search

dice-group / AGDISTIS

AGDISTIS - Agnostic Named Entity Disambiguation
http://aksw.org/Projects/AGDISTIS.html
GNU Affero General Public License v3.0
140 stars 37 forks source link

Links to things #45

Closed renespeck closed 6 years ago

renespeck commented 7 years ago

Agdistis links Mcnabb Inc to dbr:McNabb that is of type Thing in DBpedia.

DiegoMoussallem commented 7 years ago

I'll have a look on this problem. As a way to alleviate this problem, you can set the agidists.properties file for only accepting Named Entity (Person, Org and Location) instead of accepting common types. Could you see if it alleviates for a while?

renespeck commented 7 years ago

Ohh, nice. Could you tell me the key I have to use for the values of the Named Entity Types in the agidists.properties please?

DiegoMoussallem commented 7 years ago

Enable to find common entities

commonEntities=false

false for finding only Named Entities and true for finding every kind of them

renespeck commented 7 years ago

Ahh thanks. :+1:

renespeck commented 7 years ago

Hi, changing 'commonEntities' value to 'true' had no influence of the linking.

Agdistis still links Mcnabb Inc to dbr:McNabb.

DiegoMoussallem commented 7 years ago

Could you write/point the sentence/text or dataset here? and I will find the bug/problem. :)

renespeck commented 7 years ago

Jim Mcnabb Inc is in the Outdoor Advertising Services industry in Taylor, TX. This company currently has approximately 5 to 10 employees and annual sales of $1,000,000 to $4,999,999.

DiegoMoussallem commented 7 years ago

Our corporation affixes list comprises of 'corp Corp ltd Ltd inc Inc co Co'

So, AGDISTIS cleans the label using these affixes, Mcnabb Inc becomes only Mcnabb, however this label is quite generic where our filter of disambiguation pages does not work. I assume it is a bug and I will have a look into more details. I come to you as soon as possible.

DiegoMoussallem commented 6 years ago

Hi @renespeck ,

Apologizes for this late response, I figured out the bug. It has happened because of redirect's method was skipping the verification of disambiguation's page. Therefore, when AGDISTIS was receiving McNabb Inc, it was removing dbr:McNabb, however, dbr:Mcnabb was also a candidate which has been redirecting to dbr:McNabb then resulting in that mistake. I fixed this little bug and you can check by downloading the master branch again.

pls, close the issue after your verification.

renespeck commented 6 years ago

Thanks for the fix, seems to work. :D