code/process_sentence2triple.py

class StochasticRAGTripleExtraction:
    def __init__(self, retriever, generator, temperature=1.0):
        self.retriever = retriever
        self.generator = generator
        self.temperature = temperature
    
    def extract_with_sampling(self, text, num_samples=10):
        """
        通过随机采样多次提取，增强三元组多样性
        """
        all_triples = []
        
        for _ in range(num_samples):
            # 随机检索不同的文档
            docs = self.retriever.search(
                text, 
                top_k=5,
                temperature=self.temperature  # 随机性检索
            )
            
            # 随机采样提示模板
            templates = [
                "Extract triples from: {text}",
                "Find all (subject, relation, object) in: {text}",
                "Identify knowledge triples: {text}"
            ]
            template = random.choice(templates)
            
            # 生成三元组（带温度参数控制随机性）
            response = self.generator.generate(
                template.format(text=text),
                temperature=self.temperature,
                do_sample=True,  # 启用随机采样
                top_p=0.9  # nucleus采样
            )
            
            # 解析并收集三元组
            triples = self.parse_triples(response)
            all_triples.extend(triples)
        
        # 基于出现频率过滤
        return self.filter_by_frequency(all_triples)
    
    def filter_by_frequency(self, triples, min_count=3):
        """
        保留多次采样中频繁出现的三元组
        """
        triple_counts = Counter(triples)
        return [
            triple for triple, count in triple_counts.items()
            if count >= min_count
        ]