Step 1: Define your research interest with precision

Broad areas like “machine learning” or “renewable energy” will return thousands of papers. Narrow it down before you search. Instead of “machine learning in healthcare,” try “CNN-based lung nodule detection from CT scans.”

Ask yourself:

• What specific problem am I solving?
• What domain and sub-domain does it fall under?
• Is there an application context (clinical, industrial, social)?

Answering these questions first will make finding the right base paper for project far faster.

Step 2: Check journal quality and indexing

Not all published papers are equally credible. For PhD work especially, your base paper in research should come from a journal indexed in at least one of the following:

• Scopus — widely accepted by Indian universities and UGC
• Web of Science (SCI/SCIE) — highest tier, recognised globally
• IEEE or ACM — respected conference and journal publishers in technology fields

Use Scimago Journal Rank (SJR) or Journal Citation Reports (JCR) to verify the journal’s impact factor and quartile (Q1/Q2 papers carry more weight).

For Anna University scholars: check that the journal appears in the relevant annexure list for your department. Your Kenfra research advisor can help verify this quickly.

Step 3: Read the abstract and conclusion first

Do not read a paper front to back when shortlisting — it takes too long. Instead:

1. Read the abstract to check topical relevance and the problem being addressed
2. Jump to the conclusion to see what the paper achieved and, crucially, what it admits it could not do
3. Skim the introduction to understand the research gap the authors identified

If the paper’s problem, domain, and methodology feel relevant to your goals after this three-minute scan, add it to your shortlist.

Step 4: Evaluate the methodology carefully

This is the most important technical step. Your base paper’s methodology will likely become the starting point for your own. Look for:

• Type of research: experimental, simulation-based, analytical, survey-based?
• Dataset or data source: is it publicly available? Can you use or extend it?
• Tools, algorithms, or frameworks: are these accessible to you?
• Evaluation metrics: accuracy, F1-score, RMSE — do these match your own evaluation plan?

A good base research paper has a methodology that is clear, replicable, and documented in enough detail that you could reproduce the core experiment. If you cannot understand what they did, it is not a good base paper for your work.

Step 5: Look for the research gap

This is the gold in any base paper. Most high-quality papers end with a limitations section or a future work paragraph. These are your research opportunities.

Look for sentences like:

• “The proposed model was not evaluated on multi-lingual datasets…”
• “Future work should extend this approach to real-time scenarios…”
• “The model’s performance degrades when sample size falls below 500…”

Each of these is a potential gap your study can address. The sharper and more specific the gap, the stronger your research justification.

Step 6: Check citation count and recency

High citation counts signal that the academic community has validated the paper’s contribution. As a rough guide:

• A paper published 2–3 years ago with 50+ citations is performing strongly
• A paper published 5+ years ago with 200+ citations is a cornerstone in its area

Use Google Scholar’s “Cited by N” link to see which papers have built upon it — this is also a useful way to find related base papers for project.

Recency matters too. For fast-moving fields like AI, NLP, or cybersecurity, a paper from 2019 may already be outdated. Aim for work published within the past 3–5 years unless the foundational concept is stable.

Step 7: Confirm full-text access

Before getting too attached to a paper, make sure you can actually read all of it. Options include:

• Your university or institutional library subscription
• ResearchGate — authors often upload their own papers; you can request a copy directly
• Open access versions on the author’s personal or institutional page
• Unpaywall browser extension — finds legal free versions automatically

You need the full paper. The abstract alone is not enough to understand the methodology or identify the true research gap.

Step 8: Discuss with your supervisor before finalising

After narrowing down to two or three strong candidates, bring them to your supervisor or research guide. They can:

• Confirm the paper’s relevance to your department’s focus areas
• Flag whether the gap you identified is genuinely novel or already addressed elsewhere
• Suggest stronger alternatives you may not have found
• Validate that the journal is acceptable to your university

Supervisor approval at this stage saves you from discovering problems six months later.

Step 9: Avoid predatory journals

Predatory journals accept and publish papers with little or no peer review. A paper from a predatory journal is not a valid base paper for research — it will undermine your credibility entirely.

Watch for these red flags:

• No indexing in Scopus, Web of Science, or IEEE
• Promises of very fast peer review (days, not weeks)
• Vague or incomplete editorial board information
• Journal name that closely mimics a reputable publication

When in doubt, check Beall’s List or ask your supervisor.