Subject: lexical semantic tagging

lexical semantic tagging a special issue of the journal of natural language engineering guest editors : marc light and martha palmer call for papers note : deadline for submissions is november 1st , 1998 computing semantic representations is crucial for many applications of natural language processing . currently , semantic lexicons and sets of semantic composition rules are hand-crafted by the designers of the nlp system . the difficulty of building such hand-crafted semantic knowledge bases has limited the field of nlp to applications that can be contained within well-defined subdomains . the likely escape from this limitation will come from the use of automated or semi-automated methods of lexical acquisition . however , the field has yet to develop a clear consensus on a computational lexicon that could provide a springboard for such methods . one of the most controversial areas has to do with polysemy : what constitutes a clear separation into senses of a word , and how can they be computationally characterized and distinguished ? the answer to this question is critical to breaking the bottleneck of broad coverage semantic representation computation . a first step towards finding an answer lies in acquiring annotated corpora that will facilitate the use of empirical methods . the topic of this special issue is tagging word tokens in corpora with lexical semantic information . a concrete example of such tagging would be specifying the sense of an ambiguous word , like { \ em bank } , being used in a particular sentence in a corpus . lexical semantic tagging is critical to extending reliable co-reference for information extraction tasks , to widening the scope of responses to information retrieval queries , to more robust dialogue understanding , and to machine translation . in machine translation in particular , it is often necessary to disambiguate a polysemous source word in order to translate it correctly , since the target language may have distinct lexical items for different senses . we solicit articles which either : ( i ) discuss the characteristics of information to be tagged and how human tagging ( hand-tagging ) can be performed so as to maximize accuracy . such articles should concentrate on the specification of the task and also deal with issues in providing sufficient quantities of accurate and diverse hand-tagged data for the automatic approaches . ( ii ) discuss attempts to develop automated methods and their operation and performance . these articles should describe actual running systems and their design and implementation . we are particularly interested in discussions of lexical semantic tagging methods which are part of actual applications . format : in the interest of providing more coverage , we will be considering papers that are somewhat shorter than traditional journal papers . this would allow us to accept a maximum of 9 articles given an average length of 10 pages . schedule : november 1st , 1998 : submission deadline february 15th , 1999 : notification of acceptance please send submissions to : martha palmer institute for research in cognitive science 400a , 3401 walnut street / 6228 university of pennsylvania philadlephia , pa 19104 telephone : ( 215 ) 898-0361 fax no . : ( 215 ) 573-9247 e-mail : mpalmer @ cis . upenn . edu
