A probabilistic approach to identifying technology vacuum: GTM-based patent map

A patent map has long been considered as a useful tool to identify technology vacuum defined as an unexplored area of technologies that may deserve intensive investigation for future new technology development. However, previous studies for identifying technology vacuum on the patent map have been subjected to intuitive and manual identification of technology vacuum. In this context, this paper proposes a generative topographic mapping (GTM)-based patent map which aims to identify technology vacuum automatically. Since GTM is a probabilistic approach to map a low-dimensional latent space onto the multidimensional data space and vice versa, it contributes to the automatic identification of technology vacuum. This study consists of three stages. Firstly, text mining is executed to transform patent documents into keyword vectors as structured data. Secondly, the GTM is employed to develop the patent map with extracted keyword vectors and discover patent vacuums which are expressed as blank areas in the map. Lastly, technology vacuums are identified by inversely mapping patent vacuums in latent space into new vectors in data space. The procedure of the proposed approach is described in detail by employing a patent database.