本文借助真实书评采样作为模型的人本化因素来源，自建可交互的智能体应用并根据多个聚焦维度生成定制化的拟人化评估结果。通过与现有 Gen AI 评估结果的对比，研究发现，在赋予人工智能“评估者”这一拟人化身份的基础上，笔者构建的应用级评估模型所获得译文评价能够兼顾人类读者的阅读偏好，相较传统模型更能彰显独特的人文魅力。该研究旨在借助评估模型创新提升网络文学英译作品的翻译质量，为其出海传播提供语言层面的支持。

Through authentic book reviews as a source of humanism, an interactive AI-based application, capable of generating customized and personified evaluation, is thus produced focusing on multiple dimensions. According to comparative analysis with existing GenAI-based evaluation results, this study demonstrates that by assigning a human-like “evaluator” role to artificial intelligence, the application-centered assessment model developed in this study creates translation evaluations that are aligned to human readers’ preferences, thereby exhibiting distinctive humanistic appeal compared to traditional models. This research aims to improve the translation quality of Internet literature in English and provide linguistic support for its international dissemination.