Use Cases

Data intelligence solutions for better business decisions.

Powerful solutions to modernize the way your team discerns insights from data.

uc1

PR & Communications

uc2

Media

uc3

Legal Tech

uc4

Computer Vision

NITL

NITL - News In Three Lines

Industry: Media


Time to Deploy: Lorem Ipsum

32

The Challenge:

European news aggregator, NITL, is reinventing the way we consume news – in three sentences or less. News articles from over 500 sources are summarized daily by the NITL team for use on its mobile apps. Not only does the summarization itself take hours of work but EU copyright regulations also prohibit the use of more than eight words in a row from the original piece. Even though NITL links back to the full, original article, even their three line summaries have to abide by this regulation. With so many resources spent on accomplishing this manual task, NITL needed a way to automate it in order to open opportunities for growth in other areas.
33

Primary Goal:

Automate summarization in accordance with EU regulations in order to streamline operations, improve efficiencies, and grow the business while improving the experience provided to readers.
34

Custom Solution:

Our team built an abstractive summarization model that works with NITL’s Chrome extension API to create summaries according to the company’s desired length and specifications. Our model can identify the type of story (e.g. editorial vs sports) in order to extract the most relevant information and include it in each summarization.
s1

Models Used:

The Process:

Day 1
Contract Signed
Day 1 – 30
Onboarding (Methodology Review, Goal setting, Customization of Summarization model output to work with copyright restrictions)
Month 2
Customized Abstractive Model deployed to API with accompanying WebApp for ease of use by NITL to speed up summary generation
TBD
Over the next 6 months NITL will have enough of its own data on customized summaries that we can further customize the Summary model to the style of NITL’s editors.
NITL
en_USEnglish