Developing a RAG-based chatbot application supporting operational efficiency in the marine domain.

Building a chatbot to assist marine professionals (ship captain, fleet owners) by answering queries on operational processes, technical issues, and providing context-specific information for voyage planning and ship maintenance.

Being a master’s student

List of projects

This project is part of the GASS (Green AI for Sustainable Shipping) initiative, funded by The Green Platform Initiative (https://www.gass-project.com/). Simula collaborates with Navtor (https://www.navtor.com/), a provider of ship operational data, which has supplied specific marine domain business use cases for which we develop tailored solutions. Potential solutions for a wide range of generic use cases have already been explored in the literature. Numerous research articles discuss ship voyage planning and optimizing fuel consumption under varying weather conditions and sea states. These studies present data-driven, physics-based, or hybrid approaches, tested across different business use cases. The documented results highlight both successes and limitations of the methods. This project aims to filter and extract the most relevant information from existing literature, aligned with the provided use cases, and propose actionable next steps in a conversational, user-friendly manner. It lays the foundation for a RAG-based system that combines knowledge retrieval with contextual (business-use case) response generation.

Goals

Prepare knowledge database: We have downloaded several research articles, relevant information needed to be extracted. Remove noise from text, split documents into smaller logically interconnected parts according to defined use cases. Also go through data provided by Navtor (ship logs, reports). Clean and preprocess it for a standardized format.
Create vectorized database
RAG on the prepared data
Generate context aware-response after feeding info. to a language model.

Learning outcome

You will gain knowledge about RAG architecture, converting textual data into vector representations for semantic search, fine-tuning LLMs and gain insights into marine operations (voyage planning, fuel optimization, ship maintenance, and operational workflows).
You will also learn how to analyze technical logs, reports and extract actionable items from it.

Qualifications

Python programming skills
Familiarity with concepts of NLP and RAG
Familiarity with machine learning frameworks

Supervisors

Akriti Sharma

Collaboration partners