A Three Layer Stacked Auto Encoders for Semantic Hashing

Abstract

Suggesting an original method for sympathetic short texts is to present a device to augment short texts with thoughts and co-occurring terms that are mined from a probabilistic semantic network. To bunch short texts by their senses, we suggest to add more semantic signals to short texts. Precisely, for each term in a short text, we get its concepts and co-occurring terms from a probabilistic information base to augment the short text. Also, we present a basic bottomless learning network entailing of a 3-layer stacked auto-encoders for semantic hashing.