We use a cluster for completion suggester. It needs to support auto-complete in multiple languages. I'm looking for best practice in index design. Here are two options:
- a single index field for words and phrases in all languages, with standard analyzer. For instance both "solar eclipse" and "éclipse solaire" would be indexed in this field.
- an index with multiple fields, each for a different language and that language analyzer applied. For instance, "solar eclipse" goes to suggester-en field which is analyzed by "english" analyzer. And "éclipse solaire" goes to suggester-fr field which is analyzed by "french" analyzer.
Any suggestions, preference? Thanks.