This section discusses some existing data mining works within the m-business domain.
MobiMine is an experimental mobile data mining system which enables intelligent monitoring of time-critical financial data from a handheld PDA. The system consists of several modules, with some of the modules utilising data mining techniques. The data mining component of the system employs a novel Fourier-analysis-based approach to efficiently represent, visualize, and communicate decision trees over limited bandwidth wireless networks (Kargupta et al., 2002; Kargupta, Sivakumar, & Ghosh, 2002).
MobiMine is a client-server application. The clients running mobile devices likehandheld PDAs and cell phones monitor a stream of financial data coming through the MobiMine server. The MobiMine server and client apply several advanced data mining techniques to offer the user a variety of different tools to monitor the stock market at any time from anywhere . The server collects stock market data from different sources on the Web and processes it on a regular basis. It employs several data mining techniques to sift through the data. MobiMine makes use of a collection of online mining techniques, including several statistical algorithms, clustering, Bayesian nets , and decision trees. The StockConnection module uses online statistical Fourier-spectrum-based decision trees and Bayesian learning techniques for detecting the interaction among the active stocks and the portfolio. The StockNuggets module applies a collection of different online clustering algorithms for identifying interesting stocks that are influenced by the current active stocks.
Computer games are still one of the most demanded applications in the electronic environment. The primary purpose of data mining in games is to identify different patterns of behaviour, structure, or content in order to improve the overall game play. With this data, organisations can pinpoint areas that need improvement to increase player satisfaction. In return, this will increase revenue and reputation. Wireless Internet revenue models are either proportional to money per time spent by each player (e.g., WAP over GSM) or money per byte served to the player (e.g., UMTS, I-Mode, and WAP over GPRS; Tveit & Tveit, 2002).
There are three different main types of data mining approaches referring to wireless games (Tveit & Tveit, 2002):
Game content mining ‚ discovery of patterns in multimedia or textual content in games (e.g., room layout)
Game structure mining ‚ discovery of structural patterns in the form of paths and connections binding the game world together (e.g., hallways between rooms)
Game usage mining ‚ discovery of human and virtual player behaviour patterns