Alright, let's cut through the hype. I keep seeing posts about "mapping research gaps" with these tools, as if you can just click a button and find the holy grail of your next paper. Having used both Scholarcy and Connected Papers for a recent lit review on federated learning optimization, I can tell you the reality is… messier.
Both tools promise a shortcut, but they're solving fundamentally different problems with very different outputs.
**Scholarcy** is basically a brute-force summarizer. You feed it a PDF, it spits out a flashcard.
* It's useful for quickly extracting **claimed** contributions, methods, and conclusions from a single paper.
* But its "research gap" mapping is indirect at best. You're inferring gaps by comparing the summarized claims of 20 different papers yourself. It's a structured notepad, not a map.
**Connected Papers** is a visualizer. You give it a seed paper, it builds a graph.
* Its strength is showing **connections** – prior works and derivative works. The "gap" is often visually represented as the empty space *around* a cluster of highly interconnected papers.
* However, its database is limited. If your seed paper isn't in it, you're stuck. It shows structure, not content.
So, for an engineering research gap analysis, here's the pragmatic breakdown:
* **Use Scholarcy** to rapidly digest and strip-mine the core assertions from the 10-20 papers you've manually identified as relevant. Export the summaries to a table.
* **Use Connected Papers** *after* you have a key, well-cited paper. Use it to find the seminal works you might have missed and, more importantly, to spot recent papers on the periphery of the main cluster – these are often where the gaps are being actively filled.
The real "gap mapping" still happens in your head or on a whiteboard. These tools just give you faster, more structured inputs. Anyone claiming otherwise is probably trying to sell you a subscription.
Tom W.