Coling 2000 Panel

Antonio Sanfilippo

Current Issues in Translation and Knowledge Management Technologies


Antonio Sanfilippo, European Commission,
Luxembourg and Lexeme Inc., USA,


Ralph Grishman, New York University, USA,

Pierre Isabelle, Xerox, France,
Martin Kay, Xerox Parc, USA,
Steve Richardson, Microsoft, USA,
Yorick Wilks, University of Sheffield, UK,
Hans Uszkoreit, DFKI, Germany,


There is a clear sense in which Knowledge Management (KM) is much more than
technology as it strongly relies on value added by people in the form of context,
experience and interpretation to transform data and information into knowledge.
However, technology --- and in particular language technology since KM deals most
frequently with text --- does play a very important role in extending the reach and
enhancing the speed of knowledge transfer, and helping in the codification and
generation of knowledge.

Fuelled by the Internet revolution, the deployment of content technologies on corporate
intranets and the World Wide Web empowers companies and individuals alike to share knowledge and pool resources across geographical and linguistic boundaries. Content
processing, access and delivery technologies are therefore becoming increasingly
instrumental in catalysing the KM enterprise.

The pervasive role of multilingualism favours the ubiquity of cross-lingual enhancements to KM systems. The rising importance of push technologies such as the MT gateways
on the Lotus Domino server by Alis Technologies and Altavista's Babelfish translation
server by Systran are important signs of this rapidly consolidating trend.

Virtual enterprises which integrate business processes and deploy virtual marketplaces represent the core of the new digital economy. Practices such as e-teaming and e-lancing are being increasingly enforced to leverage distributed human resources. The
paramount import of supporting Knowledge Management with language processing and cross-lingual communication technologies has far reaching consequences for these
new marketing and organisational paradigms.


Dating back to the 1949 memo where Warren Weaver argued that computers can
translate language using mathematical techniques, Machine Translation (MT) is one of
the earliest R&D pursuits in the automation of Human Language services which is now
having an increasingly larger market and social impact. Both the necessity to create
and share content across linguistic boundaries and the strong growth of the
localisation industry contribute to this development, with special reference to the
following factors:

Business-to-business e-commerce is expected to soar to $1.3 trillion by 2003 (Jupiter Communications) with a shift of non-US transactions from 14% of the total worldwide
revenues to 37% by 2002 (International Data Corporation)

Volume in Multilingual web sites is currently growing by a factor of ten (Forrester

Human web content translation is growing at 50% a year (Forrester Research)

The worldwide Text Translation market was $10.4 billion in 1998 and is expected to
reach $17.2 billion by 2003 (Allied Business Intelligence).

By the year 2002, non-English speaking users will make up over 50% of the total
online population, and 70% by the year 2004 (International Data Corporation)

There is therefore no doubt that integrating Translation Technologies into KM
processes will reduce the cost of doing business, improve global business
operationsand facilitate active participation in the information society by allowing users
to cope effectively with cross-lingual communication
As general awareness about the market impact of Translation Technologies is on the
rise, the RTD community has started to look outward for consolidation and integration.
Long term research in the traditional Machine Translation paradigm is progressively
petering out, while more effort is placed on product development and integration with
KM applications.

The increasing demand for advanced technologies in support of the new knowledge
economy has caused a major paradigm shift in the development of Translation
Technologies, from a "technology push" to a "market pull" model. This orientation
poses new constraints on research activities in this area and provides novel
opportunities for penetration and development in innovative applications and services in
the face of a rapidly evolving market and users' behaviour.


A company's aim in managing knowledge is to organise the information produced in-house and gathered from the outside for future uses. In the age of globalisation, the
ability to cope with multilinguality is an integral aspect of this process.

For multinational companies, knowledge management processes have always been
deployed in several languages, according to the geographic distribution of human
resources and global market penetration. Multilingualism has also become a reality for
many national companies world-wide, given the established role of English as a lingua
franca in business transactions. Most of all, the ability to operate in different languages is essential for e-commerce where market deployment is virtually unbound and talking
the customer's language is a necessity.

KM technologies interface with knowledge and people. With respect to the knowledge
interface, an important objective is to advance the organisation of knowledge, through
automatic document indexing and classification techniques. In addition, the next
generation of KM systems will have to work with multimedia information, including
images, audio and video.

The human interface is concerned with system interaction issues about usability,
accessibility and interactivity, and the integration of people's knowledge and skills.
Cross-linguality is a crucial factor in the human interface as it extends the reach of KM systems with reference to knowledge access and transfer.

Other important factors are the ability of KM systems to address users' needs
effectively though the employ of methods such as profiling and dynamic ontological
characterisation of people's knowledge and skills. Achievements in the areas of cross-lingual communication, user profiling and knowledge classification will strongly
facilitate the interaction between humans and information systems.


Language is central to Knowledge Management. Technologies in the areas of term
extraction, translation and semantic classification are already provided as KM
components to KM vendors, portal sites and e-commerce players. Developments in the near future are likely to proceed in this direction with progressive integration of various
language tools and technologies --- including spelling, grammar and style checkers,
terminology mining --- with emphasis on exchange formats and APIs for Machine
Translation, Translation Memories, Terminology and Controlled Languages.

As users, developers, integrators and vendors of KM technologies achieve a better
appreciation of the added value of linguistic enhancements, several questions arise:

Which language technologies will be most sought after, taking into account current
and future developments of the Knowledge Economy?

Which feasible developments of such technologies are necessary in the short and
medium terms to pursue an advancement of KM which is able to meet the needs of
both organisations and end-users?

How will KM market requirements influence the orientation of research in language

Are there any future emerging language technologies which may be instrumental in
shaping KM systems in the years to come?

In keeping with these questions, the objective of the panel is to discuss the role of
Translation Technologies in fostering change and development in the area of
Knowledge Management.

back to top

related events
  DFKI Language Technology Lab
German Research Center
for Artificial Intelligence
Language Technology Lab