learn-awesome/learn

Automatic topic identification from URL

Open

#189 opened on Aug 4, 2020

View on GitHub
 (0 comments) (0 reactions) (0 assignees)HTML (46 forks)batch import
Hacktoberfest

Repository metrics

Stars
 (338 stars)
PR merge metrics
 (No merged PRs in 30d)

Description

Currently, we are unable to import a lot of book summaries, because we'd need to tag those books with topics they cover. This topic tagging must remain high-quality, otherwise we'll end up with thousands of links under vague categories such as "business" or "personal-development" / "self-help".

However, the topic taxonomy itself cannot be static. While we might not have a topic as granular as "History of Tuberculosis in China", we may need to decide that at some point "Medicine in China" might be worthy of a topic. Some of this discussion is in #14

There's back-and-forth tension between assigning topics/tags to an item using the current taxonomy, and keeping tha taxonomy itself up-to-date by adding/removing topics/tags. So we need a system that can change both the taxonomy and the topic tagging appropriately while removing the manual tagging as a bottleneck. Would love to have more ideas here.

Contributor guide