Building and Maintaining Halls of Fame over a Database

TL;DR

This paper presents a method for automatically detecting and ranking significant changes in Hall of Fame rankings derived from database data, with a prototype system evaluated on basketball statistics.

Contribution

It introduces a novel approach to identifying and ranking noteworthy events in database-driven Hall of Fame rankings using SQL queries and change detection techniques.

Findings

Effective detection of ranking changes in real-world data

A prototype system using triggers and middleware

Successful evaluation on basketball statistics dataset

Abstract

Halls of Fame are fascinating constructs. They represent the elite of an often very large amount of entities---persons, companies, products, countries etc. Beyond their practical use as static rankings, changes to them are particularly interesting---for decision making processes, as input to common media or novel narrative science applications, or simply consumed by users. In this work, we aim at detecting events that can be characterized by changes to a Hall of Fame ranking in an automated way. We describe how the schema and data of a database can be used to generate Halls of Fame. In this database scenario, by Hall of Fame we refer to distinguished tuples; entities, whose characteristics set them apart from the majority. We define every Hall of Fame as one specific instance of an SQL query, such that a change in its result is considered a noteworthy event. Identified changes (i.e., events) are ranked using lexicographic tradeoffs over event and query properties and presented to users or fed in higher-level applications. We have implemented a full-fledged prototype system that uses either database triggers or a Java based middleware for event identification. We report on an experimental evaluation using a real-world dataset of basketball statistics.