Time Series Analysis of Constraint Effects on Information Search Behavior
Takayuki Arita (AY 2015)
Today, the development of Web search systems has made it possible for anyone to obtain a wide variety of information. On the other hand, there has been a gradual increase in the number of studies that attempt to improve information retrieval behavior by setting restrictions in the search environment to positively influence users. In the previous studies, it was clarified that setting restrictions affects users' attitudes, behaviors, and outcomes. At the same time, we conducted time series analysis of users' behaviors and outcomes, and found that the behaviors and outcomes change over the course of the work. However, time series analysis of users' awareness has not been conducted, and it has not been clarified whether users' awareness changes during the work. In addition, existing studies have compared the effects on users between different types of constraints, but there is no study that compares the strength of the same type of constraints that can be expressed quantitatively, and it has not been clarified how the effects on users change depending on the strength of the constraints.
Therefore, in this study, we conducted a time-series analysis including the aspect of awareness and an investigation of the effects of quantitatively expressible constraints. Based on the obtained findings, the purpose of this research is to contribute to the improvement of information retrieval behavior by determining the timing and appropriate strength of constraints to support users. In this study, we conducted a human subject experiment on a total of 30 students at the University of Tsukuba. In the experiment, we prepared five conditions with different strengths of constraints on the number of characters entered in the search query and the number of result documents displayed, and asked all participants to perform the task five times for 10 minutes in each condition, once in each condition. Based on the log data and questionnaires collected during the task, we analyzed the effects on users' attitudes, behaviors, and outcomes. The results of the time series analysis of users' attitudes showed that the influence of the constraints on their attitudes changed little during the task, although a little was observed in the early phase of the task. It is also clear that the stronger the constraint, the more it affects the user's awareness and behavior. It is also clear that users can achieve the same level of results in the strongly constrained condition as in the weakly constrained condition by changing their search strategies and behaviors.
Our future goal is to establish a method for users' search strategies and behaviors under strong constraints, and to contribute to the improvement of information retrieval behaviors under normal conditions and to the improvement of information literacy in educational settings.
(Translated by DeepL)