search

dbpedia / lookup

Outputs a list of ranked DBpedia resources for a search string.
Apache License 2.0
185 stars 39 forks source link

How to create refCounts? #26

Closed vierja closed 8 years ago

vierja commented 8 years ago

I am trying to build an index for Spanish data, but I can't seem to find a way to build refCount information needed to properly sorting the data.

All I found was a page-links_es.nt.bz2 file containing wikiPageWikiLink triples.

sandroacoelho commented 8 years ago

Hi @vierja

In order to build a new index, you just need to concatenate all data sorted by URI and run the indexer command as in the README file.

Unfortunately, we don't have i18n for Lookup, but you can do changing the namespace "http://dbpedia.org/" for the target language (Sorry for that). We will improve it soon and you are invited to help us, if you want :+1:

*I saw what you're looking for somewhere. I will try to find it... Once I have the answer, I will let you know here

HugoSousa commented 8 years ago

Is the refCount information included in the pre-built indexes? I tried to locally mirror the service (for English, the default language) following the instructions in the README, but the refCount and the description seem to be empty.

sandroacoelho commented 8 years ago

Hi @vierja, @HugoSousa , I have an answer to you:

A few years ago, RefCounts was provided by pignlproc project and seems that this project does not active development anymore.

However, @jodaiber (From DBpedia Spotlight) told us that a startup in the US (diffbot) rewrote it to run on a single machine with the same output. The code is available here and it runs automatically for most languages once a month. These dumps are available here.

@HugoSousa : I will rebuild our index asap

Best,