LINEAR | Linear Notation Interface John Esch1 , Maurice Pagnucco2 and Michel Wermelinger3 1 2
Paramax Systems Inc. (A Unisys Company), P.O. Box 64525, MS U1N28, St. Paul, MN 55164, USA. email:
[email protected]
Knowledge Systems Group, Basser Department of Computer Science, Madsen Building, F09, University of Sydney, Sydney, NSW 2006, AUSTRALIA. 3
email:
[email protected]
Departamento de Informatica, Universidade Nova de Lisboa, 2825 Monte da Caparica, PORTUGAL. email:
[email protected]
1 Introduction The purpose of the LINEAR subgroup is to coordinate the development of a standard linear form grammar for Conceptual Graphs and to develop related tools. The main aim of this paper is to present a linear form grammar for conceptual graphs which we hope will form the basis for future standardization eorts. We also suggest some possible extensions to this grammar and contrast it with the grammar used by PEIRCE. In the following section we present a linear form grammar based on the work of Esch [2] while section 3 discusses some suggested extensions to the linear form grammar. Section 4 presents the linear form grammar adopted by the current version of PEIRCE (version 0.1) and discusses some of the more signi cant dierences between this grammar and that of section 2. In appendix A we include Sowa's comments on possible future directions with a discussion on integrating the CG linear form grammar with that of KIF[3], the Knowledge Interchange Format developed by the Interlingua Committee of the DARPA-sponsored Knowledge Sharing Eort.
2 Linear Form Grammar | Current Status This section contains extended BNF productions de ning a grammar for Conceptual Graph Linear Forms and some Extended Linear Forms. It is based on [2]. The underlying character set is ASCII4 and the case of letters is only signi cant within strings (production 260). However, actual implementations may provide for user-de nable options, like typelabels in uppercase and variables in lowercase. The user should also be able to choose whether the Extended or the Contracted Form (see section 3) is to be used on output. Furthermore, in production 356 parsers may let the arc before an r-node be optional, as well as the last comma before the period terminating the graph. 4
However, for a true mapping between Natural Language and CGs the international ISO character set should be used.
2
2.1 Meta-Language productions Number Name Production 100 Terminal Terminal symbols are written between quotes, e.g. \0". 110 Non-Terminal Non-Terminal symbols are denoted by italics, e.g. concept . 120 Alternatives Alternatives are separated by a vertical bar j, e.g. \0" j \1" means a 0 or a 1. 130 Groupings Groupings are contained in curly braces f and g, e.g. fA B C g means A followed by a B followed by a C , and fA j B j C g means either an A, or a B or a C . 150 Options Options are contained in square brackets [ and ], e.g. [rlink ] means an optional rlink and [\?>" j \