The book introduces concept invention and foundations. This includes the idea of concept blending, defined as the mechanism where two concepts or more are combined into a blended concept. The book also brings together the topics of concept amalgamation and concept coherence. The topics discussed are of importance to any researcher or developer who is interested in crafting systems that deal with forms of artificial intelligence (AI), including natural language processing (NLP) and signal processing. The book is a well-researched multidisciplinary study that is useful for a wide audience interested in the practical aspects of concept invention and expansion and manipulation techniques.
The book is divided into three parts: “Mathematical and Computational Foundations,” “Cognitive and Social Aspects,” and “Concept Invention System and Applications.” The audience includes (but is not limited to) AI specialists, academics, mathematicians, and information retrieval specialists. The study is presented over ten chapters, to cater to the needs of the diverse audience. The work includes a good level of detail, a rich set of references, and pointers to the tools and systems prototyped and presented, making the reader capable of setting up several of the example projects presented. Each chapter is self-contained.
The book starts with conceptual blending theory and the foundations of the theoretical basis. The approach used is relevant to the current boom fueled by AI and NLP, such as semantic query expansion, case-based reasoning, and knowledge graphs. A reference model presented early in the book introduces readers to the conceptual blending of Fauconnier and Turner. Necessary mathematical definitions and notations are presented, including morphism and conceptual blending amalgams, all in an intuitive approach.
The book presents a model for concept invention and creation, followed by an evaluation for newly invented concepts. Thagard’s computational theory is presented for concept invention, based on two phases: 1) the order-sorting of terms, and then 2) description logic. Query expansion and case-based reasoning are two examples of the application of concept discovery. Further, applications and prototypes include distributed ontology building where the concept of blending is illustrated to fuse multiple concepts.
The authors present further image schemas as examples of blending two concept families visually--an interesting idea that has wide applications and implications for explainable AI and reasoning.
An analogy-making methodology is presented that supports case-based reasoning. An important topic for a wide range of NLP systems, including question answering (QA) systems. The analogy-building mechanism is divided into two phases: 1) the identification of proper concepts, and 2) the mechanism of generalization.
Dialogue mapping and comprehension is presented, which is a process that humans usually carry out naturally to understand and navigate a dialogue. Inquiry, persuasion, information seeking, deliberation and negotiation, and conflict identification are the core mapping labels; interestingly enough, we can find such elements in recent smart systems, making the topic highly relevant and current to recent efforts in research and industry. It’s specifically relevant to topic modeling and text summarization.
Two prototypes are presented in the latter chapters: 1) mathematical reasoning, and 2) melodic harmonization, which is used to build a prototype to automate the harmonization to blend different melodic concepts. The last chapter presents a wealth of resources and discussions about the evaluation of creativity, which starts with a historical survey and ends with the necessary characteristics of evaluation for creative systems.
Some interesting books were recently published and are relevant to the same topic; readers can explore Image schemas and concept invention: cognitive, logical, and linguistic investigations  and Springer’s “Computational Synthesis and Creative Systems” series for more depth. While most of the examples in this book are mathematical or linguistics based, the concepts described can benefit anyone interested in developing or studying systems that deal with concepts, regardless of application.