A fundamental challenge in artificial intelligence is the construction of machines capable of reasoning with common sense. Common sense reasoning methods can help machines make more robust decisions, based on consistent assumptions about the real world, and can significantly simplify human-machine communication.
This challenge is one of the most difficult problems in building machines with human level intelligence. The scientific community in artificial intelligence has proposed for several decades partial solutions to build machines with common sense (e.g., considering extensions of classical first order logic, as well as other types of approaches, such as qualitative representations, etc.). The recent technological advances, such as big data, machine learning, natural language processing, etc., have facilitated the proposal of new methods of representation and extraction of common sense knowledge.
The goal of
this course is to present the main areas of common sense reasoning with special
attention to the recent advances of this field within artificial intelligence. First,
the course describes inference methods and algorithms that simulate common
sense reasoning (logic-based methods, physical reasoning, etc.). Then, the
course describes how to build common sense knowledge bases, reviewing different
approaches that cover both manual and automatic methods. Finally, the course
presents applications of common sense reasoning in areas such as
question-answering systems, natural language understanding, etc.