<p>User reviews are a vital resource for mobile app development, offering insights for quality improvement and feature enhancement. However, the sheer volume and diverse nature of reviews often lead to conflicting feedback, posing challenges for developers in the requirements elicitation process. This paper addresses the problem of detecting and resolving conflicts within app user reviews through a hybrid evaluation methodology combining quantitative benchmarking and qualitative practitioner assessment. We propose a semantic rule-based framework that leverages knowledge representation and linguistic analysis to automatically identify various types of conflicts. The framework is built upon a comprehensive taxonomy of conflict types derived from extensive analysis of 60,975 app reviews from three distinct sources. Comparative benchmarking against state-of-the-art transformer models (RoBERTa, LLaMA-2, DistilBERT) demonstrates that the proposed framework achieves superior precision (0.89 vs. 0.77–0.81) while maintaining competitive recall (0.86). Component extraction accuracy exceeds 85% across all semantic elements, with object identification reaching 96.2%. A practitioner study with five industry experts validates the actionability of system-generated conflict resolutions, yielding mean correctness ratings of 4.42/5.0 (CVI = 0.88). The findings underscore the potential of domain-specialist approaches over semantic generalists for requirements engineering tasks, providing developers with interpretable, actionable insights to refine development practices and enhance app quality through more effective management of contradictory user feedback.</p>