We introduce an expectation maximizationtype (EM) algorithm for maximum likelihood optimization of conditional densities. It is applicable to hidden variable models where the distributions are from the exponential family. The algorithm can alternatively be viewed as automatic step size selection for gradient ascent, where the amount of computation is traded off to guarantees that each step increases the likelihood. The tradeoff makes the algorithm computationally more feasible than the earlier conditional EM. The method gives a theoretical basis for extended Baum Welch algorithms used in discriminative hidden Markov models in speech recognition, and compares favourably with the current best method in the experiments.

Abstract. We explore the use of eye movements as a source of implicit relevance feedback information. We construct a controlled information retrieval experiment where the relevance of each text is known, and test usefulness of implicit relevance feedback with it. If perceived relevance of a text can be predicted from eye movements, eye movement signal must contain information on the relevance. The result is that relevance can be predicted to a considerable extent with discriminative hidden Markov models, and clearly better than randomly already with simple linear models of time-averaged data. 1

We introduce a new search strategy, in which the information retrieval (IR) query is inferred from eye movements measured when the user is reading text during an IR task. In training phase, we know the users ’ interest. We learn a predictor which links eye movements related to a term to the role of that term in the query. Assuming this predictor is universal with respect to the users ’ interests, it can also be applied to infer the implicit query when we have no prior knowledge of the users ’ interests. The result of an empirical study is that it is possible to learn the implicit query from a small set of read documents, such that relevance predictions for a large set of unseen documents are ranked significantly better than by random guessing. 1

In this paper we show that it is possible to make good predictions of text relevance, from only features of conscious eye movements during reading. We pay special attention to the order in which the lines of text are read, and compute simple features of this sequence. Artificial neural networks are applied to classify the relevance of the read lines. The use of ensemble techniques creates stable predictions and good generalization abilities. Using these methods we won the first competition of the PASCAL Inferring Relevance from Eye Movement Challenge [1]. 1

movements 1