ECOLE: a look-ahead editor of controlled language

ECOLE: A Look-ahead Editor for a Controlled Language Rolf Schwitter, Anna Ljungberg, David Hood Centre for Language Technology Macquarie University Sydney, NSW 2109, Australia {schwitt|anna|dhood}@ics.mq.edu.au Abstract This paper presents ECOLE, a look-ahead text editor that supports authors writing seemingly informal specifications in PENG, a computer-processable controlled natural language. ECOLE communicates via a socket interface with the controlled language processor of the PENG system. After each word form entered the look- ahead editor displays appropriate look- ahead categories. These syntactic hints tell the author what kind of word or syn- tactic structure can follow the current in- put string and reduce thereby the cogni- tive burden to learn and remember the controlled language. While the author types the text word by word and adds un- known content words on the fly to the lex- icon, a discourse representation structure and a paraphrase is built up dynamically for the text in a completely compositional manner. The arising specification can be checked automatically for consistency and informativity with the help of third-party reasoning services. 1 Introduction Controlled natural languages are - more or less - well-defined subsets of natural languages that have been restricted with respect to their grammar and lexicon. Grammatical restrictions result in less complex and less ambiguous texts. Lexical restrict- ions reduce the size of the vocabulary and the meaning of the lexical entries for a particular ap- plication domain. In general, these restrictions im- prove the readability and the processability of a document. However, different tasks need different types of controlled languages. For example, con- trolled languages for technical documentation re- quire good readability (AECMA, 2001), while con- trolled languages for machine translation focus on processability and therefore have other restrictions (Mitamura and Nyberg, 2001). Controlled natural languages not only make life a lot easier with tech- nical documents and machine translation, but they can also improve the whole knowledge acquisition process for any kind of intelligent system (Sowa, 2002). It is well known that writing documents in a controlled natural language can be a slow and pain- ful process, since it is hard to write documents that have to comply with the rules of a controlled lan- guage (Goyvaerts, 1996; Huijsen, 1998). One of the deciding factors for the acceptability of a con- trolled language is the availability of tools that check the compliance with the definition of a con- trolled language and support the writing process (Wojcik and Hoard, 1996). Without such tools that help the author to stick to the defined vocabulary and to the grammar rules, it is nearly impossible to produce informative and consistent documents. To guarantee the painless and efficient use of our controlled natural language PENG (Process- able ENGlish) (Schwitter, 2002), we have designed and implemented ECOLE, a look-ahead text editor that helps authors to write precise and seemingly informal specifications in controlled natural lan- guage. ECOLE guides the writing process and guarantees well-formed syntactic structures that can be translated deterministically into first-order logic via discourse representation structures (Kamp and Reyle, 1993). …

Content truncated to 500 words