Toward the Accumulation of Experience Records: A Comparison between Re:commend-demo Narratives and EDC-form Experience Descriptions
Ryosei Kawaguchi / Kwansei Gakuin University
Shin Hirayama / Kwansei Gakuin University
Haruhiro Katayose / Kwansei Gakuin University
In the Entertainment Computing research domain, description frameworks for sharing experiences in a retrospectively referable form remain an ongoing concern. This study compares Re:commend-demo experience-evaluation narratives with LLM-generated Entertainment Design Catalog (EDC)–style experience descriptions derived from content descriptions. Using 22 Re:commend-demo cases, we extracted ten categories of experiential viewpoints and qualitatively examined their correspondence with LLM outputs through category distributions and representative examples. Our results suggest that LLM-based descriptions can help explicate implicit assumptions and design elements, improving reusability, but may also introduce paper-driven inferences that diverge from actual experiences. Moreover, usefulness judgments grounded in social/industrial implications and on-site sensibilities are difficult to formalize, indicating a remaining role uniquely served by Re:commend-demo narratives.
読み込み中...
情報処理学会エンタテインメントコンピューティング研究会で 実施されたRe:commend-demo(2023, 2024年度)の語りからどのように主張が抽出されたか、 そして、そこで表彰された作品に対してどのようなEDCの体験記述が付与されたかを示すスプレッドシートを示す。
スプレッドシートでは、各々の語りとEDCの体験記述がどの観点から記述しているかのカテゴリの分類も同時に示している。