An Israeli startup has succeeded in developing a robot that can automatically write and design posts on Facebook, Instagram, and Twitter, which have received more exposure and user engagement than similar posts written by humans.
The development by the startup Keywee scans millions of posts, deciphers the content structure and user responses, and knows which words, phrases, and emojis to use to increase exposure.
Israeli technology was put to use to increase user engagement in posts about the series Game of Thrones, the Oscars, Emmys and Grammys, and helps media outlets increase exposure to articles including BBC, CNN, New York Times, AOL, Forbes and more.
Want more news, videos and stories? Join the Haredim 10 WhatsApp channel >>
The software knows how to write different posts about the same news item for different audiences.
For example, it turned out that a sports news item that uses the concept of "In case you missed it..." within the sports channel results in an average decrease of 10% in user interest, while the same wording when appearing on news and current affairs sites will actually increase clicks by 25%.
Measuring user responses shows that using emojis in fashion posts will increase clicks by 10%.
In the sports field, posts offering to read insights from games receive 69% more clicks, and posts that raise the question of whether users agree with statements made by players or coaches increase clicks by 54%.
In a similar way, the system weighs thousands of additional variables and writes the optimal posts. All data is "real-time" and changes rapidly, so the system constantly adapts itself to changes in user interest.
According to Yaniv Makover, founder and CEO of Kiwi, adapting posts to different types of audiences on social networks is done using natural language processing technologies, deep learning, machine learning, and algorithms based on information accumulated from textual analysis of tens of millions of articles.
