Literature Search and Paper Collection: A Practical Guide for New Researchers

If you’re just starting out in research, one of the first — and most underrated — skills you need is knowing how to find the right papers. It sounds simple, but it’s not. Good literature search isn’t about typing a phrase into Google and grabbing the first ten results. It’s a skill that directly shapes the quality of your research: the papers you find determine the questions you ask, the methods you consider, and the gaps you’re able to identify.

This guide walks through a practical workflow for searching, deciding what to read, reading efficiently, and organizing academic literature.

1. Start with the Right Mindset

Before touching a search engine, get clear on why you’re searching. Literature search usually serves one of a few purposes:

• Orientation — you’re new to a topic and need a broad overview.
• Targeted search — you need papers on a specific method, dataset, or result.
• Comprehensive review — you’re writing a survey or the related-work section of a paper and need to be exhaustive.
• Staying current — you want to track new papers in a field you already know.

Each purpose calls for a different strategy, so knowing which one you’re doing will save you a lot of wasted time.

2. Know Your Paper Types

Not all papers carry the same weight or serve the same purpose. Before you decide whether a paper is worth reading, understand what kind of paper it is.

Why this matters: if you cite a preprint as if it were peer-reviewed, or treat a conference paper as less rigorous just because it’s not a journal, you’re misjudging the evidence. Different fields also weight these differently — in ML/CS, conferences often lead; in biomedical fields, journals (and systematic reviews) dominate.

3. A Decision Tree: What Should You Read First?

New researchers often jump straight into narrow, recent papers and end up confused because they’re missing context. Instead, follow a top-down path:

Are you new to this topic?
│
├── YES → Start with a Systematic Review / Survey Paper
│         (Search: "[topic] systematic review" or "[topic] survey")
│         Goal: understand the landscape, terminology, and open questions.
│         │
│         └── Then identify the MILESTONE papers
│             (the highly-cited, field-defining works the review keeps referencing)
│             Goal: understand the foundational ideas and methods everyone builds on.
│             │
│             └── Then move to RECENT papers (last 1–3 years)
│                 Goal: see where the field is now and what's still unsolved.
│
└── NO, I know the field → Go directly to targeted/recent search
          (Section 5: Building Better Search Queries)
          + track new work via alerts (Google Scholar alerts, arXiv daily digests)

How to find milestone papers:

• They’re the ones a review paper cites repeatedly, or calls “seminal,” “foundational,” or “landmark.”
• They usually have very high citation counts relative to their age (a 2015 paper with 5,000 citations is a milestone; a 2023 paper with 5,000 citations might just be trendy).
• Search engines like Google Scholar often surface them near the top when you sort by relevance for a broad query.

Rule of thumb: Review paper → milestone papers → recent papers. Skipping the review step is the single most common reason students feel “lost” reading a new field.

4. Where to Search

Don’t rely on a single source. Different databases index different things, and combining them gives you much better coverage.

Tip: Use Google Scholar or Semantic Scholar as your entry point, then switch to a field-specific database once you know the key terms and authors in your area.

5. Building Better Search Queries

This is where most beginners struggle. A weak query returns irrelevant or overwhelming results. A strong query is built deliberately.

a) Identify your core concepts

Break your research question into 2–4 key concepts. For example:

“Does exercise improve memory in older adults?” → Concepts: exercise, memory, older adults

b) Use synonyms and related terms

Different authors use different vocabulary. Combine variants with OR:

("physical activity" OR exercise OR "aerobic training")

c) Combine concepts with Boolean operators

("physical activity" OR exercise) AND (memory OR cognition) AND ("older adults" OR elderly)

d) Use filters

Narrow by publication date, document type, or field to cut noise — especially useful when a topic has decades of literature.

e) Use quotation marks for exact phrases

"deep reinforcement learning" returns very different results than deep reinforcement learning without quotes.

6. Finding Similar Papers: The Snowball Technique

Once you find one good paper, don’t stop there — use it as a hub to find everything around it. This is often more effective than keyword search, because it follows the field’s actual citation network instead of guessing at vocabulary.

• Backward snowballing (what it cites): Open the reference list. These are the foundations the authors built on — often includes milestone papers and closely related prior work.
• Forward snowballing (what cites it): Use “Cited by” on Google Scholar (or Semantic Scholar / Web of Science) to see newer papers that build on, extend, or critique it. This is how you find the most recent developments on a specific idea.
• Author / research group tracking: Look up the authors’ other publications and their lab or group page. Research groups tend to work on a connected thread of problems for years — one paper often opens the door to 5–10 closely related ones from the same group.
• Venue tracking: Note which journals or conferences keep appearing; browse their recent issues/proceedings directly for more relevant work.
• Visual tools: Use Connected Papers or Semantic Scholar’s citation graph to see this network visually instead of clicking through one by one.

A good routine: for every important paper you find, spend 5 minutes checking its “cited by” list and its authors’ recent publications before moving on.

7. Two Modes of Reading: Skim First, Deep Read Second

You cannot deep-read everything — you’ll run out of time before you run out of papers. Use a two-pass system.

Pass 1: Quick Skim (2–5 minutes per paper)

Goal: decide if the paper is worth your time.

1. Title & abstract — What’s the claim? Is it relevant?
2. Figures and tables — Often tell you the core result faster than the text.
3. Introduction’s last paragraph — Usually states the contribution directly.
4. Conclusion — Confirms what was actually found/achieved.

If it passes this filter, tag it for a deep read. If not, file it away (you may still cite it briefly) or discard it.

Pass 2: Deep Read (30–90 minutes per paper)

Reserved for papers directly relevant to your work. Read in this order, not top to bottom:

1. Abstract & Conclusion — reconfirm the big picture.
2. Introduction — understand the motivation and how they position their work against prior work.
3. Methods — understand exactly what they did (this is where you evaluate rigor).
4. Results — check whether the results actually support the claims.
5. Discussion/Limitations — see what the authors themselves admit is uncertain.
6. Related work / References — mine this for more papers (see Section 6).

Tip: Take notes while reading, not after. A one-paragraph summary written immediately after finishing is far more useful later than trying to recall the paper from memory weeks later.

8. Evaluating What You Find

Not every paper deserves a deep read. Quickly triage using both the skim pass above and these checks:

1. Paper type — journal, conference, preprint, review? (Section 2) Adjust your trust accordingly.
2. Venue/journal reputation — Is it peer-reviewed? Reputable in this field?
3. Citation count relative to age — a highly-cited old paper is likely foundational; a lightly-cited old paper may be less influential.
4. Recency — Is it still relevant, or has it been superseded by later work?
5. Methodology — Does the method actually support the claims being made?

9. Organizing What You Collect

Finding papers is only half the job — you also need a system so you don’t lose track of them.

• Reference managers: Use tools like Zotero, Mendeley, or EndNote to store PDFs, citations, and notes in one place.
• Tagging: Tag papers by theme, method, paper type, or relevance (e.g., “background,” “milestone,” “method-comparison,” “must-cite”).
• Annotate as you read: Write a 2–3 sentence summary of each paper’s contribution right after reading it. Future-you will thank present-you.
• Keep a running spreadsheet: Columns like Author, Year, Venue/Type, Method, Key Finding, and Relevance make comparing papers much easier when you start writing.

10. Common Mistakes to Avoid

• Stopping after one search engine. Different tools surface different papers.
• Skipping the review paper step. Diving into narrow recent papers without first understanding the landscape leads to confusion.
• Only reading the abstract. Abstracts can be misleading — always check the actual results/methods before relying on a claim.
• Ignoring older “classic” papers. Foundational/milestone work is often still cited and expected knowledge.
• Confusing preprints with peer-reviewed work. Always note whether a paper has been peer-reviewed.
• Not saving search strategies. Write down what queries you used and where — you’ll need to repeat or refine the search later.
• Hoarding papers without reading. A folder of 200 unread PDFs helps no one. Read, summarize, and file as you go.

11. A Simple Weekly Workflow

For students building this habit, a lightweight routine helps:

1. Set a specific search question for the week.
2. If it’s a new topic, find a systematic review/survey first, then identify milestone papers.
3. Search 2–3 databases with refined Boolean queries for anything more targeted.
4. Snowball (backward + forward + author/group tracking) from your best 2–3 hits.
5. Skim-triage everything found; select 5–10 papers worth a deep read.
6. Deep-read and log summaries in your reference manager.
7. Note any recurring authors, methods, or gaps for next week’s search.

Final Thought

Literature search isn’t a one-time task before you start research — it’s an ongoing skill you’ll use throughout your entire career. The sooner it becomes second nature — knowing what type of paper you’re looking at, where to start in an unfamiliar field, how to trace a citation network, and how to read efficiently — the more time you’ll have for the part that actually matters: thinking deeply about your own ideas.