MIT Researches Take A Step Forward With Natural Language To Code Translation

Creating software that would allow programmers, or your average two-finger typist who wants to create the next million selling app, to write in natural language instead of a special purpose coding language has had met with little success. MIT takes a step forward.

It is the dream of many a programmer, and researchers for that matter, to be able to simply tell you computer what you want to program using your native language. Some strides have been made, but most of the effort has been put into specific languages that are "more natural like" and not actually a conversion of vernacular to code.

One of the big issues is that natural languages do not map well to coding languages. While some portions of code map rather nicely, a lot of what would need to be done is almost impossibly complex for a computer to determine actually needs to be written in the code.

Researches at MIT have release papers recently that detail how they took natural language input and converted it into regular expressions. For those you not familiar with regular expressions, they are a specific collection of symbols to perform search functions on digital information.

While this would seem to be of little use to the most programmers and hobbyists, it's another step forward in finding ways to bridge the natural language/code language barrier that prevents many normal computer users from being able to program their computers. While you may never see your Uncle Chester hammering out the next WoW killer MMO, all research in this field is welcome as it has the potential to facilitate better standards and improved productivity.

This writer is waiting patiently for software to translate "OMG WTF BBQ" in all its contexts. I won't be holding my breath.

Social
Source MIT
Channels Science, Software
Topics MIT, Programming, Natural Language, Software, Translation
Related