Federated Search was a session at IndieWebCamp Utrecht 2019.
Notes archived from: https://etherpad.indieweb.org/fedsearch
IndieWebcamp Utrecht 2019
Session: Federated Search
When: 2019-05-18 14:55
Project Searsie.org from Djoerd Hiemstra
- Example of research done on search engine manipulation. Search is influential. Research by Robert epstein and Ronald Robertson.
- People who got results skewed towards a specific candidate were leaning towards that candidate in an election.
Search invades privacy. We have to trust the ppl/org who make the search engine
Search is expensive.
Searsia's approach: Federated search. Little search services combined from several searches. Wiki/Amzon/Stack Overflow/Youtube/Twitter/Facebook/Reddit/Instagram
Aggregate to specific SERP
There is no Federated Search accepted standard, design or how to share SERP (OpenSearch is one...)
Wrote their own scraper, uses API, scrape HTML
Designed with as little administrative overhead as possible
Stores SERP from previous queries and when not busy, it will send random queries to engines to see what comes back
Q: What comes on top? How do you decide what comes on top?
Demo of search engine for University of Twente (https://search.utwente.nl)
It doesn't crawl the web but it learns from samples and previous searches.
Referrals in SERP can be turned on and off.
Q: Could we have a Indieweb search engine? What if we want to go through specific sites we want to search? Do the site needs some specific markup?
- You might need a search endpoint. We could get the RSS feed, we could do a local search. You make a config file for your own search engine.
Serversoftware is written in Java. You can run it yourself. See http://searsia.org/protocol.html
wordpress API often disabled because "security".
hook into existing search features in sites?
Couldn't we make our own configfiles and connect them, make an indiesearch engine? Should be possible
Searsia looks as very configurable