Status

Hack day project idea(s), inspired by the data science session this morning. Look at a random sample of comments across WordPress.com and…

  • Classify their content (e.g. how they’re responding to the post).
  • Do a topical classification of post content and compare against comment word count or frequency.
  • Calculate diversity of commenters for a site as a function of unique email addresses to number of comments.
  • Build a network graph indicating correlation between commenters across different sites.

The big takeaway: with any given dataset, play with visualizations first before trying to draw a conclusion.

WordPress.com idea: Tweets as comments

Often, when a post is publicized to Twitter (or Facebook), the ensuing conversation then happens on the other platform. The challenge with this is two-fold: the conversation happens out of context of the original piece, and isn’t as accessible as time goes on.

It would be neat to pull in responses to or retweets of a publicize action back into the context of the original post. Furthermore, those external reactions should be ingested in a structured manner, and the comments iA should reflect the nature of type of reaction.

This isn’t a new idea as it’s been done before but it’s still something to be vastly improved.