Web Search Engine

In productionUniversity

Domain-restricted crawler with TF-IDF ranking, stemming, and fuzzy matching.

belief propagation convergenceSample data

1,284 documents · 41 ms · BM25 + positional index

All1,284Papers612Notes431Code241

Loopy belief propagation on factor graphsnotes/inference/loopy-bp.md · …convergence is not guaranteed on graphs with cycles, but in practice…
Notes0.941
Damping schedules for message passingpapers/mp-damping.pdf · …a damping factor of 0.5 stabilised convergence across all tested…
Papers0.887
index/postings.pycode · def merge_postings(a, b): walk two sorted postings lists in one pass…
Code0.812
Variational alternatives when BP divergespapers/variational-notes.pdf · …mean-field gives a bound rather than an approximation…
Papers0.774
Crawl frontier and politeness rulesnotes/crawler/frontier.md · …per-host delay derived from robots.txt, default 1s…
Notes0.520

PrototypeAn interface sketch of this system, not a screenshot. Every figure, name and record in it is invented.