Target-Dependent Sentiment Analysis of Tweets

Identifier:  TFM:225

Handle: https://hdl.handle.net/20.500.11797/TFM225

Authors:  Fadi Abdulfattah Mohammed Hassan

Abstract:
TFM Summary: The task of target-dependent sentiment analysis aims to identify the sentiment polarity towards a certain target in a given text. All the existing models of this task assume that the target is known. This fact has motivated us to develop an end-to-end target-dependent sentiment analysis system. To the extent of our knowledge, this is the first system that identifies and extract the target of the tweets. The proposed system is composed of two main steps. First, the targets of the tweet to be analizad are extracted. Afterwards, the system identifies the polarities of the tweet towards each extracted target. We have evaluated the effectiveness of the proposed model on a benchmark dataset from Twitter. The experiments show that our proposed system outperforms the state-of-the-art methods for target-dependent sentiment analysis.