Online community forums, increasingly popular over the past decade, face challenges in maintaining quality due to the sheer volume of user questions. Manual moderation is impractical, and traditional methods relying on handcrafted features or community feedback are often inadequate. This study introduces a deep learning approach to assess question quality at creation time. Evaluated on the StackOverflow dataset, which includes 60,000 questions categorized by quality, the proposed model achieves a high F1 score of 0.92, demonstrating its effectiveness in real-time quality monitoring.