Arabic Morphological Analysis and Generation

Welcome to the Arabic Home Page of the Xerox Research Centre Europe (XRCE) in Grenoble, France. An alternative URL to reach this page is http://www.arabic-morphology.com/.

This is the gateway to the Xerox Arabic Morphological Analyzer and Generator, which was built using Xerox Finite-State Technology. The system accepts Modern Standard Arabic words and returns morphological analyses and English glosses. Arabic words are displayed in Arabic script using Java applets.

Run the Arabic Morphological Analyzer (Research Version)

Keyboard Input Method Cut-and-Paste Input Method

Arabic words can now be entered using either keyboards or cut-and-paste, using a variety of encodings and transliterations. Your browser must be Java-enabled.

Research vs. Commercial Versions

The commercial version of the Arabic morphological analyzer is not visible from these web pages. For commercial licensing information see below.

What you see in the web demo is

1. The research version of the Arabic morphological analyzer, mostly dating from 1996 and 1997.

2. A research application, written on top of the morphological analyzer, that returns English glosses.

3. A research GUI (graphical user interface) that allows users to input Arabic words and see them rendered in Arabic script.

The research demo, version 0.3, is now effectively frozen. It is made available for non-commercial testing and consultation with no guarantees or promises of any kind, and the usual legal notices apply. In this demo, the amount of input is artificially limited to prevent commercial (ab)use.

Known Problems

Since 2003 we have received a number of reports of access problems from users of Internet Explorer; the problem appears to be tied to some version or versions of the Java Virtual Machine that insert an extraneous ":-1" port into the query URL when input is sent to the Xerox server for processing. Some users have solved the problem by installing a different version of the Java Virtual Machine, or simply using Mozilla. The following combination of JVM and Internet Explorer 6 have been reported to insert the extraneous ":-1" port.

JVM: Version 1.4.2_04 (build 1.4.2_04-b05)
And Explorer: 6.0.2800.1106.xpsp2.030422-1633

Commercial Licensing

Work on Arabic recommenced in 2002 and the result was a commercial version of the Arabic Morphological Analyzer that has much wider coverage and much improved analyses compared to the research version. For commercial licensing information contact Mr. Mathieu Chuat Mathieu.Chuat@xrce.xerox.com Tel: +33 4 76 61 51 87. (France)

People

Lexicography and Arabic-language consultation for the research system were provided by Tim Buckwalter. The Arabic "Yarb" font used in the interfaces was created by Yannis Haralambous. System design, interfaces, Arabic-script rendering, and computational linguistics were done by Ken Beesley. Arabic language consulting and lexicography for the commercialization in 2002 was provided by Martine Pétrod.

Some publications

Arabic Finite-State Morphological Analysis and Generation Postscript, 6 pages, 137k bytes. COLING-96.

Arabic Morphological Analysis on the Internet Postscript, 10 pages, 115k bytes. Proceedings of the 6th International Conference and Exhibition on Multi-lingual Computing, Cambridge, 17-18 April, 1998.