Abstract
Keeping the dialogue state in dialogue systems is a notoriously difficulttask. We introduce an ontology-based dialogue manage(OntoDM), a dialoguemanager that keeps the state of the conversation, provides a basis for anaphoraresolution and drives the conversation via domain ontologies. The banking andfinance area promises great potential for disambiguating the context via a richset of products and specificity of proper nouns, named entities and verbs. Weused ontologies both as a knowledge base and a basis for the dialogue manager;the knowledge base component and dialogue manager components coalesce in asense. Domain knowledge is used to track Entities of Interest, i.e. nodes(classes) of the ontology which happen to be products and services. In this waywe also introduced conversation memory and attention in a sense. We finelyblended linguistic methods, domain-driven keyword ranking and domain ontologiesto create ways of domain-driven conversation. Proposed framework is used in ourin-house German language banking and finance chatbots. General challenges ofGerman language processing and finance-banking domain chatbot language modelsand lexicons are also introduced. This work is still in progress, hence nosuccess metrics have been introduced yet.