In this thesis, we studied the topic of Search-Based and Supervised Text Generation. Supervised encoder-decoder models require huge aligned datasets to be trained. The necessary data is not yet available for several tasks such as RDF triples verbalization or paraphrase generation. First, we explored the data-to-text task of RDF verbalization. We trained supervised Transformer models on a newly released version of the WebNLG dataset and studied in depth several pre-training strategies to overcome the small size of the aligned corpus. Then, we studied the paraphrase generation task. We have trained Transformer models on aligned corpora to directly compare with the literature model. An important contribution of the thesis was to propose a unif...
Techniques for generating and recognizing paraphrases, i.e., semantically equivalent expressions, pl...
The humanity has long been passionate about creating intellectual machines that can freely communica...
This thesis focuses on the relationships between what is uttered and human-machine spoken dialogue s...
Les modèles supervisés encodeurs-décodeurs nécessitent de grands datasets alignés pour être entraîné...
International audienceA good paraphrase is semantically similar to the original sentence but it must...
Paraphrase Generation is one of the most important and challenging tasks in the field of Natural Lan...
International audienceWe study the automatic generation of syntactic paraphrases using four differen...
This paper presents a paraphrase acquisition method that uncovers and exploits generalities underlyi...
Tutors: Leo Wanner i Simon MilleTreball fi de màster de: Master in Intelligent Interactive SystemsDa...
This paper presents FISKMÖ, a project that focuses on the development of resources and tools for cro...
Paraphrase generation is an important problem in NLP, especially in question answering, information ...
In this paper we investigate the automatic generation of paraphrases by using machine translation te...
Le thème support de cette thèse la génération de paraphrases sur support neuronal. Nos perspectives ...
Abstract — Monolingual text-to-text generation is an emerging research area in Natural Language Proc...
Paraphrasing is rooted in semantics. We show the effectiveness of transformers (Vaswani et al. 2017)...
Techniques for generating and recognizing paraphrases, i.e., semantically equivalent expressions, pl...
The humanity has long been passionate about creating intellectual machines that can freely communica...
This thesis focuses on the relationships between what is uttered and human-machine spoken dialogue s...
Les modèles supervisés encodeurs-décodeurs nécessitent de grands datasets alignés pour être entraîné...
International audienceA good paraphrase is semantically similar to the original sentence but it must...
Paraphrase Generation is one of the most important and challenging tasks in the field of Natural Lan...
International audienceWe study the automatic generation of syntactic paraphrases using four differen...
This paper presents a paraphrase acquisition method that uncovers and exploits generalities underlyi...
Tutors: Leo Wanner i Simon MilleTreball fi de màster de: Master in Intelligent Interactive SystemsDa...
This paper presents FISKMÖ, a project that focuses on the development of resources and tools for cro...
Paraphrase generation is an important problem in NLP, especially in question answering, information ...
In this paper we investigate the automatic generation of paraphrases by using machine translation te...
Le thème support de cette thèse la génération de paraphrases sur support neuronal. Nos perspectives ...
Abstract — Monolingual text-to-text generation is an emerging research area in Natural Language Proc...
Paraphrasing is rooted in semantics. We show the effectiveness of transformers (Vaswani et al. 2017)...
Techniques for generating and recognizing paraphrases, i.e., semantically equivalent expressions, pl...
The humanity has long been passionate about creating intellectual machines that can freely communica...
This thesis focuses on the relationships between what is uttered and human-machine spoken dialogue s...