Science

Language detecting technology struggles with George R. R. Martin's Game of Thrones


Language detecting technology struggles with George R. R. Martin’s Game of Thrones because the bizarre names don’t behave ‘normally’

  • Quirky names, such as Daenerys, don’t look or resemble most names
  • Algorithms are developed and trained to detect names by studying newspapers
  • A vastly different writing style is found in non-fiction novels and makes them hard to detect  

Game of Thrones characters and many other fantasy novels pose an issue for technology designed to decipher languages and the written word. 

Quirky names, such as Daenerys and Grey Worm, don’t look or resemble most names from the real world and are often not picked up by technology as they don’t behave in a normal manner. 

The algorithms are developed and trained to detect names by studying newspaper articles.

A vastly different writing style is found in non-fiction novels and makes the detection of fictional names almost impossible. 

Scroll down for video  

Names are contextualised in stories and this also adds another layer to the thorny issue. It is for example possible to refer to Daenerys Targaryen as Daenerysm but she has a plethora of other monikers. She is also known as Dany, Daenerys Stormborn, Mother of Dragons and Khaleesi

Names are contextualised in stories and this also adds another layer to the thorny issue. It is for example possible to refer to Daenerys Targaryen as Daenerysm but she has a plethora of other monikers. She is also known as Dany, Daenerys Stormborn, Mother of Dragons and Khaleesi

The researchers tested the ability of four different Natural language processing (NLP) tools tools at recognising popular characters’ names in 40 novels, including A Game of Thrones.

Natural language processing (NLP) tools are commonly used in many day-to-day applications such as Siri and Google.

The technology is imperfect however and fails when it is faced with other worldly names and events. 

Researchers from the Vrije Universiteit Amsterdam published their analysis in PeerJ, highlights types of names and texts that are particularly challenging. 

It is aware of certain telltale signs of a name, such as it beginning with a capital letter or being preceded with a pronoun.  

Game of Thrones characters and many other fantasy novels pose an issue for technology designed to decipher languages and the written word. Quirky names, such as Daenerys (pictured) and Grey Worm, don't look or resemble most names from the real world

Game of Thrones characters and many other fantasy novels pose an issue for technology designed to decipher languages and the written word. Quirky names, such as Daenerys (pictured) and Grey Worm, don’t look or resemble most names from the real world

Fiction authors can make up their own names, such as Tywin or R’hllo which have very little similarities to real names. 

Issues also arise when character names appear straight from the dictionary as nouns or adjectives – such as Grey Worm. T

These names do not behave like ‘normal’ names and NLP systems have difficulty recognising them in a text.  

Names are contextualised in stories and this also adds another layer to the thorny issue. 

It is for example possible to refer to Daenerys Targaryen as Daenerysm but she has a plethora of other monikers.

She is also known as Dany, Daenerys Stormborn, Mother of Dragons, Khaleesi, the Unburnt and Mhysa.  



READ SOURCE

Leave a Reply

This website uses cookies. By continuing to use this site, you accept our use of cookies.