Scholarlink

Large language models (LLMs) are reshaping software and transforming software paradigms. However, developing LLMs is both costly and frequently susceptible to trustworthiness issues. In this paper, we analyze whether these issues can be effectively mitigated by applying software engineering principles. In particular, we analyze the value of applying the Waterfall Model’s phases to the LLM engineering process. We believe this study can provide valuable insights and guide the future development of LLMs toward greater cost-efficiency and trustworthiness.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

An Analytical Perspective on Software Engineering for Large Language Models

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Tianlin Li,
  • Qiang Hu,
  • Chong Wang,
  • Jian Zhang,
  • Wei Ma,
  • Aishan Liu,
  • Jingyi Wang,
  • Yang Liu

摘要

Large language models (LLMs) are reshaping software and transforming software paradigms. However, developing LLMs is both costly and frequently susceptible to trustworthiness issues. In this paper, we analyze whether these issues can be effectively mitigated by applying software engineering principles. In particular, we analyze the value of applying the Waterfall Model’s phases to the LLM engineering process. We believe this study can provide valuable insights and guide the future development of LLMs toward greater cost-efficiency and trustworthiness.