LEMP: Exact and Approximate Maximum Inner Product Search

LEMP is a tool for maximum inner product search, i.e., efficiently retrieving large entries in the product of two given large matrices. This problem arises in  a number of data  mining and information retrieval tasks.  A detailed description of LEMP and its applications can be found in our SIGMOD 2015 paper and our TODS 2016 article.

Datasets

IE-NMF (Rank 10, 50, and 100)

IE-SVD

Netflix (factorization with and without averaging) 

Glove

Source code

The source code is hosted on LEMP GitHub repository. The SIGMOD version is available here. The TODS version is available here.