000K utf8 1100 $c2021 1500 ger 2050 urn:nbn:de:hbz:464-20210507-122901-3 2051 10.17185/duepublico/74398 3000 Zesch, Torsten 3010 Horbach, Andrea 3010 Laarmann-Quante, Ronja 4000 Künstliche Intelligenz in der Bildung$dWie Sprachtechnologie Lehrende unterstützen kann [Zesch, Torsten] 4209 Am Ende jedes Semesters das gleiche Bild: ein riesiger Stapel Klausuren wartet schon seit Wochen darauf benotet zu werden. Für jede Antwort ein ermutigendes und korrigierendes Feedback zu schreiben, würde noch mehr Zeit kosten. Außerdem kursiert die Klausur im Internet. Für nächstes Jahr müssen dann doch mal neue Aufgaben her. 4209 Artificial intelligence, especially methods of natural language process-ing, can be used in educational con-texts in order to support the teaching process. In this article, we focus on three main application types that we are currently working on: we automatically generate exercises for various question types that are hard or tedious to formulate for humans, and assess the difficulty of exercises we created. We automatically score learner answers, be it answers to short answer questions, which are scored based on their content, or longer essays, scored based on lan-guage use and argumentation quality. Finally we give learners feedbackabout their texts, for example by analyzing the spelling errors pro-duced by school children in order to inform them about the orthographic phenomena they still struggle with. 4950 https://doi.org/10.17185/duepublico/74398$xR$3Volltext$534 4950 https://nbn-resolving.org/urn:nbn:de:hbz:464-20210507-122901-3$xR$3Volltext$534 4961 https://duepublico2.uni-due.de/receive/duepublico_mods_00074398 5051 300