Semantic ambiguity in spoken language understanding poses a significant challenge to the accuracy of intent detection and slot filling. This challenge arises due to overlapping word meanings, weak contextual cues, and vague slot boundaries. To address this issue, the present study proposes the Dual-View Intent-Slot Interaction (DVISI) model, which integrates profile information to enhance contextual understanding and disambiguation. Building on this foundation, the model introduces two core components. The Window-Token Intent-Slot Interaction Module (WTIM) is centred on achieving fine-grained semantic alignment between intent and slot features at both window and token levels. This effectively resolves ambiguous expressions by capturing local contextual cues. The Multi-Head Slot-to-Intent Attention Module (MSIA) has been developed to further enhance disambiguation by dynamically modelling the varying contributions of slot features to intent understanding. The experimental findings on the PROSLU dataset demonstrate that DVISI achieves SOTA performance, exhibiting a marked superiority in comparison to extant baselines across the full spectrum of metrics.