Modeling users' web search behavior and their cognitive styles

Task You recently heard about the Bermuda Triangle mystery, and you are curious and want to know more about it. So you want to search any relevant information (articles, images and videos) about it and what effect it has on the travellers in the region. Table 1: Search Tasks This is a preprint of an article accepted for publication in Journal of the American Society for Information Science and Technology copyright © 2013 (American Society for Information Science and Technology) JOURNAL OF THE AMERICAN SOCIETY FOR INFORMATION SCIENCE AND TECHNOLOGY—•• 2013 DOI: 10.1002/asi 7 Internet access. Although the participants were never stopped while performing their search tasks, it was recommended that they spend between 10 and 15 minutes on each search task. During the Web search experiment, in order to break a hierarchical level of task complexity, the exploratory task of second level complexity was issued first to the participants, followed by the factual and abstract tasks. All participants completed the set of three search tasks in the same order. This study used a multiple data collection instruments, including the cognitive style analysis test, Web search session logs and think-aloud. User’s demographic information was collected using a pre-search questionnaire. A detailed description of data collection can be found in Kinley (2013). Cognitive Style Analysis: Riding’s (1991) Cognitive Style Analysis (CSA) test was used to measure participants’ wholistanalytic and verbal-imagery cognitive styles (Riding & Cheema, 1991). The CSA is a computer presented test. It indicates the position of an individual on each of the fundamental style dimensions by means of a ratio. The ratios typically range from 0.4 through to 4.0 with a central value around 1.0. The two fundamental dimensions of cognitive styles are both continual, but for descriptive convenience, the dimensions may be divided into groups by dividing the population on each dimension into two groups. The cut-off points of the ratios on each dimension are given in Figure 3. Based on the CSA test, participants were classified as wholist or analytic and verbaliser or imager. Think-Aloud Protocols: A highly effective method of reading cognitive thinking of a user is to ask him or her to verbalize what he or she is thinking. This method is known as think-aloud or protocol analysis. It is used as data collection and analysis mainly to understand the thoughts of users while they are performing some assigned tasks (Ericsson & Simon, 1993). In this research, participants’ cognitive thinking was collected through a think-aloud method. They were asked to think-aloud their thinking, actions and emotions as they interact with the Web search engine while completing their search tasks. The data was manually entered into a log file and then transcribed to create observation records. Web Search Session Logs: This study used Web search session logs to examine the interactions between the participants and the search engines. A standard search log file format with the following fields, similar to those of Jansen (2006), was adopted (see Table 2): • User Identification: A unique number used to identify a participant • Date: The date of the interaction • The Time: The duration of the interaction • The URL: The URL of the Web site visited • Search Terms: The query terms as entered by the user Figure 3: CSA Map (Dichotomous Classifications – two groups) User_ID Date Time URL Search Terms 40 03/02/10 14:00 www..google.com.au Bermuda Triangle + effects it has on travellers in the region 40 03/02/10 14:00 www..google.com.au Bermuda Triangle + effects on travellers 40 03/02/10 14:02 www..google.com.au hypoxia and prevention 40 03/02/10 14:03 www..google.com.au hypoxia Table 2: Examples of Web Search Session Logs This is a preprint of an article accepted for publication in Journal of the American Society for Information Science and Technology copyright © 2013 (American Society for Information Science and Technology) JOURNAL OF THE AMERICAN SOCIETY FOR INFORMATION SCIENCE AND TECHNOLOGY—•• 2013 DOI: 10.1002/asi 8 Participants’ Web interactions, including think-aloud and Web search session logs, were captured by using a monitoring program, Camtasia Studio software (TechSmith, 2009). The output of the program is a video record that can be played and replayed at any time for transcription and analysis.