Sarah Hoffmann
Sarah Hoffmann
Indeed. If this feature ever gets implemented, it's much more likely to use the data from the lokku project.
That's great. Next I would suggest that you consistently add the "search_analyser" to the mappings and then get rid of setting the analyser in the query builder. It should then...
Yes, that's what I meant. > It is not enough only to use edge_gram. We have to tokenize Japanese words with Kuromori analyzer. Maybe Chinese and Korean language are also...
> In fact, I think it is necessary to add some source code in PhotonQueryBuilder. I added some raw searching for Japanese language here Checking the raw collector is already...
> First, let me check whether it can work without adding collector.default.raw_ja. However, I still think we need to add Japanese index for collector.default.raw because there are a lot of...
@JinIgarashi Unfortunately you have now created a completely different search algorithm. I will have a closer look at the changes you have done and if some or all of them...
You can still find the effect on small islands, for example https://nominatim.openstreetmap.org/ui/details.html?osmtype=W&osmid=64971378
This is not fixable in general. The final order of the results is determined after the requested number of results have been retrieved from the database. And the criteria for...
We could just the increase the number of results retrieved but often this will just be a band-aid to an underlying problem. It's better to address the root cause why...
Can I bump this issue a bit? It would be really helpful to have more consistent reports from people.