Paper Published in Requirements Engineering

Together with Jane Cleland-Huang, our paper
“Automated Classification of Non-Functional Requirements Twenty Years Later: Why This Isn’t a Solved Problem”
has been published in Requirements Engineering.

DOI: 10.1007/s00766-026-00479-4

Automated Classification of Non-Functional Requirements Twenty Years Later: Why This Isn’t a Solved Problem

This paper revisits “Automated Classification of Non-Functional Requirements,” published in 2007, and reflects on how the problem has evolved over two decades. We review 55 papers and replicate representative classification approaches, finding that modern LLM- and agent-based techniques outperform earlier approaches on the PROMISE dataset. However, better accuracy has not solved the underlying challenge: quality concerns increasingly arise in conversational fragments, implementation discussions, and architectural tradeoffs rather than explicit requirements statements.

We argue that requirements engineering tools should move beyond classifying documents to identifying, surfacing, and preserving architecturally significant quality concerns throughout development. We outline a research agenda covering emerging and missing quality concerns, selective intervention, living documentation of quality decisions, and evaluation beyond classification accuracy. The paper also reflects on dataset lineage, experimental comparability, and benchmark evolution.