Unsupervised query segmentation using click data and dictionaries information

Abstract

We describe results of experiments with an unsupervised framework for query segmentation, transforming keyword queries into structured queries. The resulting queries be used to more accurately search product databases, and potentially improve result presentation and query suggestion. The key to developing an accurate and scalable system for task is to train a query segmentation or attribute detection system over labeled data, which be acquired automatically from query and click-through logs. The main contribution of work is a improving method to automatically acquire such training data — resulting in significantly higher segmentation performance, compared to previously reported methods.