An evaluation of semantic fisheye views for opportunistic search in an annotated image collection

Abstract

Abstract. Visual interfaces are potentially powerful tools for users to explore a representation of a collection and opportunistically discover information that will guide them toward relevant documents. Semantic fisheye views (SFEVs) are focus + context visualization techniques that manage visual complexity by selectively emphasizing and increasing the detail of information related to the user's focus and deemphasizing or filtering less important information.In this paper we describe a prototype for visualizing an annotated image collection and an experiment to compare the effectiveness of two distinctly different SFEVs for a complex opportunistic search task. The first SFEV calculates relevance based on keyword-content similarity and the second based on conceptual relationships between images derived using WordNet. The results of the experiment suggest that semantic-guided search is significantly more effective than similarity-guided search for discovering and using domain knowledge in a collection.

Concept Expansion Using Semantic Fisheye Views

“…Many of these use geometric distortions to magnify objects near the focus [15]. SFEVs, on the other hand [3], calculate conceptual distance from the focus within one or more data models, and are therefore independent of a particular visual representation [5].…”

“…However, Pirolli et al [19] point out that "squeezing" more information into the display does not necessarily "squeeze" more information into the mind, but that strong information scent cues and focus + context techniques can enable users to navigate and search through information at more than twice the rate of the user of a normal browser. VisAmp uses several emphasis techniques to align the visual weight of objects with their semantic importance in a particular context so that the "most interesting" objects are immediately apparent (i.e., "pop out"), and "less interesting" objects fade to the background (e.g., through dimming, shrinking in size and detail, or filtering) [3]. The relative contrast creates a visual ordering that allows a user to rapidly and opportunistically access the most important contextual information, i.e., visual emphasis corresponds to information "scent".…”

“…The formal evaluation described in [3] investigated differences in exploratory search behavior and sensmaking tasks between VisAmp prototypes using keyword-or concept-based similarity metrics. Users performed significantly better with the semantic metrics, and commented that they strongly appreciated the semantic structure for exploring and making sense of the unstructured collection.…”

“…An important component of this process is "sensemaking," where users construct and refine their mental schemas of the concepts and relationships they encounter in the documents of a collection [2]. In this paper, we describe the implementation details of Semantic Fisheye View (SFEV) [3], a focus + context technique that interactively guides a user's attention over a potentially dense visualization of information to the objects that are the most semantically related to their current focus.…”

“…We described the architecture of SFEV in [5], focusing on the strategies used to support real time display of semantic fisheye views and how interaction can be rapidly and smoothly handled. In a related paper [3], we described an experimental evaluation where users were significantly more effective at sensemaking tasks with an interface that revealed semantically related information rather than one that revealed keyword co-occurrence. The results of the experiment do not discount the usefulness of keyword-similarity or imply that semantic models are more effective in all cases.…”