Post date: Sep 27, 2016 2:06:32 PM
Antônio Mourão will be presenting our short paper on how to extend Cassandra to support similarity searches at the Brazilian Symposium on Databases, to be held in Salvador in October. The work tackles two main challenges: place similar content in close nodes and efficiently retrieve similar data at each node reached by the query. Our solutions consist employing Locality Sensitive Hashing for data distribution, in a fashion similar to HammingDHT's, and in using Locality Sensitive Bloom Filters to rule out SSTables that do not possess similar data.
[pdf][slides][poster]