| Most information systems commonly assume that information seeking is discrete. Such an assumption conforms to neither real-life information seeking processes nor the phenomenology perspective: “life is a journey constituted by continuous acquisition of knowledge.” Thus, this study is concerned with the following questions. In what situations could a user search for essentially the same information problem across sessions? How can objects encountered and actions performed previously help the current search episode? What are the implications of the answers to these questions for the design of information systems?; To answer these questions, this research reviews literature to create a four-dimension conceptual model, Multiple Information Seeking Episodes (MISE), explaining why and how successive searches take place in terms problematic situation, information problem, information seeking process, episodes. Eight modes of successive searches are consequently identified and specified with MISE dimensional properties, which serve as the basis for a cognitive task analysis. Potential solutions for each problem specified with the MISE properties become requirements of the proposed system, PERsonalized and Successive Information Seeking Toolkit (PERSIST).; To investigate the validity of MISE and PERSIST requirements, this study focuses on one particular mode of successive searches, prototyping PERSIST on the Web. It postulates that if MISE is accurate and sufficient, it will help derive a set of appropriate system features to support successive searches across sessions. Twenty subjects are recruited to participate in the three-session experiment, using PERSIST to search for information to plan a vacation. Interview questionnaires and search logs are used to collect the data.; The results indicate the following. First, the original MISE model is highly accurate, but less sufficient. The revised MISE model is shifted from the user-centered to the interaction-centered perspective, taking into account factors of searcher, system, search activity, search context, information attainment, and other supplemental activities. Second, PERSIST supports users very well in navigation of the information space, estimation and projection of the search progress, and organization of search thoughts. However, the non-association between search performance and productivity suggests revise the conceptual requirements for PERSIST with more proactive support features in finding useful information objects. |
