Aspect-based sentiment analysis in smart devices: A comprehensive and specialized dataset.

Area of Science:

• Computer Science
• Human-Computer Interaction

Background:

• Understanding user sentiment is vital for smart device development.
• Existing datasets may lack specificity for nuanced aspect analysis.

Purpose of the Study:

• To introduce a novel aspect-based sentiment analysis dataset for smart devices.
• To provide a foundational resource for sentiment analysis research in this domain.

Main Methods:

• Compiled 2370 user reviews from major US e-commerce platforms (Amazon, Target, Walmart).
• Utilized a consultative approach with industry professionals and end-users to identify key sentiment aspects.
• Covered 10 different smart devices and approximately 30 distinct aspects.

Main Results:

• The dataset features a balanced sentiment distribution: 842 positive, 800 negative, and 728 neutral reviews.
• Identified key aspects such as 'Clock,' 'Connectivity,' and 'Sound' through expert and user consultation.
• The dataset serves as valuable training data for machine and deep learning models.

Conclusions:

• This aspect-based sentiment analysis dataset offers comprehensive insights into smart device user experiences.
• It is a foundational resource for advancing sentiment analysis research and improving smart device technology.