Topics: Text, Query, Triage, Dyslexia, Pattern
We hope to add to this body of knowledge by recording the eye gaze patterns of searchers with and without dyslexia as they design queries, triage results, and extract information in this study. The experimenter recounts the backstory of the search task aloud during this period. “You recently watched a show on the Discovery Channel about fish that can dwell so deep in the ocean that they spend much of their lives in darkness,” for example. This piqued your interest in the ocean’s deepest point. “What is the name of the ocean’s deepest point?” After they have completed the search task, the researcher gives them an oral version of the NASA task load index and asks them about their prior domain experience, or how much they know about this issue before they start the task. This is an example of one of the Phase One Tasks, in which we asked them to explain how the ATP rankings are determined. They then moved on to the title of the second search result, clicked it, and were taken to the specific web page. However, we controlled for query formulation in the second and third phases because we were particularly interested in search results triage in the second phase, so we controlled for query formulation by presenting a prepopulated search results page with a fixed query and observing how people choose which parts of the search results page are more relevant to answer the prompt. In that instance, we show them the pre-populated search results page for the fixed query, just as we did in phase two. However, we asked them to open at least three online sites from the search results page, in the hopes that there would be an open overlap in the number of people looking at specific web pages, allowing us to evaluate information extraction behavior. In that instance, we show them the pre-populated search results page for the fixed query, just as we did in phase two. Looking at the search logs, we can see that when compared to the control group, searchers with dyslexia issued more queries and made more errors per inquiry, especially at the phonetic level as opposed to the typographic level. The page outlines of the individual page elements are shown in black, and the different fixations where the participants’ eyes stop to digest information are shown in red. The red marks on the commitment fixation pattern virtually cover the entire body of the page in this example. Text on the left and near the top of the page is read more frequently than text on the right or at the bottom. When we looked at how searchers with dyslexia and those without dyslexia examined a search engine results page, we discovered that searchers with dyslexia usually employed the commitment scanning pattern. The control group, or searchers without dyslexia, on the other hand, uses a F shape pattern, which implies a speedy perusal of the search engine results based on titles and focusing mostly on the top left corner. This could be because they used the commitment fixation pattern to analyze the search engine results page more thoroughly. We plan to investigate the impact of dyslexic searchers using various tools during web searches in the future. We would also like to investigate how different subsets of dyslexia symptoms affect web search in the future.