RESEARCH

Overview

High Utility Itemset Mining (HUIM) is an important knowledge discovery technique in data mining. It aims to discover all itemsets that have high value in the data. Most previous works focussed on finding these itemsets in transactional databases, and did not take into account the spatiotemporal characteritics of an item in the data. Consequently, HUIM fails to discover useful information hidden in very large spatiotemporal data.

We propose a generic model of Spatial High Utility Itemset (SHUI) that may exist in a spatiotemporal database. The significance of the proposed itemsets has been demonstrated with a case study on traffic congestion data.

(Collaboration with NICT, Japan. NICT has provided the traffic congestion and rainfall data.)

Publication:
R. Uday Kiran, P. P. C. Reddy, Koji Zettsu, Masashi Toyoda, Masaru Kitsuregawa, P. Krishna Reddy: Efficient Discovery of Weighted Frequent Neighborhood Itemsets in Very Large Spatiotemporal Databases. IEEE Access 8: 27584-27596 (2020)

R. Uday Kiran, Koji Zettsu, Masashi Toyoda, Philippe Fournier-Viger, P. Krishna Reddy, Masaru Kitsuregawa: Discovering Spatial High Utility Itemsets in Spatiotemporal Databases. SSDBM 2019: 49-60

Rage Uday Kiran, P. P. C. Reddy, Koji Zettsu, Masashi Toyoda, Masaru Kitsuregawa, P. Krishna Reddy: Discovering Spatial Weighted Frequent Itemsets in Spatiotemporal Databases. ICDM Workshops 2019: 987-996