News & Events

Dialogue Systems Group
Department of Chinese, Translation and Linguistics

Research Seminar

Segmentation, Anchoring, and Extraction for the Spatio-Temporal and Event-Related Annotation of Multilingual Dialogues

Presented by

Prof. Kiyong Lee

Professor Emeritus of Linguistics, Korea University

Date: 8 April 2009, Wednesday
Time: 3:00 – 4:00pm
Venue: B6605 (Lift 3, 6/F, Blue Zone), Academic Building, CityU

Abstract

This talk is an interim report on an on-going joint work on the semantic annotation of multilingual dialogues for spatio-temporal and event-related information.   · First, as a way of motivating the joint work, the current status of ISO standards on language resource management under development will be browsed through the ISOTC Server. The project portal under ISO/TC 37/SC 4 shows the list of 11 projects under current development. All these standards aim at making the best use of current achievements in the area of Computational Linguistics for developing language resources that are needed for language technology. · Second, as an interim report, the rest of the time for presentation will be spent on introducing the working draft on preliminaries to spatio-temporal and event-related annotation of transcribed dialogue text, namely the specification of character segmentation and anchoring of markables for standoff annotation. Each character in UTF8 is identified as a pair of integers that represent segment boundaries. · Third, the final part of today's presentation discusses some ways of extracting relevant information from available encoded corpora such as BNC and ANC. The header part of each file may provide information relevant to dialogue participants and settings, both temporal and spatial. POS-tagging may also provide information on tense and aspect necessary for the proper annotation of events.

Speaker