Readability and Style Metrics for Human and AI-Generated Academic Abstracts

150 human-written and 150 AI-generated abstracts from linguistics and computer science journals were analyzed for readability and writing style. Yumei Zou created this dataset, which includes metrics from a computational Readability Scoring System and expert evaluations. The dataset was last updated in April 2026.

Use Cases

• Compare readability scores across eight metrics between human and AI-generated text.
• Analyze discipline-specific differences in writing style metrics for linguistics and computer science abstracts.
• Train classifiers to detect AI-generated content using features like lexical diversity and complexity scores.
• Benchmark AI text generation models against human writing on metrics from the Readability Scoring System.

Strengths

• 300 total text samples (150 human, 150 AI) provide a balanced comparison.
• Analysis covers two distinct academic disciplines: linguistics and computer science.
• Data includes both computational metrics and expert evaluation labels.

Limitations

• The dataset is very small at 5.5 KB, limiting statistical power for complex modeling.
• Sample is restricted to abstracts from high-impact journals in only two fields, reducing generalizability.
• Row count and specific column names are unknown, complicating initial assessment.

Provenance

Source

Figshare, authored by Yumei Zou.

Collection Method

Abstracts were sourced from high-impact journal articles, with AI counterparts generated from the same corpus. Metrics were computed using a Readability Scoring System.

Time Range

null

Freshness

Last updated in April 2026.

Geography

null