Categorization of Known-Item Search Terms in a TV Archive
Anne-Stine Ruud Husevåg · 2017
This paper reports on a preliminary explorative study that categorizes search terms provided by 50 users after having watched short clips from TV programs from The Norwegian Broadcasting Corporation archive. The aim of this study is to improve indexing by gaining knowledge about users preferred search terms. One clip is from a news program, and two clips come from a literature program. The search terms are categorized according to The Panofsky-Shatford mode/facet matrix. The study shows that specific search terms like named entities are used the most for both genres, and that generic and abstract terms are more important in the literature clips than the news clip. The search terms provided by users were matched with text from the subtitles from the clip, and the results showed that 17%, 32% and 40% of the terms were found for literature clip 1, literature clip 2 and news clip respectively. The percentage was higher when the terms were matched with subtitles from the whole program, and would have been even higher if additional on-screen text was included.