Ruprecht-Karls-Universität Heidelberg
Institut für Computerlinguistik

Bilder vom Neuenheimer Feld, Heidelberg und der Universität Heidelberg

Text-to-text generation: Paraphrasing and Text Simplifcation

Module Description

Course Module Abbreviation Credit Points
BA-2010 AS-CL 8 LP
Master SS-CL, SS-TAC 8 LP
Lecturer Katja Markert
Module Type Hauptseminar
Language English
First Session 25.10.2018
Time and Place Thursday, 14:15-15:45, SR 2, INF 327
Commitment Period tbd.

Prerequisite for Participation

  • good knowledge of statistical methods
  • advanced BA students or MA students

Assessment

  • regular, constructive participation; completion of exercises
  • presentation
  • project or seminar paper

Inhalt

Whereas standard text generation methods start from abstract meaning representations, databases or knowledge bases, in text-to-text generation existing text is   transformed into new text. Important applications for text-to-text generation include to make texts simpler, summarizing texts, adapting text to different audiences via  wording changes or to generate dialogue turns by rephrasing prior ones.

 

There are (at least) four different types of text-to-text generation, namely

  • paraphrasing
  • text simplification (including sentence splitting)
  • sentence compression (and summarization)
  • sentence fusion

This seminar will investigate the first two, paraphrasing and text simplification, with a follow-on seminar in one of the next two semester looking at sentence compression and sentence fusion.
The methods we look at range from symbolic over statistical to neural methods, specifically sequence-rewriting methods.

Module Overview

Agenda

Date Session Materials

Literature

Will be announced at beginning of semester.

» More Materials

zum Seitenanfang