With the emergence of Artificial Intelligence technologies like machine and deep learning, data analysis is increasingly associated with executing so-called Inference Queries over unstructured data such as text, images, and video, which are an important aspect of modern data analytics. Executing such queries involves the evaluation of predicates based on user-defined functions (UDF) containing a so-called oracle in the form of packaged deep models or other procedures, such as labeling data through human effort. For example, querying large video datasets for given objects like humans or wheelchairs requires the invocation of a deep object detector on every frame of the video. The main aim of this chapter is to introduce the reader to the problem of unstructured data querying in the realm of big data analytics. Starting from the relevant problem formulations, we will introduce the challenges associated with integrating AI models and other oracle types (like human inference) in big data analytics workloads. We will then proceed with introducing the main research directions in the data management community in order to mitigate the impact of those issues. The chapter includes representative use cases, examples, and a literature review of the state of the art, highlighting current and future research directions in the area.