Experimental Ratings for AI-Assisted Animation Using a Structured Kano-AHP Framework

Twenty practitioners produced anime-style clips under two workflows, yielding 20 final clips. Five experts and 60 audience members rated the clips on five quality dimensions, including narrative coherence and visual style consistency. The dataset likely contains the resulting ratings and was published by Jiayi Wang on figshare in 2026.

Use Cases

• Compare the effectiveness of structured versus baseline AI media workflows based on expert and audience ratings.
• Analyze the relationship between workflow structure and perceived quality dimensions like emotional expressiveness and motion smoothness.
• Benchmark inter-rater reliability for evaluating AI-generated short-form animation.
• Investigate the impact of requirement definition and criterion weighting on final AI-generated outputs.

Strengths

• Includes ratings from both domain experts (5) and audience members (60), providing multiple perspectives.
• Reports specific effect sizes (Hedges' g ranging from 0.74 to 1.32) and inter-rater reliability metrics (ICC(2,k) from 0.70 to 0.86).
• Compares two distinct production workflows (baseline and structured package) in a controlled experiment.

Limitations

• Row count and column-level documentation are absent; field semantics must be inferred after download.
• The dataset is very small (19.3 KB), indicating limited scope.
• The structured package is evaluated as a whole, so the separate effects of the Kano model and AHP cannot be disentangled.

Provenance

Source

figshare

Collection Method

Experimental study where 20 practitioners produced clips under controlled conditions, which were then rated by experts and audience members.

Freshness

Last updated 2026-05-28 06:17:28; freshness should be verified.