OWS for Science WebApplication Prototype

The science search engine prototype is a web application developed using Vue for the front end and Python for the back end. 

The science search engine is a vertical search engine for a specific scientific domain, earth observation. The user enters a query as a natural language text, which is fed to a query analyzer that efficiently extracts three different attributes: geospatial, temporal expressions, and keywords.  These attributes are then sent to the Retrieval Engine which searches and retrieves the top hits via three different indices structures.

• Exploratory search: A graph database that links EO publications to STAC API items via keywords
• Specialized Search: STAC Items represent an EO-specialized dataset, which is fetched via API, saved, and then accessed via the search engine.
• Web Search: A traditional index (created via ChatNoir for this version) contained predefined web blogs and pages for EO.

The Dashboard displays the top hits, designed to show different facets that fit different search purposes and, therefore, fit users with different expertise. In addition, the toggle elements (e.g., only one search genre appears at a time) avoid information overload. The \textit{Dashboard} contains two main interface component views: A document list and an interactive map.

This version contains the following updates:

• A new user interface.
• The initial deployment of Query optimization and suggestions using zero-shot prompting.
• The initial deployment of Snippets generation using zero-shot prompting.