Route-Explanation-System
A knowledge-based AI system that not only routes between points in Bloomington but explains why, grounding each recommendation in a formal argumentation framework over real traffic data.
Navigation apps tell you a route, not why.
Given a start and end point in Bloomington, Indiana, this system generates three candidate routes, optimized for speed, ease of driving, or a balanced trade-off, and produces a structured natural-language explanation of why a particular route was chosen.
Built for CSCI-B 551 (Elements of AI) at Indiana University, it deliberately uses no neural network: explicit knowledge representations, real traffic data, case-based reasoning, and formal argumentation justify each recommendation.
Arguments attack each other; grounded semantics decide.
A knowledge base of parameterized rules blends road-stress scores with observed City of Bloomington traffic counts (AADT), turn-difficulty rules, and time-of-day multipliers. A multi-objective router computes three geometrically distinct routes via weighted shortest path with an edge-penalty diversity guarantee.
For each route and dimension (time, road stress, turn complexity, CBR evidence) the system generates pro/con arguments, resolves their attacks with a Dung grounded/preferred/stable semantics solver, then traces the winning argument chain into natural language, complete with a verdict sentence, counterfactual, decisiveness score, and an interactive SVG argument graph.