We already met Parsey McParseface, along with new SyntaxNet models in May when Google unveiled them. SyntaxNet is a part of Tensor Flow, Google’s open-source framework for deep learning and with Parsey they in turn helps build foundation for Natural Language Understanding (NLU) . Today Google debuted Parsey’s family, with his 40 cousins who are open-sourced pre-trained models for parsing text in 40 languages.
For those of you who are unfamiliar with Parsey McParseface and parsing in general, let us give you an insight into what it is. Parsing involves breaking down a sentence and identifying it’s components as nouns, verbs, adjectives, adverbs and so on. It simply labels the parts of speech present in the composition of a sentence. This is done so that computer systems can understand and ‘read’ human language in order to intelligently process it as a command. It might not sound like anything significant but Mc Parseface works at an enormous scale in Google and helps break down and understand queries of a web search done by users. Now Google has made this technology available in 40 languages, thus helping hoards of researchers all over the world.
Fluency in several languages wasn’t just the dimension Google was working on, besides that, they have strengthened the underlying SyntaxNet NLU library. Parsey can now detect different meanings based on difference in spellings, which is better known as morphology. In English placing the alphabet ‘s’ after a word generally translates into plurality, that however isn’t the case in other languages for example German and Russian both are heavily morphed languages. Through Parsey, Google aims at improving deep learning which is a type of artificial intelligence involving humongous quantities of data surging through artificial neural networks, in an attempt to train them in understanding inferences related to new data and processing strings of words.
Parsey and deep learning have brought about quite a few changes in the sharing businesses, one of them being the wave of chat bots. With parsing expanding by breaking language hurdles, On-Demand platform could take a hit. Since in a On-Demand platform placing an order requires a detailed description of what is desired and entails taking more time, an alternate to which is an API integration, which again is a complex process for a layman. Parsing will ensure ease of communication with natural language. Since the computer will understand the statement made on chat and analyze what the command is without the use of a detailed statement or an API key, thus replacing On-Demand with chats and allowing platforms like Fugu to grow.
