This paper takes an in-depth look at the relationship between mechanically extracted keywords and 'Top Ten News of the Year' compiled by the news editors. A previous study that briefly touched on the topic concludes there does not seem to exist any meaningful connection between the two. In this paper, we set up a more elaborate way of comparing and connecting the two, and argue that there is a certain reasonably good converging point. The corpus we make use of for our experiment is a subset of the Trend 21 corpus which is a collection of Korean major newspapers (2000-2013). For keyword extraction, log-likelihood ratio was made use of. Extraction of collocation for each keyword was needed, for which a version of Mutual Information was utilized. Finally a detailed comparison of the top ten news with the top 100 keywords was conducted from several points of view.
展开▼
