Natural Language Processing in Action

Другая компьютерная литература, Программы

Год: 2019

Добавил: Admin 12 Янв 21

Проверил: Admin 12 Янв 21

Формат: PDF (9938 Kb)

Currently 0/5

Рейтинг: 0/5 (Всего голосов: 0)

Аннотация

Around 2013, natural language processing and chatbots began dominating our lives.
At first Google Search had seemed more like an index, a tool that required a little skill in order to find what you were looking for. But it soon got smarter and would accept more and more natural language searches. Then smart phone autocomplete began to get sophisticated. The middle button was often exactly the word you were looking for.1 In late 2014, Thunder Shiviah and I were collaborating on a Hack Oregon project to mine natural language campaign finance data. We were trying to find connections between political donors. It seemed politicians were hiding their donors' identities behind obfuscating language in their campaign finance filings. The interesting thing wasn’t that we were able to use simple natural language processing techniques to uncover these connections. What surprised me the most was that Thunder would often respond to my rambling emails with a succinct but apt reply seconds after I hit send on my email. He was using Smart Reply, a Gmail Inbox “assistant” that composes replies faster than you can read your email.
So I dug deeper, to learn the tricks behind the magic. The more I learned, the more these impressive natural language processing feats seemed doable, understand-able. And nearly every machine learning project I took on seemed to involve natural language processing.