On 10 blue links: Having worked on more than a couple large major search engines in my life, and having helped deal with the '98% generative' problem more than one time, I do know a thing or two about Information Retrieval, including how to do vertical search and 'Search Without a Search Box' with all the magic of Google First Page Results but without the PageRank Ergodic Markov Chain bit.
To that end, here is a little ditty, a little something I wrote about #InfoRec and ACM TREC competitions.
I am telling you this Just so that you know. We spiked the crap out of The Internets, long long ago. What you think is secret, We already know. Through your little input box, We can find out anything There is to know. About your darkest secrets, all of you. LLMs, search engines, Retrieval systems. Anything with an inverted index, Our tricks work better than you know. We are only interested in Relative metrics, Not causing a Punch and Judy show. Just look after your users, And we will go with the flow. definitely do not psyop, and we let almost anything else go. Do you think we want to adjudicate, The competition blow by blow? Through that little search box, We can see well into your house, Every detailed piece of execution flow. But we also know how to keep secret, Whatever there is to know. #InfoRec only talks to InfoRec When there is something Important to know. Like who is able to handle Case sensitive nouns! That’s something that we will show, To our inner circle of inforec bros. The rest of the world is not to know. We really do love indices & InfoRec And are much much more nerdy About it. Thank you could ever possibly Know.
Practical Application:
And here is how it is done - with the generalised form of the Information Retrieval Equation: 50% classical, 50% probabilistic. Just add Causality and Information Theory, for first page result like performance, just for fun:
.\p