Understanding and executing natural language instructions in a grounded domain is one of the hallmarks of artificial intelligence. In this paper, we focus on instruction understanding in the blocks world domain and investigate the language understand
ing abilities of two top-performing systems for the task. We aim to understand if the test performance of these models indicates an understanding of the spatial domain and of the natural language instructions relative to it, or whether they merely over-fit spurious signals in the dataset. We formulate a set of expectations one might have from an instruction following model and concretely characterize the different dimensions of robustness such a model should possess. Despite decent test performance, we find that state-of-the-art models fall short of these expectations and are extremely brittle. We then propose a learning strategy that involves data augmentation and show through extensive experiments that the proposed learning strategy yields models that are competitive on the original test set while satisfying our expectations much better.
Manipulation-relevant common-sense knowledge is crucial to support action-planning for complex tasks. In particular, instrumentality information of what can be done with certain tools can be used to limit the search space which is growing exponential
ly with the number of viable options. Typical sources for such knowledge, structured common-sense knowledge bases such as ConceptNet or WebChild, provide a limited amount of information which also varies drastically across different domains. Considering the recent success of pre-trained language models such as BERT, we investigate whether common-sense information can directly be extracted from semi-structured text with an acceptable annotation effort. Concretely, we compare the common-sense relations obtained from ConceptNet versus those extracted with BERT from large recipe databases. In this context, we propose a scoring function, based on the WordNet taxonomy to match specific terms to more general ones, enabling a rich evaluation against a set of ground-truth relations.
Due to the prominent place composition holds in teaching French
to non-native speakers, this article examines the effectiveness of
formative evaluation in improving the writing skills of second year
students in the department of French language. I
n addition, by
showing the difference in the scoring results between the students
who followed the formative evaluation and those who didn’t, this
article aims to illustrate the impact of distributing copies of the
grading scale to students during the training period.
In class, the teachers provide constantly instructions, and the learners must
continually perform. But, some of these instructions are not followed ! It is the same of
exercises or examination subjects. The instruction poses several challenges both
in terms
of its formulation or that of his understanding.
Formulate an instruction require a great effort, great care and special skills, because
of the quality of the latter depends in large part the quality of work performed.
This study aimed to examine students’ attitude and achievement in
relation to their use of Moodle. A Mixture qualitative and quantitative
methods were applied in order to investigate the aims of the study.
Questionnaires and interviews were used f
or collecting data from the
students and the lecturer. Students’ achievement was measured by the grades they obtained in the observed module. Students’ attitudes towards WebCT were measured by using a Likert scale. The data was analysed by using SPSS program. The results of this study showed that the students have positive attitude towards Moodle and they appreciated the flexibility of using it anytime and from anyplace. The module leader valued Moodle as a supporting tool to traditional face-to-face learning but not as an effective tool on students’ learning. The results obtained from the interview and the questionnaire and the quantitative data (tracking system data) can be considered new and promising.
