Turn titles into clean URL slugs with Arabic transliteration, diacritic removal, stop words and a length limit
Keyword Intent Clusterer
Paste a keyword list to classify its intent with Arabic and English vocabulary rules and cluster it by Jaccard similarity on stems, with a suggested page per group, cannibalization detection and CSV export.
Workspace
Groups
How the result is calculated
Intent uses vocabulary rules in priority order: transactional (buy, price, سعر, اشتري…), then commercial (best, review, أفضل, مقارنة…), then navigational (login, official, تسجيل دخول, or the brand name in two words or fewer), otherwise informational (how, what, كيف, شرح…). Clustering: each keyword is tokenised after light normalisation (removing “ال”, the prefixes و/ب/ل/ف and English plural s/ies) and stopword removal, then added to the first group whose representative reaches the Jaccard similarity threshold (0.5 by default). A group with mixed intents is flagged as possible cannibalization.
Limits:
An honest alternative to “keyword difficulty”: no search volumes, no difficulty and no SERP data (unavailable locally). Clustering is lexical, not semantic: “iphone price” and “iphone cost” will not group. Classification uses short dictionaries and may err on ambiguous words. Tune the threshold to your list.
FAQ
What is cannibalization?
Two pages targeting the same intent compete; a mixed-intent group needs different pages.
How do I tune the threshold?
Raise it for narrower groups and lower it for broader ones; start at 0.5.
