Microsoft natural language processing researchers heading to ACL 2016

Cortana

Microsoft has sent more than twenty of their researchers to the annual meeting of the Association for Computational Linguistics (ACL). Among them is Bill Dolan, the Principal Researcher at Microsoft Research. In his blog post today, he described a bit about what Microsoft was willing to share during the ACL conference and how it could affect technology moving forward.

As Dolan states, a major focus of Microsoft’s work has been to break down the language barriers between people and machines alike. Machine learning is a growing market where a massive shift of developer interest has swayed. Microsoft Research intends to find a way to turn natural language into a means of communicating and collaborating with their devices in a seamless manner. Some examples of this project is generating questions based on an image or photograph and a  publication exploring the translation between natural language to code.

Natural language is the translation of human to human interaction whereas machines speak in a completely different way with their coding. In natural language, the tone, mood, and inflections of words and syllables can change the entire meaning of a conversation that machine learning hasn’t quite grasped as of yet. Dolan stated that their most recent publication will also be presented during ACL that touches on this very subject.

“A Persona-Based Neural Conversation Model” will explain the research of mimicking human interaction with that of artificial intelligence. The idea is to respond to the linguistic behavior of the person to create a subtle characteristic of natural human conversation. The research, in particular, is aimed towards creating immersive virtual reality characters and agents. The full publication is already available to download as a PDF.

This year, the conference is being held at Humboldt University in Berlin, Germany. While it’s already started for the week, it will be continuing until August 12th.

Share This
Further reading: , , ,