Automated Extraction of a Game Model using Natural Language Processing: National Strategies of the Asian Infrastructure Investment Bank

摘要

This article presents a method for representing highly strategic events using natural language processing. In 2013, China proposed the creation of the Asian Infrastructure Investment Bank (AIIB) as the country continues to capitalize on its economic growth and makes moves to gain influence in regional development. International society is facing a big shift in power balances, where the existing sphere of influence wielded by financial institutions led by the US and Europe are challenged by new orders being established by the developing countries. The method proposed in this article automates the analysis of written material so that the strategies taken by people or organizations involved can be modelled. The method uses clue expressions and supervised machine learning to identify and extract sentences containing strategic relations from a corpus. The accuracy of extraction, defined as the sentences including strategic relations in the sentences that included clue expressions, was 60%. Coverage was calculated as the ratio of sentences correctly extracted in the sentences containing strategic relations within the sentences including clue expressions, and was 67%. In addition, a method for modelling the strategies extracted is proposed and the relations of strategies contained within news articles are represented in a strategic map. Strategic maps of a complex initiative such as the AIIB has revealed the links between various events that intuitively may be considered irrelevant, such as the AIIB and the annexation of Crimea. Furthermore, the change in game structure before and after UK participation in the AIIB became visible as the consequences of UK participation onto other players' preferences has been revealed.