Peter Johnson

hi @arnesetzer there is an issue running CI for external contributors. I hopefully have it fixed in https://github.com/pelias/parser/pull/160 Please rebase master against your branch and push to trigger the CI

@arnesetzer the unit tests are failing, you can run them locally with the command `npm test`

one of the issues with this method is that there will be 'jitter' for partially complete inputs, eg: ```bash 'foo & bar' (0.80) ➜ [ { venue: 'foo & bar'...

This is the plan to roll this out without breaking backwards compatibility: 1. Update schema to add `street.ngram` field 2. Update queries to use `multi_match` 3. Queries should now work...

This is a really cool feature but I didn't realise how much work would be involved in order to change the way `name.default` works. In particular, there would need to...

The increase in disk space for the whole planet was

These articles are great, especially in the second one: - https://www.elastic.co/blog/practical-bm25-part-1-how-shards-affect-relevance-scoring-in-elasticsearch - https://www.elastic.co/blog/practical-bm25-part-2-the-bm25-algorithm-and-its-variables

Some examples of existing TF/IDF scoring I don't like (which I'll update over time): | search term | indexed term | description | |---|---|---| | "mcdonalds" | "mcdonalds - mcdonalds...

I merged https://github.com/pelias/schema/pull/430 a few months back so it should be available on all indices built this year. In that PR I demonstrated how to close an index, change the...

One possibility worth mentioning is to create a custom `similarity` (in Java) which adjusts the `lengthNorm` calculation. Disabling/enabling 'norms' is already possible via `norms: false` but we could potentially improve...