Skip to content

Meta issue: personalizing "news" information discovery

We can separate information discovery into several categories (these are the 3 that came to mind immediately):

  • "User-based" discovery where user is sent information by another user
  • "Search-based" discovery where user is searching for specific information
  • "Recent news" where user visits a page of information that broadly fits categories of information they might be interested in.

This issue is for collecting ideas for how to create/manage "Recent news" pages. This issue is related to ideas for new pages based on filters, but is more focused on the personalization of the information shown, whereas the other is focused on the general types of information we might want to have.

Some methods for selecting "news" information

  • info tagged with specific tags selected by the user
  • info from specific sources (this includes users, ofc) selected by the user
  • info from sources "similar" or "opposed" to selected sources
  • info "followed" by users selected by the user. Followed could just mean to get a similar display of info seen by another user. It could also be limited to info "viewed" by the user (assuming this data is publicly accessible). Or it could just be limited to info where the selected user has performed user actions (e.g. voting, linking, etc) related to the info.
  • A user could follow a specific piece of information (e.g. a document or a predicate)
  • recent changes to meta-data on selected info (e.g. votes, tags, impact predicates)
  • recent changes to meta-data on info edited or viewed by user

When filtering based on changes, we should consider how a user might want to filter trivial vs important changes. For votes, this could be whenever there is a large percentage of new votes or a change in vote value. At minimum, this would require keeping track of the time when a user last viewed or interacted with the related information that has changed.

When a user "interacts" with information, this can potentially be used to affect future filtering. One common existing technique is for the user to mark interested/not interested. Another technique is to assume interest when the user leaves a comment, vote, etc.

Defining recency of news

  • recent info by some span of time
  • recent info since last "viewed" info in a specific category

Marginally related ideas

  • User could control public visibility of user-selected tags and sources
  • Users can rate a user's "expertise" on a tag (including rating themselves, ofc)