An important aspect of Natural Language Generation (NLG) as an area of Artificial Intelligence is the question on the representation formalisms of the knowledge structures which serve as input to a generator. In the past, considerable attention has been given to "deep", language-independent formalisms. Recent generators show a tendency of using XML to encode input structures. The Semantic Web (SW) initiative bears an enormous potential for NLG with respect to the powerfulness and standardization of the input structure descriptions across different approaches to generation. This potential has hardly been explored as yet. The purpose of this workshop is to bring together researchers that are working in the area of Natural Language Generation and Semantic Web.

The Semantic Web (SW) is an emerging technology that wants to bring content into the current Web by associating meta-data to arbitrary web resources, like web pages, pictures, data sheets etc. SW consists of meta-data and links to global ontologies, which serve as structural vocabularies for the interpretation of domain-specific terms. It is widely accepted that Language Technology, especially Information Extraction is a core technology for the extraction and creation of semantic content. LT hence gains increased importance in the SW-community. As long as the human user is in the "Internet Loop", Natural language (NL) will remain to be the core Human-SW communication device. In the future, humans will continue to exchange knowledge via NL documents - with semantically annotated documents serving as the core Human-SW interface. In this context, NLG methods and applications are needed for allowing an effective machine-user content-oriented interaction, as well as a context sensitive production of human-readable and understandable content in form of "natural" NL, e.g., NL-summarization from heterogeneous resources, dynamically created reports, magazines, newspapers, or the personal digital memory. So far, NLG has not gained much attention in the context of the Semantic Web, possibly also due to the very initial state of meta-data creation. However, with the emergence of ontology standards and the development of large-scale annotated resources, NLG from meta-data will receive increased importance. The workshop will be of major interest to researchers interested in Language Technology, NLG, and Semantic Web. Furthermore, we expect that it will attract people working on Information Extraction, Question Answering and other related fields.

Note:
The workshop is open to all members of the AI community, but the number of participants is strictly limited (maximum of 40). Participants are encouraged to register for the main IJCAI conference in addition to the workshop.
Those wishing to attend without making a presentation should email a short (less than one page) statement of interest to nlg-sw-ijcai2005@dfki.de not later than 15th May, 2005.

The program committee welcomes submissions of long papers for full plenary presentation, as well as short papers for presentation during a poster session. All papers must describe original, previously unpublished research. Papers that are being submitted to other conferences or workshops must indicate this explicitly on the title page. Long papers should be no more than 8 pages inclusive of all references and figures, and short papers should be no more than 4 pages inclusive of all references and figures. The programme committee may suggest the acceptance of a long paper as a short paper.

All papers must be submitted in PDF. It is the responsibility of the authors to ensure that the submitted papers print correctly on a variety of printers. If any special fonts are used, they must be included in the submission.

The deadline for the submission of both long and short papers is 15th March 2005. All submissions must be made electronically to nlg-sw-ijcai2005@dfki.de. The style and format to use for papers and abstracts is the same as for IJCAI with the difference that papers need not be anonymised. Please follow the IJCAI formatting instructions. The title page (no separate title page is needed) should include the following information:

• Title
• Authors' names, affiliations, and email addresses
• 3-5 relevant keywords
  
• Abstract (short summary up to 5 lines)