dice-group / n3-collection

N3 - A Collection of Datasets for Named Entity Recognition and Disambiguation in the NLP Interchange Format

n3-collection

N3 - A Collection of Datasets for Named Entity Recognition and Disambiguation in the NLP Interchange Format

n3-collection by http://aksw.org is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.

n3-collection's Issues

Type annotations

Since there are named entities with another type than person, organization or location, the types should be annotated giving other researchers the possibility to filter the names entities using these types.

While it seems to be clear to use dbpedia URIs:
dbpedia-owl:Organisation
dbpedia-owl:Place
dbpedia-owl:Person
it is not really clear, how we could formulate that something is none of these three types.

reuters.xml

ERROR: 'ParseError at [row,col]:[2350,3]
Message: The element type "SimpleTextPart" must be terminated by the matching end-tag "".'

ERROR: 'ParseError at [row,col]:[2350,3]
Message: The element type "TextWithNamedEntities" must be terminated by the matching end-tag "".

"Sun Co." is short for "Sun Refining and Marketing_Co" in Reuters-128

The marking for "Sun Co." is wrong and should be fixed.

wrong URI vor "Reuter" in reuters-128

At the end of some articles, there is an entity with the label "Reuter". While it was marked with the URI "http://dbpedia.org/resource/Reuter" it seems to be clear that the news agency Reuters is mentioned. The URI should be replaced with "http://dbpedia.org/resource/Reuters".

"Burlington Industries" marked as "Dominion Textiles" in Reuters-128

"Burlington Industries" (http://dbpedia.org/resource/Burlington_Industries) was marked as "Dominion Textiles" (http://dbpedia.org/resource/Dominion_Textile).

Check whether some entities got different URIs

Check the URIs and if they are refering to the same or different entities:

http://aksw.org/notInWiki/Bernhard_Schulte_Management http://aksw.org/notInWiki/Bernhard_Schulte_Shipmanagement

Encoding problem

In document 38 the URI http://aksw.org/notInWiki/J��rg_Feddern has an encoding error.

Advanced Micro is short for Adcanced Micro Devices Inc. in reuters-128

Thus, both should get the same URI.
Since we want to move to DBpedia 3.10 / 4.0 both should get the URI http://dbpedia.org/page/Advanced_Micro_Devices

Only "Bureau" instead of "Bureau of Reclamation"

In document http://aksw.org/N3/RSS-500/110#char=0,310 only the word "Bureau" has been marked while "Bureau of Reclamation" should have been marked for the meaning http://dbpedia.org/resource/United_States_Bureau_of_Reclamation

Missing URI for Christopher Clark in reuters-128

A man named Christopher Clark got "http://aksw.org/notInWiki" as URI. The URI should be extended to "http://aksw.org/notInWiki/ChristopherClark"

Fix the NIF creation

Error report for News-100.ttl

line 3448: Invalid characters in angle-bracketed name: nif:sourceUrl <\n    > .
line 11028: Invalid characters in angle-bracketed name: nif:sourceUrl <\n    > .

nif:beginIndex is typed xsd:int, should be typed xsd:nonNegativeInteger
Number of errors: 1755
nif:endIndex is typed xsd:int, should be typed xsd:nonNegativeInteger
Number of errors: 1755

---

Error report for Reuters-128.ttl

line 1930: invalid whitespace in URL: itsrdf:taIdentRef <http://aksw.org/notInWiki/Colorado Rural Electric Association> .

nif:beginIndex is typed xsd:int, should be typed xsd:nonNegativeInteger
Number of errors: 1008
nif:endIndex is typed xsd:int, should be typed xsd:nonNegativeInteger
Number of errors: 1008

---

Error report for RSS-500.ttl

line 5601:  invalid whitespace in URL: itsrdf:taIdentRef <http://dbpedia.org/resource/Bureau of Land Management> ;

nif:beginIndex is typed xsd:int, should be typed xsd:nonNegativeInteger
Number of errors: 1500
nif:endIndex is typed xsd:int, should be typed xsd:nonNegativeInteger
Number of errors: 1500

URIs yield "OntoWiki Error",  example: http://aksw.org/N3/RSS-500/197#char=0,184 

Rename XML-files

Rename the XML files to match the corpus naming.

Only "Tan" has been marked instead of "Tan Ching-ting"

In document http://aksw.org/N3/RSS-500/368#char=0,129, only "Tan" has been marked (http://aksw.org/N3/RSS-500/368#char=19,22) while it should have been "Tan Ching-ting" at the beginning of the document (Hospital spokesman Tan Ching-ting ...).

Underlying Knowledge Base for Linking

Obviously, the corpus has some entities like "Advanced Micro Devices" marked as not in DBpedia, which is wrong[1].

[1] http://en.wikipedia.org/wiki/AMD