Sentiment Classification Method Based on Blending of Emoticons and Short Texts

Area of Science:

• Natural Language Processing
• Artificial Intelligence
• Computational Linguistics

Background:

• Short texts, including those with emoticons, are prevalent for information exchange.
• The brevity of short texts poses challenges for accurate sentiment analysis.
• Existing methods struggle to capture the full sentiment due to limited information content.

Purpose of the Study:

• To propose an effective sentiment classification method for short texts.
• To leverage both textual content and emoticons for improved sentiment analysis.
• To enhance the accuracy of sentiment analysis in the context of fast-paced digital communication.

Main Methods:

• Transforming short-text content and emoticons into vector representations.
• Creating a sentence matrix by connecting word vectors and emoticon vectors.
• Utilizing a convolution neural network (CNN) classification model for sentiment analysis.

Main Results:

• The proposed method demonstrates improved accuracy in sentiment analysis.
• Blending emoticons with short-text content enhances analytical performance.
• The convolution neural network effectively classifies sentiment from the combined data.

Conclusions:

• The developed method offers a more accurate approach to short-text sentiment classification.
• Integrating emoticons is crucial for capturing nuanced sentiment in brief messages.
• This research contributes to advancements in natural language understanding for social media and digital communication.