The first Computational Semantics in Clinical Text (CSCT) workshop is a forum for the exploration and application of medically-oriented computational semantics. Clinical text has unique semantic and pragmatic characteristics -- prototypically describing a physician-patient encounter with entities and events in the real world. Furthermore, large-scale semantic resources (e.g., UMLS Metathesaurus) are well-developed and frequently used, text corpora are increasingly available (e.g., i2b2 NLP Challenge data), and the possibility of tangible medical benefit is of broad interest. These characteristics make the medical domain an ideal proving ground for computational semantics theory and practice.
Submissions Due: Novemeber 16th, 2012 (Fri)
Notifications: December 14th, 2012 (Fri)
Final Version Due: January 4th, 2013 (Fri)
Workshop: March 19th, 2013 (Tue)
Topics of Interest
CSCT invites submissions on a broad range of computational semantics topics within the domain of clinical text, such as representations of meaning, the semantics-pragmatics interface, deep semantic understanding, distributional semantics, semantic ontologies, semantic web and NLP, and semantic resource development.
The clinical focus of submissions may be exhibited in various ways, including but not limited to:
- Domain lexicons for clinical text
- Semantic variability across subdomain, medical institution
- Semantic similarity in biomedical terminologies
- Novel semantic spaces
- Composition in biomedical semantics
- Discourse and pragmatics in clinical narratives
- Cognitive models of participants
- Text summarization of clinical narratives
- Medical knowledge bases, logic, and inference
- Semantic annotation of clinical text
- Language-driven interfaces in clinical care
- Language-related disorders and therapies (e.g., for aphasia)
- Clinical use cases benefiting from semantic analysis
- Integrating clinical semantics into other NLP tasks
Two types of submission are solicited. Both types of papers should be anonymous and should be submitted no later than November 16, 2012.
- Long papers, up to 10 pages.
- Short papers, up to 5 pages.
Both types of papers can be submitted here.
Long papers should describe original research and must not exceed 10 pages (including references). Short papers are typically system or project descriptions, or ongoing research, and must not exceed 5 pages.
Following review, successful long papers will be published as workshop articles in the IWCS Proceedings, then presented at CSCT 2013. Accepted short papers will be published in the conference proceedings, and will have a lightning talk at the conference, followed by a poster/demo session for discussion.
Please follow IWCS's information for authors for the formatting of both long and short papers.
In addition to being published in the IWCS conference proceedings, long papers accepted to CSCT may be slightly extended and published in Biomedical Informatics Insights (BII) as a Special Issue on "Computational Semantics in Clinical Text". BII is a PubMed-indexed journal that allows open access to all articles, and there is a $1,478 fee to publish in this special issue. Selected short papers may also be eligible for journal publication, with extension and revision.
Historically, corpora of clinical text have not been available for research due to concerns about patient confidentiality. However, de-identified annotated corpora are now being developed to overcome this barrier in data access, including cooperative projects such as SHARP, ShARe, THYME, MiPACQ, and i2b2/VA. In particular, 5 years of i2b2/VA NLP Challenge annotated data sets are readily available with the execution of a Data Use Agreement.
Extensive resources are available in the biomedical domain. Biomedical ontologies and terminologies are available via the National Center for Biomedical Ontology and the National Library of Medicine's Unified Medical Language System (UMLS) Metathesaurus.