Peter Johnson

A setting of `5` shows some improvement, but I think we might need to go a little deeper and configure two additional values of [field_value_factor](https://www.elastic.co/guide/en/elasticsearch/reference/current/query-dsl-function-score-query.html#function-field-value-factor), the `factor` and `modifier` options...

Sounds like a reasonable change, it's always a tradeoff, this approach will of course eliminate some legitimate street results, such as `street: Wrigley Field` or `street: Union Square` in other...

One other thought I had was that it might introduce jitter, although on reflection it's likely not an issue since the parser likely detects `place` and `street` classifications on the...

this issue appears to be more complex than I had hoped, it would take some more time to investigate: - whether `clean.text` is producing the correct value - whether `_sanitizer/_tokinizer.js`...

note: there is already some code which is fairly similar looking but subtly different from the changes in this PR: https://github.com/pelias/api/blob/e6179c5734d76fedc130a3905da521303c675b6f/sanitizer/_tokenizer.js#L41-L45

Coming back to this now after thinking it over... I think it's good to merge, the concept that the parent names at the same layer as the feature can be...

Here are some test scripts I wrote to experiment with this setting: https://gist.github.com/missinglink/a520d23a855d4b84e236ff1fa3dfce5b From what I can see, enabling `search.low_level_cancellation` and setting an explicit `timeout` property on the query will...

One final thought is that the `search.low_level_cancellation` may have been much slower prior to Elasticsearch@6 because of the work done in https://github.com/elastic/elasticsearch/pull/25776

I think they've been available (albeit off by default) since v5, so we could take advantage of them immediately.

I'm not sure why the tests are failing when they pass on my local machine ❓