Keybot - Translation Search Machine
Keybot is a web-based translation search machine operated by Keybot LLC, based in Geneva. It indexes multilingual
websites in over 140 languages and is constantly adding new languages as well. As an application, Keybot helps users
find fitting sample translations online, thereby making it easier to find specialist terms not included in conventional
online dictionaries. Keybot not only works with language pairs, like other services of this type, but also allows its
users to perform searches in any number of language combinations and have the answers displayed beneath each other
in the search results. This makes Keybot useful not just for comparative linguists, but also for translators seeking
to derive new terms from the most commonly used languages in less widespread regional languages.
Features
The search engine offers more than 10,000 language combinations, so it covers many minority languages that are under
pressure from majority languages and are not included by commercially oriented language platforms. In addition to
European languages such as Romansh, Corsican and Breton, the index includes numerous African and 15 Indian languages.
Keybot is aimed primarily at professional translators who have a basic command of the language combinations they
seek and are capable of evaluating the sample translations it finds. The order in which results are displayed
can be influenced by using settings to define 'priority domains'. This feature enables users to configure the
search engine so that selected web domains containing the terms they are looking for are displayed at the top
of the list. Search results are displayed as snippets with the hits presented in both the source and target
languages. In addition, all pages can be downloaded as plain text or in HTML format. The parallel display of
pages in the source and target languages enables context-sensitive searches for specialist terms.
Web-to-TM
A translation memory (TM) is a
database of structured translations that is the main component of applications for
computer-aided translation (CAT). Keybot
works like a giant translation memory that individually analyses and indexes every multilingual
web domain. Translators working with
CAT tools including a TM can download translated segments from the Web in a standardised
TMX format for free, integrate
these into their system and in this way quickly take on board new customers' terminology.
How it works
Keybot has a specially programmed web crawler linked to a
four-stage language recognition module capable of identifying even short texts consisting of just a few words.
To divide up the text and match results with their translation, the crawler relies mainly on the formatting
characters in the HTML code. The language elements are
fully indexed by a proprietary system and can be retrieved via the Keybot web pages or downloaded using Web-to-TM.
Back story
The translation agency TTN Translation Network has been developing Keybot since 1999.
Originally, Keybot only indexed customers' orders, to enable (different) translators to translate their terms
consistently. In 2006, TTN launched an Internet search engine under the name Translation Search Machine (TSM).
In 2010, the search engine was renamed Keybot and its operation was outsourced to the startup Keybot LLC. In 2019,
the Web-to-TM tool was added to the search engine.