GNOSS Chatbot allows you to design, build and train conversational assistants powered by the Knowledge Graph of an organization, to establish dialogues with humans in their conversational areas: call-center, health, financial, etc.

Conversational assistant based on Knowledge Graph

Knowledge Graph is the ideal data source for the construction of a conversational assistant (Chatbot) that is capable of answering complex questions related to a conversational or knowledge field. A Knowledge Graph allows to operate effectively on the relationships of the data and entities in this area. These relationships will be implicit in the intentions and entities of people's questions, as long as they accumulate a minimum of complexity.

Conversational assistant based on Knowledge Graph -- Based on Knowledge Graph

Contextual dialogue

In addition to answering their questions, the assistant must be able to dialogue with a person using contextual information appropriate to their first request so that the conversation can progress with meaning in human terms.

A Knowledge Graph allows complex queries and interactions with data that are not managed a priori, which in practice means that the conversation can adapt quickly to unforeseen paths.

See more
Contextual dialogue -- Contextual dialogue

The expressiveness of the Knowledge Graph allows the realization of all kinds of queries, predefined or not, about the relationships between its data and entities and it is what gives the conversation this agility; on the other hand, its extensibility enables its ontological extension and, consequently, the organic incorporation of new entities or data without the need to refactor a predefined database schema. This quality of graphs makes it possible to extend conversational areas with little effort.

Dialogue appropriate to the intention

GNOSS Chatbot generates the responses and projects them in a dialogue appropriate to the intention and identified entities. It does so on the basis of both the training received and the learning it is capable of developing from the activity with its human interlocutors and the feedback they provide on its behavior.

See more
Dialogue appropriate to the intention -- Dialogue appropriate to the intention

The conversational domains inherent in a knowledge graph are defined in GNOSS Chatbot by means of intents, entities, and dialogues. Thus, an assistant is capable of translating natural language sentences into SPARQL language, after analyzing the intention and entities of the user's question using Natural Language Understanding (NLU) techniques, within one or more areas in which the graph has been has been trained.

Training should be understood as the set of processes that enables the incorporation of expert knowledge that characterizes a conversational environment to the general framework for the construction of assistants provided by GNOSS Chatbot. In practice, it is about training GNOSS Chatbot in the Knowledge Graph of a given conversational field.

SPARQL is the W3C standard for the RDF graph interrogation language. It allows, by means of shorter and clearer expressions than those provided by the relational SQL language, the execution of queries that relate multiple entities and their data, which makes it an ideal method to translate the database queries into a computer language. natural language questions that involve various entities and their relationships, which are precisely the type of questions that humans ask and that an assistant must answer.