One of the biggest challenges in interacting with machines is their ability to understand the way we meat bags naturally speak. Thankfully, Google has just open sourced a tool that was constructed to tackle this precise problem. And standing in stark contrast to the tool’s quite serious purpose is its exceptionally silly name: Parsey McParseface.
Parsey McParseface is actually the English language parser that is itself part of a larger framework for understanding natural language, called SyntaxNet, that Google has open sourced. Google touts Parsey McParseface as the most accurate model of its kind, noting its potential benefits to developers and researchers now that it’s open source. From Google:
Parsey McParseface is built on powerful machine learning algorithms that learn to analyze the linguistic structure of language, and that can explain the functional role of each word in a given sentence. Because Parsey McParseface is the most accurate such model in the world, we hope that it will be useful to developers and researchers interested in automatic extraction of information, translation, and other core applications of NLU.
In fact, Google says that Parsey McParseface’s dissections of English language sentences are 94% accurate. Given the inherent ambiguity in the meaning of many sentences — something that Google says is one of its main challenge — that’s a pretty impressive statistic.
This follows Google’s recent move to open source TensorFlow, the company’s overall machine learning platform to which Parsey McParseface and SyntaxNet belong. For much more on how Parsey McParseface works, be sure to check out Google’s full blog post.
