On The Spatiotemporal Burstiness of Terms

TL;DR

This paper introduces a novel approach to simultaneously analyze the spatiotemporal burstiness of terms in document streams and applies this to improve search engine relevance by identifying influential events with strong spatiotemporal signals.

Contribution

It is the first to jointly measure and track spatiotemporal term burstiness and leverages this for enhanced document search relevance.

Findings

Effective detection of spatiotemporal bursts in real and synthetic datasets

Improved ranking of documents discussing influential events

Demonstrated efficiency and scalability of the proposed methods

Abstract

Thousands of documents are made available to the users via the web on a daily basis. One of the most extensively studied problems in the context of such document streams is burst identification. Given a term t, a burst is generally exhibited when an unusually high frequency is observed for t. While spatial and temporal burstiness have been studied individually in the past, our work is the first to simultaneously track and measure spatiotemporal term burstiness. In addition, we use the mined burstiness information toward an efficient document-search engine: given a user's query of terms, our engine returns a ranked list of documents discussing influential events with a strong spatiotemporal impact. We demonstrate the efficiency of our methods with an extensive experimental evaluation on real and synthetic datasets.