David Millard

I have a confession. I have been a researcher for over 20 years, and I have published more than 240 papers with hundreds of collaborators, but I have – until very recently – never found a reference manager that I could use.

It’s not from lack of trying.

As a PhD student back in the day my options were pretty limited. I probably started out with nested folders of PDF files, and a master bibtex file, and that works with a single project (although to be honest, not that well!) But over a career that approach simply becomes unmanageable, and the options have got much better. Over the years I’ve tried Papers for the Mac, Mendeley, and Zotero (and probably some others that I have forgotten), but they never seemed to stick. I’d use them for one writing project, but in the next I would drift back into the habit of relying on my own past library of publications and the bibtex files that went with them.

The thing that drew me in to these tools in the first place was the capture process. The bit where you make a magic keystroke while browsing in a digital library and it pulls in all the meta-data and builds your item meta-data for free. It kills a bit of drudge work, and encourages you to reference more widely as a result, both of which are good things!

The thing that drove me away from these tools in the end was the organisation. I always seemed to end up with a massive list of sources that I’d struggle to navigate and thus wasn’t incentivised to refer back to. No matter what combination of tags I’d use it always seemed easier to think ‘now which paper did I reference that in’ and then just refer back to my own library. It wasn’t efficient and it wasn’t pretty, but it worked, and the overhead was never that bad.

But in the last few years I was faced with a couple of writing projects that required more significant effort in finding and organising new sources (big literature reviews, or new areas where I wasn’t familiar with the work that was out there), so – ever the optimist – I decided to try the reference managers again. And this time one of them clicked.

The one I have found that finally works for me is Zotero. The open source ugly duckling that has nevertheless become my workhorse for reference management. And the secret? It was finding a way to manage my resources that worked for me, and that actually meant that the tool became a useful way to navigate and organise my work. It may be something that others found on their own too, but when researching I never saw anything about how best to use Zotero to organise your library, so that is what we are doing here – preventing new researchers from bouncing off Zotero just because its not entirely obvious what you should do.

Collections vs. Tags

Zotero has two main ways of organising. Collections are a hierarchy of folders shown at the top of the left sidebar, with ‘My Library’ as the root. Items (papers, chapters, or whatever, captured with its citation data) automatically appear in the root folder but can be dragged or copied into any subfolder. Clicking on a folder shows all the items inside it, as well as all the items in its subfolders.

Tags are lists of keywords that can be attached to items. By default Zotero will try to create tags for you on import, but you are free to change or expand these. Tags are shown as a cloud in the bottom of the left sidebar. Clicking on a tag will filter the items shown from the currently selected folder based on this tag, you can click on multiple tags to get more specific. You can right click on a tag to assign it to be a coloured tag. This is a bit like a sticky tag that gets moved to the top of the tag window. It also causes a square icon (of the colour you choose) to be placed next to the name of an item in the main list. Its visually appealing and lets you easily see items that have these important tags.

Zotero gives a nice account of this on its help pages.

Opinion is clearly mixed on how to use these features. Collections are hierarchical, tags are not. Multiple tags can be selected at once, but only one collection at a time. Collections determine what is shown in the main list of items, but tags can augment and highlight things in that list. Oh my. Choices, choices.

A lot of people seem to use the collections as topics. They might have a collection for a general area, and then within that they might have subtopics for more specific aspects of that area. The hierarchy then gives you a nice overview of how your items are structured and the topics that they cover. I suspect that if you are certain type of person curating that hierarchy brings you deep pleasure, but that if you are a different type of person it is only deeply painful.

Another common tactic is to use tags for status. For example, you might use coloured tags to flag items as ‘to read’ or ‘current project’ and they are then easy to see in the item list. Clicking on that tag will filter the list and show you only the stuff that is relevant to your current task.

It all seems very reasonable. But don’t do either of these things. People who do these things are wrong.

What about Projects?

The problem with these two approaches is that it doesn’t account for that fact that you will have multiple writing projects, and you want to be able to use your reference manager to manage these separately. Zotero should be the place where you gather items for a project, and where you can organise them according to your thoughts.

Right. I’m not Colonel Sanders. Here’s the secret recipe:

1. Use Collections for Projects. Your highest level collections should be years, and within those years you should break things down by writing project, and within those projects you can organise as you like (e.g. sections of a paper, key themes, whatever is helpful to you whilst you are writing it – its local to this writing project so you change what you do to suit without worrying about the overall structure or other projects). You can collapse the hierarchy so only the bit you are working on is visible. I have a few other high level collections that I find useful. Inbox is where items first go when they are captured, Books is my list of digital textbooks, Teaching is like projects but for teaching rather than writing projects, and I have a Scratch Space that I use for meta-organising things – like creating a list of any items that I want to read in a particular session, or items whose metadata needs some TLC. But the Writing Projects is the heart of this system. Use this to give yourself the space to use the tool to think.

2. Use tags for topics – and be atomic. Tags are much better way to handle topics than collections. For a start this is compatible with the automatic tags that Zotero generates, but you should feel free to alter or extend those if its helpful. The main advantage is that you can select multiple tags. This effectively gives you just in time topics, some of which you wont even know you have created until you need them (much better than curating a big hierarchy). For example, if you have tagged a bunch of items as being about ethnography, and have separately tagged a bunch of items as being concerned with museums, you can can select both to find any papers about ethnography in museums. To make this work you should try to make your tags as atomic as possible (one word if you can). For example, I fought the temptation to create a tag called ‘mixed reality game’ and instead created a tag called mixed reality and a tag called game. I can select both to see the same set of items, but I could also select game and virtual reality to see those, or mixed reality and museum to see any papers about mixed reality (games or not) in museum spaces. Finding these little serendipitous overlaps is a little moment of delight, and who doesn’t want that in their software.

3. Use coloured tags for broad areas. Unless you are one of those super focused research superstars I suspect that your work will fall into several different broad areas. For example, amongst other things I’ve worked in Hypertext, E-learning, Web Science, and Mixed Reality. When I add a paper I try to make sure that it gets tagged with at least one of these as well as the more specific topic tags. This has the visual appeal of creating an item list that shows you a colourful guide as to what broad areas are being shown at any particular time, but more functionally it puts these key tags at the top of the tags list, and gives you an easy to reach first filter when looking for things.

The screenshot below shows what this all looks like in my copy of Zotero for Mac (click to embiggen), showing a particular writing project I was working on last year (around the topic of mixed reality ethics).

You can see that the hierarchy of collections gives me a nice working space, the tags allow for powerful filtering and searching, and the coloured tags are easy to get to as a first filter, and show quickly on the list what mix of areas I happen to be looking at.

I now have a system that not only allows for quick capture, and easy look up of sources from past papers (I can look in the relevant collection), but that also gives me a thinking space for new writing projects where I can use sub-collections for information triage and organisation, and a powerful search system based around area and topic tags.

I hope that is useful to someone out there. Once you have been using it for a while it does have an unfortunate side effect as when you see someone else using Zotero with collections for topics you will scream ‘you are using it wrong!’ and die a bit inside, but that is a small price to pay for actually having a useful reference manager that feels like a help rather than a chore.

Good luck!

(Header image is from Wikipedia Commons)