Category: 

What is Fuzzy Matching?

Article Details
  • Written By: John Lister
  • Edited By: Kristen Osborne
  • Last Modified Date: 30 October 2016
  • Copyright Protected:
    2003-2016
    Conjecture Corporation
  • Print this Article
Free Widgets for your Site/Blog
President Richard Nixon had an entire speech prepared in case the Apollo 11 astronauts became stranded on the Moon.  more...

December 8 ,  1965 :  Pope Paul VI promulgated Vatican II into ecumenical law.  more...

Fuzzy matching is a method used in some computer programs, most notably those designed to automatically translate text between different languages. It works in situations where the database used by the software does not have an exact match for a piece of text. Instead, the database looks for close matches, with human revision usually checking that this match is acceptably accurate.

Most automated translation programs simply compare text to a database of previous translations. This database is usually known as a translation memory (TM). Each new and confirmed translation is added to the database, meaning that future translations are usually more accurate and quicker.

Normally, TM-based translation works with entire sentences. For example, if a user has previously approved a particular sentence in French as being an accurate translation of, "The shop is open between 9 a.m. and 5 p.m. on weekdays, but closes at 2 p.m. on Wednesday," then the next time the software is asked to translate that sentence, it will automatically use the same French wording as before.

Ad

The fuzzy matching tool works with situations where a particular phrase has not come up in documents translated by the system before. The tool involves either the creator of the software or the end user setting an acceptable fuzzy matching level in the form of a percentage match. If the phrase is close enough to one previously translated to meet this percentage match, the software will advise the user of the match, give the previous translation, and ask the user to confirm or edit the proposed translation.

To follow up the previous example, if the fuzzy matching threshold was set at 90% and the new English document contained the sentence, "The shop opens between 9 a.m. and 5 p.m. on weekdays, but closes at 2 p.m. on Wednesday," the software would suggest the previously used translation and the user would likely click to approve it. If the new document had the sentence, "The shop is open between 9 a.m. and 5 p.m. on weekdays, but closes at 2 p.m. on Thursday," the software would suggest the previously used translation and the user would correct this translation to conclude with the French word for Thursday. If the new document had the sentence, "The shop is open from 9 a.m. to 5 p.m. on every weekday except Wednesday, when it shuts at 2 p.m.," the software would not consider this a fuzzy match, despite the meaning being the same, and thus would not offer an automatic translation for the sentence, leaving the user to translate it manually.

Ad

You might also Like

Recommended

Discuss this Article

Post your comments

Post Anonymously

Login

username
password
forgot password?

Register

username
password
confirm
email