Ask about Search feature and some important questions on Stardog

Hi, my app's all about search and inference.
Stardog uses Lucene to provide search feature. Why dont you connect to ES to provide search feature. Because Lucene don't have much feature we expect (for example: like custom tokenizer Edge ngram - if i want to use this feature i have to custom a lot on Stardog) and also scalability (index and search words on the same server with inference feature). Does it provide the best performance ?.
I hope Stardog have the way to connect to ES and index data to on it for searching. (This feature like GraphDB Ontotext).

When i search on multiple Thing VALUES (Person,Address,Email ,...)
with properties like that:

  • Person has name
  • Address has fullAddress
  • Email has emailAddress
  • Person hasAddress and hasEmailAccount
    By using tag:stardog:api:property:textMatch, i only receive uri, that property, its value.
    Is there a way that i can extract all the aforementioned properties in single search query ????????

Another question, i want to ask about the database limit (2^32 or 2^64 or even more) ? How many entities that Stardog can save (expect unlimited data saved on Stardog) ? Can i save data in different datasources (like each repo) but still query and infer from them ?

Thanks for reading.

Up. No one repliess !!!

It's been 7 hrs. Chill. If you haven't noticed there's a bit of a world crisis at the moment so I hope you'll excuse people if they don't respond in a timely manner.

They probably use embedded Lucene because it doesn't require someone to setup a completely separate database for full text search and it avoids a network round trip.

You can search Elastic with the Stardog virtual graph feature

From the example in the docs

?s ?p ?l.
(?l ?score) <tag:stardog:api:property:textMatch> ('mac' 100).

After binding the literal you can make the rest of the query as complex as you'd like. It doesn't have to be just ?s ?p ?l

Nothing can store unlimited data except maybe /dev/null. I believe Stardog can scale to 50B triples.

Yes, but it would be a federated query and would be less performant than if it was in a single database.


Zach hit all the important points in your list of questions. The question about how to "extract all the aforementioned properties in single search query" was not quite clear to me. If you don't have a sufficient answer, please share more detail about the requirements. I would also like to add that we are considering support of Elasticsearch to store Stardog's search index.


This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.