Browse Prior Art Database

Pivot language determination system

IP.com Disclosure Number: IPCOM000012014D
Original Publication Date: 2003-Apr-02
Included in the Prior Art Database: 2003-Apr-02
Document File: 3 page(s) / 78K

Publishing Venue

IBM

Abstract

This application describes a means to best determine the translation path from source to target, through a pivot language, or multiple pivot languages, based on cost and estimated semantic degradation (and optionally, estimated turnaround time). Where a number of translation options exist for a given source-translation pair, each can be searched for appropriateness with regard to price/degradation, optionally given a particular content type. A repository is provided where estimated language degradation on a language-pair basis can be provided, so that the total pivot language degradation can be calculated. This can be allied to the cost of translation (on a language-pair basis) to calculate the total cost/degradation. A final repository is also provided where language translation path (eg. English->Japanese->Korean) information is stored and can be updated based on experience. This system allows both approximate determination and tracking of cost/degradation of translation though one or more pivot languages.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 48% of the total text.

Page 1 of 3

Pivot language determination system

Translation through a pivot language is occasionally required, due to the unavailability or expense of translators with direct source to target translation capability. As companies extend into smaller markets, with less widely spoken languages, this will become more of an issue.

     Pivot languages are already used in projects relating to Domino Global Workbench (English is the pivot language used at the following DGW supported site in MIT http://gssd.mit.edu ), but their usefulness is not currently tracked/determined. Note : The term 'Pivot Language' is more commonly used to describe an artificial intermediate language used for machine translation, or in multilingual search engines (Cross Language Information Retrieveal eg. http://www.ilc.cnr.it/pesystem/eurosearch.htm), however it is also used for intermediate 'real' languages, and that is the meaning used here.

     This disclosure describes a means to best determine the translation path from source to target, through a pivot language, or multiple pivot languages, based on cost and estimated semantic degradation (and optionally, estimated turnaround time). Where a number of translation options exist for a given source-translation pair, each can be searched for appropriateness with regard to price/degradation, optionally given a particular content type.

     The basic idea is to provide a repository where estimated language degradation on a language-pair basis can be provided, so that the total pivot language degradation can be calculated. This can be allied to the cost of translation (on a language-pair basis) to calculate the total cost/degradation. Content-type (eg. Legal/Engineering) can be added as to provide more accurate cost/degradation information.

     A final repository is also provided where language translation path (eg. English->Japanese->Korean) information is stored and can be updated based on experience.

     This system allows both approximate determination and tracking of cost/degradation of translation though one or more pivot languages. Defining the translation options

Defining information about a given language pairing translation option

For each option for language-pair translation (eg. English->French using machine translation, or Japanese->Korean using high quality translators with knowledge of the Computer Industry), the following information is submitted (the turnaround time is optional). -

Price per word Degradation Turn Around

20c 8% 2 days

   Different translation options may refer to different translation vendors, different translators, translation qualities, or even machine translation. Defining translation options

Conceptually, a 2 dimensional array is used to define the quality options for any language pairing (the names in the title are purely descriptional)

Page 2 of 3

Content type is a text description that can be used to describe any kind of specific content with which a translator may need familiarity to be able to accurately trans...