Molina A.,University of Avignon |
SanJuan E.,University of Avignon |
SanJuan E.,Brain and Language Research Institute |
Torres-Moreno J.-M.,University of Avignon |
And 2 more authors.
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) | Year: 2013
This paper deals with a new strategy to evaluate a Natural Language Processing (NLP) complex task using the Turing test. Automatic summarization based on sentence compression requires to asses informativeness and modify inner sentence structures. This is much more intrinsically related with real rephrasing than plain sentence extraction and ranking paradigm so new evaluation methods are needed. We propose a novel imitation game to evaluate Automatic Summarization by Compression (ASC). Rationale of this Turing-like evaluation could be applied to many other NLP complex tasks like Machine translation or Text Generation. We show that a state of the art ASC system can pass such a test and simulate a human summary in 60% of the cases. © 2013 Springer-Verlag.
Pope S.M.,Georgia State University |
Pope S.M.,Aix - Marseille University |
Meguerditchian A.,Aix - Marseille University |
Meguerditchian A.,Brain and Language Research Institute |
And 4 more authors.
Animal Cognition | Year: 2015
Cognitive set can be both helpful and harmful in problem solving. A large set of similar problems may be solved mechanically by applying a single-solution method. However, efficiency might be sacrificed if a better solution exists and is overlooked. Despite half a century of research on cognitive set, there have been no attempts to investigate whether it occurs in nonhuman species. The current study utilized a nonverbal, computer task to compare cognitive set between 104 humans and 15 baboons (Papio papio). A substantial difference was found between humans’ and baboons’ abilities to break cognitive set. Consistent with previous studies, the majority of humans were highly impaired by set, yet baboons were almost completely unaffected.Analysis of the human data revealed that children (aged 7–10) were significantly better able to break set than adolescents (11–18) and adults (19–68). Both the evolutionary and developmental implications of these findings are discussed. © 2015 Springer-Verlag Berlin Heidelberg
Fagot J.,Aix - Marseille University |
Fagot J.,Brain and Language Research Institute |
Gullstrand J.,Aix - Marseille University |
Gullstrand J.,Brain and Language Research Institute |
And 3 more authors.
American Journal of Primatology | Year: 2014
Fagot and Paleressompoulle [Fagot and Paleressompoulle (2009) Behav Res Methods 41: 396-404] described a new automated learning device for monkeys (ALDM) to test the cognitive functions of nonhuman primates within their social groups. However, the impact of the ALDM procedure on animal well-being needs to be investigated. The present study assessed the consequences of ALDM testing on the behavioral repertoire of Guinea baboons (Papio papio) and their stress levels as inferred from measurements of saliva cortisol. Accessibility to ALDM test computers reduced the number of resting periods as well as the number of stereotypies. Lower cortisol levels were also found during ALDM testing. These findings and others demonstrate that ALDM testing has a positive impact on animal well-being and can be considered as a means for behavioral enrichment in captive primates. Am. J. Primatol. 76:56-64, 2014. © 2013 Wiley Periodicals, Inc.