1.2. Finding scientific literature
Searching for scientific literature
- Microsoft Academic: better organized and filtered, plus useful visual and networking tools.
- ResearchRabbit: powerful AI-based search.
- SemanticScholar: best ordering and interesting insights on citation flow + helps you obtain the PDF, but incomplete citation data.
- Google Scholar: most complete, but a lot of noise (low-quality papers) and suboptimal ordering. I tend to recommend novice researchers to start with Microsoft Academic, to avoid ending up with many lower-quality publications.
In comparison with search engines, databases include less publications. They have more strict inclusion criteria for journals and publications, which tend to leave out less important publishers, languages and documents.
Main limitation: most require a paid subscription by the university.
Especially useful if needing to establish a robust, finite and replicable set of publications (e.g., for a research synthesis). Otherwise, you probably don’t need them: use a good search engine like Microsoft Academic.
Organizing papers: reference management
Reference management software:
- Zotero (recommended)
- auto import citation data from websites (with Zotero Connector)
- auto identification of PDF files (finds good citation metadata based on the PDF, but still needs some manual check afterwards)
- integration with Word
- free & open source
- powerful, endless possibilities of tagging, notes, exports, collaboration (with Zotero Groups)
- also install the Zotero Connector browser extension (for Firefox, Chrome, Safari or Edge)
- Quick start guide
- Recommended extensions:
- Mendeley (also quite good, but in my opinion less powerful than Zotero).
One-time bibliography generation:
How to access full-text papers? (e.g., in PDF)
Many quality papers from international journals are sadly behind paywalls: if your university does not have a subscription (which is very rare in Ecuador), you have to pay for the PDF. Do not pay for one paper! It’s too expensive for a single publication, especially if you’re not sure how good or relevant it’s going to be. And authors don’t get a single cent from scientific publications.
There are solutions:
- Open access journals/papers: some papers are free to download (🔓).
- Repositories: many authors now publish a full version (often a preprint) on a personal website or an institutional repository (a website from their university which contains all publications by the university’s faculty), or on “academic social networks” such as ResearchGate.
- University library and proxy access for some subscribed content:
- Access the publisher’s website through your university proxy, if your university has a subscription (sadly very rare in Ecuador and the Global South, because too expensive for most universities).
- As a last resort, e-mail the first author asking for the PDF: most people will gladly send it to you.
Not fully legal solutions:
- Sci-Hub for papers (not legal, but considered legitimate by the large majority of the scientific community).
- Library Genesis for books (a bit more problematic in terms of copyright infringement than Sci-Hub).
Prioritising the literature
Identifying key sources (when starting researching a topic)
- Citation count/year
- When comparing publications' citation count, normalize by year since publication: a paper from 2020 with 10 citations is actually more influential (10 citations/year) than one from 2000 with 100 citations (4.7 citations/year).
- Most recent papers will refer to other important recent publications.
- ResearchRabbit can help to identify most cited/relevant papers
- Highly influential papers (SemanticScholar)
- Snowballing references
Source: J. Hayton, Day 12: How to filter the academic literature (2018)
- A: best + most relevant → read and re-read
- B: high quality + relevant, but not essential
- C: maybe interesting (ok quality, not clearly relevant) → keep for later
- D: not relevant at all / low quality → throw away/do not spend time