State-of-the-Art and Challenges in Timeline
                  Summarization

                                 Katja Markert

                      Institute of Computational Linguistics
                             University of Heidelberg
                               Heidelberg, Germany



Abstract
Timeline Summarization (TLS) creates an overview of long-running events via
dated daily summaries for the most important dates and is essential to keep
track of a flood of information on, for example, crises data. TLS differs from
standard single and multi-document summarization in the importance of date
selection, interdependencies between summaries of different dates, the lack of
large-scale human training data and a very low compression rate, i.e very short
summaries in comparison to the number of corpus documents. In this talk, I
will discuss the impact these properties have on (i) optimization algorithms
and objective functions for timeline summarization algorithms (ii) evaluation of
timeline summarization outputs and (iii) the dependence of TLS on information
retrieval components.


Short Biography

Katja Markert is Chair of Computational Linguistics at Heidelberg University
(Germany) since 2016. After her PhD at Freiburg University (Germany) in 1999,
she worked as Postdoc and Emmy Noether-Fellow at The University of Edin-
burgh, UK, as well as Lecturer and Reader at the University of Leeds, UK,
during that time also having longer research stays at the University of Hannover
as well as the Heidelberg Institute of Theoretical Studies. She works both in more
theoretical computational linguistic areas, such as the automatic recognition of
anaphora and figurative language, as well as application areas such as senti-
ment analysis and summarization. She also is co-director of the Leibniz-Science-
Campus ”Empirical Linguistics and Computational Language Modeling”, which
concentrates on NLP models for German, in a joint project between Heidelberg
University and the Institute for the German Language, Mannheim.




  Copyright c 2020 for this paper by its authors. Use permitted under Creative Com-
  mons License Attribution 4.0 International (CC BY 4.0).