During a livestreamed event this afternoon, Google detailed the ways it’s applying AI and machine learning to improve the Google Search experience.
Soon, Google says users will be able to see how busy places are in Google Maps without having to search for specific beaches, parks, grocery stores, gas stations, laundromats, pharmacies, or other business, an expansion of Google’s existing busyness metrics. The company also says it’s adding COVID-19 safety information to business profiles across Search and Maps, revealing whether they’re using safety precautions like temperature checks, plexiglass, and more.
An algorithmic improvement to “Did you mean,” Google’s spell-checking feature for Search, will enable more accurate and precise spelling suggestions. Google says the new underlying language model contains 680 million parameters — the variables that determine each prediction — and runs in less than three milliseconds. “This single change makes a greater improvement to spelling than all of our improvements over the last five years,” Prabhakar Raghavan, head of Search at Google, said in a blog post.
Beyond this, Google says it can now index individual passages from webpages as opposed to whole pages. When this rolls out fully, it will improve roughly 7% of search queries across all languages, the company claims. A complementary AI component will help Search capture the nuances of what webpages are about, ostensibly leading to a wider range of results for search queries.
“We’ve applied neural nets to understand subtopics around an interest, which helps deliver a greater diversity of content when you search for something broad,” Raghavan continued. “As an example, if you search for ‘home exercise equipment,’ we can now understand relevant subtopics, such as budget equipment, premium picks, or small space ideas, and show a wider range of content for you on the search results page.”
Google is also bringing Data Commons, its open knowledge repository that combines data from public datasets (e.g., COVID-19 stats from the U.S. Centers for Disease Control and Prevention) using mapped common entities, to search results on the web and mobile. In the near future, users will be able to search for topics like “employment in Chicago” on Search to see information in context.
On the ecommerce and shopping front, Google says it has built cloud streaming technology that enables users to see products in augmented reality (AR). With cars from Volvo, Porsche, and other “top” auto brands, for example, they can zoom in to view the steering wheel and other details in a driveway, to scale, on their smartphones. Separately, Google Lens on the Google app or Chrome on Android (and soon iOS) will let shoppers discover similar products by tapping on elements like vintage denim, ruffle sleeves, and more.
Above: Augmented reality previews in Google Search.
In another addition to Search, Google says it will deploy a feature that highlights notable points in videos — for example, a screenshot comparing different products or a key step in a recipe. (Google expects 10% of searches will use this technology by the end of 2020.) And Live View in Maps, a tool that taps AR to provide turn-by-turn walking directions, will enable users to quickly see information about restaurants including how busy they tend to get and their star ratings.
Lastly, Google says it will let users search for songs by simply humming or whistling melodies, initially in English on iOS and in more than 20 languages on Android. You will able to launch the feature by opening the latest version of the Google app or Search widget, tapping the mic icon, and saying “What’s this song?” or selecting the “Search a song” button, followed by at least 10 to 15 seconds of humming or whistling.
“After you’re finished humming, our machine learning algorithm helps identify potential song matches,” Google wrote in a blog post. “We’ll show you the most likely options based on the tune. Then you can select the best match and explore information on the song and artist, view any accompanying music videos or listen to the song on your favorite music app, find the lyrics, read analysis and even check out other recordings of the song when available.”
Google says that melodies hummed into Search are transformed by machine learning algorithms into a number-based sequence representing the song’s melody. The models are trained to identify songs based on a variety of sources, including humans singing, whistling, or humming, as well as studio recordings. They also take away all the other details, like accompanying instruments and the voice’s timbre and tone. This leaves a fingerprint that Google compares with thousands of songs from around the world and identify potential matches in real time, much like the Pixel’s Now Playing feature.
“From new technologies to new opportunities, I’m really excited about the future of search and all of the ways that it can help us make sense of the world,” Raghavan said.
Last month, Google announced it will begin showing quick facts related to photos in Google Images, enabled by AI. Starting in the U.S. in English, users who search for images on mobile might see information from Google’s Knowledge Graph — Google’s database of billions of facts — including people, places, or things germane to specific pictures.
Google also recently revealed it’s using AI and machine learning techniques to more quickly detect breaking news around crises like natural disasters. In a related development, Google said it launched an update using language models to improve the matching between news stories and available fact checks.
In 2019, Google peeled back the curtains on its efforts to solve query ambiguities with a technique called Bidirectional Encoder Representations from Transformers, or BERT for short. BERT, which emerged from the tech giant’s research on Transformers, forces models to consider the context of a word by looking at the words that come before and after it. According to Google, BERT helped Google Search better understand 10% of queries in the U.S. in English — particularly longer, more conversational searches where prepositions like “for” and “to” matter a lot to the meaning.
BERT is now used in every English search, Google says, and it’s deployed across languages including Spanish, Portuguese, Hindi, Arabic, and German.
AI tools could improve fake news detection by analyzing users’ interactions and comments
In a paper published on the preprint server Arxiv.org, researchers affiliated with Microsoft and Arizona State University propose an approach to detecting fake news that leverages a technique called weak social supervision. They say that by enabling the training of fake news-detecting AI even in scenarios where labeled examples aren’t available, weak social supervision opens the door to exploring how aspects of user interactions indicate news might be misleading.
According to the Pew Research Center, approximately 68% of U.S. adults got their news from social media in 2018 — which is worrisome considering misinformation about the pandemic continues to go viral, for instance. Companies from Facebook and Twitter to Google are pursuing automated detection solutions, but fake news remains a moving target owing to its topical and stylistic diverseness.
Building on a study published in April, the coauthors of this latest work suggest that weak supervision — where noisy or imprecise sources provide data labeling signals — could improve fake news detection accuracy without requiring fine-tuning. To this end, they built a framework dubbed Tri-relationship for Fake News (TiFN) that models social media users and their connections as an “interaction network” to detect fake news.
Interaction networks describe the relationships among entities like publishers, news pieces, and users; given an interaction network, TiFN’s goal is to embed different types of entities, following from the observation that people tend to interact with like-minded friends. In making its predictions, the framework also accounts for the fact that connected users are more likely to share similar interests in news pieces; that publishers with a high degree of political bias are more likely to publish fake news; and that users with low credibility are more likely to spread fake news.
To test whether TiFN’s weak social supervision could help to detect fake news effectively, the team validated it against a Politifact data set containing 120 true news and 120 verifiably fake pieces shared among 23,865 users. Versus baseline detectors that consider only news content and some social interactions, they report that TiFN achieved between 75% to 87% accuracy even with a limited amount of weak social supervision (within 12 hours after the news was published).
In another experiment involving a separate custom framework called Defend, the researchers sought to use as a weak supervision signal news sentences and user comments explaining why a piece of news is fake. Tested on a second Politifact data set consisting of 145 true news and 270 fake news pieces with 89,999 comments from 68,523 users on Twitter, they say that Defend achieved 90% accuracy.
[W]ith the help of weak social supervision from publisher-bias and user-credibility, the detection performance is better than those without utilizing weak social supervision. We [also] observe that when we eliminate news content component, user comment component, or the co-attention for news contents and user comments, the performances are reduced. [This] indicates capturing the semantic relations between the weak social supervision from user comments and news contents is important,” wrote the researchers. “[W]e can see within a certain range, more weak social supervision leads to a larger performance increase, which shows the benefit of using weak social supervision.”
Let’s block ads! (Why?)
Big Data – VentureBeat