It is critical that students learn how to retrieve useful information in hypermedia environments, a task that is often especially difficult when it comes to image retrieval, as little text feedback is given that allows them to reformulate keywords they need to use. This situation may make students feel disorientated while attempting image searching. This study thus designed an image navigation tool, location-based hierarchical navigation support (LHINS), which can dynamically construct a compact WordNet-based hierarchy augmented by location. Using this tool, learners can assimilate new information based on their existing knowledge structure, thus avoiding cognitive overload so as to scaffold their metacognitive skills. Sixty-four high school students were invited to take part in an experiment to test the efficacy of the proposed tool compared to a normal keyword-based search (NKBS) system. The experiment evaluated not only the students' task completion time in the NKBS and LHINS groups, but also their keyword reformulation process, in order to determine the differences in their metacognitive skills. The results revealed that the LHINS group tended to complete the tasks faster and develop better metacognitive skills related to keyword reformulation as compared to the NKBS group. This finding suggests that an image search engine, enhanced by a compact hierarchical navigation tool, can help learners develop better search strategies. When examining how learners with different cognitive styles used the tool, the results showed that learner performance depends on cognitive style, as well as the image retrieval system used, and thus a more detailed investigation of the interaction between the tool and cognitive styles was conducted. Based on these results, several suggestions are derived for designing a more powerful image navigation tool.
All Science Journal Classification (ASJC) codes