Search Engine Relevance
Relevance is such a broad topic, I will group some bullet points into 3 buckets: query understanding and query rewrite, query evaluation and ranking, and relevance feedback. Query understanding and...
View ArticleQuery Demand for Ranking
When user types a query, for example “iphone 6s”, it helps if we can detect the intention based on history demand. The above query matches “iphone 6s white 64gb verizon” better than “pink case for...
View ArticleRegression Model for eCommerce Relevance
Nowadays, all sophisticated eCommerce search engines are harvesting user data to guide relevance ranking. First of all, we need a lot user behavior data, then we can mine the data and build machine...
View ArticleSignals and Signal Handling in C Multi-Threads
A refreshment for myself, and copy-paste from online: Signals are the system’s way of informing a process about various events. There are two types of signals, synchronous and asynchronous. Synchronous...
View ArticleMemory Management in our Search Engine
Here are how we manage memory in query nodes of our search engine, in two memory use cases: index memory, and, query memory. Index Memory We load all index into main memory for performance, the index...
View ArticleOracle “Snapshot Too Old”
I have seen Oracle “Snapshot Too Old” error for too many times, here are some learnings. Oracle maintains rollback segments so it can rollback failed transactions, the rollback segment is also used to...
View ArticleLearning HugePages
We are using dedicated bare metal machines with 76GB RAM to run our application, in most cases, 60+GB can be consumed by one process, so mem management is very critical. The most important factor is...
View ArticleDiscrimination in Online Marketplace
Discrimination does happen in any online marketplace, for example http://www.tnooz.com/articles/discrimination-on-Airbnb-do-whites-make-more-than-blacks/, it tells: … a recently released study from the...
View ArticleClik here to view.
jemalloc vs tcmalloc vs dlmalloc
“jemalloc” seems to be the newer allocator for c/c++ applications, it is definitely more sophisticated than e.g. tcmalloc:...
View ArticleClik here to view.
multi tenant service
It is a reasonable design choice to have tenant list for a web service, it will save capex, also fight intentional or unintentional attacks with Quality-of-Service.
View Article
![Cloonee, Yandel & ATLAC – Movimiento – Single [iTunes Plus M4A]](http://i.imgur.com/Q8tO40p.jpeg)