01 · The Question
What Should a Tag Actually Tell You About a Paper?
You begin with a few sensible tags: “AI,” “higher education,” and “qualitative.” Then come “student perceptions,” “interviews,” “positive findings,” “theory,” “background,” “important,” “maybe cite,” and several variations of terms you have already used.
Soon, organizing the tags takes almost as much thought as organizing the literature.
The problem is not that topic, method, finding, or argumentative role is the wrong way to tag a paper. Each captures a different and potentially useful dimension. The real question is which dimensions will help you retrieve and use literature later, and whether you can apply them consistently enough to justify maintaining them.
03 · What You Need to Know
Different Tags Answer Different Research Questions
Topic Tags Tell You What the Paper Is About
Topic tags describe substantive subject matter: “academic integrity,” “AI literacy,” “doctoral supervision,” “feedback,” or “technology acceptance,” for example.
They are useful when your future question is something like, “What have I collected about AI literacy?” or “Which papers discuss feedback?”
Topic tags are often the easiest to understand but also the easiest to overproduce. One paper can contain dozens of concepts, and tagging every term that appears in it creates noise rather than retrieval value.
A useful topic tag should therefore represent a concept you are likely to search across papers, not merely a word that happens to occur in one.
Method Tags Tell You How the Research Was Conducted
Method tags describe features such as research design, data source, analytical approach, or methodological tradition. Depending on your work, useful tags might include “randomized trial,” “cross-sectional survey,” “interviews,” “structural equation modeling,” “thematic analysis,” or “mixed methods.”
These become valuable when you need to compare methodological choices across the literature.
Suppose you later ask, “How have researchers measured AI literacy?” or “Which studies used interviews rather than surveys?” Topic tags alone will not answer those questions efficiently.
Finding Tags Tell You What the Research Reported
Finding tags attempt to characterize results. These can be useful, but they require particular care because findings are usually more nuanced than labels.
A tag such as “positive effect” may conceal what improved, compared with what, in which population, under what conditions, using which outcome, and with what uncertainty. The tag can help retrieve potentially relevant studies, but it should not replace your actual notes or extracted evidence.
Watch Out
A finding tag is an index, not a conclusion. Do not reduce a conditional, mixed, uncertain, or outcome-specific result to a categorical tag that becomes stronger than the evidence reported in the paper.
If you need detailed findings, keep them in notes or structured extraction fields where the claim can remain attached to its source and context. Tags work better as retrieval handles than as miniature literature reviews.
Argument Tags Tell You Why You May Use the Paper
A different kind of tag describes the paper's possible role in your own reasoning: “supports rationale,” “counterevidence,” “method precedent,” “theoretical foundation,” “definition,” “boundary condition,” or “discussion comparison.”
These tags can be particularly useful during writing because they answer a different question from subject classification: “Why might I need this paper?”
A paper about AI feedback, for example, might carry the topic tag “AI feedback,” the method tag “experiment,” and an argumentative-role tag such as “counterevidence.” Those labels are not redundant. They describe different properties.
Separate What the Paper Is From What You Have Done With It
One useful distinction is between descriptive tags and workflow statuses.
Descriptive tag
Describes the source or its possible intellectual role: topic, method, finding, theory, or argumentative function.
Workflow status
Describes your relationship with the source: to read, screened, read, included, excluded, or another project-specific state.
Mixing these categories is not inherently wrong, but you should know what each label is doing. The distinction among found, screened, read, included, and cited papers is fundamentally about workflow and research decisions rather than subject matter.
Zotero explicitly supports tags for topics, methods, status, ratings, and workflows such as “to-read.” This flexibility is useful, but it also means the software will not design the taxonomy for you.
Collections and Tags Solve Different Organizational Problems
In Zotero, collections allow items to be grouped hierarchically, while tags provide more granular characterization and filtering. A single item can belong to multiple collections without being duplicated, and an item can have multiple tags.
That suggests a useful practical distinction: collections can represent relatively stable containers such as projects or broad bodies of work, while tags can represent characteristics that cut across those containers.
For example, a paper might belong to your “Doctoral Education” project collection while carrying tags for “qualitative,” “supervision,” and “theoretical framework.” Another project can use the same underlying item without creating another copy.
This is particularly helpful when you need to preserve one paper across several projects.
Use a Controlled Vocabulary Where Consistency Matters
The usefulness of tags depends heavily on consistency.
If some papers are tagged “qualitative,” others “qualitative research,” others “qual,” and others “interview study,” you may fail to retrieve relevant items together even when that was your intention.
You do not need a formal controlled vocabulary for every personal research library, but you do need enough discipline to avoid accidental synonyms, inconsistent capitalization, singular-plural variations, and tags that mean different things on different days.
Before creating a new tag, ask whether an existing one already represents the concept adequately.
Do Not Tag Information You Would Never Retrieve by Tag
This is perhaps the simplest test for whether a tag deserves to exist.
Imagine yourself six months from now. Can you formulate a plausible retrieval action involving the tag?
“Show me all longitudinal studies.” Useful.
“Show me papers providing counterevidence to this claim.” Potentially useful.
“Show me papers tagged interesting.” Perhaps less useful unless “interesting” has a stable operational meaning in your workflow.
Tags should reduce future search effort. If assigning and maintaining them consumes more effort than they save, the taxonomy has become its own research project, which is rarely the intended methodology.
Automatic Tags May Not Match Your Personal Taxonomy
Reference managers may import subject terms from external records. Zotero, for example, can automatically add keywords or subject headings when items are saved from the web. Its documentation distinguishes these automatic tags from manually added tags and allows users to hide or remove automatic tags or disable their automatic addition.
Imported terms can sometimes be useful, but they were not necessarily created for your research questions. If they overwhelm your own vocabulary, they can make filtering less useful rather than more useful.
