HTX’s xData & Q-Team have developed a solution – NEMO – to help Home Team officers monitor emerging news topics with ease.
With NEMO’s user-friendly dashboard, observing topic trends is a breeze. (Photo: HTX)
Have you ever felt overwhelmed looking at the news? There often seems to be a barrage of new information surrounding developing events, and it is always difficult to know where to start reading about a topic. And even if you did know where to start reading, news articles can often be long and convoluted, with key points being obscured somewhere within the text.
If catching up on a single event is frustrating for you, imagine how much more tiring it is for Home Team officers who pore through thousands of articles across many news platforms. While this task is arduous and time-consuming, Home Team officers must keep abreast of new developments and events that could potentially jeopardise Singapore’s security.
To make the whole process less labour-intensive, HTX’s xData & Q-Team have developed a solution to help Home Team officers monitor emerging news topics with ease. NEMO not only automatically collates relevant articles for Home Team officers and sorts them into neat topics, but it also distils the most salient points of the articles. NEMO also stands apart from other online news monitoring tools as it allows Home Team users to train “custom topics” and observe the trends that matter most to them.
The art of synthesis
NEMO scours the internet for news articles that best match the chosen topic. (Photo: HTX)
NEMO uses multiple artificial intelligence (AI) techniques to automate news monitoring for the Home Team.
Firstly, to get the content it needs to summarise, NEMO monitors content from 2,000 news sites, both local and foreign. NEMO also conducts automated scraping every six hours to provide updated trends and information to Home Team officers.
Once NEMO has gotten the information it needs, it summarises the articles using Machine Learning (ML). These summarised articles are then categorised into topics according to their subject matter, making it easier for Home Team officers to delve deeper into specific areas of interest.
A personal touch
Want to observe more specific trends like Internet-related scams? With NEMO, you can monitor the topics that matter most to you. (Photo: HTX)
Apart from using pre-set topics, users can also train NEMO to monitor custom topics by submitting five positive and five negative training articles (i.e. examples). While positive examples signal to NEMO the sort of articles the user wants to see, the negative training examples help NEMO understand what to avoid. From there, training NEMO is as simple as clicking the “train topic” button!
Now, when NEMO encounters news articles, it assesses their similarity to the positive examples. To find a good match, NEMO goes beyond identifying similar keywords between the examples and the articles; it “learns” what the articles are trying to convey.
Most online news monitoring tools use keyword matching to sort articles into topics. For example, a news article will be classified under the topic ‘climate change’ as long as those keywords are in the article, even if the article barely touches on climate change. However, NEMO uses positive examples to identify patterns and similarities between the examples and the article it needs to classify, ensuring that it only classifies relevant articles.
Normally, training an AI is like training a pet in that it requires a lot of time and effort. For example, most AIs require hundreds or even thousands of sources of training material to “learn” how to pick out the right articles. However, NEMO utilises Few Shot Learning (FSL) to generalise information from a limited number of examples while also using the context and semantic meaning of those examples to discover additional articles in that area.
NEMO and ScamBuster were both part of the Eye on the Internet exhibit recently showcased at the Milipol Asia-Pacific – TechX Summit exhibition held at the Marina Bay Sands Convention Centre from 3-5 April 2024.