2017/04/04 by Feras A. Saad, Saad, Feras, Leonardo Casarsa +3
Computer Science · #Advanced Database Systems and Queries #Artificial Intelligence (cs.AI) #Bayesian Methods and Mixture Models #Bayesian Modeling and Causal Inference #Data Management and Algorithms #Databases (cs.DB) #FOS: Computer and information sciences #Machine Learning (cs.LG) #Machine Learning (stat.ML)
paper · pdf · doi:10.48550/arxiv.1704.01087
openalex publication_date 2017/04/04 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
Databases are widespread, yet extracting relevant data can be difficult.\nWithout substantial domain knowledge, multivariate search queries often return\nsparse or uninformative results. This paper introduces an approach for\nsearching structured data based on probabilistic programming and nonparametric\nBayes. Users specify queries in a probabilistic language that combines standard\nSQL database search operators with an information theoretic ranking function\ncalled predictive relevance. Predictive relevance can be calculated by a fast\nsparse matrix algorithm based on posterior samples from CrossCat, a\nnonparametric Bayesian model for high-dimensional, heterogeneously-typed data\ntables. The result is a flexible search technique that applies to a broad class\nof information retrieval problems, which we integrate into BayesDB, a\nprobabilistic programming platform for probabilistic data analysis. This paper\ndemonstrates applications to databases of US colleges, global macroeconomic\nindicators of public health, and classic cars. We found that human evaluators\noften prefer the results from probabilistic search to results from a standard\nbaseline.\n