Text and Speech to SQL System for Enterprise Data Access using NLP

Enterprise data contain lot of useful rich information, but not everyone in an enterprise has technical abilities for access and manipulation of database. This problem makes it difficult for organization to work and coordinate within itself. So, there is need for simple methods to access company database. Current methods of Text to SQL and Speech to SQL that do so do not have specialized features for enterprise data access. So, we propose this project for specifically enterprise data access. This architecture and components like domain specific representations, dynamic schema encoding, multi-level semantic validation and reinforcementdriven refinement. Together, these components help the system better understand user intent, generate SQL queries that match database. All this is integrated with intent understanding which makes it easier for users to access data contextually. This architecture composed components natural language processing and Dual stage error correction and to continuously improve user interaction which reduce human efforts to learn technical skills. After the finding the system can accomplish result accuracy from 85%- 90% per cent. This allows business users to ask questions in a more natural way and get most useful insight without needing strong technical skills.