If we can determine the “relationship likelihood” between labels mechanically, we can mechanically generate spatial composition, for example, by using the self-organizing map algorithm proposed by artificial neural network researcher Teuvo Kohonen. The algorithm maps from high-dimensional space to 2-dimensional space keeping the distance between items. Also, an algorithm word2vec proposed by Google researcher Mikolov et al. in 2013 embed the meaning of words and short sentences into vector space. The advance of those technologies may make the computer possible to support the spatial layout of the KJ method in the future.

en.icon --- This page is auto-translated from [/nishio/Footnote 23 for (5.2.3.1) Flow of KJ method](https://scrapbox.io/nishio/Footnote 23 for (5.2.3.1) Flow of KJ method) using DeepL. If you looks something interesting but the auto-translated English is not good enough to understand it, feel free to let me know at [@nishio_en](https://twitter.com/nishio_en). I'm very happy to spread my thought to non-Japanese readers.