Architecture of a Web-based Predictive Editor for Controlled Natural Language Processing

TL;DR

This paper presents the architecture of a web-based predictive editor designed for controlled natural language PENG$^{ASP)$, enabling users to write non-monotonic specifications with real-time guidance and flexible lexical support.

Contribution

It introduces a novel web-based predictive editor architecture that interacts asynchronously with a controlled natural language processor for enhanced specification writing.

Findings

Supports multiple lookahead categories simultaneously

Enables addition of new content words dynamically

Facilitates writing of non-monotonic specifications

Abstract

In this paper, we describe the architecture of a web-based predictive text editor being developed for the controlled natural language PENG$^{ASP)$. This controlled language can be used to write non-monotonic specifications that have the same expressive power as Answer Set Programs. In order to support the writing process of these specifications, the predictive text editor communicates asynchronously with the controlled natural language processor that generates lookahead categories and additional auxiliary information for the author of a specification text. The text editor can display multiple sets of lookahead categories simultaneously for different possible sentence completions, anaphoric expressions, and supports the addition of new content words to the lexicon.