tag on category
if "Exam" in category: tags.add("#Exam")
if "Scholarship" in category: tags.add("#Scholarship")
if "Job" in category or "Career" in category: tags.add("#Job")
if "Policy" in category: tags.add("#Policy")
if "Abroad" in category: tags.add("#StudyAbroad")
if "Tech" in category: tags.add("#Tech")
# Specific keyword tags
if "nta" in text or "jee" in text or "neet" in text: tags.add("#CompetitiveExams")
if "cbse" in text or "board" in text: tags.add("#BoardExams")
if "internship" in text: tags.add("#Internship")
if "hackathon" in text or "coding" in text: tags.add("#Coding")
return list(tags)[:4] # max 4 tags
def _extract_dates(self, text: str) -> List[str]:
# Very lightweight date extraction regex (e.g. 15th Jan, March 20, 2024-05-12)
dates = []
# Basic pattern for "DD Month" or "Month DD"
month_pattern = r'(Jan|Feb|Mar|Apr|May|Jun|Jul|Aug|Sep|Oct|Nov|Dec)[a-z]*'
date_pattern = rf'\b(\d{{1,2}}(?:st|nd|rd|th)?\s+{month_pattern}|{month_pattern}\s+\d{{1,2}})\b'
matches = re.findall(date_pattern, text, re.IGNORECASE)
# matches returns tuples because of groups in regex, clean it up
for match in re.finditer(date_pattern, text, re.IGNORECASE):
dates.append(match.group(0).title())
return list(set(dates))[:2] # Top 2 dates extracted
def _extract_authority(self, text: str) -> str:
for auth in self.AUTHORITIES:
if auth.lower() in text:
return auth
return "General"
def _determine_urgency(self, text: str) -> str:
for level, keywords in self.URGENCY_KEYWORDS.items():
if any(kw in text for kw in keywords):
return level
return "Low"
def _generate_summary(self, content: str) -> str:
# Fast 3-line summary approximation by splitting sentences
# Not perfect NLP, but extremely fast
if not content:
return ""
# Clean text slightly
text = re.sub(r'\s+', ' ', content).strip()
# Split by periods roughly
sentences = [s.strip() + "." for s in text.split('.') if len(s.strip()) > 15]
if len(sentences) <= 3:
return " ".join(sentences)
# Take first 2 sentences and try to find one with numbers/dates for the 3rd
summary_sents = sentences[:2]
best_third = sentences[2]
for s in sentences[2:6]:
if re.search(r'\d', s):
best_third = s
break
summary_sents.append(best_third)
return " ".join(summary_sents)
def _calculate_trend_score(self, text: str, urgency: str) -> int:
score = 10
if urgency == "High":
score += 40
elif urgency == "Medium":
score += 20
# Boost score based on strong keywords
hot_keywords = ["deadline", "released", "declared", "announce", "breaking", "major"]
score += sum(15 for kw in hot_keywords if kw in text)
return min(score, 100)
def _assign_profiles(self, text: str) -> List[str]:
profiles = []
for profile, keywords in self.PROFILES.items():
if any(kw in text for kw in keywords):
profiles.append(profile)
# default to all if none matched
if not profiles:
profiles.append("General Student")
return profiles
def _extract_links(self, text: str) -> List[str]:
# Extract http/https links
urls = re.findall(r'https?://[^\s<>"]+|www\.[^\s<>"]+', text)
return list(set(urls))[:2]
|