ROBOKOP has a proof-of-concept user interface to facilitate creation and exploration of blackboards, and to showcase the importance of such exploration to practical use of the Reasoner tool. The user interface is web-based and is available at tyrosine.ncats.io:5000.
When you first launch the user interface, you will be greated with a simple launch screen. Simply click through to get started.
In ROBOKOP, a collection of blackboards are stored in a common database. A blackboard is a knowledge graph corresponding to a particular user query. ROBOKOP currently has a database with many of the challenge questions detailed in the NCATS Reasoner challenge. From this initial user interface, we can search or browse existing blackboards, or build a new one.
New blackbaords are created using a custom user interface to define a sequence of steps between specified node types. Nodes can be specified by a specific search term, as a specific type, or as an unspecified number of steps. Once the build process is started, the knowledge graph is constructed in the background and the user is notified when it is finished.
Completed blackboards can be explored by clicking on the entry in the blackboards list. The initial interface shows general information about the original query, the knowledge graph, and the knowledge sources that were used to construct the knowledge graph.
A visualization of the knowledge graph can be found using the tabbed interface on the right.
In ROBOKOP, answers to the user query are extracted and ranked from the knowledge graph in real-time. To start the answer-generation process, click the button. For larger knowledge graphs this process can take a while (see current limitations below), but this process typically takes less than 1 minute.
ROBOKOP ranks answers extracted from the knowledge graph using several factors calculated from the supporting information. A list of the top 50 answers are shown as a list. Each answer can be visualized and explored using the graphical interface. The meta-data found for each edge can be explored and PubMed articles can be opened.
ROBOKOP extracts answers from the knowledge graph in real-time. For large knowledge graphs, this process may be very slow. In these cases, a warning message will be displayed when it is anticipated that answer extraction will take longer than usual.
