Skip to content

Finding Papers

Start with Google Scholar. It is broad, easy to use and usually the quickest way to learn which words a research community uses.

The goal is not to read everything that comes up. The goal is to find a few good starting papers, then use them to explore the literature around your question.

1. Turn the topic into keywords

Suppose your project is broadly about using XR for robot teleoperation. Searching that whole sentence is not especially useful. Break it into groups of related terms:

Part of the topic Possible keywords
Tech virtual reality, augmented reality, mixed reality, immersive, XR
Task teleoperation, remote manipulation, mobile manipulation, telepresence, education
Question workload, performance, situational awareness, depth perception, latency

Mix terms from different columns:

"virtual reality" robot teleoperation workload
"mixed reality" remote manipulation
immersive teleoperation latency

Write down useful keywords as you find them. The first few papers will often teach you better terminology than the words you started with.

2. Look for recent and influential work

For a fast-moving technical topic, I usually begin by looking at papers from roughly the last ten years, with extra attention to recent work. Among those, citation counts can help identify papers that other researchers have built on.

This is a starting heuristic, not a law:

  • a new paper has not had time to collect many citations;
  • an older paper may be the original or foundational work;
  • citation practices vary a lot between fields; and
  • a highly cited paper is not automatically a good paper—or relevant to your question.

Use the year and citation count to help decide what to inspect, not to decide what is true.

Tip

A recent survey or literature review can be a useful map of a new area. Use it to learn the main terms, methods and papers, then read the important primary sources yourself.

3. Start with a few seed papers

A seed paper is simply a paper useful enough to explore from. Good seeds might be:

  • a recent paper very close to your project;
  • a well-cited foundational paper; or
  • a review or survey paper that gives you a broad view of the topic.

You only need a few. Once you have them, stop relying entirely on keyword search and start following the connections between papers.

4. Follow the citation trail

There are two basic directions:

  • Backward: look through the paper's reference list to see what it builds on.
  • Forward: use Cited by in Google Scholar to see which later papers built on, compared with or criticised it.

Also check the authors. A research group may have published several versions of a system, and the paper you found may sit in the middle rather than at the beginning or end.

5. Go down the rabbit hole

A tool I usually use is ResearchRabbit. It lets you add seed papers and explore their citation network visually. If you are a visual thinker like me, it is much easier to see the connections between papers than in a list of search results.

A simple workflow is:

  1. create a collection for your topic;
  2. add two to five good seed papers;
  3. inspect related, earlier and later work;
  4. add genuinely relevant papers to the collection; and
  5. save the ones you intend to keep into Zotero.

ResearchRabbit is excellent for exploration. Zotero should still be the main home of your reference library.

Warning

Do not mistake a recommendation for evidence of quality. ResearchRabbit helps you find connected papers; you still need to judge the venue, method, evidence and relevance yourself. ResearchRabbit might miss a paper that is important to your question, so do not rely on it alone.

6. Start with established research venues

When you are new to a topic, start with papers from established publishers and research venues. For technical fields, good places to begin include:

For supported publisher sites, set up the Zotero and UTS proxy workflow first. UTS may already provide access to papers that otherwise appear to be behind a paywall.

This is only a starting shortcut. A well-known publisher does not automatically make every paper good. You should still check the specific journal or conference, and whether its method and evidence are convincing.

What about MDPI?

Some MDPI journals publish perfectly useful research, but the quality and reputation can vary considerably between journals. I would not reject a paper simply because it was published by MDPI, but I also would not use MDPI papers as my main starting point when learning a completely new topic.

If an MDPI paper looks relevant, check the journal, the authors, the methodology and how the work has been received. Ask me if you are unsure.

What about arXiv?

arXiv is useful for finding very recent work, particularly in robotics, computer science and AI. However, papers on arXiv are preprints and may not have been peer reviewed. The quality ranges from excellent papers awaiting publication to work that contains serious problems. It is the Wild West.

When you are just starting research, I would not recommend using arXiv as your main source of papers. If you find something useful on arXiv, check whether a peer-reviewed version has since been published. Once you have more experience evaluating methods and evidence, arXiv becomes much more useful for keeping up with the latest work.

7. Save it properly -- or close the tab

If a paper looks useful:

  1. save it to Zotero with the browser connector;
  2. check the metadata;
  3. place it in the right collection;
  4. add a to-read tag; and
  5. add one short note explaining why you saved it.

For example:

Compares point-cloud and video interfaces for remote manipulation; may provide a baseline and workload measures for our study.

That sentence is much more helpful in three months than a forgotten PDF.

If you cannot explain why you are saving a paper, you probably do not need to save it yet.

When do I stop searching?

You are probably ready to pause when:

  • new results mostly repeat themes you already know;
  • you have useful sources for the main claims you need to make; and
  • you can clearly describe what you still do not know.

You will search again as the project changes.

Life, and research, goes on. You will not read every paper, and you will not read every paper in depth.

So, let's decide how deeply each paper deserves to be read.