Home

The framework_code repository contains code for the core modules of the general NLU system. These modules are all implemented to facilitate easy retargeting of the system to new application domains. Even without retargeting, the core modules function as an integrated starter application.

These modules can be divided into three categories: language-side, action-side, and communication.

Language-side

As a block box, the language-side modules receive text/speech as input, and output an n-tuple. More specifically, a user types or speaks a statement, which is interpreted by a parser called the ECG Analyzer (Bryant 2008). This parser uses a grammar we’ve developed (called Embodied Construction Grammar) which incorporates theories/representations from linguistics, and outputs a Semantic Specification (SemSpec). The Specializer crawls the SemSpec data structure, extracting relevant information for a given task, and produces an n-tuple.

• Core UI-Agent
• ECG Analyzer
• Core ECG Grammar
• Core Specializer

Action-side

• Core Problem Solver

Communication

The Communication modules bridge the language and action sides of the system. Specifically, the n-tuple templates guide the Specializer in determining which information to extract from the SemSpec, and the Core-Agent/Transport modules establish a simple message-passing paradigm by which n-tuples can be communicated between agents (Trott, Appriou, Feldman, & Janin, 2015).

• Transport
• Core Agent
• N-tuple templates