A recent post by Moustafa Hammad and Mohamed Elhawary, two engineers at Google search quality group, captured my attention. The post discusses recent improvements to Google Arabic search engine.
The improvements involve searches where users repeat the same letter or forget to add spaces between words.
“Our algorithm employs rules of Arabic spelling and grammar along with signals from historical search data to decide when to leave out spaces between words or when to remove unnecessarily repeated letters.”
Since Arabic is my native tongue, I decided to give this a try.
Arabic letters are cursive, which means a letter can change shape based on the position of the letter in the word (initial, middle, final or isolated):
Arabic Alphabets. Source: Wikipedia
Following the same example in the post. The word [قصيدة رائعة الجماال] (amazingly beautiful poem) will look like this if you take spaces out: [قصيدهرائعةالجمال]
Google returned corrected results and I got [Did you mean: قصيدة رائعة الجماال] which is the correct spelling.
To see how smart Google Arabic search is, I wanted to try using a word that naturally have the same letter repeated such as [معلومات ووقائع] which mean [information and facts]. By misspelling the word [معلومات ] and doing a search on [معلوماتو ووقائع] Google was able to correct the spelling and give me correct results and suggesting the correct spelling. Even when I completely misspelled the word and avoided putting spaces [معلوماتوووقائع], Google search was able to find the correct spelling.
Comparing the same search query [معلوماتوووقائع ] with Yahoo’s new Arabic search engine from Maktoob.com, I received no results:
Just like in any language, forgetting a letter in some instances can completely change the word meaning. For example, when I tried to search for [محرك بحث عربي] or [Arabic search engine] I missed the letter B in Arabic and wrote [عري] which now means [search engine nudity].
I would assume that’s a popular search query but at the same time it is a popular misspell. Google was kind enough to correct me and present me with the correct results.
Considering the complexity of the Arabic language, Google’s engineers have done a great job. Other than Yahoo’s Maktoob, the Arabic search landscape has been impotent.
According to Google, the new improvements will impact 10% of Arabic language queries.
Tariq Ali is an avid follower of the search engine and Internet marketing industry for the past 10 years. When he is not working or playing with tech gadgets, you will find him swimming, biking or running.
Read other articles by Tariq Ali
Thank you Tariq, your article has been mentioned in Google’s Webmaster Forum :)
http://www.google.com/support/forum/p/Webmasters/thread?tid=12e3b27834c825bc&hl=ar
Adel, thank you so much. You guys are doing an amazing job.