PENG Processable ENGlish

Writing in PENG

In contrast to other controlled languages, the author of a PENG text does not need to know the restrictions of the controlled natural language explicitly. PENG uses an intelligent authoring tool that communicates with a language processor that generates lookahead information for each word form that the author enters while the specification text is written. During the parsing process an incremental chart parser generates a chart which can be used to harvest the lookahead information for the text editor.

The authoring tool of the PENG system provides a standard mode and a web feed mode. The standard mode can be used to write normal specification texts in controlled natural language. The web feed mode is specially designed to annotate individual web pages with machine-processable information and web sites to which these individual web pages belong to with ontological knowledge.

Figure 1 shows the editor in the standard mode:

Figure 1: Standard mode

Figure 2 shows the editor in the web feed mode with an active summary pane and the FOL representatation in the message field.

Figure 2: Web feed mode (summary pane)

Figure 3 shows the editor in the web feed mode with an active ontology pane and the DRS representation in the message field.

Figure 3: Web feed mode (ontology pane)

Instead of typing an approved word form into the description field of the editor, the author can alternatively select a word form from the currently active look-ahead categories via a context menu as Figure 4 illustrates:

Figure 4: Context-menu with available word forms

Not only can approved word forms be inserted in this way, but also all noun phrases which are accessible in the specification text. Accessible noun phrases occur in the context menu and can be selected from there. Figure 5 shows that after the processing of the two sentences in the description field of Figure 2, the following three noun phrases are available in the context menu:

Figure 5: Context-menu with accessible noun phrases

If the author enters a word that is unknown then the system checks whether the word is misspelled using a set of heuristics. The author can add new content words to the lexicon with the help of a lexical editor that requires minimal linguistic knowledge as Figure 6 shows:

Figure 6: Lexical editor

For each content word the author can define a number of synonyms. Whenever the system detects a synonym during parsing then it replaces the synonym by the main word and indicates this replacement in the paraphrase.

[Back]

---

© Rolf Schwitter, Macquarie University, Australia
Last modified: 28th February 2007