# Topic Maps for Written Content In the book [[Fluent Forever]] the author suggests learning the top 600-1000 words, by frequency, for a language and that these are different per language. Not only are the words different, but so are the associations and connotations of each word, and the topics which are discussed. This would show up in the differences of those list of words. We can take this a step further by creating maps, using something like [[UMAP]] to create easy to visualise representations of the content and topics and connections of written content, and in this case use it to understand what the most important words, topics and conversations are and in what context they are discussed. This doesn't have to be limited to learning languages. We can generalise this concept to mapping any set or group of written content, and understanding what's important and how it relates. As an example, we could use this to [[Build a better Goodreads]] by mapping all topics written about in books, seeing which books are related, how many topics there are, in what areas of the map we have read extensively and what areas of the map are sparse, or where our reading has been sparse. This would be helpful in trying to find books with contrary opinions but on the same topic. They would be in the same neighborhood but disconnected (in terms of references or community of readers). We can also use this to map the topics and discussions happening in a specific form of content such as news. This would give us something similar to the narrative machine used by [[Epsilon Theory]]: ![[Epsilon Theory Narrative Map.png]] We could also use it in a specific niche such as scientific papers on economics. There are two great graphics in the book [[From Galileo to Modern Economics]] using this concept: - The first one maps the four volumes of [[Pareto]]'s Treatise and shows what groupings are formed from the topics discussed in the four different volumes and what topics each volume focuses on. - ![[Paretos Treatise Topic Map.png]] - The second one looks at the clustering of topics in the subfield of econophysics over the course of almost two decades.