July 2006 Volume 9 Number 3 - CiteSeerX
July 2006 Volume 9 Number 3 - CiteSeerX
July 2006 Volume 9 Number 3 - CiteSeerX
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
To analyze the convergence behavior of the particles, we testify whether the swarm evolves to the same<br />
optimization goal. We propose the information entropy for measuring the similarity convergence among the<br />
particles as follows. Let pij be the binary value of the jth bit for the ith particle, i = 1, 2, …, R, and j = 1, 2, …,<br />
NK, where R is the swarm size. We can calculate probj as the conditional probability that value one happens at<br />
the jth bit given the total number of bits that take value one in the entire swarm as follows.<br />
prob<br />
R<br />
∑i=<br />
∑ ∑<br />
j = R<br />
p<br />
1 ij .<br />
NK<br />
i=<br />
1 h=<br />
1<br />
The particle entropy can be then defined as<br />
Entropy = −<br />
NK<br />
prob log prob .<br />
∑ j=<br />
1<br />
p<br />
ih<br />
j<br />
2<br />
( )<br />
j<br />
The particle entropy is smaller if the probability distributions are denser. As such, the variations of particle<br />
entropy during the swarm evolution measure the convergence about the similarity among all particles. If the<br />
particles are highly similar to one another, the values of the non-zero probj would be high, resulting in denser<br />
probability distributions and less entropy value. This also means the swarm particles reach the consensus about<br />
which test items should be selected for composing the test sheets.<br />
Figure 6 shows the variations of particle entropy as the number of generations increases. It is observed that the<br />
entropy value drops drastically during the first 18 generations since the particles exchange information by<br />
referring to the swarm’s best solution. After this period, the entropy value is relatively fixed due to the good<br />
quality solutions found and the high similarity among the particles, meaning the particles are resorting to the<br />
same high quality solution as the swarm converges.<br />
6. Conclusions and Future work<br />
Figure 6. The particle entropy as the number of generations increases<br />
In this paper, a particle swarm optimization-based approach is proposed to cope with the serial test sheet<br />
composition problems. The algorithm has been embedded in an intelligent tutoring, evaluation and diagnosis<br />
system with large-scale test banks that are accessible to students and instructors through the World-Wide Web.<br />
To evaluate the performance of the proposed algorithm, a series of experiments have been conducted to compare<br />
the execution time and the solution quality of three solution-seeking strategies on twelve item banks.<br />
Experimental results show that serial test sheets with near-optimal average difficulty to a specified target value<br />
can be obtained with reasonable time by employing the novel approach.<br />
For further application, collaborative plans with some local e-learning companies are proceeding, in which the<br />
present approach is used in the testing and assessment of students in elementary school and junior high schools.<br />
13