Second Workshop on Multilingual Surface Realization

Hong Kong, November 3rd, 2019

MSR'19 - A SIGGEN event

    The workshop on multilingual surface realization aims at bringing together people who are interested in surface-oriented Natural Language Generation problems such as word order determination, inflection, functional word determination, paraphrasing, etc. It will accommodate for the presentation of the results of the Surface Realization Shared Task 2019 and of a number of technical papers on the topic. The workshop will be held at EMNLP-IJCNLP'19 in Hong Kong, on November 3rd 2019. It is endorsed by the ACL Special Interest Group on Natural Language Generation (SIGGEN).

Call for papers

    Natural Language Generation (NLG) is in the ascendant both as a stand-alone task (e.g. data-to-text and text-to-text generation) and as a component task in embedding applications (e.g., abstractive summarization, dialogue-based interaction, question answering, etc.). In 2017 alone, three ‘deep’ NLG shared tasks that focused on language generation from abstract semantic representations were organized: WebNLG, SemEval Task 9 , E2E – although for English only. In 2018, the first shared task that focused on multilingual surface realization (SR’18) was organized. However, when compared to NLP areas such as parsing or machine translation, NLG can still not register the same degree of success, lacking comparable methodological progress. Thus, while recent years have seen a paradigm shift in parsing and MT from traditional supervised machine learning techniques to deep learning, with associated substantial improvements in quality, NLG is not quite there yet.
    In parallel to the boost from deep learning techniques, the last few years have seen a push in the annotation of multilingual treebanks with Universal Dependencies (UD), such that now resources for a number of languages are available: Currently, with version 2.3, 129 treebanks covering 76 languages can be downloaded freely. UD Treebanks facilitate the development of applications that work potentially across all of the UD treebank languages in a uniform fashion, which is a big advantage for system developers. As has already been seen in parsing, these treebanks are also a good basis for multilingual shared tasks: a system that has been built for some of the languages may work for most of the other languages as well. But again, NLG is a few steps behind. SR’18 was the first attempt to change this state of affairs. It took place as an ACL 2018 Workshop, with a focus on the presentation of the results of the shared task by the (eight) participating teams. The MSR Workshop is intended as an event with a broader scope, seeking to provide a forum not only for the presentation of the outcome of the second multilingual surface realization task (SR’19), but also for a more general consideration of the role of UD structures and the linkages they afford to the generation and parsing fields. Like its predecessor, SR’19 is designed as a generation challenge from UD treeebanks. The proposed second MSR workshop aims to attract new participants who did not participate in SR’19, and to bring together two communities that have hardly collaborated so far, namely parsing and generation, in that it will address NLG from structures of an abstraction that is targeted by state-of-the-art parsing, such that the challenge to reverse neural network parsing algorithms for generation becomes a plausible research question; and it will explore to what extent the notion of Universal Dependencies, which is about to become the dominating linguistic format across CL applications, and, in particular, in parsing, serves for NLG.
    To complement the presentation of the SR’19 results, MSR-WS solicits contributions on all topics that are related to surface realization in NLG, specifically including and encouraging reversible methods. Cutting edge recent works will be sought that address problems of surface-oriented generation such as grammatical and/or information structure-driven word order determination, inflection, functional word determination, paraphrasing, etc. The presented works are expected to be a clear contribution to the progress in robust multilingual surface generation, i.e., be language-independent or easily portable from one language to another and clearly scalable. The topics of interest include, but are not limited to:

Shared Task

    The second Surface Realization Shared Task is currently open and will run until August 19th, 2019. Details can be found on the SR'19 page.

Important Dates


    We invite long papers (8 pages) and short papers (4 pages). Both long and short papers have unlimited references, and their final versions will be given one additional page (up to 9 respectively 5 pages in the proceedings and unlimited pages for references).
    MSR 2019 uses a double-blind review process. Submissions must conform the official style guidelines, and be in PDF format, using the Softconf START conference management system. The paper submission deadline for both long and short papers is August 19th, 2019 for the workshop, and September 9th, 2019 for the SR'19 system descriptions.
    To encourage inclusiveness and the presentation of speculative and recent work, inclusion in the conference proceedings will be made optional. The author’s preference should be indicated with the final submission.
    Multiple submission policy:
    Note that dual submissions are prohibited among EMNLP-IJCNLP 2019 workshops. In addition, the workshop will not consider any paper that is under review in a journal or another conference at the time of submission.
    Templates, guidelines, other policies:
    Please refer to the EMNLP conference website for new policies for submission, review, and citation, and official style guidelines.


For registration information, please visit https://www.emnlp-ijcnlp2019.org/registration/


    The workshop will consist of technical presentations, a poster session with WS papers and ST systems, the presentation of the shared task results, invited talks and a discussion session.
    Invited talks: We are happy to announce that there will be two invited speakers at the workshop:

8:45 Opening

Invited Talk: Claire Gardent


Oral presentation

The Second Multilingual Surface Realisation Shared Task: Overview and Evaluation Results
Simon Mille, Anja Belz, Bernd Bohnet, Yvette Graham, Leo Wanner
10:30Coffee break





Oral presentations (25 min each)

Learning to Order Graph Elements with Application to Multilingual Surface Realization
Wenchao Du and Alan W Black

DepDist: Surface Realization via RegEx and Dependency-distance Tolerance
William Dyer

BME-UW at SR'19: Surface realization with Interpreted Regular Tree Grammars
Ádám Kovács, Evelin Ács, Judit Ács, Andras Kornai and Gábor Recski

Realizing Universal Dependencies Structures Using a Symbolic Approach
Guy Lapalme
12:40Lunch break

Invited Talk: Maurice Benayoun (a.k.a. Moben) 1


Oral presentation

IMSurReal: IMS at the Surface Realization Shared Task 2019
Xiang Yu, Agnieszka Falenska, Marina Haid, Ngoc Thang Vu and Jonas Kuhn

Poster Session (including break)

Surface Realization Shared Task 2019 (SR'19): The Tilburg University Approach
Thiago Castro Ferreira and Emiel Krahmer

The Concordia NLG Surface Realizer at SR'19
Farhood Farahnak, Laya Rafiee, Leila Kosseim and Thomas Fevens

The OSU/Facebook Realizer for SR'19: Seq2Seq Inflection and Serialized Tree2Tree Linearization
Kartikeya Upasani, David King, Jinfeng Rao, Anusha Balakrishnan and Michael White

Improving Language Generation from Feature-Rich Tree-Structured Data with Relational Graph Convolutional Encoders
Xudong Hong, Ernie Chang and Vera Demberg

Back-Translation as Strategy to Tackle the Lack of Corpus in Natural Language Generation from Semantic Representations
Marco Antonio Sobrevilla Cabezudo, Simon Mille and Thiago Pardo

The DipInfoUniTo Realizer at SR'19: Learning to Rank and Deep Morphology Prediction for Multilingual Surface Realization
Alessandro Mazzei and Valerio Basile

LORIA / Lorraine University at Multilingual Surface Realisation 2019
Anastasia Shimorina and Claire Gardent

The Second Multilingual Surface Realisation Shared Task: Overview and Evaluation Results
Simon Mille, Anja Belz, Bernd Bohnet, Yvette Graham, Leo Wanner
17:00 Panel/Discussions
18:00 Closing


    You can download the proceedings from the ACL Anthology and see the details of the task results and participating systems:

Programme Committee


    Please send us an email at msr.organizers@gmail.com if you have any question.

Organizing committee

Simon Mille TALN Pompeu Fabra University,
Barcelona, Spain
Anya Belz University of Brighton
Brighton, UK
Bernd Bohnet Google Research,
London, UK
Yvette Graham ADAPT Center, Dublin City University,
Dublin, Ireland
Leo Wanner TALN Pompeu Fabra University and ICREA,
Barcelona, Spain


    (1) Science Foundation Ireland (sfi.ie) under the SFI Research Centres Programme co-funded under the European Regional Development Fund, grant number 13/RC/2106 (ADAPT Centre for Digital Content Technology, www.adaptcentre.ie) at Dublin City University;
    (2) the Applied Data Analytics Research & Enterprise Group, University of Brighton, UK; and
    (3) the European Commission under the H2020 via contracts to UPF, with the numbers 825079-STARTS, 779962-RIA, 700475-IA, 7000024-RIA.

tensor tensor tensor tensor

Photo at the top of the page by Steve Huntington on Unsplash.