Numerical Range Facets Partition: Evaluation Metric and Methods

Faceted browsing is a very useful interface component provided in many of today's search engines. It is especially useful when the user has an exploratory information need or have a clear preference for certain attribute values. Existing work has tried to optimize faceted browsing systems in many aspects, but little work has been done on optimizing the partitions of numerical facet ranges (e.g., price ranges of products). In this paper, we introduce the new problem of numerical facet range partition and formally frame it as an optimization problem. To enable quantitative evaluation and reuse of search log data, we propose an evaluation metric based on user's browsing cost when using the suggested ranges for navigation. We further propose and study multiple methods to computationally optimize the partition by leveraging search logs. Experimental results on a two-month search log from a major e-Commerce engine show that learning can significantly improve the performance over baseline.