Item Selection Algorithm Based on Collaborative Filtering for Item Exposure Control

In computerized adaptive testing, overexposure of items in the bank is a serious problem and might result in item compromise (e.g., Veldkamp et al., 2010). We develop an item selection algorithm that utilizes the entire bank well and reduces overexposure of items. The algorithm is based on collaborative filtering (Su & Khoshgoftaar, 2009) and selects an item in two stages. In the first stage, a set of candidate items whose expected performance well matches the examinee’s current performance is selected. In the second stage, an item that is the approximately matched to the most recently administered item is selected from the candidate set. The expected performance of an examinee on an item is predicted by autoencoders (Goodfellow et al., 2016). Simulation results show that the proposed algorithm outperforms existing item selection algorithms in terms of item exposure while incurring only a small loss in measurement precision.