[4D2-GS-12] Human interface, education aid: Fundamental human interface

[4D2-GS-12-02] Parallel Maximum Clique Algorithm for Uniform Test Assembly

〇Kazuma Fuchimoto¹, Maomi Ueno¹ (1. The University of Electro-Communications)

Keywords:e-testing, Uniform test, Item response theory, Maximum clique problem, Integer programming

Recently, the necessity of ''uniform test forms'' for which each form comprises a different set of items, but which still has equivalent measurement accuracy has been emerging. For uniform test assembly, one of the most important issues is how to assemble uniform tests as many as possible. For this purpose, this study proposes a new uniform test assembly: Hybrid Maximum Clique Algorithm with Parallel Integer Programming (HMCAPIP). The first step seeks a maximum clique as large as possible up to a computer limitation. The second step repeats the parallel search of the vertices connected with all the vertices of the current clique using Integer Programming. Our method assembles nearly twice the number of uniform tests of the traditional methods.