| 摘要 |
<p>The invention relates to the generation of fill-in-the-blank exercises for language learning. Each word (MT) from an initial text (FTI) is associated with a label (ET) containing the grammatical category and at least one morpho-syntactic property of the word, in order to create annotated words (MA). Textual models (PT) which are at least representative of one word (MT) and which contain one label (ET) of the represented word are predefined by a teacher or administrator and prestored (M2) such that they are associated with the recommendations (RC) in accordance with the substitution rules (RG). The labels (ET) in the annotated words and the textual models are compared in order to mark the word (MT) in an annotated word (MA) when the grammatical categories (CG) are identical and the morpho-syntactic properties are compatible in the labels. The marked words (MM) are replaced by blanks (TR) followed by recommendations (RC) which are associated with the models by the rules.</p> |
