In this paper, we study the problem of answering questions of type "Could X cause Y?" where X and Y are general phrases without any constraints. Answering such questions will assist with various decision analysis tasks such as verifying and extending presumed causal associations used for decision making. Our goal is to analyze the ability of an AI agent built using state-of-the-art unsupervised methods in answering causal questions derived from collections of cause-effect pairs from human experts. We focus only on unsupervised and weakly supervised methods due to the difficulty of creating a large enough training set with a reasonable quality and coverage. The methods we examine rely on a large corpus of text derived from news articles, and include methods ranging from large-scale application of classic NLP techniques and statistical analysis to the use of neural network based phrase embeddings and state-of-the-art neural language models.
Gathering the right kind and the right amount of information is crucial for any decision-making process. This book presents a unified framework for assessing the value of potential data gathering schemes by integrating spatial modelling and decision analysis, with a focus on the Earth sciences. The authors discuss the value of imperfect versus perfect information, and the value of total versus partial information, where only subsets of the data are acquired. Concepts are illustrated using a suite of quantitative tools from decision analysis, such as decision trees and influence diagrams, as well as models for continuous and discrete dependent spatial variables, including Bayesian networks, Markov random fields, Gaussian processes, and multiple-point geostatistics. Unique in scope, this book is of interest to students, researchers and industry professionals in the Earth and environmental sciences, who use applied statistics and decision analysis techniques, and particularly to those working in petroleum, mining, and environmental geoscience.
We propose a method for computing the value of information in petroleum exploration, a field in which decisions regarding seismic or electromagnetic data acquisition and processing are critical. We estimate the monetary value, in a certain context, of a seismic amplitude or electromagnetic-resistivity data set before purchasing the data. The method is novel in the way we incorporate spatial dependence to solve large-scale, real-world problems by integrating the decision-theoretical concept of value of information with rock physics and statistics. The method is based on a statistical model for saturation and porosity on a lattice along the top reservoir. Our model treats these variables as spatially correlated. The porosity and saturation are tied to the seismic and electromagnetic data via nonlinear rock-physics relations. We efficiently approximate the posterior distribution for the reservoir variables in a Bayesian model by fitting a Gaussian at the posterior mode for transformed versions of saturation and porosity. The value of information is estimated based on the prior and posterior distributions, the possible revenues from the reservoir, and the cost of drilling wells. We illustrate the method with three examples.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.