SPIDER

SecureChain and Interactive Elements

Say the user inputs the question "What are the known security issues with qt versions 6.3.2 and later?"

• The system recognizes that this is a question that needs to be answered with data from the "SecureChain" module and then proposes a SPARQL query to the user.
• The system also provides a plain English interpretation of that SPARQL.
• Upon the user's acceptance, the system executes the query.
• The results, specifically the software and version along with specific vulnerabilities, are displayed as a table. Additionally, the system recognized that a software dependency graph might be useful.
• Hence an interactive, explorable graph is displayed as well. Each node is a software package, an edge indicates a dependency. Clicking on a node explores that node's dependencies.

SUD-OKN and follow up queries

The system also supports multi-turn queries. In this example we successively build up a query by first pulling a specific state into view (North Dakota) and then asking for more detailed information about this state, in this case about businesses supplying the mining industry.

SOC-KG and multiple data views

Multiturn queries can also target retrieved values. In this example about Soil Organic Carbon (SOC) the user asks, "What is the average change in SOC stock (kg C ha-1 yr-1) in 20-40 cm soils?" This is a rather involved query and it takes several seconds until the system generates the data from the underlying database.

We see the user then requests the data to be shown as a grouped bar plot. The system is aware of the retrieved data, its dimensionality, how many distinct values are in each dimension/axis and the data type (string, integer, float, etc.). Based on that, the system then generates a grouped bar plot, adhering to basic design principles.

CollabNext and Similarities

In this example the user provides this input: "There is an environmental emergency at the Gulf of Mexico (coordinates are 29°45′46″N 95°22′59″W). I need to assemble a team of experts geographically close who have scientific expertise in algae so that we can provide a rapid response to an emergency."

• After identifying that this is a query for CollabNext, the system recognized that this involves a field "algae". In CollabNext, however, concepts and topics are fixed terms and there is no guarantee that the user provided field exists in the underlying database. Hence, a RAG lookup first retrieves the most similar existing concept "Brown algae".
• Then, a SPARQL query is again presented to the user and upon acceptance executed. Finally, the latitude and longitude values provided by the user are used to rank the SPARQL output by distance.

WEN-OKN and external sites

Some Theme 1 teams provide their own visualizations or advanced analysis functionalities. If our system detects that a query would best be answered by such a site, it directs the user to that site.