Automatically recognizing the e-learning activities is an important task forimproving the online learning process. Probabilistic graphical models such ashidden Markov models and conditional random fields have been successfully usedin order to identify a Web users activity. For such models, the sequences ofobservation are crucial for training and inference processes. Despite theefficiency of these probabilistic graphical models in segmenting and labelingstochastic sequences, their performance is adversely affected by the imperfectquality of data used for the construction of sequences of observation. In thispaper, a formalism of the possibilistic theory will be used in order to proposea new approach for observation sequences preparation. The eminent contributionof our approach is to evaluate the effect of possibilistic reasoning during thegeneration of observation sequences on the effectiveness of hidden Markovmodels and conditional random fields models. Using a dataset containing 51 realmanipulations related to three types of learners tasks, the preliminaryexperiments demonstrate that the sequences of observation obtained based onpossibilistic reasoning significantly improve the performance of hidden Marvovmodels and conditional random fields models in the automatic recognition of thee-learning activities.
展开▼
