Typical approaches to XML authoring view a XML document as a mixture o1' structure (tile tags) and surt'acc (text between the tags). We advocalc a radical appr(>ach where the surface disappears from lhc XML document altogether to be Dandled exclusively by rendering mech- anisns. This move is based on the view that lhc author's choices when authoring XML documents are best seen as language-neutral semantic decisions, that the strucltll 'O Cttll then be viewed as inletlingual content, tllct that the textual OUll)Ut should be derived lom this content by languagc-spccilic realization mechanisms. thus assimilating XMI. authoring lo Multilingual l)ocuncnt Authoring. However; standard XMI. tools have imporlant limitations when used IBr such a purpose: (1) they al'e weak at propagttting semantic dependencies belwecn dil'lrcnt parts of the structure, and, (2) current XMI rendering tools arc ill-suited for handling the grammatical combil atiol of lcxttal units. We present two related proposals for overcoming these limitations: one ((;1:) oriinatin in lhc tradition of mathematical proof cctitcrs and conslructivc type llleory, the other (l(J), a specialiT. atio off l)clinilc Clause G'allmlt's strongly inspired by GF.

Abstract not found

We describe techniques based on natural language generation which allow a user to author a document in controlled language for multiple natural languages. The author is expected to be an expert in the application domain but not in the controlled language or in more than one of the supported natural languages.

In reconstructive surgery, multiple interventions during one surgical episode are common. Each intervention must be explained, its intended and potential consequences articulated, and informed consent of the patient secured. Although the pre-surgical encounter between the patient and the surgeon is the opportunity to accomplish this, it

Document normalization is an interactive process that transforms raw legacy documents into semantically well-formed and linguistically controlled documents with the same communicative intention content. A paradigm for content analysis has been implemented to select candidate semantic representations of the communicative content of an input document. This implementation reuses the formal content specification of a multilingual controlled authoring system. As a consequence, a candidate semantic representation can not only be associated with a text in the language of the input document, but also in all the languages supported by the system. This paper presents how multilingual versions of an input legacy document can be obtained interactively with a proposed implementation, and discusses the advantages and limitations of this kind of normalizing translation. 1

This paper introduces document nor- malization, and addresses the issue of whether controlled document authoring systems can be used in a reverse mode to normalize legacy documents. A paradigm for deep content analysis using such a system is proposed, and an architecture for a document normaliza- tion system is described.

A paradigm for the deep content analysis of documents in restricted domains is proposed, along with an implementation technique relying on the emergent field of interactive natural language generation.