Elasticsearch - 通配符 vs 前缀 vs vs 正则表达式 vs query_string 差异和性能

Question

I've just started using Elasticsearch for my project and I want to search like the sql keyword 'like%' does.

1. Could anyone please explain what are the differences between Wildcard , Prefix , query_string and Regexp ?

2. Which one can search with the best performance ?

3. Is it the best way to search rapidly between product tags with defining the keyword field and search with one of the three ways which I mentioned?

I appreciate whoever answers me.

Thanks.

Answer 1

Which query is better?

Let me answer point by point.

1. Could anyone please explain what are the differences between Wildcard, Prefix, query_string and Regexp?

They are very similar in what they can do and how they are implemented. In fact, wildcard , prefix , and regexp are equally bad in performance . query_string query allows for a more complex queries, for instance using logical operators like OR and AND . query_string supports wildcards , which I believe are the same wildcards as wildcard query does.

2. Which one can search with the best performance?

Bottomline: they all are bad.

3. Is it the best way to search rapidly between product tags with defining the keyword field and search with one of the three ways which I mentioned?

I believe that if you use a boolean filter query to reduce amount of matched documents then performance should improve, but the performance gain might still be overwhelmed with how inefficient regexp-like queries from above are.

How can I implement autocomplete with Elasticsearch?

As you mentioned in the comment above, you would like to implement an autocomplete, so the search_as_you_type data type might be exactly what you are looking for.

Another similar feature is suggesters . Both of them will try to use a more efficient term-based matching, and not regexp-based.

Hope that helps!