Articles

. In this paper, we focus on shell nouns, a class of items involved in the signaling of discourse organisation. These signals have been little studied in Natural Language Processing and within discourse analysis theories. The main goal is to study their participation to discourse organisation, especially when they occur …...

. Our work is in the field of the validation of term candidates occurrences in context. The textual data used in this article comes from the freely available corpus SCIENTEXT. The term candidates are computed by the platform TTC-TermSuite and their occurrences are projected in the texts. The main issue …...

. The robustness of probabilistic parsing generally comes at the expense of grammaticality judgment – the grammaticality of the most probable output parse remaining unknown. Parsers, such as the Stanford or the Reranking ones, can not discriminate between grammatical and ungrammatical probable parses, whether their surface realisations are themselves grammatical …...

. The process of understanding a document uses both visual, logic and discursive information along with the mutual relationships between those levels. Most approaches studying those relationships were conducted in the frame of text generation, where the text visual properties are infered from logical and discursive elements. We chose in …...

. We provide a new weighted parsing algorithm for deterministic context free grammar parsing inspired by LR (Knuth, 1965). The parser is weighted by a discriminative model that allows determinism (Collins, 2002). We show that the discriminative model allows to take advantage of morphological information available in the data, hence …...

. Our work addresses the automatic detection of paraphrastic rephrasing in spoken corpus. The proposed approach is syntagmatic. It is based on paraphrastic rephrasing markers and the specificities of the spoken language. Man- ual annotation performed by two annotators provides fine-grained and multi-dimensional description of the reference data. Automatic method …...

. We present in this article the results of experiments we led concerning disfluencies in the discourse of schizophrenic patients (in remediation). These experiments are part of a larger study dealing with other levels of linguistic analysis, that could eventually help identifying clues leading to the diagnostic of the disease. …...

. Statistical approaches used in machine translation (MT) require the availability of large parallel corpora for the task at hand. The relative scarcity of thes resources makes domain adaptation a central issue in MT. In this paper, a study of thematic adaptation for News texts is presented. All data are …...

. Bilingual lexicons play a vital role in cross-language information retrieval and machine translation. The manual construction of these lexicons is often costly and time consuming. Word alignment techniques are generally used to construct bilingual lexicons from parallel texts. Aligning single words and nominal syntagms from parallel texts is relatively …...

. If neural networks play an increasingly important role in natural language processing, training issues still hinder their dissemination in the community. This paper studies different learning strategies for neural language models (including two new strategies), focusing on the adaptation of the learning rate. Experimental results show the impact of …...

. Vector space models implement the distributional hypothesis relying on the repetition of information oc- curring in the contexts of words to associate. However, these models suffer from a high number of dimensions and data sparseness in the matrix of contextual vectors. This is a major issue with specialized corpora …...

. Morphological disambiguation of Arabic words consists in identifying their appropriate morphological analysis. In this paper, we present three models of morphological disambiguation of non-vocalized Arabic texts based on possibilistic classification. This approach deals with imprecise training and testing datasets, as we learn from untagged texts. We experiment our approach …...

. In this paper, we address the issue of building and improving a distributional thesaurus. We first show that existing tools from the information retrieval domain can be directly used in order to build a thesaurus with state-of-the-art performance. Secondly, we focus more specifically on improving the obtained thesaurus, seen …...

. Démonette is a lexical database whose vertices (lexical entries) and edges (morphological relations between the vertices) are annotated with morpho-semantic information. It results from an original design incorporating two radically different approaches: Morphonette, a resource based on derivational analogies and DériF, an analyzer based on linguistic rules. However, Daemonette …...

. Incremental dialogue is at the heart of current research in the field of dialogue systems. Several archi- tectures and models have been published such as (Allen et al., 2001; Schlangen & Skantze, 2011). This work has made it possible to understand many aspects of incremental dialogue, however, in order …...

. This work is motivated by the will of creating a new part-of-speech annotated corpus in French from an existing one. We propose a general and operational definition of the comparability relation between annotated monolin- gual corpora. We also propose a comparability measure and a procedure to build semi-automatically a …...

. When Part-of-Speech annotated data is scarce, e.g. for under resourced languages, one can turn to cross- lingual transfer and crawled dictionaries to collect partially supervised data. We cast this problem in the framework of ambiguous learning and show how to learn an accurate history-based model. This method is evaluated …...

. This paper presents a knowledge base containing triples involving pairs of verbs associated with seman- tic or discourse relations. The relations in these triples are marked by discourse connectors between two adjacent instances of the verbs in the triple in the large French corpus, frWaC. We detail several measures …...

. G-TAG is a formalism dedicated to text generation. It relies on the Tree Adjoining Grammar (TAG) for- malism and extends it with several specific notions allowing for the construction of a surface form from a conceptual representation. This conceptual representation is independent from the target language. The goal of …...

This paper describes the participation of the SINAI research group in Task 4 of the 2014 edition of the International Work- shop SemEval. This task is concerned with Aspect Based Sentiment Analysis and its goal is to identify the aspects of given target entities and the sentiment ex- pressed towards …...