Brief fact sheet...
Type:
DLL, .OBJ, .LIB and WCF web service.
Platforms:
MS-Windows XP, Vista, 7 and 8. MAC OS 10.4.x and 10.5.x. Unix
and Linux.
Code structure:
32 and 64 native code.
API: World class designed API
that enable the developers to check, suggest, replace, skip and add.
Required resources: 430KBytes
of storage (distributable zip size is 220KB), full functioning RAM requirement is
154KByte of RAM.
Speed:
Checks over 4 million word per minute. Suggests over 150 proper suggestion per second.
Al-Khaleej News Paper in 1990 applauding Abgad Hawaz breakthrough.
Abgad Hawaz© family of products and technologies goes back to the late eighties when
Abgad Hawaz first release was introduced in 1990. After a successful assignment
with Apple© Inc. that had started in the year 1987, Info Arab had delivered the world
first ever Arabic spelling checker in 1989 on the Apple Macintosh© platform. Followed
by two major licensing agreements with Diwan© Science and Technologies and Word Perfect©
in 1991 and 1992, Abgad Hawaz spell checker has reached its maturity stage in the
beginning of the nineties. And since then, Abgad Hawaz had been licensed to many
international software companies to be included as part of their software applications.
Abgad Hawaz family of technologies is mainly concerned with the proper format of
Arabic words. Their proper spelling checker is only one of the dimensions (though
the most famous) that Abgad Hawaz technologies provide.
The Arabic language has many unique features that would have required
the consumption of huge computer resources
in terms of memory, storage and processing power except if smart algorithms are
implemented. Through a very powerful morphological analysis technologies, phonological
based algorithms and many more, Abgad Hawaz has succeeded to minimize the required
resources to small fractions of its existing competitors and at the same time provides
multiples of speed when it comes to the ability to scan millions of words per minute.
Bench mark of the current version of Abgad Hawaz spelling checker (on a data sample
of 11’559’673 Arabic words from social networks' user generated text, News agency
news clips and typical business correspondence and documents)…
- Accuracy of correct words…
The speller was able to identify 100.00% of the wrong words.
Not a single miss-spelled word was skipped.
- Accuracy of the wrong words...
0.1% of the wrong words were an acceptable Arabic legal word.
- Suggestions performance…
The speller was able to give suggestions for 83% of the words
marked as wrong.
- Suggestions quality…
Suggestions algorithms were able to suggest the correct word
in 90% of the suggestion cases.
- Suggestions relevance…
The proper word of suggestion was in the average of 2.6 (i.e.
the second or the 3rd word of suggestions).