See Episode 71 of Learning Machines 101 for additional discussion of Markov Logic Nets!!!
Markov Logic: An Interface Layer for Artificial Intelligence
Synthesis Lectures on Artificial Intelligence and Machine Learning
Morgan and Claypool Publishers
2009
Paperback
145
About the Book:
Chapter 2: Overview of basic idea of how first-order logic is implemented in a Markov Random Field. The book begins with a short introduction to first-order logic and Markov random fields. It then provides a more in-depth discussion of many of the topics covered in this podcast.
Chapters 3, 4, and 5 cover more advanced topics in Markov Logic Nets which we did not have an opportunity to discuss in this podcast.
Chapter 6 discusses the applications of Markov Logic Nets to several real-world problems. Examples include: labeling objects, analyzing social networks such as Facebook and Linked-In for the purpose of determining what types of properties connect people together, and information extraction where the goal is to process raw text or semi-structured data sources and extract key pieces of information which can then be stored in a standard relational computer database. In addition, they also discuss the use of Markov Logic Nets for “robot mapping”. In robot mapping, a robot collects information about its environment using laser beams which provide information about the location of points in three-dimensional space. The Markov Logic Net is then used to construct a map of the environment for the robot which indicates the location of doors and walls to support robot navigation. And another really exciting application of a Markov Logic Net is to support semantic network extraction from text. This application is totally cool. The goal of this type of research is to avoid having humans type in common sense knowledge and facts into a big database such as CyC. Rather, the learning machine would simply go surfing on the web and read millions of web pages looking for little facts which were mentioned on those web pages. Once it finds a fact, the learning machine would add it automatically to the common sense knowledge database!
Free Software is discussed in the Appendix!!! In the Appendix A of the book “Markov Logic: An Interface Layer for Artificial Intelligence” the authors describe an open-source free software package which can be used by researchers and engineers to build their own Markov Logic Nets. They call this system ALCHEMY. If you go to the show notes of the episode, I provide a hyperlink to this software so that you can download this free software system for building and evaluating Markov Logic Nets!!
Who Should Read This Book:
Although this book is not really designed for the novice machine learning student, it is written in a relatively clear manner and is relatively self-contained so it should be accessible and useful to scientists and engineers interested in an introduction to exploring Markov Logic Net technology. The mathematical prequisites are minimal but you might to do some supplemental background reading on first-order logic, Markov fields, and stochastic optimization methods.
