This document provides an overview of linear-chain conditional random fields (CRFs), including how they relate to logistic regression and how they can be used for tasks like part-of-speech tagging and speech disfluency detection. It explains that linear-chain CRFs are a type of log-linear model that uses a graph structure to represent relationships between input features and output labels. Feature functions in CRFs can capture dependencies between neighboring output labels. The document provides examples of how CRFs are trained and tested for sequence labeling tasks.