Sentiment Analysis or in particular social network analysis (SNA) is a new research area which is increased explosively. This domain has become a very active research issue in data mining and natural language processing. Sentiment analysis (opinion mining) consists in analyzing and extracting emotions, opinions or attitudes from product’s reviews, movie's reviews, etc., and classify them into classes such as positive, negative and neutral, or extract the degree of importance (polarity). In this paper, we propose a new hybrid approach for classifying tweets into classes based on fuzzy logic and a lexicon based approach using SentiWordnet. Our approach consists in classifying tweets according to three classes: positive, negative or neutral, using SentiWordNet and the fuzzy logic with its three important steps: Fuzzification, Rule Inference/aggregation, and Defuzzification. The dataset of tweets to classify and the result of the classification are stored in the Hadoop Distributed File System (HDFS), and we use the Hadoop MapReduce for the application of our proposal.

中文翻译：

---

一种使用 SentiWordNet 的基于混合多语言模糊的情感分析方法

情感分析，特别是社交网络分析（SNA）是一个爆炸式增长的新研究领域。该领域已成为数据挖掘和自然语言处理领域非常活跃的研究课题。情感分析（观点挖掘）是从产品评论、电影评论等中分析提取情感、观点或态度，并将其分为正面、负面和中性等类别，或提取重要程度（极性）。在本文中，我们提出了一种新的混合方法，用于基于模糊逻辑将推文分类，以及使用 SentiWordnet 的基于词典的方法。我们的方法包括根据三个类别对推文进行分类：正面、负面或中性，使用 SentiWordNet 和模糊逻辑及其三个重要步骤：模糊化，规则推理/聚合和去模糊化。要分类的推文数据集和分类结果存储在 Hadoop 分布式文件系统 (HDFS) 中，我们使用 Hadoop MapReduce 来应用我们的提案。