Sema is one example of an AI agent that helps you find relevant reports on SDU’s Analysis Portal. Sema does more than a standard keyword search. It is designed to understand the meaning of your query and can therefore return more accurate results more quickly.

How does SEMA work?

Sema is built by SDU Analytics in Python and uses two technologies for its search function: BM25, which matches your words with the content of each report, and semantic search, which understands the meaning of your query and finds reports on the same topic even when different words are used. Sema combines the two methods and assigns an overall score to each report. The most relevant results appear at the top, so you can find the right web reports faster. As an AI agent, Sema evaluates your query, chooses the best search method and explains why it selected the results. You can ask questions in plain English and get clear answers. This provides a better experience than a standard search box because the results are easier to understand and act on.

Sema on SDU’s Analysis Portal

You can find Sema via the red AI icon on the SDU Analysis Portal homepage, in the bottom-right corner. Sema’s direct URL is: https://kvaser.analytics.sdu.dk/sema/

Sema is the first of several AI agents being developed by SDU Analytics. Future agents will focus on specific analytical tasks, such as finding specific information and creating data visualisations for PowerPoint presentations in SDU’s brand colours.