Comparing MMVIS to a timeline for temporal trend analysis of video data

Abstract

Our MultiMedia Visual Information Seeking (MMVIS) environment provides an exploratory visual paradigm for temporal trend analysis. In this paper, we present the results of a user interface study evaluating the utility of MMVIS. We compare MMVIS to a timeline-based approach for analyzing temporal trends in real video data. We evaluate the quantity, complexity and accuracy of temporal trend observations made within each interface, compare the number of positive versus negative trends found, and collect feedback on user satisfaction. Our results show that subjects made interesting and complex observations of temporal trends using either interface. The results also indicate some advantages and biases of each interface, such as 1) timeline subjects make more errors during analysis and 2) timeline subjects are biased against identifying negative trends such as exceptions. At the same time, however, subjects appreciate the familiarity of timelines. Because we designed the MMVIS architecture to provide users with a library of visualizations, we thus include a discussion on enhancing the utility of MMVIS through incorporating a timeline into it in the future.