After a careful study & analysis of few association rule algorithms, Rosa Meo's method[11] was found to be suitable for Intension mining environment. The algorithm was implemented by Uma B. on Pentium-II machine, with 128 MB RAM & Linux operating system[3]. The programming language used was C++. The proposed association rule algorithm was successfully integrated into the Intension Mining System.
A synthetic database was generated using datagen package developed by Quest, IBM Almaden Research Center. These synthetic transactions mimic the transaction in a retailing environment. The name of the database was taken as ``test'' and the database schema of ``test'' database was specified as follows :
create table out
(
0 boolean not null
1 boolean not null
2 boolean not null
3 boolean not null
4 boolean not null
5 boolean not null
6 boolean not null
7 boolean not null
8 boolean not null
..........
99 boolean not null
)
The Knowledge Discovery Schema to mine association rules was specified as follows using the KDSE.
| Schema_Name is kds01 |
|---|
| Accumulate Association Rules |
| Use database test |
| related to * |
| from out |
| use mining Rosa Meo |
After that the KDSC was used to compile the above Knowledge Discovery Schema and the metadata was created successfully.
When the mining was done, we could see considerable amount of efficiency. The efficiency increases as the database grows in size which is evident from the results shown in the table below.
Input - A synthetic database ``test'' with
no_of_trans - The number of transactions added to the database.
total_trans - The total number of transactions present in the database.
Taccumulation - Time taken in milliseconds by accumulation phase of the intension mining system.
Tintensional - Time taken in milliseconds to mine the entire database using Rosa-Meo's algorithm in the intension mining system.
Tdirect - Time taken in milliseconds to mine by Rosa Meo method when it is applied directly on entire database.
| no_of_trans | total_trans | Taccumulation(msec) | Tintensional(msec) | Tdirect(msec) |
|---|---|---|---|---|
| 10,000 | 10,000 | 550 | 150 | 880 |
| 10,000 | 20,000 | 770 | 390 | 1970 |
| 10,000 | 30,000 | 1090 | 460 | 3050 |
| 10,000 | 40,000 | 1240 | 570 | 3850 |
| 10,000 | 50,000 | 1550 | 800 | 5140 |
| 10,000 | 60,000 | 1870 | 920 | 5820 |
| 10,000 | 70,000 | 2000 | 1040 | 7170 |
| 10,000 | 80,000 | 2300 | 1480 | 8340 |
| 10,000 | 90,000 | 2350 | 1590 | 9530 |
| 10,000 | 100,000 | 2670 | 1790 | 10,740 |
