Ruprecht-Karls-Universität Heidelberg
Institut für Computerlinguistik

Bilder vom Neuenheimer Feld, Heidelberg und der Universität Heidelberg

Neural Text Summarization


Studiengang Modulkürzel Leistungs-
BA-2010 AS-FL 8 LP
BA-2010[100%|75%] CS-CL 6 LP
BA-2010[50%] BS-CL 6 LP
BA-2010[25%] BS-AC, BS-FL 4 LP
Dozenten/-innen Julius Steen
Veranstaltungsart Proseminar / Hauptseminar
Sprache Englisch
Erster Termin 23.10.2019
Zeit und Ort Mittwoch, 14:15-15:45
INF 326 / 27
Commitment-Frist tbd.

Fachliche Voraussetzungen

Master: Expertise comparable to course "Neural Networks: Architectures and Applications for NLP" ( Bachelor: Successful completion of "Statistical Methods for Computational Linguistics"; ideally also "Neural Networks: Architectures and Applications for NLP"


(i) Active Participation
(ii) Presentation
(iii) Second presentation, term paper or implementation project


Automatic text summarization systems generate concise summaries of key information extracted from one or more documents.
As in many other NLP fields, a lot of recent progress in text summarization has been driven by neural methods.
This includes both the traditionally more common extractive summarization, where systems identify relevant sentences in an input document to serve as a summary, as well as abstractive summarization, where the entire summary is generated from scratch.
Especially the latter has profited immensely from the availability of powerful neural seq2seq modeling tools, which have made the free-form generation of text summaries a viable proposition.

In this seminar, we will first briefly study the task of text summarization in general, including common evaluation methods. We will then take a look at the various neural architectures that have been suggested for both abstractive and extractive summarization.
While the course will focus on the task of generic single document summarization, we will also study some of the recent proposals to enable neural multi-document summarization.



Datum Sitzung Materialien


» weitere Kursmaterialien

zum Seitenanfang