WMTrans: Tuning Dictionary Lookup - the LEO Case
The search function is essential in any online dictionary. Canoo's WMTrans Lemmatizer offers the technology to increase the number of relevant search results during dictionary lookup. The online service Canoo.net complements this technology by providing in-depth information on German grammar, enabling users to comprehend and elaborate on dictionary content.
The Customer
With over 1.5 million queries per day, LEO is the most frequently used English-German online dictionary. It can be accessed free of charge, worldwide. Both in size and processing speed, the LEO dictionary is unparalleled in the Internet community. In the last two years LEO faced a rapid growth in access rates, from 6 million to 38 million accesses per month.
The Challenge
As a service to their users, the LEO team wanted to offer a possibility to recognize inflected word forms and spelling variants, and to transform these queries into valid search terms. A large-scale morphological component was required, which had to fulfil the highest standards regarding processing speed, quality and volume. An additional key requirement was the availability of linguistic background information to explain these word transformations.
The WMTrans / Canoo.net Solution
In the LEO scenario, the WMTrans Lemmatizer was the product that offered exactly what was needed. The Lemmatizer determines the base form of every input word, e.g. the infinitive form for verbs, the nominative singular form for nouns. The Canoo.net online service provides related grammatical information, which makes such word transformations intelligible and which provides a wealth of self-learning opportunities to the user, e.g. the entry bringst in relation to the infinitive bringen.
The Results
LEO benefits from WMTrans products in the following ways:
- Recognition of inflected query input and lookup of the corresponding base form in LEO
- Hyperlinking of query input and query results with additional information regarding inflection and spelling as provided by Canoo.net
The enhancement of LEO's functionality using WMTrans was one of the reasons why LEO was chosen Yahoo Site of the Year in December 2001.
>> Download full case study as PDF document