CAI Skills
AI is a confluence of many kinds of knowledge, expertise, and skill.
Here I list the different components or areas of knowledge and skill I believe are necessary for creating the best conversational AI (CAI) or CAI applications, or for becoming a well-rounded expert in this new and necessarily-interdisciplinary field. To the best of my knowledge, all of these skills and terms predate CAI but are now working together to produce CAI. For each skill, I also list how it is generally applied within CAI, some of the sub-skills you may have heard of, and a link to a list of published research in that area. Also see my post called CAI Terminology for definitions of terms.
Agent planning, multi-agent systems, Markov decision processes, reinforcement learning: Improve dialog outcomes
Cloud environment: Deploy a scalable infrastructure
Deep learning: Text classification, information extraction, graph + text inference.
Graph database: Knowledge representation, knowledge graph
Intrapreneurship or entrepreneurship: Getting something new started inside or outside of an existing business.
Named entity recognition (NER): Entity and relation extraction from text. Fill slots.
NL dialog management:
NL Understanding: Translate language to Cypher queries
Planning and scheduling
Semantic parsing: Translating a natural language utterance into a representation of its meaning that consists of grounded entities and their relationships. This consists of entity recognition, entity resolution, relation extraction. The most promising area of research here is what I call semantic parsing by graph matching.
Sentence generation: Generate responses
Text classification: Classify intent of a human utterance.
Reasoning: Enhanced language understanding
