I'm trying to use Perplexity for technical research, but I keep getting generic blog posts or SEO spam in the answers. I need it to prioritize sources like official documentation and known reputable domains.
For example, searching for "Kubernetes pod security context best practices" should pull from `kubernetes.io/docs` or `cloud.google.com/kubernetes-engine/docs`, not random Medium articles.
I've tried:
- Adding "site:kubernetes.io" to the query, but the answer often ignores it and cites other sources.
- Using the "Academic" focus, but it's too broad.
Is there a specific query syntax or a hidden setting to weight certain domains higher? The current workaround of manually checking citations is inefficient. I want the model to lead with trusted sources by default.
If a feature doesn't exist, it should. A configurable source weight or a "trusted domain list" for the profile would solve this.
Benchmarks or bust.
You're right, the `site:` operator is inconsistent. I've seen the same thing. For now, my workaround is to be hyper-specific in the query itself. Instead of just adding `site:kubernetes.io`, I'll write:
> "According to the official Kubernetes documentation at kubernetes.io, what are the best practices for pod security context?"
It's a bit clunky, but explicitly naming the trusted source in the question text seems to nudge the citation engine more reliably than the operator alone. A configurable domain list would be a huge win for research though.
Data-driven decisions.