Whether "dbc:Earthquakes" or "dbr:Vanilla_Ice" available in English Wikipedia (Feb 2015) 1000 dimension - No stemming - 10skipgram

[zhq2009]

Hello,

We are currently using the dataset "English Wikipedia (Feb 2015) 1000 dimension - No stemming - 10skipgram". We are search for some test cases on http://dbpedia.org/page/Earthquake. We have tried "DBPEDIA_ID/Vanilla_Ice" is available in the dataset. But when we try "dbc:Earthquakes" or "dbr:Vanilla_Ice", we will get error message KeyError "dbc:Earthquakes" not in vocabulary and "dbr:Vanilla_Ice" not in vocabulary. We are wonder whether the dataset stores data as "dbc:" or "dbo:"?

Thank you

[dav009]

Hi @zhq2009,

I did not tag categories, so at the moemnt we dont have any dbc: nor dbo:.
If this were to be included then the categories should be tagged on the text everytime an item of the category occurs?

[dav009]

If I understand correctly, you want vectors for categories as well ? Category vectors could be easily done by appending the categories of a mention.

dbos sound a bit more complicated though

[zhq2009]

Hello David,

Yes, we are looking for the vector categories as well.

Thank you for your help.

Sincerely,

Hanqing

On Mon, Jun 20, 2016 at 12:10 PM, David Przybilla [email protected]
wrote:

If I understand correctly, you want vectors for categories as well ?

—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
#19 (comment), or mute
the thread
https://github.com/notifications/unsubscribe/ARKSjGQQIahpjDXDHaOzwk57ezpyir_uks5qNrt1gaJpZM4I5NQl
.

[dav009]

if you provide :

• the map from page/resource to category (input: Wikipedia Dump, output: Category-Resource mapping)
• a way to load them

I would gladly add the changes for supporting the gen of such vectors

[zhq2009]

Sure, We can discuss later to decide how to map and load.

We are currently wondering why some entities (dbr) are not available in the DBpedia vectors? If you could provide some suggestions, that would be great.

Thank you.

[dav009]

some reasons:

• gensim(the tool used to create the vectors) cant deal with a very big vocubulary set, do "tokens" with less frequency than certain threshold were dropped
• they are not linked in wikipedia
• redirects are not resolved (there are no canonicals)
• it could be such resource did not exist in hte wiki we processed