作者: Lillian Lee
关键词: Range (statistics) 、 Distributional similarity 、 Computer science 、 Data mining 、 Function (mathematics) 、 Similarity (network science) 、 Econometrics 、 Proxy (statistics)
摘要: We study distributional similarity measures for the purpose of improving probability estimation unseen cooccurrences. Our contributions are three-fold: an empirical comparison a broad range measures; classification functions based on information that they incorporate; and introduction novel function is superior at evaluating potential proxy distributions.