This is how I use the Mendeley system for managing my phd. Mendeley has the ability to full-text search downloaded academic papers, and features for adding notes, tags, keywords and more.
The advantage is that I can quickly retrieve a paper that is relevant to a given current work, and filter on how many times I've read the given paper, and what the topic of the paper is. Very useful indeed.
How I use tags
Tags are for reading queues.
I organize the papers I have downloaded into a set of reading queues. Each paper I'm writing is associated with a queue. Each queue has three buckets where a paper is placed depending on how far I am reading it. These buckets are pass0 (never read), pass1 (read abstract,introduction,conclusion), pass2 (read whole thing), pass3 (read whole thing in excruciating detail).
So to summarize, I use two tags. One tag for the queue-name (e.g. "reading_queue_xyz"), one tag for the progress (e.g. "pass1").
Example tag list for a paper:
tags: reading_queue_xyz, pass1
I have a special reading queue called "culture", which is simply for stuff I want to know in general. If a paper is not in a queue, I should delete it!
The use of tags could be extended to include a priority, like "highprio", "medprio", "lowprio"...
How I use notes
Notes are for "how will I use this".
I use notes to capture the purpose of why I reading the paper. How will I use it in my current work? Is it for the related works section? Is it part of the motivation?
Use this paper for related work section in mapping paper
How I use keywords
Keywords are for objective descriptions of content.
This is not related to how I intend to use a given paper in my work. I use a mix of general keywords and very specific ones.
databases distributed systems fault tolerance
paxos spatial join
How I use the "favorite star"
When I'm in a hurry I'll use the favorite star to indicate that a paper is related to current work.