Google Releases Cloud Natural Language API

The stakes for Machine Learning APIs are increasing by the day. Major vendors are making moves towards releasing ML Platform APIs that will give developers a jumpstart into incorporating ML functionality into their applications. At the latest GCP Next and Google IO events held a few months back, Google placed its bets on a Machine Intelligence driven future and presented plans that provide a spectrum of services ranging from 'run it on your own infrastructure' v/s using Google’s machine learning models that they have developed over the years. One such release is the Cloud Natural Language API, a fully managed API from Google that provides developers the ability to investigate the structure and meaning of text via REST APIs.

Announcing the release of Cloud Natural Language API, the functionality is divided into 3 key areas:

  • Sentiment Analysis : Analyze the text for overall sentiment.
  • Entity Recognition : Identify entities in the text and get classification like persons, organization, products, etc.
  • Syntax Analysis : Earlier this year, Google had announced SyntaxNet, which analyzes grammatical structure of text. This feature of the API helps to break down the text into tokens, parts of speech and complex patterns out of it. 

The Google Cloud Natural Language API currently supports 3 languages : English, Japanese and Spanish. Often the blocks of text that you might want to analyze are not always available in text format. For example, you could have several speeches that you might want to anaylze in a audio file, or you might want to simply do an OCR of text documents and then push them into the processing pipeline utilizing the Cloud Natural Language API. To facilitate these situations, there are multiple ways that you could do that:

  • Cloud Natural Language API has integrated with Google Cloud Storage (GCS). So once your files are present in text form and stored in GCS, it can pick it up from there. 
  • Along with Cloud Natural API entering Beta, another API announced in the blog post is Google Speech API. This API provides Speech to Text conversion capabilities in more than 80+ languages. Imagine combining the Speech API to analyze your audio files and then feeding that directly to the Cloud Natural Language API for text analysis. 

The blog post highlights an interesting article that uses the Cloud Natural Language API to analyze Harry Potter and New York Times stories. It uses features like Entity Detection and Sentiment Analysis, then feeds those results into another Google Cloud Data Platform product, BigQuery to analyze and visualize the results. 

The Cloud Natural Language API is currently in Beta and pricing includes a free tier. All the features (Sentiment Analysis, Entity Recognition and Syntax Analysis) are free till 5K units/month. The pricing details are available here and will be effective from August 1 this year. 

Be sure to read the next Machine Learning article: Affectiva Emotion Measurement API Now Free to Startups