Active Mining Project: Overview

Abstract

Active mining is a new direction in the knowledge discovery process for real-world applications handling various kinds of data with actual user need. Our ability to collect data, be it in business, government, science, and perhaps personal, has been increasing at a dramatic rate, which we call “information flood”. However, our ability to analyze and understand massive data lags far behind our ability to collect them. The value of data is no longer in “how much of it we have”. Rather, the value is in how quickly and effectively can the data be reduced, explored, manipulated and managed. For this purpose, Knowledge Discovery and Data mining (KDD) emerges as a technique that extracts implicit, previously unknown, and potentially useful information (or patterns) from data. However, recent extensive studies and real world applications show that the following requirements are indispensable to overcome information flood: (1) identifying and collecting the relevant data from a huge information search space (active information collection), (2) mining useful knowledge from different forms of massive data efficiently and effectively (user-centered active data mining), and (3) promptly reacting to situation changes and giving necessary feedback to both data collection and mining steps (active user reaction). Active mining is proposed as a solution to these requirements, which collectively achieves the various mining need. By “collectively achieving” we mean that the total effect outperforms the simple add-sum effect that each individual effort can bring.