LSI is just a SEO term for related, similar or synonyms words. Before this term has came into picture the only thing to determine relevancy of article was keyword density. Which means if your post is about “abc” and you have used it 10 times throughout the article than your competitor needs to use it 11 time to beat your position in Google search, but let me ask you, is it fair?? Same Thing applies for search engine, instead of taking high keyword density into consideration, search engine bot looks for related keywords (Latent semantic indexing keywords) throughout the article. It doesn’t mean that we shouldn’t maintain keyword density for target keyword in our article, but over optimization can cause issues. I usually keep it between 1- 3 %.
LSI makes a lot of sense too. For example – Search engine discovers a page with the word “Apple” on it, how does it differentiate whether the Apple is a fruit, Apple is a brand or Apple is name of any pet? It uses Latent semantic indexing keywords, if there are words like fruit, taste, flavor in the article then it would be easy for search engine to determine that the article is all about the “Apple fruit”. Hence in this example fruit, taste and flavor can be termed as LSI keywords for keyword “Apple”.