webscout 7.1__py3-none-any.whl → 7.3__py3-none-any.whl

webstoken/classifier.py

@@ -1,189 +1,189 @@
-"""
-Text classification module using rule-based and statistical approaches.
-"""
-
-from typing import Dict, List, Set, Tuple
-from collections import Counter
-import math
-import re
-
-from .normalizer import TextNormalizer
-from .tokenizer import WordTokenizer
-
-
-class TextClassifier:
-    """Simple text classifier using TF-IDF and cosine similarity."""
-    
-    def __init__(self):
-        self.word_tokenizer = WordTokenizer()
-        self.normalizer = TextNormalizer()
-        self.documents: Dict[str, List[str]] = {}  # category -> list of documents
-        self.vocabulary: Set[str] = set()
-        self.idf_scores: Dict[str, float] = {}
-        self.category_vectors: Dict[str, Dict[str, float]] = {}
-    
-    def train(self, documents: Dict[str, List[str]]) -> None:
-        """
-        Train the classifier on labeled documents.
-        
-        Args:
-            documents: Dict mapping categories to lists of documents
-        """
-        self.documents = documents
-        
-        # Build vocabulary and document frequencies
-        doc_frequencies: Dict[str, int] = Counter()
-        total_docs = sum(len(docs) for docs in documents.values())
-        
-        for category, docs in documents.items():
-            for doc in docs:
-                # Normalize and tokenize
-                doc = self.normalizer.normalize(doc)
-                tokens = self.word_tokenizer.tokenize(doc)
-                
-                # Update vocabulary and document frequencies
-                unique_tokens = set(tokens)
-                self.vocabulary.update(unique_tokens)
-                doc_frequencies.update(unique_tokens)
-        
-        # Calculate IDF scores
-        self.idf_scores = {
-            word: math.log(total_docs / (freq + 1))
-            for word, freq in doc_frequencies.items()
-        }
-        
-        # Calculate TF-IDF vectors for each category
-        for category, docs in documents.items():
-            category_vector: Dict[str, float] = {word: 0.0 for word in self.vocabulary}
-            
-            for doc in docs:
-                # Get term frequencies
-                doc = self.normalizer.normalize(doc)
-                tokens = self.word_tokenizer.tokenize(doc)
-                term_freqs = Counter(tokens)
-                
-                # Update category vector with TF-IDF scores
-                for word, tf in term_freqs.items():
-                    if word in self.idf_scores:
-                        category_vector[word] += tf * self.idf_scores[word]
-            
-            # Average the scores
-            for word in category_vector:
-                category_vector[word] /= len(docs)
-            
-            self.category_vectors[category] = category_vector
-    
-    def _calculate_vector(self, text: str) -> Dict[str, float]:
-        """Calculate TF-IDF vector for input text."""
-        # Normalize and tokenize
-        text = self.normalizer.normalize(text)
-        tokens = self.word_tokenizer.tokenize(text)
-        term_freqs = Counter(tokens)
-        
-        # Calculate TF-IDF scores
-        vector = {word: 0.0 for word in self.vocabulary}
-        for word, tf in term_freqs.items():
-            if word in self.idf_scores:
-                vector[word] = tf * self.idf_scores[word]
-        
-        return vector
-    
-    def _cosine_similarity(self, vec1: Dict[str, float], vec2: Dict[str, float]) -> float:
-        """Calculate cosine similarity between two vectors."""
-        dot_product = sum(vec1[word] * vec2[word] for word in vec1)
-        norm1 = math.sqrt(sum(score * score for score in vec1.values()))
-        norm2 = math.sqrt(sum(score * score for score in vec2.values()))
-        
-        if norm1 == 0 or norm2 == 0:
-            return 0.0
-        return dot_product / (norm1 * norm2)
-    
-    def classify(self, text: str) -> List[Tuple[str, float]]:
-        """
-        Classify text into categories with confidence scores.
-        
-        Returns:
-            List of (category, confidence) tuples, sorted by confidence
-        """
-        if not self.category_vectors:
-            raise ValueError("Classifier must be trained before classification")
-        
-        # Calculate vector for input text
-        text_vector = self._calculate_vector(text)
-        
-        # Calculate similarity with each category
-        similarities = [
-            (category, self._cosine_similarity(text_vector, category_vec))
-            for category, category_vec in self.category_vectors.items()
-        ]
-        
-        # Sort by similarity score
-        return sorted(similarities, key=lambda x: x[1], reverse=True)
-
-
-class TopicClassifier:
-    """Rule-based topic classifier using keyword matching."""
-    
-    def __init__(self):
-        # Define topic keywords
-        self.topic_keywords = {
-            'TECHNOLOGY': {
-                'computer', 'software', 'hardware', 'internet', 'programming',
-                'digital', 'data', 'algorithm', 'code', 'web', 'app', 'mobile',
-                'cyber', 'robot', 'ai', 'artificial intelligence', 'machine learning'
-            },
-            'SCIENCE': {
-                'research', 'experiment', 'laboratory', 'scientific', 'physics',
-                'chemistry', 'biology', 'mathematics', 'theory', 'hypothesis',
-                'study', 'discovery', 'innovation', 'analysis', 'observation'
-            },
-            'BUSINESS': {
-                'company', 'market', 'finance', 'investment', 'stock', 'trade',
-                'economy', 'business', 'corporate', 'startup', 'entrepreneur',
-                'profit', 'revenue', 'management', 'strategy', 'commercial'
-            },
-            'POLITICS': {
-                'government', 'policy', 'election', 'political', 'democracy',
-                'parliament', 'congress', 'law', 'legislation', 'party',
-                'vote', 'campaign', 'president', 'minister', 'diplomatic'
-            },
-            'SPORTS': {
-                'game', 'team', 'player', 'competition', 'tournament',
-                'championship', 'score', 'match', 'athlete', 'sport',
-                'win', 'lose', 'victory', 'defeat', 'coach', 'training'
-            },
-            'ENTERTAINMENT': {
-                'movie', 'film', 'music', 'song', 'concert', 'actor',
-                'actress', 'celebrity', 'show', 'performance', 'art',
-                'entertainment', 'theater', 'dance', 'festival', 'media'
-            }
-        }
-        
-        # Compile regex patterns for each topic
-        self.topic_patterns = {
-            topic: re.compile(r'\b(' + '|'.join(re.escape(kw) for kw in keywords) + r')\b', re.IGNORECASE)
-            for topic, keywords in self.topic_keywords.items()
-        }
-    
-    def classify(self, text: str) -> List[Tuple[str, float]]:
-        """
-        Classify text into topics with confidence scores.
-        
-        Returns:
-            List of (topic, confidence) tuples, sorted by confidence
-        """
-        # Count keyword matches for each topic
-        topic_matches = {
-            topic: len(pattern.findall(text))
-            for topic, pattern in self.topic_patterns.items()
-        }
-        
-        # Calculate confidence scores
-        total_matches = sum(topic_matches.values()) or 1  # Avoid division by zero
-        topic_scores = [
-            (topic, count / total_matches)
-            for topic, count in topic_matches.items()
-        ]
-        
-        # Sort by score
-        return sorted(topic_scores, key=lambda x: x[1], reverse=True)
+"""
+Text classification module using rule-based and statistical approaches.
+"""
+
+from typing import Dict, List, Set, Tuple
+from collections import Counter
+import math
+import re
+
+from .normalizer import TextNormalizer
+from .tokenizer import WordTokenizer
+
+
+class TextClassifier:
+    """Simple text classifier using TF-IDF and cosine similarity."""
+    
+    def __init__(self):
+        self.word_tokenizer = WordTokenizer()
+        self.normalizer = TextNormalizer()
+        self.documents: Dict[str, List[str]] = {}  # category -> list of documents
+        self.vocabulary: Set[str] = set()
+        self.idf_scores: Dict[str, float] = {}
+        self.category_vectors: Dict[str, Dict[str, float]] = {}
+    
+    def train(self, documents: Dict[str, List[str]]) -> None:
+        """
+        Train the classifier on labeled documents.
+        
+        Args:
+            documents: Dict mapping categories to lists of documents
+        """
+        self.documents = documents
+        
+        # Build vocabulary and document frequencies
+        doc_frequencies: Dict[str, int] = Counter()
+        total_docs = sum(len(docs) for docs in documents.values())
+        
+        for category, docs in documents.items():
+            for doc in docs:
+                # Normalize and tokenize
+                doc = self.normalizer.normalize(doc)
+                tokens = self.word_tokenizer.tokenize(doc)
+                
+                # Update vocabulary and document frequencies
+                unique_tokens = set(tokens)
+                self.vocabulary.update(unique_tokens)
+                doc_frequencies.update(unique_tokens)
+        
+        # Calculate IDF scores
+        self.idf_scores = {
+            word: math.log(total_docs / (freq + 1))
+            for word, freq in doc_frequencies.items()
+        }
+        
+        # Calculate TF-IDF vectors for each category
+        for category, docs in documents.items():
+            category_vector: Dict[str, float] = {word: 0.0 for word in self.vocabulary}
+            
+            for doc in docs:
+                # Get term frequencies
+                doc = self.normalizer.normalize(doc)
+                tokens = self.word_tokenizer.tokenize(doc)
+                term_freqs = Counter(tokens)
+                
+                # Update category vector with TF-IDF scores
+                for word, tf in term_freqs.items():
+                    if word in self.idf_scores:
+                        category_vector[word] += tf * self.idf_scores[word]
+            
+            # Average the scores
+            for word in category_vector:
+                category_vector[word] /= len(docs)
+            
+            self.category_vectors[category] = category_vector
+    
+    def _calculate_vector(self, text: str) -> Dict[str, float]:
+        """Calculate TF-IDF vector for input text."""
+        # Normalize and tokenize
+        text = self.normalizer.normalize(text)
+        tokens = self.word_tokenizer.tokenize(text)
+        term_freqs = Counter(tokens)
+        
+        # Calculate TF-IDF scores
+        vector = {word: 0.0 for word in self.vocabulary}
+        for word, tf in term_freqs.items():
+            if word in self.idf_scores:
+                vector[word] = tf * self.idf_scores[word]
+        
+        return vector
+    
+    def _cosine_similarity(self, vec1: Dict[str, float], vec2: Dict[str, float]) -> float:
+        """Calculate cosine similarity between two vectors."""
+        dot_product = sum(vec1[word] * vec2[word] for word in vec1)
+        norm1 = math.sqrt(sum(score * score for score in vec1.values()))
+        norm2 = math.sqrt(sum(score * score for score in vec2.values()))
+        
+        if norm1 == 0 or norm2 == 0:
+            return 0.0
+        return dot_product / (norm1 * norm2)
+    
+    def classify(self, text: str) -> List[Tuple[str, float]]:
+        """
+        Classify text into categories with confidence scores.
+        
+        Returns:
+            List of (category, confidence) tuples, sorted by confidence
+        """
+        if not self.category_vectors:
+            raise ValueError("Classifier must be trained before classification")
+        
+        # Calculate vector for input text
+        text_vector = self._calculate_vector(text)
+        
+        # Calculate similarity with each category
+        similarities = [
+            (category, self._cosine_similarity(text_vector, category_vec))
+            for category, category_vec in self.category_vectors.items()
+        ]
+        
+        # Sort by similarity score
+        return sorted(similarities, key=lambda x: x[1], reverse=True)
+
+
+class TopicClassifier:
+    """Rule-based topic classifier using keyword matching."""
+    
+    def __init__(self):
+        # Define topic keywords
+        self.topic_keywords = {
+            'TECHNOLOGY': {
+                'computer', 'software', 'hardware', 'internet', 'programming',
+                'digital', 'data', 'algorithm', 'code', 'web', 'app', 'mobile',
+                'cyber', 'robot', 'ai', 'artificial intelligence', 'machine learning'
+            },
+            'SCIENCE': {
+                'research', 'experiment', 'laboratory', 'scientific', 'physics',
+                'chemistry', 'biology', 'mathematics', 'theory', 'hypothesis',
+                'study', 'discovery', 'innovation', 'analysis', 'observation'
+            },
+            'BUSINESS': {
+                'company', 'market', 'finance', 'investment', 'stock', 'trade',
+                'economy', 'business', 'corporate', 'startup', 'entrepreneur',
+                'profit', 'revenue', 'management', 'strategy', 'commercial'
+            },
+            'POLITICS': {
+                'government', 'policy', 'election', 'political', 'democracy',
+                'parliament', 'congress', 'law', 'legislation', 'party',
+                'vote', 'campaign', 'president', 'minister', 'diplomatic'
+            },
+            'SPORTS': {
+                'game', 'team', 'player', 'competition', 'tournament',
+                'championship', 'score', 'match', 'athlete', 'sport',
+                'win', 'lose', 'victory', 'defeat', 'coach', 'training'
+            },
+            'ENTERTAINMENT': {
+                'movie', 'film', 'music', 'song', 'concert', 'actor',
+                'actress', 'celebrity', 'show', 'performance', 'art',
+                'entertainment', 'theater', 'dance', 'festival', 'media'
+            }
+        }
+        
+        # Compile regex patterns for each topic
+        self.topic_patterns = {
+            topic: re.compile(r'\b(' + '|'.join(re.escape(kw) for kw in keywords) + r')\b', re.IGNORECASE)
+            for topic, keywords in self.topic_keywords.items()
+        }
+    
+    def classify(self, text: str) -> List[Tuple[str, float]]:
+        """
+        Classify text into topics with confidence scores.
+        
+        Returns:
+            List of (topic, confidence) tuples, sorted by confidence
+        """
+        # Count keyword matches for each topic
+        topic_matches = {
+            topic: len(pattern.findall(text))
+            for topic, pattern in self.topic_patterns.items()
+        }
+        
+        # Calculate confidence scores
+        total_matches = sum(topic_matches.values()) or 1  # Avoid division by zero
+        topic_scores = [
+            (topic, count / total_matches)
+            for topic, count in topic_matches.items()
+        ]
+        
+        # Sort by score
+        return sorted(topic_scores, key=lambda x: x[1], reverse=True)