The National Institute of Korean Language said on the 31st it has released six language resources on “Everyone’s Corpus” that can be used for artificial intelligence development and language research.
The newly released materials include four datasets: an “argumentative writing summary corpus” that summarizes newspaper editorials; a “collaborative dialogue summary corpus” that summarizes conversations; and separate evaluation corpora for each set of summaries. The institute also released two additional resources: a “context inference corpus,” which contains inference statements written based on context or common sense, and a “knowledge graph” that structures dialogue context. The institute said the resources can help AI better understand context embedded in Korean, produce summaries and make inferences grounded in common sense and Korean culture.
Including the six released this time, the institute has made public 140 Korean-language resources for AI training to date. Anyone seeking to use the corpora for AI development and research or Korean-language studies can download them from the Everyone’s Corpus website.
An institute official said it plans to release a total of 36 Korean language-and-culture knowledge resources this year to support development of AI specialized for Korean.
* This article has been translated by AI.
Copyright ⓒ Aju Press All rights reserved.