Speculative execution of similarity queries: Real-time parameter optimization through visual exploration

The parameters of complex analytical models often have an unpredictable influence on the models' results, rendering parameter tuning a non-intuitive task. By concurrently visualizing both the model and its results, visual analytics tackles this issue, supporting the user in understanding the connection between abstract model parameters and model results. We present a visual analytics system enabling result understanding and model refinement on a ranking-based similarity search algorithm. Our system (1) visualizes the results in a projection view, mapping their pair-wise similarity to screen distance, (2) indicates the influence of model parameters on the results, and (3) implements speculative execution to enable real-time iterative refinement on the time-intensive offline similarity search algorithm.