Developed as part of the X-Media project, to visualise the results of search,
K-Views is based on the prefuse visualisation library, and is to be maintained as
an open-source project.

K-Views was extended in the second part of the project to support visual,
ontology-guided hypothesis investigation. In collaboration with the industrial
X-Media consortium member, Rolls-Royce plc, K-Views was evaluated with
target end users, to determine the value it could bring to root cause analysis
during issue resolution. The outcomes of our research have since been adopted
in a Rolls-Royce in-house tool. The results of further research done in the
second stage of the project are to be introduced as new features in the same tool,
in line with company IT policy.

K-Views is currently being extended to browse linked data, an early prototype
was used as an example in the ESWC 2010 tutorial Essential HCI for the Semantic Web.

For more information or to obtain a working copy email "aba-sah" followed by "@dcs" and then ".shef.ac.uk"
License: LGPL, uses prefuse

Relevant Publications

• Dadzie, A.-S., Rowe, M., Petrelli, D. (2011). Hide the Stack: Toward Usable Linked Data, Proceedings (Part I), The Semantic Web: Research and Applications, the 8th Extended Semantic Web Conference (ESWC 2011), pp.93-107. [bibtex] [presentation]
• Dadzie, A.-S. (2010). Abstract & Poster: Visual, Ontology-Guided Hypothesis Analysis. In Proceedings (Poster Track), ESWC 2010.
• Dadzie, A.-S. & Petrelli, D. (2009). Extended Abstract & Poster: Visual Knowledge Exploration and Discovery from Different Points of View. In Proceedings of the IEEE Symposium on Visual Analytics Science and Technology (VAST 2009), pp.227-228.
• Dadzie, A.-S., Lanfranchi, V., Petrelli, D. (2009). Seeing is Believing: Linking Data With Knowledge. In the Special issue of the Information Visualization Journal on Human-Centered Information Visualization, 8:3, pp.197-211.

back to top

Semantic Network

• a visual representation of the results of a semantic search using a knowledge cloud - ideally, the network structure should be mapped to an ontology
• icons and/or text are used to summarise each piece of information of interest (an assertion) in leaf nodes, each of which is linked to a parent node (that represents a concept in the backing ontology, if this exists)
• colour and size of nodes can also be used to encode node properties, e.g.,
  • the strength of the relationships between parent and child nodes
  • the relative strength of assertions that contribute to a specified argument
  • the confidence that the information retrieved represents an assertion being made
• information on provenance (source documents) may also be retrieved from a node of interest.
  
  
• up to four AND queries may be built, to filter/highlight nodes of interest, based on properties attached to nodes (see right-hand pane - click on image for higher resolution).

Related Publications

• Petrelli, D., Dadzie, A.-S. and Lanfranchi, V. (2009). Mediating Between AI and Highly Specialized Users. In AI Magazine 30(4), pp.95-102.

back to top

Tree Graph

• used to display (sub-parts of) ontologies
  • e.g., the snapshot below shows part of an ontology used to describe isses pertaining to bicycle brakes

back to top