Master projects in Computer Science and Natural Language Processing
Suitable as either a 30 or 60 ECTS master's project
The projects are done in collaboration with Belgingur ehf., renumerated by a grant, and are suitable as either a 30 or 60 ECTS master's project, depending on the research's scope and depth.
#1 Building predictive models for the energy production of solar-farms
Renewable resources for producing green electricity include windmill- and solar-farms. Although suitable for generating environmentally-friendly energy, such farms are highly dependent on favorable weather conditions, both in the short- and the long-term. Thus, their production capacity can vary from one day (hour) to the next.
This project will use deep-learning neural-network to build predictive models for solar-farms' short-term energy production, based on high-resolution weather forecasts. The forecasts will be made accessible through an online web-portal.
#2 Generating natural-langue-based weather forecasts from structured weather data.
The use of artificial-intelligence-based software systems to monitor and report on various events in real-time autonomously is on the rise. Such service includes writing news about stock-market development, reporting earthquakes in real-time, or even commentating on ongoing sports events.
This project will use natural-langue generation (NLG) techniques, including machine learning, to translate structured weather data (grid-based) into a natural language (in several languages). The task requires choosing self-consistent textual representation from a pool many potential representations and combine them into concise and coherent paragraphs. The natural-language-based weather forecasts will be made accessible in real-time through an online web-portal.
If you are interested in either of the projects, please contact Dr. Yngvi Björnsson in the Department of Computer Science at Reykjavik University (email: yngvi@ru.is).