I am curious about a possible “suggested links” based on natural language: The “unlinked mentions” is totally dependent as I understand it on the exact title of a note. In Tinderbox (Mac OS) from Eastgate and Devonthink they use tf-idf to come up with suggested links based on word frequency. One step further would be to use natural language analysis and deep learning to go beyond crude words and frequencies but also look at context. Are you considering this for later versions? You would then have Linked Mentions, Unlinked mentions (referring to the exact title of a note) and Suggested mentions based on context. That would set you apart from, for example, Roam Research.
By the way, I am not in any way linked to Tinderbox or Devonthink other than that I am a long time user of these apps. No commercial ties.
4 Likes