Julien Gosme

The structure of unseen trigrams and its application to language models: A first investigation

In a series of preparatory experiments in 4 languages on subsets of the Europarl corpus, we show that a large number of unseen trigrams can be reconstructed by proportional analogy with trigrams having the lowest frequencies. We derive a very simple smoothing scheme from this empirical result and show that it outperforms Good-Turing and Kneser-Ney smoothing schemes on trigrams models in all 11 languages on the common multilingual part of the Europarl corpus, except Finnish.

Estimating the proximity between languages by their commonality in vocabulary structures

This article proposes a possible way of measuring proximity between languages: it consists in measuring the commonality of structures between the vocabularies of two languages. Experiments conducted on a multilingual lexicon of nine European languages acquired from the Acquis communautaire confirmed usual knowledge on the closeness or remoteness of these languages.