Semantic Zoom View: A Focus+Context Technique for Visualizing a Document Collection

Candidate: Dustin Dunsmuir
Type: Master of Science (MSc), School of Interactive Arts and Technology
Date: March 11, 2011
Senior Supervisor: Dr. John Dill
Thesis: Download Thesis Document

Abstract

In the field of visual analytics, analysts need overviews of large amounts of data. This becomes a challenge when working with non-numerical data such as document collections. This thesis describes the design and use of a new visualization technique called Semantic Zoom View (SVZ), which provides an interactive overview of a document collection combined with a detailed view of entities contained in the documents (people, places, etc.) and full text of each document. SZV lets analysts easily and quickly see the main topics of a document collection. Any subset of documents can be semantically zoomed to show increasing detail as the zoom level increases, while keeping surrounding documents visible to supply context. This tight integration of focus within context encourages and facilitates the iterative process of finding relevant documents and reading them. An evaluation compares the described technique to an overview+detail technique for finding answers within a document collection.