Veronica Dahl

The Third International Workshop on Applications of Natural Language to Information Systems took place during June 26-27, at the Harbour Centre Campus of Simon Fraser University, Vancouver, B.C. Canada. Some of the work presented is based on a logic programming approach, and has been singled out in this report.

Its keynote speakers were:

- Dr. James Hoard (Program Manager, Natural Language Processing, Boeing Information and Support Services, Boeing) who discussed Sharing Meaning in a Global Collaborative Environment: The Crucial Role of Natural Language in Effective Process and Information Management

- Dr. Nick Cercone (University of Regina), who talked about Natural Language/ Database interfaces: New Hope for the Dead or a Procrustean Bed?

An interesting emphasis on Controlled English was observed in several presentations, including Dr. Hoard's. There were also several contributions regarding conceptual modelling ("Conceptual Models as Knowledge Resources for Text Generation, by Frohlich and van de Riet; "SAX: Generating Hypertext from SADT models; "Color-X: Validating Linguistically-based Conceptual Models, by J.F.M. Burg and R.P. van de Riet); and regarding integration: of linguistic tools within a given knowledge based system ("A Linguistic Application to Knowledge Management", by Edith Brousses); of database views ("View Integration as a Way to Build a Semantic Dictionary for a Data Warehouse", by Isabelle Comyn-Wattiau and Elisabeth Metais; of fuzzy concepts transfering from NL to knowledge bases ("Natural Language and Essential Imprecision in Requirements for Information Systems", by A. T. Berztiss).

NL processing for indexing large technical documents was covered in "Objects and Actions: Two Concepts of Major Interest in Information Retrieval in Full-Text Databases", by Viviane Clavier, Christel Froissart and Celine Paganelli. Concrete NL consultable database systems were addressed in "Natural Language Interface to an Object-Relational Database Management System", by M. Bastawala and P. Bhattacharyya, and error recognition was studied in "Error Detection in Several Languages fro an OCR-Generated Multilingual Database, by Nic Gearailt and J.G. Byrne.

Semantic similarity concepts were explored by Ray Richardson("On Applying Large-Scale On-line Linguistic Resources to Information Processing") and by Stephen Green ("Building Hypertext Links in Newspaper Articles using Semantic Similarity"). A language for configuring linguistically oriented tools for other frame spaces was described by Jeroen Hoppenbrouwers ("Generative Specification of a Template-Based Textual Analysis Tool").

The contributions with most LP related content were:

- A logic representation of linguistic patters for Spanish was proposed within the work of A. M. Moreno and R. P. van de Riet, "Justification of the Equivalence between Linguistic and Conceptual Patterns for the Object Model".

- A preliminary version of the QDMT system for translating from English into Thai, developed using Sicstus Prolog by Nick Cercone and Kanlaya Naruedomkul, tests the hypothesis that machine translation can benefit from a multi-phase architecture: quick and dirty machine translation (QDMT), conceptual comparison and repair, and iteration ("Justification of the Equivalence between Linguistic and Conceptual Patterns for the Object Model").

- In Uses of Assumption Grammars (AGs) for Knowledge Based Systems, Veronica Dahl, Paul Tarau, Pablo Accuosto, Stephen Rochefort and Marius Scurtescu argue that the logic grammar formalism AGs, recently shown useful for NL processing problems such as anaphora, free word order and coordination, is also useful for knowledge based systems, and that some of the techniques developed for natural language processing can naturally and usefully be transferred to knowledge based systems applications. Surprisingly, some of these techniques also extend to coordination languages (Linda) with similar results.

- Concept-Based Retrieval using Controlled Natural Language, by Osmar R. Zaiane, Andrew Fall, Stephen Rochefort, Veronica Dahl and Paul Tarau, presents a method for retrieving concepts from web search queries and from candidate documents on the web, to help determine which of these documents are semantically (rather than simply key-word wise) related to the query. This method combines hypothetical reasoning, which is used both for natural language analysis and for concept extraction, and domain-oriented taxonomies of concepts to guide the system's reasoning.

- Dealing with Exceptions in Textual Databases, by Renwei Li, Veronica Dahl, Luis Moniz Pereira and Marius Scurtescu, addresses the semantic issue of exceptions in natural language texts. Extended logic programs are used for meaning representation and inference engine in natural language processing. Deductive parsing is used to generate semantics of texts by using definite clause grammars with focus on meaning of exceptions. The method is especially applicable to incomplete textual databases, given the non- monotonicity of extended logic programs, but it is also generally applicable to the analysis of natural language descriptions in view of conceptual modelling and to natural language interfaces for database querying. It also provides a method for conflict resolution when the conflict arises from the exceptions.

Veronica Dahl

Simon Fraser University Vancouver, B.C. Canada

E-mail: veronica@cs.sfu.ca