Probabilistic logics and probabilistic networks (Q2638162)

A fundamental question in classical logics is that of inference, in particular if premises \(\varphi_1,\dots,\varphi_n\) entail a conclusion \(\psi.\) This book investigates this question in the framework of probabilistic logic. For probabilities \(X_1,\dots,X_n,Y\subseteq[0,1]\), the question treated is which probability/ies \(Y\) make \[ \varphi_1^{X_1},\dots,\varphi_n^{X_n}{\mathrel|\joinrel\approx}\,\psi^Y \] true, where \(\chi^X\) is interpreted as ``the probability of \(\chi\) lies in \(X\)''. The authors refer to this schema as the \textit{Fundamental Question of Probabilistic Logic}. Structurally, the book consists of two main parts; the first part (Chapter 2 to Chapter 7), the second part (Chapter 8 to Chapter 14) and a conclusion (Chapter 15). In the first part, the authors show how the above schema can be interpreted in several probabilistic frameworks: standard probabilistic semantics, probabilistic argumentation, evidential probability, statistical inference, Bayesian statistical inference and objective Bayesian epistemology. The second part shows how Bayesian and credal networks are in a canonical way calculi for these five frameworks. Summing up, the authors give the following simple recipe (page 85): {\parindent5mm \begin{itemize}\item[1)] Formulate a question as in the above schema. \item[2)] Decide upon appropriate semantics. \item[3)] Construct a network for the left-hand side of the above schema. \item[4)] Use the machinery presented in in this book to solve the question posed in the first step. \end{itemize}} The underlying theme of this book is to give a framework unifying various approaches of combining probability and logic with the goal of practical applications.