Augmenting faceted browsing of the Isis bibliography of the history of science with small multiple time series visualizations

Abstract

Bibliographic data can be defined as content of a library catalog, union catalog or abstract and indexing service which connects users to the information they need. A bibliographic record helps users find and access the resources they need. Many bibliographic records have key words or indexes that can be used to search these records. These records are often contained in bibliographic databases, or historically, on paper.

IsisCB is one such bibliographic database which started off on paper like most historical bibliographies. The website we are working on, IsisCB Explore, is a user interface that allows its users to access the large amounts of bibliographic data present in the IsisCB database. The main focus of this thesis is improving the usability of the IsisCB Explore user interface by adding small multiple time series visualizations.

The way the user interface is designed right now, it takes a lot of time and effort for a user to able to perform tasks such as finding specific information, filtering information, and seeing and navigating thorough the data. All of these types of tasks allow users to find patterns in the data available on IsisCB. Most users of IsisCB Explore use the interface for analyses of these types. Currently, the user needs to go through different steps and pages to construct complex queries using a series of mouse clicks that map to these tasks. This makes the IsisCB Explore user interface complicated to use, especially on a regular basis or for someone with less familiarity. Through our work we want to make this process easier and more user friendly.

Visualization is one of the most common ways of exploring and analyzing data, especially when we have large amounts of data. By adding visualizations to the IsisCB user interface, we hope to make it possible for the user to analyze different subsets of the bibliographic data much faster and in a less error-prone way. This will result in improved user satisfaction and productivity for existing IsisCB users. By incorporating the technique of small multiple time series visualizations in the IsisCB Explore interface, we aim to improve a user's ability to find patterns from the bibliographic entries in the database over time by displaying them graphically.

Towards demonstrating this thesis, we utilized the raw data available on IsisCB Explore and we worked with experts on designing and implementing small multiple visualizations. A discovery user study was performed to understand the needs of a regular IsisCB user, and analyze our design against those needs. The focus of the user study was to help define the design space of the data to visualize and visual representations to potentially add. We wanted to see how interested and needed these visualizations are to our users in relation to the design space and time-oriented data from our discovery study.

From the process that we followed and from the results of our discovery user study, we conclude that small multiple visualizations provide an interactive way of understanding data and enable IsisCB Explore users to see different kinds of patterns in the IsisCB data that sometimes are not apparent by looking at data in textual. From these patterns, IsisCB users can gain meaningful insights to further their everyday work.