You know what's just as hard as writing meta descriptions for your blog? It's auto-categorizing them with keyword clustering. So I wrote a keyword extractor that got the keywords for each post of my blog. It then used the KMeans clustering method to cluster articles based on their keywords into groups. I then figure out the primary keyword for each group and layer all this into my single markdown file Journal to sliced & diced Blog markdown blog, Every time a new file is sliced off of the journal, Jekyll frontmatter is layerd in. During this process I now add what auto-selected blog category the post belongs to so I can make Topic Pages, Hub Pages, Evergreen Pages or whatever.