A fine-grained labeled dataset for textual sentiment analysis in technical education.

Area of Science:

• Natural Language Processing
• Pattern Recognition
• Machine Learning
• Data Science

Background:

• Sentiment analysis is crucial for understanding user feedback in various domains.
• Existing datasets lack comprehensive coverage of the technical education sector.
• The All India Council for Technical Education (AICTE) oversees a vast network of technical institutions in India.

Purpose of the Study:

• To present a novel, large-scale, and meticulously curated dataset for sentiment analysis specific to technical education.
• To provide a publicly available resource for training and evaluating Natural Language Processing and deep learning models.
• To facilitate nuanced understanding and analysis of feedback within the technical education landscape.

Main Methods:

• Data collection involved an online application distributed to approximately 10,000 technical institutions under AICTE.
• Over 14,000 records were manually entered by representatives from these institutions over one year.
• The dataset was categorized into seven distinct sentiment labels and further classified into 10 modules.

Main Results:

• A dataset comprising 14,272 records was successfully created and curated.
• The dataset is the first of its kind publicly available for the technical education domain.
• It offers a rich resource with multi-class, multi-labeled data suitable for deep learning applications.

Conclusions:

• The developed dataset is a valuable asset for advancing research in sentiment analysis and Natural Language Processing within technical education.
• It enables the development and benchmarking of more robust and accurate machine learning models.
• The dataset's quality and comprehensiveness support diverse applications, including sentiment classification and feedback analysis.