<p>Recommending applications to users is a dynamic and challenging task that evolves with time. Traditional recommendation systems often fail to capture real-time changes in user preferences, leading to suboptimal outcomes. To address this, we propose a novel Artificial Neural Network (ANN)-based Recommendation System (ARS) that integrates diverse factors such as user behaviour patterns, location-based information, and recently introduced applications. Unlike existing systems, which focus mainly on static user profiles or historical data, our approach dynamically updates application attributes by analyzing explicit features and hidden insights from user reviews. This enables the ARS to uncover features that may not be immediately apparent and continuously refine the recommendation process. In addition, our system tracks user activities during a session, such as visited apps, liked applications, and platform trends, allowing real-time adaptability to changing preferences. The system regularly updates the application vector by incorporating new trends, ratings, and rankings to ensure timely and accurate suggestions. The performance was evaluated using a dataset from Kaggle containing 2.5 lakh applications across multiple categories, attributes, and reviews with polarity values. The ARS achieved an 85% success rate, outperforming the top-ranked applications in the charts, with accuracy measured using a precision-recall metric.</p>