Generic Inverted Index on the GPU

Data variety, as one of the three Vs of the Big Data, is man-ifested by a growing number of complex data types such asdocuments, sequences, trees, graphs and high dimensionalvectors. To perform similarity search on these data, exist-ing works mainly choose to create customized indexes fordifferent data types. Due to the diversity of customized in-dexes, it is hard to devise a general parallelization strategyto speed up the search. In this paper, we propose a genericinverted index on the GPU (called GENIE), which can sup-port similarity search of multiple queries on various datatypes. GENIE can effectively support the approximate near-est neighbor search in different similarity measures throughexerting Locality Sensitive Hashing schemes, as well as sim-ilarity search on original data such as short document dataand relational data. Extensive experiments on different real-life datasets demonstrate the efficiency and effectiveness ofour system.