Core Conversational AI Tech Open Sourced
- By John K. Waters
COTOBA Design, a Tokyo-based conversational AI products and services startup, has released its core software to open source.
The company's namesake product is a dialog description language processing engine written in Python 3. It contains programs for building interactions using the Artificial Intelligence Markup Language (AIML).
AIML is an XML dialect for creating natural language software agents. The COTOBA Agent OSS is designed to be available in multiple languages, the company says, and was available in Japanese and English at launch. It's available now on GitHub.
Agent OSS is designed to allow users to "embrace industrial conversational AI as a white box" with its ability to utilize sensor information from IoT with external APIs. It's designed to adapt a rule-based architecture to control dialog and generate responses, and introduces state-of-the-art deep learning technologies to interpret diverse intents to absorb natural language fluctuations in practice that are difficult to achieve with a classic rule-based approach.
The package being contributed includes Dialog Control (in Japanese and English), which provides a sample AIML and an AIML interpreter with COTOBA Agent original enhancements for industrial applications; and Intent Analysis (in Japanese and English), which provides functions of learning and inference of intent interpretation by deep learning, training data sample (Japanese), and models trained with large data sets (Japanese)
With this contribution to open source, the company says it hopes to promote the formation of a conversational AI development community and the formation of an ecosystem for dialog-agent development, operation and distribution. The goal is to advance "the realization of the content of dialog agents and the distribution mechanism."
"You can develop your own Siri or Alexa alternative from scratch with COTOBA Agent OSS for a wide-range of enterprise applications," said the company's CEO Mick Etoh in a statement. "Those are not chat bots, but voice user interfaces which share your feelings with intelligent real systems, such as robots, autonomous vehicles, factory automation controllers, and digital signages. However, applying such an interactive interface to a full-fledged industrial application is not easy."
John K. Waters is the editor in chief of a number of Converge360.com sites, with a focus on high-end development, AI and future tech. He's been writing about cutting-edge technologies and culture of Silicon Valley for more than two decades, and he's written more than a dozen books. He also co-scripted the documentary film Silicon Valley: A 100 Year Renaissance, which aired on PBS. He can be reached at email@example.com.