The short answer is "almost all" :).
How does it work?
What matters from the search perspective is if we are able to identify word boundaries. As long as we are able to do it, Klevu can provide search for that language. Fortunately, most of the languages around the world use space and other punctuations as word delimiters. The exceptions to this method are the languages such as Chinese, Japanese and Korean but for them as well we have special parsers to recognise word boundaries.
What features of Klevu Search are language independent?
Most of the features provided by Klevu Search are language independent. This includes features such as its capability to self-learn the ranking of products based on consumer activity, fuzzy match to handle error tolerance, synonyms, URL redirects, product promotions, normalisation of model numbers and SKU codes, accented characters, handling stopwords, partial match etc.
Are there any features language dependent? What are they and what languages are they supported for?
There are a few features that are language dependent.
- Normalising inflected words: This feature allows us to normalise inflected words. For example, plurals to singulars (e.g. batteries -> battery), identifying base forms (e.g. singing -> sing). Indexing both the inflected form and the normalised form allows shoppers a flexibility to search for any form. Currently, this feature is supported for the following languages: English, Finnish, Swedish, Danish, Hungarian, Russian, Turkish, Dutch, French, Italian, Portuguese, Spanish, German, Norwegian, Romanian.
- Automated Enrichment: Based on the context of a word, our automated enrichment algorithm identifies relevant synonyms, which it then adds to the search index. For example, if a word is "smart", we may add additional words such as "intelligent" and "clever" to the search index. If the word is a "curtain", we may add the word "drape". Automated addition of relevant synonyms allows shoppers to locate a product not only by using the original word but with additional synonyms added by our algorithm. Currently, this feature is supported for the following languages: English, Finnish, Swedish, Spanish, French, Danish, Italian, Dutch, Polish, Portuguese, and Hebrew languages.
- Decoupling Compound Words: If a word is "röttvinglas" (i.e. red wine glass in English), we are able to decouple it and index it as three separate words "rött", "vin" and "glas". This allows shoppers to fire a query such as "glas för rött vin" (i.e. glass for red wine). Currently, this feature is supported for the following languages: English, Finnish, Swedish, Estonian, German, Spanish and Danish.
- Advanced Query Processing: Queries such as "formal tops under 100 USD" requires understanding the intent behind the query and fetch the appropriate products. Currently, the feature is available for the following languages: English, Finnish, Spanish, French, and Swedish languages.
Note: If this page does not answer your question on a language-specific feature, please write to us at firstname.lastname@example.org.