Box Introduces Box Skills and Box Skills Kit Machine Learning Tools

Box, a cloud content management platform provider, has introduced Box Skills, a framework that makes it possible to take applications of machine learning such as computer vision, natural language processing, and sentiment analysis and apply them to content stored within the Box platform. At BoxWorks 2017, the company previewed three initial Box Skills currently in development; Audio Intelligence, Video Intelligence, and Image Intelligence. These three Box Skills leverage machine learning tools from Google Cloud, IBM Watson, and Microsoft Azure. Box also introduced the Box Skills Kit which is an SDK that developers can use to build custom skills. The Box Skills and Skills Kit will be available in beta in early 2018.

Image Credit: Box

The Audio Intelligence skill creates and indexes a text transcript from audio files, and the transcript can be searched and manipulated in a variety of ways. The Video Intelligence skill analyzes video and provides transcription, topic detection, and people detection capabilities. Video Intelligence allows users to search for and find specific information in a video. The Image Intelligence skill provides a number of computer vision capabilities such as object and concept recognition, optical character recognition (OCR), and automatic image classification and tagging.

Accompanying Box Skills is the Box Skills Kit which allows developers to build custom skills and chain together multiple skills to build intelligent business processes. In the Box platform, there is a storage area, compliance, security, and other features. The Box platform also includes a presentational area and collaboration capabilities. The Skills Kit includes a skills engine, skills metadata, and metadata preview feature, and the SDK inserts these components into the Box platform making it easy for developers to build custom skills. While the initial Box Skills currently in development use machine learning services from Google Cloud, IBM Watson, and Microsoft Azure, developers can use any third-party machine learning service to build custom skills.

We reached out to Box Chief Product Officer Jeetu Patel who provided ProgrammableWeb some details about the upcoming Box Skills and Box Skills Kit. "We think that the large amount of skills that we build will be custom skills as we move forward, where you can have any developer enabled to build a custom skill," said Patel. "Not only do you build a custom skill, you can actually chain a bunch of algorithms together so that you can use, for example, transcription from Microsoft, and translation from Google, and sentiment analysis from IBM."

Patel described a use case for chaining skills together to build a specific business process. He explained that call centers or customer support operations often have recordings of the phone conversations that have taken place with customers. In this case, an IBM Watson audio transcription skill could be chained with an IBM Watson sentiment analysis skill, or the transcription and sentiment analysis services used could be from different providers. This skills chain could be used to automatically apply tags so that it is easy to discover at what points in a conversation the customer expressed frustration, agreeableness, or other types of emotions. These skills not only make it possible to jump right to a specific emotion in the conversation but also provide a complete transcript for analysis.

Box also showcased at BoxWorks 2017 the company’s Box Graph technology which is an intelligent network of content, relationships, and activity that the company will use to provide new services to Box users. "The Skills are designed so that you get the most amount of value from a single file," said Patel. "Box Graph is designed so that you can get what happens around the file; the patterns that are starting to emerge, and what the relationships are between content and content, or content and people, or people and people, you pull those together."

Patel said that "with Skills, our goal is to make sure that every piece of content is more valuable in Box than it is outside of Box." He also said regarding Box Skills and Box Graph that "the combination of those two create the beginning of the intelligence strategy for us. We want to make Box more and more intelligent innately in the way that you use it."

For more information about the Box platform, visit Box.com.

Janet Wagner is a technical writer and contributor to ProgrammableWeb who covers breaking news and in-depth analysis. She specializes in creating well-researched, in-depth content about APIs, machine learning, deep learning, computer vision, analytics, and other advanced technologies.
 

Comments