With this software we offer the possibility to simulate the childish process of learning the meaning of words. This implementation is based on an algorithm published by Afra Alishahi [1] which models the learning process by iteratively redefining the probability distribution for a word of mapping it to a visually perceived object (image). Although a variety of approaches has been published in literature, the scientific community still lacks a well designed framework for comparing different algorithms.
Therefore, we aim at providing a software with an easy to use and minimalistic interface which lets the researcher monitor the behaviour of a learning algorithm. To this end, the interface features various views containing both general and word specific information. The former can be viewed as the percentage of words learned over time. To further increase the user's focus on important words of the input file, words can be marked. Consequently, additional information on these words, like the progression of comprehension score, will be shown. Attributable to the iterative nature of the computation, the navigation is realised by a standard compliant navigation bar.
Subsequently, the user can save the generated output to a compliant XML file for viewing the results at a later time point or optional postprocessing steps.
Since this software will be subject to continuous developement it is especially designed with respect to extensibility. This means, that the architectural design offers the possibility to implement and integrate new algorithms, stemmers, or log-formats.
[1] Fazly, Alishahi, and Stevenson (2008). A probabilistic incremental model of word learning in the presence of referential uncertainty. Proceedings of the 30th Annual Conference of the Cognitive Science Society, Washington, D.C.