Call for Papers
Coreference resolution, the task of determining the mentions in a text or dialogue that refer to the same entity in the real world, has been at the core of natural language understanding since the 1960s. Owing in large part to the public availability of several coreference-annotated corpora since the 1990s, such as MUC, ACE, and OntoNotes, significant progress has been made in the development of corpus-based approaches to coreference resolution.
Advances in modeling have outpaced advances in feature engineering for coreference resolution, however. While the development of large lexical databases (e.g., WordNet, Wikipedia, FrameNet, YAGO, and Freebase) and the progress made in corpus-based lexical semantics research in the past 15 years have enabled researchers to automatically extract and employ sophisticated knowledge for coreference resolution, Durrett and Klein (2013) have shown that the addition of shallow semantic features to the morpho-syntactic feature set employed by their state-of-the-art resolver failed to improve its performance. Nevertheless, recent results suggest that the performance of knowledge-lean coreference resolvers is plateauing, and that performance gains beyond the current state of the art will likely come from the incorporation of sophisticated knowledge sources.
To encourage work on advancing the state of the art in coreference resolution using sophisticated knowledge sources, we invite contributions on topics related to knowledge-rich coreference resolution, including but not limited to the following areas:
Employing semantic and world knowledge for coreference resolution. Can state-of-the-art coreference resolvers benefit from new kinds of features that encode semantic and world knowledge? Can such knowledge be induced from raw text, or can it be robustly extracted from large-scale knowledge bases? Do new methods need to be designed to represent such knowledge so that it can be profitably exploited by coreference resolvers?
Leveraging domain resources for domain-specific coreference resolution. What kind of domain resources can benefit domain-specific coreference resolution? Can domain knowledge be reliably learned from raw text? Can we design domain adaptation methods for coreference resolution so that resources for one domain can be profitably reused for another, possibly related, domain?
Training and operational speed of knowledge-rich coreference resolution systems. Can state-of-the-art coreference resolvers operate in an efficient manner given their increasing complexity with respect to system architecture and the knowledge sources they rely on? What learning algorithms need to be developed so that learning-based resolvers can be efficiently trained on coreference corpora that are much larger than existing ones (e.g., OntoNotes)?
We particularly welcome submissions that demonstrate how sophisticated knowledge can be used to improve coreference resolution for less-investigated coreference tasks (e.g., bridging anaphora resolution, event coreference resolution, resolution to abstract entities), as well as submissions that address the challenges involved in the development and application of knowledge-rich approaches to coreference resolution for less-investigated and/or low-resource languages (e.g., issues that could complicate the extraction, induction, and/or use of sophisticated knowledge for coreference resolution in a low-resource setting or in a specific language). For submissions that involve a standard coreference task (e.g., English identity coreference resolution), it is imperative that an empirical comparison be made against the state of the art, possibly with a systematic analysis of what types of errors made by state-of-the-art resolvers are being addressed.
Articles submitted to this special issue must adhere to the Journal Style Guidelines. Style Guide and LaTeX style files can be found here. We encourage authors to keep their submissions below 30 pages.
Manuscripts will be processed via the journal submission system. Please register as an author and select the article type "Special Issue: Knowledge-rich Coreference Resolution". Additional submission details will be announced closer to the deadline.
- Maciej Ogrodniczuk, Institute of Computer Science,
Polish Academy of Sciences
- Vincent Ng, The University of Texas at Dallas
Guest Editorial Board
- Antonio Branco, University of Lisbon, Portugal
- Chen Chen, Apple, USA
- Dan Cristea, A. I. Cuza University of Iaşi, Romania
- Pascal Denis, INRIA, France
- Sobha Lalitha Devi, AU-KBC Research Center, Anna University, Chennai, India
- Greg Durrett, The University of Texas at Austin, USA
- Lars Hellan, Norwegian University of Science and Technology, Norway
- Veronique Hoste, Ghent University, Belgium
- Yufang Hou, IBM, Ireland
- Ryu Iida, National Institute of Information and Communications Technology, Japan
- Sandra Kübler, Indiana University, USA
- Ekaterina Lapshinova-Koltunski, Universität des Saarlandes, Germany
- Emmanuel Lassalle, Citadel, UK
- Sebastian Martschat, Heidelberg University, Germany
- Costanza Navarretta, University of Copenhagen, Denmark
- Anna Nedoluzhko, Charles University in Prague, Czech Republic
- Constantin Orasan, University of Wolverhampton, UK
- Massimo Poesio, University of Essex, UK
- Sameer Pradhan, cemantix.org and Boulder Learning Inc., USA
- Agata Savary, François Rabelais University Tours, France
- Manfred Stede, Universität Potsdam, Germany
- Veselin Stoyanov, Facebook, USA
- Olga Uryupina, University of Trento, Italy
- Yannick Versley, Heidelberg University, Germany
- Bonnie Webber, University of Edinburgh, UK
- Nianwen Xue, Brandeis University, USA
- Bishan Yang, Carnegie Mellon University, USA
- Heike Zinsmeister, Universität Hamburg, Germany