Google Newscasts – Topic Related News Cards Presented Together
Google is taking steps to bring us news information to make it easier to find information about topics that may interest us, and provide rich coverage of such topics in the form of a newscast, combining multiple articles about a specific topic.
I wrote about top news stories in January, in the post Google Top Stories Are Chosen By Importance Scores
A new visual format referred to as newscasts are described in a Verge Article from 2018: Google News is getting an overhaul and customized news feeds. It tells us about this new Google Newscasts format in these words:
There’s a new visual format called newscasts, which uses natural language understanding to give you a collection of articles, videos, and quotes on a single topic. This lets you get the basics by browsing different sources so you can decide if you want to look further into a story.
A Google patent granted at the start of June of this year tells us about this Google newscasts presentation. Specifically, it is about an interactive user interface visually scrolling cards associated with content items (e.g., news articles) that relate to a topic (e.g., a news story).
The patent tells us the problem that it is intended to solve:
Consuming information around a news story is difficult. For example, for a prominent news story, there may be thousands of news articles published by hundreds of publishers. Further, in addition to traditional textual news articles, today’s media landscape includes news content in different media formats such as videos, blogs, social media posts, etc. that are scattered across the web and not consolidated together in a single place.
Distilling the information from this large number of different sources is an extremely challenging task. In particular, gaining a nuanced understanding of a news story involves not just reading one article but having enough information regarding different perspectives around the news story to make a well-informed judgment about the issues underlying the news story.
Currently, there is no simple yet effective way for a user to get a sneak-peek into the content present in content items around a story that helps the user identify, understand, and read interesting documents. For example, in certain existing systems, a user is provided with only a list of unrelated news articles and the user must examine each different article to determine whether the article relates to a topic in which the user is interested in understanding. If the article does relate to the topic, the user is then required to load and read the whole article to understand its content.
The patent tells us that readers of news can find it difficult to understand all the dimensions of a story and that they do “not have the time to scout the Internet for relevant information to be well informed about a news story.”
The patent tells us about the advantages behind following the process it describes:
One example aspect of the present disclosure is directed to a computer-implemented method to provide an informational display. The method includes selecting, by one or more computing devices, a plurality of content items that relate to a topic. The method includes generating, by the one or more computing devices, one or more content descriptors for each of the plurality of content items. The method includes providing for display, by the one or more computing devices, a user interface that cycles through a plurality of cards respectively associated with the plurality of content items. For each of the plurality of cards, the user interface presents at least one of the one or more content descriptors generated for the content item that corresponds to such card.
This Google newscasts topics patent can be found at:
Systems and methods for presentation of content items relating to a topic
Inventors: Ziad Sultan and Vidhya Ramesh Bhat
Assignee: Google LLC
US Patent: 10,671,267
Granted: June 2, 2020
Filed: May 6, 2019
The present disclosure provides systems and methods that provide a lean-back, interactive experience that enables a user to browse and understand content items (e.g., news articles or reactions) that provide different perspectives around a topic (e.g., a news story). In particular, the systems and methods can provide context around a topic by assembling a wide range of interesting content around a topic and seamlessly allowing users to scan through articles by surfacing unique information contained in them. The systems and methods of the present disclosure can read, understand, and organizes thousands of documents around a topic, hence allowing easy consumption of news.
Google Newscasts from the Google news app
These Google newscasts are available on Google news and the Google News App. In the news app, you see the “Full Coverage” icon that allows to access a range of stories related to a specific topic, and you can choose one of the stories, and scroll to one side or the other and read other stories on the same topic. An example of the “Full Coverage” landing page for news on the topic of the Las Vegas Raiders:
The idea behind a newscast such as this is to allow for a range of information (e.g., news articles or reactions) providing for different perspectives around a topic.
The patent tells us about such newscasts:
In particular, the systems and methods can provide context around a topic by assembling a wide range of interesting content around a topic and seamlessly allowing users to scan through articles by surfacing unique information contained in them. The systems and methods of the present disclosure can read, understand, and organize thousands of documents around a topic, hence allowing easy consumption of news.
The patent provides us with a peek at an algorithm behind the choosing of stories for Google newscasts, aiming at diversity:
Thus, in some implementations, the computing system can perform a backend algorithm that identifies and clusters content items related to a topic from a wide range of sources of content items (e.g., publisher websites, blogs, social media platforms, video platforms, and/or the like). The content item selection algorithm can ensure that there is sufficient diversity in both actual contents (e.g., perspective) and also a type of content (e.g., video versus textual social media post), which prevents redundant information being shown to a user while also allowing the user to explore various differing perspectives around the topic. The algorithm can also identify and cluster other useful information that is necessary to provide more context around the story such as social media posts from key people in the story and videos that are relevant to the story. Thus, by selecting a diverse array of content types and perspectives, the computing system can objectively curate content items related to a core theme or topic.
The patent tells us that one implementation of the newscast involves swiping through available cards like is available now. It hints as other approaches, such as the addition of a fast-forward button to make scanning through the cards quick and easy.
The patent provides more information about how Google newscasts could evolve in the future, including showing snippets from different stories included in a collection of articles and social media reactions in a newscast.
Machine Learning and Google Newscasts
The patent also tells us how Google could use machine learning approaches in newscasts which are aimed at improving available newscasts:
In some cases, systems of the type disclosed herein may learn through one or more various machine learning techniques (e.g., by training a neural network or another machine-learned model) a balance of the types of content items, perspectives, sources, and/or other attributes that are preferred, such as based on different types of content, different user populations, different contexts such as timing and location, etc. For example, data descriptive of actions taken by one or more users (e.g., “clicks,” “likes,” or similar) concerning the user interface in various contextual scenarios can be stored and used as training data to train (e.g., via supervised training techniques) one or more machine-learned models to, after training, generate predictions which assist in providing content in the user interface which meets the one or more users respective preferences. In such a way, system performance is improved with reduced manual intervention, providing fewer user searches and further conserving processing, memory, and network resources of the computing system (whether server device, client device or both).
Google Newscasts Takeaways
I have been reading most of my news using Google Discover, with occasional use of the Google News App.
I had not noticed the “Full Coverage” icons and landing pages leading to newscasts that can be swiped through.
It is interesting seeing a wider range of related content about a news topic.
I will likely use it more frequently now that I know more about it, and why it is set up the way that it is.
I like that I will be getting a more diverse set of information about topics that I see in the news.
Have you tried the Google News app? Which do you like better, Google News or Google Discover?