The task of Visual Question Generation (VQG) is to generate human-like questions relevant to the given image. As VQG is an emerging research field, existing works tend to focus only on resource-rich language such as English due to the availability of datasets. In this paper, we propose the first Bengali Visual Question Generation task and develop a novel transformer-based encoder-decoder architecture that generates questions in Bengali when given an image. We propose multiple variants of models - (i) image-only: baseline model of generating questions from images without additional information, (ii) image-category and image-answer-category: guided VQG where we condition the model to generate questions based on the answer and the category of expected question. These models are trained and evaluated on the translated VQAv2.0 dataset. Our quantitative and qualitative results establish the first state of the art models for VQG task in Bengali and demonstrate that our models are capable of generating grammatically correct and relevant questions. Our quantitative results show that our image-cat model achieves a BLUE-1 score of 33.12 and BLEU-3 score of 7.56 which is the highest of the other two variants. We also perform a human evaluation to assess the quality of the generation tasks. Human evaluation suggests that image-cat model is capable of generating goal-driven and attribute-specific questions and also stays relevant to the corresponding image.

Nos travaux abordent la question de la mesure de la complexité d’un texte vis-à-vis d’une cible de lecteurs, les enfants en âge de lire, au travers de la mise en place d’une chaîne de traitements. Cette chaîne vise à extraire des descripteurs linguistiques, principalement issus de travaux en psycholinguistique et de travaux sur la lisibilité, mobilisables pour appréhender la complexité d’un texte. En l’appliquant sur un corpus de textes de fiction, elle permet d’étudier des corrélations entre certains descripteurs linguistiques et les tranches d’âges associées aux textes par les éditeurs. L’analyse de ces corrélations tend à valider la pertinence de la catégorisation en âges par les éditeurs. Elle justifie ainsi la mobilisation d’un tel corpus pour entraîner à partir des âges éditeurs un modèle de prédiction de l’âge cible d’un texte.

Children have less linguistic skills than adults, which makes it more difficult for them to understand some texts, for instance when browsing the Internet. In this context, we present a novel method which predicts the minimal age from which a text can be understood. This method analyses each sentence of a text using a recurrent neural network, and then aggregates this information to provide the text-level prediction. Different approaches are proposed and compared to baseline models, at sentence and text levels. Experiments are carried out on a corpus of 1, 500 texts and 160K sentences. Our best model, based on LSTMs, outperforms state-of-the-art results and achieves mean absolute errors of 1.86 and 2.28, at sentence and text levels, respectively.

Le système présenté permet la construction automatisée d’une base de connaissances sur des personnes et des organisations à partir d’une collection de documents. Il s’appuie sur de l’apprentissage distant pour l’extraction d’hypothèses de relations entre mentions d’entités qu’il consolide avec des informations orientées graphe.