Discovering core terms for effective short text clustering

This thesis aims to address the current limitations in short texts clustering and provides a systematic framework that includes three novel methods to effectively measure similarity of two short texts, efficiently group short texts, and dynamically cluster short text streams.