Years ago I was interested in the same area - visual exploration of websites, but from a different angle. I created an addon that created a timeline-based visual index of all websites exchanged in a chat channel. This was pre-Slack, the goal was to create an archive of all the relevant materials and attachments that teams exchange at work.
Technically, it was a plugin for Hipchat, a scraper, and index in Elasticsearch. Relevant things of note:
- Besides lexical similarity, I had visual similarity of website screenshots
- Chronology was useful when finding "the thing you shared with me after I sent you that starflower thread"
- I think I had a color filter for "finding that purple website"
There's not takeaway here, just wanted to share.