New Online Proofreader for Czech

Autoři

HLAVÁČKOVÁ Dana HRABALOVÁ Barbora MACHURA Jakub MASOPUSTOVÁ Markéta MRKÝVKA Vojtěch VALÍČKOVÁ Marie ŽIŽKOVÁ Hana

Druh Článek ve sborníku
Konference Horák, Aleš; Rychlý, Pavel; Rambousek, Adam (eds.): Slavonic Natural Language Processing in the 21st Century
Fakulta / Pracoviště MU

Filozofická fakulta

Citace
Klíčová slova checker; grammar; agreement; error; punctuation; mistake; tool; Czech
Popis This paper focuses on the new web-based language checker – tool currently developed to help users to produce Czech texts with correct grammar and spelling. The paper describes the inner workings of the existing prototype, its modular structure and aims at its upcoming language-checking compounds. Currently, the punctuational module is able to insert nearly two-thirds of all commas into particular types of texts. If a sentence contains a connector, it is easier to find a position where the comma should be placed. However, to detect a boundary between two clauses without the presence of the connector or two members of the complex multiple element represents a harder task. Another module aims at an agreement, especially at two types of agreement: agreeing pre-/post-nominal adjectives and at subject-predicate agreement. The paper also introduces a module dealing with the selected small mistakes in the Czech language. Whereas all modules and the tool itself as well need quality testing data, building of an annotated database of authentic sentences and errors is described as well.
Související projekty: