Saturday, July 27, 2024
HomeAINippon Television, Docomo and other companies develop news article automatic summary service,...

Nippon Television, Docomo and other companies develop news article automatic summary service, start offering beta version

Forecast Communication, NTT DoCoMo, and Nippon Television Network Corporation (Nippon Television) have jointly developed an automatic summary service for news articles using AI, “Automatic Summary AI (β version)” .

It is expected to streamline the work of summarizing news articles, which requires a large amount of work and specialized skills, and to reform the way media work.

Table of Contents 

  • A summary service that continues to evolve with users
  • Nittele article and Docomo’s technology

A summary service that continues to evolve with users

“Automatic summary AI (beta version)” is a site that can summarize news articles. When you enter the “text you want to summarize” on the left side of the screen, the “summary text” will be displayed on the right side. In addition to the “character number specification function” and “sentence number specification function” that summarizes based on the number of characters and sentences specified in advance, and the “hint function” that specifies the keywords you want to reflect in the summary results, it is the first in Japan (* 1). It is provided with a “title function” and “location specific function”. “Title function” can increase the importance of the words contained in the title and create a summary related to the title. We plan to collect user feedback and continue to improve the service through a learning loop.
*1 As of December 21, 2021 (according to Docomo research).

Nittele article and Docomo’s technology

“Automatic Summarization AI (beta version)” uses the automatic summarization technology of NTT DoCoMo Beijing Laboratories. This summarization technology is realized by a neural network (DNN: Deep Neural Network) independently developed by NTT Docomo Beijing Research Laboratories using AI technology, and compared to existing technology, it only increases the comprehensiveness of important information. Instead, it selects and generates necessary conjunctions, etc., to generate easier-to-read sentences.

From September 2018, NTT DoCoMo and Nippon Television will train the nearly 200,000 past article data and human-summarized data sets that Nippon Television possesses in large quantities to improve the accuracy of summarization. In the demonstration experiment that started from , we expanded the functions of the system, implemented the interface, and developed a web application type prototype for use in the field.

Furthermore, from 2021, NTV group company Forecast will join us to develop a user interface for “automatic summary AI (β version)”.

Going forward, the three companies will collect feedback from users of “Automatic Summarization AI (beta version)”, analyze and improve it, and aim to start providing commercial services in FY2022.

RELATED ARTICLES

What is an AI algorithm?

LEAVE A REPLY

Please enter your comment!
Please enter your name here

Recent Posts

Most Popular

Recent Comments