The visualization of evolving searches

It is a common misconception that all web searches can be answered with a single query. It is true that when users have a clear idea of what they are searching for, they can specify an accurate and efficient query to the search engine and find pertinent results in the first 10 search results returned. However, studies of search engine usage by Jansen et al. show that, on average, one out of every three users cannot satisfy their information need with a single search, and are forced to perform multiple searches. This might be because they begin with limited information, or only a vague information need, and subsequently must browse their data, moving through a variety of different sources. Bates theorized that as users search and reformulate their queries, they are affected by each new piece of information that they view, giving them new ideas and directions to follow. Their information needs are evolving. Furthermore, she noted that each new query contributed information towards the users final knowledge set, and as a result aided in fulfilling the information need. She postulated that a query is not satisfied by a single final retrieved set, but instead by different pieces of information gathered throughout the search. She called this the berry-picking model (an analogy to picking berries in the forest). However, performing evolving searches can be time-consuming and difficult, and Bates outlined a set of interface issues that needed to be addressed in order to allow users to perform efficient evolving searches. As of yet no research has focused on the development of tools to directly aid users in performing evolving searches. This thesis will investigate the visualization aspects of evolving searches, looking at each of Bates's interface issues and considering different tool designs, techniques and technologies that would prove beneficial to a user performing an evolving search. The outcome of this thesis is the design, implementation and evaluation of a tool for aiding users in performing evolving searches.