Question-Answering with Logic Specific to Video Games

Question-Answering with Logic Specific to Video Games Corentin Dumont, Ran Tian, Kentaro Inui Tohoku University, 6-6-05 Sendai, Miyagi 980-8579, Japan {corentin-d, tianran, inui}@ecei.tohoku.ac.jp Abstract We present a corpus and a knowledge database aiming at developing Question-Answering in a new context, the open world of a video game. We chose a popular game called ‘Minecraft’, and created a QA corpus with a knowledge database related to this game and the ontology of a meaning representation that will be used to structure this database. We are interested in the logic rules specific to the game, which may not exist in the real world. The ultimate goal of this research is to build a QA system that can answer natural language questions from players by using inference on these game-specific logic rules. The QA corpus is partially composed of online quiz questions and partially composed of manually written variations of the most relevant ones. The knowledge database is extracted from several wiki-like websites about Minecraft. It is composed of unstructured data, such as text, that will be structured using the meaning representation we defined, and already structured data such as infoboxes. A preliminary examination of the data shows that players are asking creative questions about the game, and that the QA corpus can be used for clustering verbs and linking them to predefined actions in the game. Keywords: question-answering, knowledge acquisition, meaning representation 1. Introduction This paper presents a corpus and a knowledge database aiming at creating a Question-Answering system specific to a new context, the open world of a video game. Un- like many QA systems that are designed to answer real world questions (Berant and Liang, 2014; Yao, 2015), the goal of this research is to build a system that can answer questions using the logic specific to the game, which may not be identical to the logic in the real world. We choose a popular game called Minecraft, whose openness pro- vides a great liberty for players, which guarantees a large number of possible questions to ask about the game, and yet the presence of a specific logic that limits the actions of players (Section 2.). We are interested in this prob- lem setting because it could provide a testbed for com- bining Natural Language Processing with advanced logi- cal inference techniques. The QA corpus is partially col- lected from quiz websites and partially written by human annotators (Section 3.). The knowledge database is ex- tracted from several wiki-like websites about Minecraft, which contain both unstructured text data and structured tables and infoboxes (Section 4.). A preliminary exam- ination of the data suggests that the knowledge database can be used for answering most of the relevant questions, but it may not be as simple as a keyword search (Section 5.), and may require to structure the unstructured part of the knowledge database, so that inference can be done by the QA system on several pieces of information in order to answer complex questions (Section 6.). We also show that the …

Content truncated to 500 words