Associative Word Relations in Natural Language Processing

Motivation for this work comes from the longest-running Serbian television quiz show called TV Slagalica and more specifically one of its games named associations. In associations game, two players attempt to guess a solution given several clue words. There is large number publicly available game scenarios that were used evaluate applicability trained artificial neural networks predict possible solutions. Material network training was obtained through unconventional sources as no professional text corpus exists language. Under outlined schemes, it observed words come up within 2% or less vocabulary, depending on method data preparation. Data preparation specifics are further demonstrate effects each technique used. Even though results below human-level performance, they can nevertheless be useful puzzle creation.