Sidath R. Liyanage and Chandima H. de Silva
Making accurate strategic decisions is vital to the outcome of a game of cricket. There are a lot of
actions and activities that take place with every delivery of a game. The actions or activities that are
observed and recorded are a fraction of the whole picture. The unrecorded data can be used to
identify hidden patterns in the data to support strategic decision making.
The proposed system has been designed and implemented to present a new approach to capture
ball-by-ball and other related data of a cricket match. Data Mining and Expert System techniques
are the main methodologies used in this system development. Unconventional data such as the
type of bowling, speed of the bowl, line of the bowl, length of the bowl, type of the shot offered by
the batsman etc. are recorded in order to improve the dependability and accuracy of data. It
provides a way of analyzing and making strategic decisions from the vast amount of data available.
Initially traditional knowledge from the domain experts was captured. Additionally it uses
data mining techniques to generate previously unknown relationships among the collected data.
Then the selected knowledge is stored in a knowledge base in the form of rules, which provides
expert advice to the strategic decision maker. The front-end of the prototype has been developed
using JBuilder and Java. Data mining was carried out using Oracle Data Mining embedded in
Oracle 9i and the Java Expert System Shell (JESS) fabricated the backend of the artifact.
The prototype developed has been tested and evaluated using a substantial amount of data. However,
more data should be entered in order to demonstrate the full functionality of the data mining
tool. A user evaluation was carried out to assess the prototype and to identify its shortcomings.
Over 60% of the evaluators were satisfied with the overall functionality of the prototype. Furthermore,
the importance of a computer based tool to support strategic decision making in the game of
cricket has been confirmed by studying the domain as well as by the positive remarks received
from the evaluators.