Ontology development often requires the participation of various col-laborators. Web-‐based ontology editors, such as WebProtégé, have been developed to provide users with collaborative support such as comments and discussions. There is a large body of work concerning ontology visualization techniques; however, less research attention has been placed on providing the necessary support for collaborative on-tology visualization. To explore this research gap, the web-‐based col-laborative ontology visualization tool BioMixer is presented in this paper. In order to assist the collaborative visualization process, Bio-Mixer provides users with sharable workspaces and embeddable visu-alizations that can be seamlessly inserted into external websites.
Department of Computer Science, University of Victoria, Canada
Visualizations can provide essential cognitive support when
trying to make sense of the semantics embedded in
ontologies. Improving cognitive support for ontology
understanding is particularly important in the domain of biomedical
ontologies, as such ontologies are typically large and rely on
collaborative development (Noy et al., 2008). Despite a
large amount of research effort in developing ontology
visualization techniques
include social data analysis websites (such as Many Eyes1
(Viégas et al., 2007)), scientific research projects (such as
National Fusion Collaboratory (Schissel et al., 2004)) and
environmental planning
In an attempt to address the need for collaborative ontology visualization, this paper presents the BioMixer3 tool that allows users to share visualization workspaces and to embed visualizations in websites. The vision for BioMixer is that collaborative ontology visualization will improve ontology authoring activities and foster collaboration across groups.
overview on related work is presented in Section 2. The design, implementation and key features of BioMixer are presented in Section 3. Finally, Section 4 outlines BioMixer’s future research directions.
one of the key challenges in visual analytics (Thomas &
Cook, 2005) and design considerations for such tools have
been recommended (Heer & Agrawala, 2007). Collaborative
visualization can be described as “the intersection of two
major research fields: traditional visualization and computer
supported cooperative working”
Expanding on the place-time matrix, Brodlie et al.
been studied in social data analysis websites such as Many Eyes (Viégas et al., 2007). The goal of Many Eyes is twofold. First, the creation and publication of visualizations can reach out to a larger audience, not just experts. Second, the social potential of web-based visualizations enables discussions among the wider audience. The advantages of collaborative visualization in social data analysis may also hold true for the domain of ontology visualization. As with collaborative support, visualizations will not only serve as a tool for sense-making, but also as a channel to stimulate discussions between users.
This trend of collaborative support is already embraced
by existing ontology editors. For instance, WebProtégé
(Tudorache et al., 2008) aims to support the collaborative
ontology development process by providing an online
environment for users to edit, discuss and annotate ontologies.
visCOntE (Vonrueden & Hampel, 2005) provides support
for searching, creating and editing ontologies among
collaborators. COVE
development tools, little research attention has focused on
providing collaborative support for ontology visualization.
An extensive review of existing ontology visualization
approaches is presented in
BIOMIXER DESIGN, FEATURES &
Inspired by previous design recommendations (Heer &
Agrawala, 2007) and tool requirements
In order to address the needs identified above, BioMixer is being designed to support visualizations in collaborative settings. In particular, BioMixer: • supports social interaction around the visualization.
A user can send an existing visualization workspace to his/her collaborators via email, as well as initiate discussions by adding notes to the visualizations. • engages a wider audience by providing a web-based interface. A user can easily access BioMixer using a web browser and does not need to download or install any software. • supports the publication of visualizations by providing interactive visualization embeds that can be easily inserted into external websites. • supports users with diverse backgrounds and preferences by presenting multiple coordinated views, which aim to engage the audience from different viewpoints.
Figure 1 illustrates an example of using BioMixer for collaborative ontology visualization, exemplifying the features listed above. In this example, the user searches for tissue and is given a list of ontologies that contain this term in the Search view (see top left window in Fig. 1)4. The user then selects the Cell Line Ontology, the RadLex Ontology, the BioTop Ontology and the Gene Regulation Ontology in the search results and subsequently creates Selection 1 in the view frame. By selecting the Graph button under Views, an empty graph view is created in the workspace. The user can then drag Selection 1 and drop this selection onto the graph view. He/she can also add comments by adding a Note view (see bottom right view in Fig. 1). To explore the nodes shown in the graph view, the user can select a node and choose to visualize its associated concepts or mappings. In Fig. 1, four ontologies (color coded) are visualized in the circle layout, the is-a relations are visualized by solid directional lines and the mappings are visualized by grey dashed lines. Visualizing different types of mappings (e.g. exact, close, related, broad and narrow mappings) are not supported in the current graph view, but will be included in a future release. This visualization can also be displayed using other layouts as shown in Fig. 1 (the panel to the right includes
of BioMixer, but is to be included in a future release.
tree, spring, grid layout, etc.). For example, the spring
layout may be appropriate when presenting an overview of
several ontologies and how the mappings relate them to one
another (an example is shown in Fig. 2), however the tree
layout may be more suitable for visualizing the hierarchical
relationships among the nodes (an example is shown in Fig.
3). As pointed out in
In BioMixer, the user can further explore any subset of the current visualization by selecting the nodes he/she is interested in viewing. In the example shown in Fig. 1, the user selects the nodes connected by mapping which subsequently creates Selection 2. By selecting the show mapping nodes checkbox in Nodes, this second selection can be dragged and dropped onto the Timeline view, which will then show the creation dates of these mappings (e.g. May 17th, 2010 in Fig. 1). If the user is only interested in an overview of all the terms used in the ontology irrespective of the relationships between them, a tag cloud can be generated in the Text view.
BioMixer also visualizes mappings between multiple
ontologies. This differs from existing tools, such as Optima
%IsPA Embed%Persistence% % EA Workspace%Persistence% G
%sscceA %taaD gaanM rkoW em sap %ten %ce
V i s u a li z a > o n s %
To share the workspace shown in Fig. 1 with a collaborator, the user can click the Share button and enter the collaborator’s email address. An email containing the URL of the workspace will be sent to the collaborator, who can then load this workspace into his/her browser by simply opening the URL. To publish visualizations online, inline frames are provided to the user, which enable the insertion of visualizations in external websites (currently, users must sign in to use this feature). This feature allows the user to quickly update visualizations on external websites when required. Figure 4 demonstrates these Share and Embed features.
top of the Google Web Toolkit5 (GWT) and the Google App Engine6 (GAE) technologies, and integrates visualization components written in Flash, Java and JavaScript. Figure 5 presents the BioMixer architecture. The three key components are the BioMixer Client, the BioMixer Server and the BioPortal REST Services7 provided by the National Center for Biomedical Ontology8 (NCBO). The BioMixer client is an ontology visualization environment that runs in the user’s browser. It is written in Java and compiled to JavaScript using the GWT. The client currently supports graph, text, timeline and note views; however, additional types of visualization can be easily integrated given the extensible architecture of BioMixer. The client also provides visualization coordination such as synchronized highlighting (brushing), filtering and selections across multiple views. In addition, it supports basic features such as window management and undo/redo. The client accesses the data stored in BioPortal through the NCBO BioPortal REST services. To enable workspace sharing and persistence, the BioMixer client uses services offered by the BioMixer server. The BioMixer
server runs on GAE and provides services that require longterm data storage, email notification and user authentication.
4
BioMixer is an on-going research effort. This paper is among the first attempts at applying collaborative support in the field of ontology visualization. The current implementation of BioMixer focuses on the visualization of ontologies from the biomedical domain. However, the underlying architecture is domain independent and could therefore be applied to visualize ontologies from other domains of interest. Future research in BioMixer includes improving its technical infrastructure as well as conducting rigorous evaluation of its usability through real-world case studies.
and asynchronous collaborative visualization on collaborative biomedical ontology development. In particular, we will continue interacting with different user groups and improve the visualization and collaboration features in BioMixer based on user feedback.
medical Ontology (NCBO), the NHGRI, the NHLBI, the
dez V., Zablith F. (2011), A novel approach to visualizing and navigating ontologies. In Proceedings of the 10th International Conference on the Semantic Web - Volume Part I, 470-486.
Noy N. F., Shah N. H., Whetzel P. L., Dai B., Dorf M., Griffith N., JonBioPortal: ontologies and integrated data resources at the click of a mouse. Nucleic Acids Res. Jul 1;37 (Web Server issue):W170-3. Epub 2009 May 29. PubMed PMID: 19483092; PubMed Central PMCID: PMC2703982. Biomedical Ontologies Collaboratively. American Medical Informatics Association Annual Symposium Proceeding, 520-524. for Visualization using Java RMI. Concurrency and Computation: PracBuilding the US National Fusion Grid: Results from the National Fu245-250. and Development Agenda for Visual Analytics. IEEE Computer Society,
Workshop on OWL: Experiences and Directions. Visualization in CSCW Systems. ICEIS 2005 – Information Systems zation of Owl Class Hierarchies. In Proceedings of the 5th International sis. In Proceedings of the Symposium on Information Visualization, 17.