geonetwork / core-geonetwork

GeoNetwork is a catalog application to manage spatially referenced resources. It provides powerful metadata editing and search functions as well as an interactive web map viewer. It is currently used in numerous Spatial Data Infrastructure initiatives across the world.
http://geonetwork-opensource.org/
GNU General Public License v2.0
433 stars 489 forks source link

Searching based on full title suggestion sometimes fails to give results #1003

Open beekhuiz opened 9 years ago

beekhuiz commented 9 years ago

When using the autosuggestion to complete the search text to the full title of the dataset, the search (sometimes) returns no results. Shorter queries (based on for example one word of the title) do return the correct datasets. I am using Geonetwork 3.0.1

fxprunayre commented 9 years ago

Could you provide a clear example ? That type of cases usually occur depending on the language of the catalogue content and the language configuration (see admin page of the catalogue).

beekhuiz commented 9 years ago

When typing in "rivier", there is the autosuggestion "Aankomsttijden water vanuit de rivier" (which is the title of a dataset). When selecting this autosuggestion (so the search term becomes "Aankomsttijden water vanuit de rivier") and pressing the Search button, there are no results. When searching on just "rivier", the dataset does show up in the results.

I think these are all the language settings: image

fxprunayre commented 9 years ago

Does "Aankomsttijden water vanuit rivier" return the record ? If yes, then "de" may interact with stopwords... You may need to check if a Lucene Analyzer exist for your language and use it.

You could also try to turn off "Prefer UI Language" too and switch to "prefer all docs in all languages" mode.

ekkelenkamp commented 9 years ago

Hi,

The same thing is happening with the english sample data that can be imported with geonetwork.

When searching for "Physiographic Map" using a dutch GUI (http://localhost:8080/geonetwork/srv/dut/catalog.search), the autosuggest will give:

Physiographic Map of North and Central Eurasia (Sample record, please remove!)

No records will be found. After removing "and" from the query the result is found.

If I switch to an English gui (http://localhost:8080/geonetwork/srv/eng/catalog.search) the autosuggest works. Here the "and" is recognized as stop word.

If I understand correctly, the stop word filtering only works if the gui language is the same as the query language? So using a dutch gui "and" is not considered a stop word in a query?

What would be the recommended settings if a user used both dutch and english queries? I suppose merging the english and dutch stopword lists might do the job?

It would be nice to make the autosuggest always work. That's what users will expect. Would it be hard to implement such a feature? I've been looking at the source code but when the search query is called, it cannot be determined if a autosuggest was used.

Regards,

Rudie.

fxprunayre commented 8 years ago

Moving to future release as there is no easy fix/configuration for this.

jodygarnett commented 9 months ago

I am not sure if this issue can be worked as described