Research on Input Characteristics in DeepSeek-Generated English Reading Comprehension Materials for China's National College Entrance Examination
Year: 2025
Type: article
Abstract: Under the digital transformation of educational evaluation in China, AI-assisted assessment has garnered unprecedented attention. This study focuses on the discourse features of English reading comprehension materials generated by DeepSeek for the National College Entrance Examination (Gaokao), employing a hybrid approach that integrates scientific prompt engineering frameworks with official documentation guidelines. The findings reveal that DeepSeek demonstrates limited proficiency in controlling micro-level textual features (e.g., length, lexical difficulty) but excels in managing macro-level features such as thematic relevance and genre alignment. This research contributes to advancing the digital transformation of educational evaluation in China, offering empirical data and methodological innovations for AI-generated assessment materials.
Author: Qiang Liu
Institution: Sichuan International Studies University
Language: English
Subfield: Information Systems
Field: Computer Science
Domain: Physical Sciences
SDG: Quality education
Open Access status: hybrid
OpenAlex ID: W4411124339