The Pokérator is an automatic generator of new Pokémon names with descriptions based on user input. The names are generated by blending words based on syllables (or characters if monosyllabic) and are evaluated using bigram language models. The description is generated by retrieving words from ConceptNet and feeding these into templates, as well as using OpenAI’s GPT-2 language generatio n model, which is trained on all Pokédex entries.
This project has been developed in the frame of the Advanced Project in Natural Language Processing at the University of Twente. The project authors are Dominique Geissler (@domi151), Elisa Nguyen (@ElisaNguyen) and Daphne Theodorakopoulos (@daphne12345).
This project contains the Python files, models and Python notebooks created.
Geissler, D., Nguyen, E., Theodorakopoulos, D., & Gatti, L. (2020). Pokérator - Unveil your inner Pokémon. In Proceedings of the 11th International Conference on Computational Creativity (ICCC’20) (pp. 500-503) http://computationalcreativity.net/iccc20/papers/159-iccc20.pdf