Autocomplete Multi-Language Search Using Ngram and EDismax Phrase Queries - Ivan Provalov, Netflix
Autocomplete presents some challenges for search in that users' search intent must be matched from incomplete token queries. Many non-Latin character based languages have additional complications. The following are some of the examples of unique language-specific issues which must be addressed in search systems in order to support these languages:
Japanese and Chinese multiple scripts (Hiragana, Katakana, Romaji, Zhuyin, Paoding)
No token-delimiters for Japanese and Chinese
Korean character composition
Arabic spelling variations of the transliterated foreign words
This talk covers these challenges in detail, describes approaches to solving them, and shares some tools used to help address them.
Presented at Lucene/Solr Revolution. Learn more: https://activate-conf.com/