SyntaxNet; Google’s new releaseSyntaxNet will change forever the way we can translate on the Internet as it can actually understand and process human language intelligently 17/05/2016
Google now has something to celebrate after opening SyntaxNet, which is a language parsing model. Parsey McParseface is the English version given to this model. This system is incredibly accurate as it understands human language. This means that in order to raise the game of AI, artificial intelligence, majority have to use this parsing model created by Google. This has further led to the realization that machines can understand sentences based on a standard database that is English language journalism. The immediate aim of this open sourcing is taking over the world.
Slav Petrov, Google’s senior staff research scientist, explained how burdening it has been spending a lot of time on computer systems figuring out how they read, understand and process human language intelligently. Fortunately, today 13th may, 2016, an open source has been released. It was implemented in Tensor Flow and it is aimed to provide the basis for Natural Language Understanding systems.
Powerful machine learning algorithms that have the capacity of analyzing linguistic structures of language and that can determine roles of words in sentences are used to make up SyntaxNet. This system determines the syntactic relationships between the words in a sentence in order to recognize the subject and the object that make up the sentence and hence represent them in a dependency parse tree.
Petrov confirmed that today’s release has all that is needed for the training of these new SyntaxNet models as far as your own data is concerned. This also applies to Parsey McParseface. Neural networks are applied in SyntaxNet and the input sentence is usually processed left to right. Parsey McParseface is 94% on the news text while human linguists are 96% or 97% accurate. However, in random sentences on mainly on the Web, Parsey McParseface is 90 percent accurate. This is according to Google.
[picture: digitaltrends .com]