I need to systematically track literature for my startup's R&D. Traditional keyword searches in academic databases are inconsistent and miss connections.
I'm testing Iris.ai to build a reproducible search methodology. The goal is to lock down a process that any team member can run quarterly, yielding the same core relevant papers. Has anyone mapped out a strict workflow with this tool? I'm looking at the Researcher Workspace. Specifically, how do you structure the initial "seed" documents and refine the context to avoid scope creep in results? The pricing jump for teams is significant, so I need to know if the reproducibility justifies the cost over manual PubMed/Google Scholar searches.
I'm actually trying to figure out something similar for tracking papers in my data engineering niche. That pricing jump is a real concern.
For the seed documents, I found that uploading 3-4 "perfect" papers you already know are core to your topic works better than just a broad abstract. It seems to anchor the context more tightly. But I've also had it suddenly pull in weirdly off-topic results after a few iterations. How do you handle the "refine context" step to lock that down? Do you exclude terms one by one?
The reproducibility is tempting for a quarterly process, but I'm wondering if a simple saved search in Semantic Scholar plus a manual deduplication step is almost as good for a small team. Have you run a cost/benefit on the time saved vs. the subscription?
rookie