The paper addresses the task of prioritizing items in information systems operating under incomplete and changing data. It presents an analysis of existing methods (Analytic Hierarchy Process, Technique for Order of Preference by Similarity to Ideal Solution, heuristics, machine learning) and shows that none of them fully meet the requirements of robustness, adaptability, and interpretability. To address this, the authors propose a new model that automatically evaluates the importance of features, handles missing data, and provides understandable results. The model also adjusts priorities based on task dependencies. A numerical experiment confirms that the model remains stable even with significant data loss. The solution is applicable in help desk systems, project management, digital platforms, and real-time processing environments.