Sergiu Nisioi

Unsupervised Classification of Translated Texts

In our paper we investigate the possibility to use an unsupervised classifier to automatically distinguish between the translated and original novels of a multilingual writer (Vladimir Nabokov) and to determine whether the authorship of a translated document can be achieved. We employ a rank-based document vector representation using only function words as features. To extract the results, we propose a generalization of Ward’s hierarchical clustering method that is compatible with any similarity metric.

Feature Analysis for Native Language Identification

In this study we investigate the role of different features for the task of native language identification. For this purpose, we compile a learner corpus based on a subset of the EF Cambridge Open Language Database - EFCAMDAT [10] developed at the University of Cambridge in collaboration with EF Education. The features we are taking into consideration include character n-grams, positional token frequencies, part of speech n-grams, function words, shell nouns and a set of annotated errors. Last but not least, we examine whether the essays of English learners that share the same mother tongue can be distinguished based on their country of origin.

Exploring Neural Text Simplification Models.

We present the first attempt at using sequence to sequence neural networks to model text simplification (TS). Unlike the previously proposed automated TS systems, our neural text simplification (NTS) systems are able to simultaneously perform lexical simplification and content reduction. An extensive human evaluation of the output has shown that NTS systems achieve almost perfect grammaticality and meaning preservation of output sentences and higher level of simplification than the state-of-the-art automated TS systems.

On the syllabic structures of Aromanian

In this paper we have investigated the syllabic structures found in Aromanian a Romance language spoken in the Balkans across multiple countries with important communities which spread from Greece to Romania. We have created a dictionary of syllabified words and analyzed a few general quantitative and phonological aspects of the dictionary. Furthermore, we have approached the syllabic complexities, the sonority patterns present in the syllable’s constituents and the degree in which the Sonority Sequencing Principle (SSP) holds for this language. Based on all the information gathered we have devised an automatic syllabification algorithm which has a 99% accuracy on the words in the dictionary. In this way we hope to extend the existing phonological studies on Eastern Romance and to spread and preserve meta-linguistic information on this endangered language.