Proposal for AI-driven Information Retrieval Framework
AI & Machine Learning
Completed

Proposal for AI-driven Information Retrieval Framework

BCMentor: Biswajit Chaki
Team: Satyapir Ghosh, Jigisha Basu
Started: 7/1/2025
Project Overview

The proposed solution introduces an AI-assisted shipment register enhancement tool that integrates seamlessly with existing ERP and export management systems. By applying NLP techniques, the system extracts contextual meaning from item descriptions and related fields, while deep learning models predict and fill in missing or inconsistent data points. The entire process ensures standardization across chemical entries, reduces manual intervention, and increases operational efficiency.

Methodology
  • NLP-based text extraction from shipment registers
  • Deep learning models for intelligent field prediction and completion
  • Integration with existing ERP/export management systems
  • Continuous learning from corrected entries for accuracy improvement
Key Results & Findings
  • Streamlined accreditation workflow with reduced manual work
  • Saves evaluation time and effort for institutions
  • Increases transparency and credibility in the grading process
Challenges & Solutions
  • Inconsistent chemical nomenclature and missing data - Fine-tuned NLP models with domain-specific chemical dictionaries
  • Handling multilingual shipment records - Integrated multi-language text preprocessing pipeline
Future Work
  • Integration with customs/export compliance APIs
  • Expansion to predictive analytics for shipment trends
  • Cloud deployment for global scalability
Project Information
Duration
7/1/2025 - 5/2/2025
Team Members
Satyapir Ghosh, Jigisha Basu
Use Cases
Shipment Data Automation
Export Compliance Checks
Intelligent Field Prediction
ERP Data Integration
Project Milestones
Requirement Analysis100%
Model Design85%
LLM Integration70%
Testing & Validation50%
Deployment30%