作者: Yi Ma , Paul A. Crook , Ruhi Sarikaya , Eric Fosler-Lussier
DOI: 10.1109/ICASSP.2015.7178992
关键词: Markov chain 、 Semantics 、 Inference 、 Knowledge graph 、 Graph database 、 Computer science 、 Semantic memory 、 Probabilistic logic 、 Natural language processing 、 Artificial intelligence 、 Spoken dialog systems 、 Graphical model
摘要: We propose Inference Knowledge Graph, a novel approach of remapping existing, large scale, semantic knowledge graphs into Markov Random Fields in order to create user goal tracking models that could form part spoken dialog system. Since include both entities and their attributes, the proposed method merges dialog-state-tracking attributes database lookup fulfill users' requests one single unified step. Using graph contains all businesses Bellevue, WA, extracted from Microsoft Satori, we demonstrate can return significantly more relevant than baseline system using lookup.