Abstract Christopher Madge 16 October 2019

From IMC wiki
Jump to: navigation, search

Talk 1

Title: Incremental Game Mechanics applied to Text Annotation


We argue that the mechanics of ‘Ville type Free-To-Play (F2P) games in general, and incremental games in particular, is es- pecially suited for Games-With-A-Purpose. We demonstrate this through WordClicker, an incremental game whose me- chanics is designed for text labelling. We believe the design and mechanics used are highly transferable to other games featuring annotation where game design is a challenge, such as serious games and language resourcing GWAPs. The game was tested with audiences from three popular indie gaming portals, achieving promising results both for entertainment value and learning.

Talk 2

Title: Progression In A Language Annotation Game With A Purpose


Within traditional games design, incorporating progressive difficulty is considered a fundamental principle. However, de- spite the clear benefits, progression is not such a prominent feature of Games-With-A-Purpose (GWAPs), nor one that is commonly evaluated. There is little evidence of the effects of progression, despite the clear benefits it can bring for training non-expert annotators to produce more complex judgements. Most current methods utilise either text based, or readability based heuristics for estimating item difficulty. There is often a substantial disconnect between readability and the complex- ity of the task itself with respect to the text in question. In this work we present an approach to progression in GWAPs that generalizes to different annotation tasks with minimal, if any, dependency on gold annotated data. Using this method we show a statistically significant increase in accuracy over randomly showing items to annotators.