North Side develops a full-coverage semantic frame system (extracting meaning from text irrespective of syntax and particular words, an idea popularized by FrameNet). Semantic frames are used to formalize our large database of knowledge about everyday life.


A knowledge-based conversation pipeline, able to converse on any topic of everyday life.

Situated language understanding (resolve references to 3D objects and spatial relations), used in our Say it, See it!™ solution and videogames.

Our Jimmy’s World videogame leverages our pipeline and a vast database of knowledge about everyday life.

In 2015, our Bot Colony videogame pioneered text-based 3D animation – the ability to control virtual character actions in English. Our 3D capable NLU powers our Say It, See It!™ product.

The robustness of our NLU enabled us to support speech-based, transactional financial applications with the same pipeline used for videogames.

The best of both worlds: Coverage and Precision

Nowadays, everyone relies on Machine Learning (ML) to ensure coverage. Our knowledge-based approach increases precision, giving deterministic answers when possible.

Our massively rule-based approach complements ML solutions. It:

  • ‣ Makes sense of more user utterances. Deals better with UMM (unsupported, misunderstood, mismatched utterances).

  • ‣ Extracts ALL the entities and relations, modality and negation from utterances.

  • ‣ Handles more out of turn utterances.

  • ‣ Handles more dialogue acts including user clarification requests when answering.

  • ‣ Co-references resolves pronouns, nouns, adverbials.

  • ‣ Uses context to resolve ellipsis (missing verbs, nouns, etc) and to do co-reference resolution.

  • ‣ Handles complex dialogue graphs, where the next prompt depends on the user's answer to a previous question.

North Side Natural Language Understanding (NLU) pipeline

Complementing or replacing traditional GUI's with a spoken interface
Add voice and text (everyday English) to your current web-based, mobile or embedded applications. Let us call your API's from our Conversation Server.

Existing Interface

Our massively rule-based approach complements ML solutions. It:

Conversational Interface

Here's how our pipeline can voice-enable and text-enable your current applications.


1. A Virtual Robot-Human Development framework