There are six classes of entity in this task:
Below are guidelines for each of these.
If you use this dataset, please cite the task paper:
Leon Derczynski, Eric Nichols, Marieke van Erp, Nut Limsopatham (2017) “Results of the WNUT2017 Shared Task on Novel and Emerging Entity Recognition”, in Proceedings of the 3rd Workshop on Noisy, User-generated Text.
Names of people (e.g. Virginia Wade). Don’t mark people that don’t have their own name. Include punctuation in the middle of names. Fictional people can be included, as long as they’re referred to by name (e.g. “Harry Potter”).
“There was a celebration for Sarkozy” correct - Sarkozy is a person’s name
“The driver** is lost” wrong, because driver isn’t the name of a particular person
“We ate with Ringo Starr and he** didn’t do any impressions” Only mark the names of people, not words that just refer to them
“Dell** shares are up 17%” In this case, Dell refers to a company, not a person
“The award went to Ben** and Jerry**” wrong, because here Ben and Jerry is an organisation
“I’m so glad @ MileyCyrus won the award” correct - this refers to a person by name
Names that are locations (e.g. France). Don’t mark locations that don’t have their own name. Include punctuation in the middle of names. Fictional locations can be included, as long as they’re referred to by name (e.g. “Hogwarts”).
“There was a celebration in London” Correct - London is a location name
“The Eiffel Tower is over 300 metres tall” Correct - both words of the location are marked
“The room** is empty” Wrong, because room isn’t the name of a particular location
“Last night’s game saw Liverpool** lose to Manchester United” Wrong, because in this case Liverpool refers to a sports team
“Liverpool played well at Old Trafford” Correct, OT is a location
“The award went to Chelsea** Clinton” Wrong, because here Chelsea refers to a person, not a location
“Just taken my son to KFC for dinner #greasybutgood” Right. In this case, KFC is used as a location.
If you can visit it, and the word’s used as a location, then mark it like one.
Names of corporations (e.g. Google). Don’t mark locations that don’t have their own name. Include punctuation in the middle of names.
“Stock in Tesla is soaring” Correct - Tesla is a corporation name
“Just taken my son to KFC** for dinner #greasybutgood” Wrong. In this case, KFC is used as a location.
“I have a Samsung** Galaxy phone” This one’s out, because it doesn’t refer to the corp, just the product.
If you can buy shares in it, or it has employees, and the word’s used as a corporation, then mark it like one.
Name of products (e.g. iPhone**). Don’t mark products that don’t have their own name. Include punctuation in the middle of names.
There may be no products mentioned by name in the sentence at all - that’s OK. Fictional products can be included, as long as they’re referred to by name (e.g. “Everlasting Gobstopper”). It’s got to be something you can touch, and it’s got to be the official name.
“There was a recall of the Lexus i700” correct - Lexus i700 is a product name
“The apple** is rotten” wrong, because apple isn’t the name of a particular product
“We love our Kindles, and never travel without them**” Only mark the product names, not words that just refer to them
“I need to go to Crosshill iPhone** Repair shop” wrong, because here iPhone is part of another name (for a location)
“Today’s plane will be an Airbus A320Correct - the A320 is a product made by Airbus
“I forgot my phone**” _That’s out - this is just the generic term, not the specific product
If you can touch it, you can buy it, and it’s the technical or manufacturer name for it, then mark it.
Names of creative works (e.g. Bohemian Rhapsody). Include punctuation in the middle of names. The work should be created by a human, and referred to by its specific name.
If it’s the specific name of a creative work, for example a movie, song or book, then mark it.
Names of groups (e.g. Nirvana, San Diego Padres). Don’t mark groups that don’t have a specific, unique name, or companies.
There may be no groups mentioned by name in the sentence at all - that’s OK. Fictional groups can be included, as long as they’re referred to by name.
“Manchester played very well on the field tonight” Correct; Manchester here refers to a sports team
“Closing 8% up on team Starbucks** today” This is out; Starbucks here is a company
“Great set from The Darkness at Wembley” This is in - referring to the rock bank
If it’s a special name that refers to a unique, specific group, then mark it.