CBPO

Monthly Archives: May 2019

Google Image Classification and Landmarks

May 20, 2019 No Comments

Image Classification in the past

Back in 2008, I was writing about how a search engine might learn from photo databases like Flickr, and how people label images there in a post I wrote called, Community Tagging and Ranking in Images of Landmarks

In another post that covers the Flickr image classification Landmark work, Faces and Landmarks: Two Steps Towards Smarter Image Searches, I mentioned part of what the Yahoo study uncovered:

Using automatically generated location data, and software that can cluster together similar images to learn about images again goes beyond just looking at the words associated with pictures to learn what they are about.

That is using metadata from images in an image collection, which is very different from what Google is doing in this post about identifying landmarks in the post, How Google May Interpret Queries Based on Locations and Entities (Tested), where it might identify landmarks based upon a knowledge of their actual location.

More Recent Image Classification of Landmarks

I mention those earlier posts because I wanted to share what I had written about landmarks, before pointing to more recent studies from Google about how they might recognize landmarks, a year apart from each other, with one being a followup to the other.

The first of these papers, Google-Landmarks: A New Dataset and Challenge for Landmark Recognition, starts out by telling us about a problem that needs solving:

Image classification technology has shown remarkable improvement over the past few years, exemplified in part by the Imagenet classification challenge, where error rates continue to drop substantially every year. In order to continue advancing the state of the art in computer vision, many researchers are now putting more focus on fine-grained and instance-level recognition problems – instead of recognizing general entities such as buildings, mountains and (of course) cats, many are designing machine learning algorithms capable of identifying the Eiffel Tower, Mount Fuji or Persian cats. However, a significant obstacle for research in this area has been the lack of large annotated datasets.

A year later, Google worked to improve the dataset that was being used for image classification when identifying landmarks, and updated the dataset that they had created the year before, as they tell us in,Announcing Google-Landmarks-v2: An Improved Dataset for Landmark Recognition & Retrieval Part of the effort behind that work came from getting a lot of help as described in the blog post announcing it:

A particular problem in preparing Google-Landmarks-v2 was the generation of instance labels for the landmarks represented since it is virtually impossible for annotators to recognize all of the hundreds of thousands of landmarks that could potentially be present in a given photo. Our solution to this problem was to crowdsource the landmark labeling through the efforts of a world-spanning community of hobby photographers, each familiar with the landmarks in their region.

Google Patent for Image Classification when Identifying Landmarks in Image Collections

image classification

Google was recently granted a patent that focuses on identifying popular landmarks in large digital image collections. Considering Google operates Google photos, that makes a lot of sense. The landmark identification efforts at Flickr sound a little similar to this effort on Google’s part. The patent does target a specific problem which it tells us is:

However, there is no known system that can automatically extract information such as the most popular tourist destinations from these large collections. As numerous new photographs are added to these digital image collections, it may not be feasible for users to manually label the photographs in a complete and consistent manner that will increase the usefulness of those digital image collections. What is needed, therefore, are systems and methods that can automatically identify and label popular landmarks in large digital image collections.

Some of it does sound similar to the Flickr efforts where it talks about working to populate and update “a database of images of landmarks including geo-clustering geo-tagged images according to geographic proximity to generate one or more geo-clusters, and visual-clustering the one or more geo-clusters according to image similarity to generate one or more visual clusters.”

How might this play into image classification and search involving landmarks?

The patent describes how it could fit into searches, with the following steps:

  • Enhancing user queries to retrieve images of landmarks, including the stages of receiving a user query
  • Identifying one or more trigger words in the user query
  • Selecting one or more corresponding tags from a landmark database corresponding to the one or more trigger words
  • Supplementing the user query with the one or more corresponding tags, generating a supplemented user query

Trigger words appearing in queries is interesting.

The patent also tells us that it could also involve a method of automatically tagging a new digital image, which would also cover:

  • Comparing the new digital image to images in a landmark image database, wherein the landmark image database comprises visual clusters of images of one or more landmarks
  • tagging the new digital image with at least one tag based on at least one of said visual clusters

The patent is:

Automatic discovery of popular landmarks
Inventors: Fernando A. Brucher, Ulrich Buddemeier, Hartwig Adam and Hartmut Neven
Assignee: Google LLC
US Patent: 10,289,643
Granted: May 14, 2019
Filed: October 3, 2016

Abstract

In one embodiment the present invention is a method for populating and updating a database of images of landmarks including geo-clustering geo-tagged images according to geographic proximity to generate one or more geo-clusters, and visual-clustering the one or more geo-clusters according to image similarity to generate one or more visual clusters. In another embodiment, the present invention is a system for identifying landmarks from digital images, including the following components: a database of geo-tagged images; a landmark database; a geo-clustering module; and a visual clustering module. In other embodiments, the present invention may be a method of enhancing user queries to retrieve images of landmarks or a method of automatically tagging a new digital image with text labels.

Even Smarter Image Classification of Landmarks

This system appears to be capable of finding very popular landmarks in photo collections across the web and storing those in a landmark database, where it might geocluster those. It’s interesting thinking about this effort. If Google Might use those landmark images in Image Search Results, it may not stop image classification at that point

I recently wrote about Google Image Search Labels Becoming More Semantic? where we were told in an updated Google Patent that images were being labeled based upon an ontology related to the topics of those images. A Google image search for a landmark like The Washington Monument shows a number of image classification labels at the top of the results that can be clicked on if you want to narrow down the results to specific aspects of those monuments.

So, image classification may include specific monuments, and then even more narrow classifications, like having the following labels applied to the Washington Monument:

Reflecting Pool
Lincoln Memorial
Washington DC
Elevator
Inside
Trump
Construction
Top
Baltimore
Earthquake
Building
interior
Capstone
original
sunrise
Observation deck
National Mall

So, Google may have smarter image classification when it comes to landmarks, but it is labeling them so that they are more meaningful, too.


Copyright © 2019 SEO by the Sea ⚓. This Feed is for personal non-commercial use only. If you are not reading this material in your news aggregator, the site you are looking at may be guilty of copyright infringement. Please contact SEO by the Sea, so we can take appropriate action immediately.
Plugin by Taragana

The post Google Image Classification and Landmarks appeared first on SEO by the Sea ⚓.


SEO by the Sea ⚓


Your First Google Ads Campaign: Everything You Should Know

May 20, 2019 No Comments

This article goes over the very basics of getting started advertising with Google Ads for someone with little to no experience.

Read more at PPCHero.com
PPC Hero


TikTok owner ByteDance’s long-awaited chat app is here

May 20, 2019 No Comments

In WeChat -dominated China, there’s no shortage of challengers out there claiming to create an alternative social experience. The latest creation comes from ByteDance, the world’s most valuable startup and the operator behind TikTok, the video app that has consistently topped the iOS App Store over the last few quarters.

The new offer is called Feiliao (飞聊), or Flipchat in English, a hybrid of an instant messenger plus interest-based forums, and it’s currently available for both iOS and Android. It arrived only four months after Bytedance unveiled its video-focused chatting app Duoshan at a buzzy press event.

Screenshots of Feiliao / Image source: Feiliao

Some are already calling Feiliao a WeChat challenger, but a closer look shows it’s targeting a more niche need. WeChat, in its own right, is the go-to place for daily communication in addition to facilitating payments, car-hailing, food delivery and other forms of convenience.

Feiliao, which literally translates to ‘fly chat’, encourages users to create forums and chat groups centered around their penchants and hobbies. As its app description writes:

Feiliao is an interest-based social app. Here you will find the familiar [features of] chats and video calls. In addition, you will discover new friends and share what’s fun; as well as share your daily life on your feed and interact with close friends.

Feiliao “is an open social product,” said ByteDance in a statement provided to TechCrunch. “We hope Feiliao will connect people of the same interests, making people’s life more diverse and interesting.”

It’s unclear what Feiliao means by claiming to be ‘open’, but one door is already shut. As expected, there’s no direct way to transfer people’s WeChat profiles and friend connections to Feiliao, and there’s no option to log in via the Tencent app. As of Monday morning, links to Feiliao can’t be opened on WeChat, which recently crossed 1.1 billion monthly active users.

On the other side, Alibaba, Tencent’s long-time nemesis, is enabling Feiliao’s payments function through the Alipay digital wallet. Alibaba has also partnered with Bytedance elsewhere, most notably on TikTok’s Chinese version Douyin where certain users can sell goods via Taobao stores.

In all, Flipchat is more reminiscent of another blossoming social app — Tencent-backed Jike — than WeChat. Jike (pronounced ‘gee-keh’) lets people discover content and connect with each other based on various topics, making it one of the closest counterparts to Reddit in China.

Jike’s CEO Wa Nen has taken noticed of Feiliao, commenting with the 👌 emoji on his Jike feed, saying no more.

Screenshot of Jike CEO Wa Ren commenting on Feiliao

“I think [Feiliao] is a product anchored in ‘communities’, such as groups for hobbies, key opinion leaders/celebrities, people from the same city, and alumni,” a product manager for a Chinese enterprise software startup told TechCrunch after trying out the app.

Though Feiliao isn’t a direct take on WeChat, there’s little doubt that the fight between Bytedance and Tencent has heated up tremendously as the former’s army of apps captures more user attention.

According to a new report published by research firm Questmobile, ByteDance accounted for 11.3 percent of Chinese users’ total time spent on ‘giant apps’ — those that surpassed 100 million MAUs — in March, compared to 8.2 percent a year earlier. The percentage controlled by Tencent was 43.8 percent in March, down from 47.5 percent, while the remaining share, divided between Alibaba, Baidu and others, grew only slightly from 44.3 percent to 44.9 percent over the past year.


Social – TechCrunch


Inside Facebook’s New Robotics Lab, Where AI and Machines Friend One Another

May 20, 2019 No Comments

The social network has a plan to merge the worlds of artificial intelligence and real-world machines, so that both may grow more powerful.
Feed: All Latest


Like *Game of Thrones* Languages? Here’s How to Make Your Own

May 18, 2019 No Comments

We got pro tips from the master—like, the actual guy who created Dothraki and High Valyrian.
Feed: All Latest


Under the hood on Zoom’s IPO, with founder and CEO Eric Yuan

May 18, 2019 No Comments

Extra Crunch offers members the opportunity to tune into conference calls led and moderated by the TechCrunch writers you read every day. This week, TechCrunch’s Kate Clark sat down with Eric Yuan, the founder and CEO of video communications startup Zoom, to go behind the curtain on the company’s recent IPO process and its path to the public markets.

Since hitting the trading desks just a few weeks ago, Zoom stock is up over 30%. But the Zoom’s path to becoming a Silicon Valley and Wall Street darling was anything but easy. Eric tells Kate how the company’s early focus on profitability, which is now helping drive the stock’s strong performance out of the gate, actually made it difficult to get VC money early on, and the company’s consistent focus on user experience led to organic growth across different customer bases.

Eric: I experienced the year 2000 dot com crash and the 2008 financial crisis, and it almost wiped out the company. I only got seed money from my friends, and also one or two VCs like AME Cloud Ventures and Qualcomm Ventures.

nd all other institutional VCs had no interest to invest in us. I was very paranoid and always thought “wow, we are not going to survive next week because we cannot raise the capital. And on the way, I thought we have to look into our own destiny. We wanted to be cash flow positive. We wanted to be profitable.

nd so by doing that, people thought I wasn’t as wise, because we’d probably be sacrificing growth, right? And a lot of other companies, they did very well and were not profitable because they focused on growth. And in the future they could be very, very profitable.

Eric and Kate also dive deeper into Zoom’s founding and Eric’s initial decision to leave WebEx to work on a better video communication solution. Eric also offers his take on what the future of video conferencing may look like in the next five to 10 years and gives advice to founders looking to build the next great company.

For access to the full transcription and the call audio, and for the opportunity to participate in future conference calls, become a member of Extra Crunch. Learn more and try it for free. 

Kate Clark: Well thanks for joining us Eric.

Eric Yuan: No problem, no problem.

Kate: Super excited to chat about Zoom’s historic IPO. Before we jump into questions, I’m just going to review some of the key events leading up to the IPO, just to give some context to any of the listeners on the call.


Enterprise – TechCrunch


Using IF functions on Google Ads to improve productivity

May 18, 2019 No Comments

Back in the days when I was learning PPC, one of the two biggest growing pains I had were:

  1. Learning the difference between segmenting campaigns out to maximize efficiency
  2. Reaching the point where the juice is no longer worth the squeeze

Rather than creating clutter and a burdensome account to manage, I’ve since learned to make use of everything I can to speed up my workflow and free up bandwidth to focus on things that actually make a difference.

IF functions are a versatile means to tailor your ads to users in real time, using either the type of device they’re browsing on or the audience segment they belong to as signals to serve up specialized ad copy. The right message at the right time can make all the difference between a conversion or another bounced visitor. Search marketing is rapidly moving towards heavy automation and personalization, so IF functions are helpful because they’re a simple way to keep your seat at the table.

Setting up IF functions

The process of setting up IF Functions is painless. You could easily set one up in the time it will take to finish this article, regardless of your comfort level with Excel formulas. And if doing it on Excel is too daunting, you can set them up directly in the Google Ads UI under the Ads tab.

The basic logic is as follows

{=IF(condition is met, show this text):If not, show this text}.

So, if you wanted specific messaging for users on mobile, the logic runs something like this:

IF the user is ON a mobile device, show mobile-friendly CTA. If not, show the general CTA.

To put that in the basic formula

{=IF(device=mobile, Call Now!):Get a Quote.}

Another common usage of IF statements is serving specific offers to specific audience segments.

The basic formula for audience-based IF functions is

{=IF(Audience IN(audience name), Audience-specific copy.):General copy}

To put the above into a sentence: “If a user is IN this specific audience segment, serve them this specific copy. Otherwise, serve this more general copy.”

Suppose you were running a tiered promotion, where Club Members were eligible for an additional 15% discount on top of a 30% off sale, that text would look something like this:

Shop Now for{=IF(Audience IN(ClubList),45%):30%} Off!

Or, if your nurture campaigns weren’t entirely broken out and you wanted to move recent visitors into booking a consultation, you might have something like:

{=IF(Audience IN(Returning Visitor 7 Days), Book Your Consultation Today!):Download Our Free Guide.

Take note that you can target multiple audience segments in the same IF function. However, you are still limited to two copy options. The syntax is the same, just with your audiences separated by commas in the Audience IN section –

{=IF(Audience IN(Segment1,Segment2,Segment3)Learn More!):Get a Quote.}

If you’re feeling overwhelmed by keeping track of all of those brackets, commas, and colons, you can also build IF functions directly in the Google Ads UI. Simply add an open bracket in an ad field, anywhere from the headline one to URL paths one or two (note that ad customizers in Final URLs are not supported) and let the system walk you through putting it together.

Things to note while using IF functions

  • The character limits for each field still apply (but only for the ad text defined in your functions).
  • Symbols in the function’s ad text options like quote marks (both single and double), commas, and colons will need to be preceded by backslashes (\) for the function to work properly. For example, rather than “SearchEngineWatch’s” your function copy would read “SearchEngineWatch/’s.”

Using IF functions for fun and profit

Although IF functions don’t offer as many options to customize ads as using a business data feed, the options they do provide are staggering.

Shaping expectations based on device type is a must. While mobile browsers have come a long way in recent years, filling out long forms on a small screen with no keyboard is a slog, and desktop users might not have the same propensity to turn into brick and mortar visitors.

Tailoring your copy for devices isn’t a replacement for setting realistic device bid modifiers and taking cross-device/cross-channel conversions into account. But it is another way to squeeze more efficiency out of your ad budget.

Beyond device-type, the real power of IF functions come from the ease with which you can target specific audience segments. If you have a large enough CRM list to make customer match audiences viable for search, great. If your lists aren’t quite big enough, have no fear, you can create details of the possible audiences in Google Analytics and import it to Google Ads, the options are endless.

Bonus: Countdown ads

Countdown ads are yet another feature that is effective and easy to use but tend to fly under the radar. Beyond highlighting promotions, I’ve seen success in highlighting shipping windows (keep that in mind for the holiday shopping season), special events (for example, store openings), and more. Just like the other customizers available, countdowns can be put anywhere in an ad except for the URL.

The syntax is pretty straightforward

  • Specify a date in Year/Month/Day, pick a time in Hour:Minute:Second
  • Specify the language you’re targeting, and how many days you’d like the countdown to run

In the below example, the countdown will end at midnight on June 7, 2019, after starting seven days prior

{=COUNTDOWN(“2019/7/7 12:00:00″,”en-US”,7)}

The future is now

Running a successful paid search campaign has always required knowing who your customers are. Ad customizers make reaching the right user with the right messaging easier, and at scale. IF functions are easy inroads towards better tailoring of your users’ experiences towards their needs. It gives you more control over your ad copy than dynamic keyword insertion or responsive search ads, with a lower likelihood of matching to undesirable search queries than dynamic search ads. And with less setup needed than the Ad Customizer feeds, IF functions ultimately give savvy search marketers a powerful tool to boost performance.

Have any queries or interesting functions you know? Share them in the comments.

Clay Schulenburg is Director of SEM at PMG.

The post Using IF functions on Google Ads to improve productivity appeared first on Search Engine Watch.

Search Engine Watch


The PPC Newsflash: Key Takeaways from Google Marketing Live

May 17, 2019 No Comments

Did you miss Google Marketing Live? Join us for an expert recap.

Read more at PPCHero.com
PPC Hero


Facebook changes algorithm to promote worthwhile & close friend content

May 17, 2019 No Comments

Facebook is updating the News Feed ranking algorithm to incorporate data from surveys about who you say are your closest friends and which links you find most worthwhile. Today Facebook announced it’s trained new classifiers based on patterns linking these surveys with usage data so it can better predict what to show in the News Feed. The change could hurt Pages that share clickbait and preference those sharing content that makes people feel satisfied afterwards.

For close friends, Facebook surveyed users about which people they were closest too. It then detected how this matches up with who you are tagged in photos with, constantly interact with, like the same post and check in to the same places as, and more. That way if it recognizes those signals about other people’s friendships, it can be confident those are someone’s closest friends they’ll want to see the most of. You won’t see more friend content in total, but more from your best pals instead of distant acquaintances.

A Facebook News Feed survey from 2016, shared by Varsha Sharma

For worthwhile content, Facebook conducted surveys via News Feed to find out which links people said were good uses of their time. Facebook then detected which types of link posts, which publishers and how much engagement the posts got and matched that to survey results. This then lets it determine that if a post has a similar style and engagement level, it’s likely to be worthwhile and should be ranked higher in the feed.

The change aligns with CEO Mark Zuckerberg’s recent comments declaring that Facebook’s goal isn’t total time spent, but time well spent with meaningful content you feel good about. Most recently, that push has been about demoting unsafe content. Last month Facebook changed the algorithm to minimize clickbait and links to crappy ad-filled sites that receive a disproportionately high amount of their traffic from Facebook. It cracked down on unoriginality by hiding videos ripped off from other creators, and began levying harsher demotions to repeat violators of its policies. And it began to decrease the distribution of “borderline content” on Facebook and Instagram that comes close to but doesn’t technically break its rules.

While many assume Facebook just juices News Feed to be as addictive in the short-term as possible to keep us glued to the screen and viewing ads, that would actually be ruinous for its long-term business. If users leave the feed feeling exhausted, confused and unfulfilled, they won’t come back. Facebook’s already had trouble with users ditching its text-heavy News Feed for more visual apps like Instagram (which it luckily bought) and Snapchat (which it tried to). While demoting clickbait and viral content might decrease total usage time today, it could preserve Facebook’s money-making ability for the future while also helping to rot our brains a little less.


Social – TechCrunch


Retail Zipline raises $9.6M from Emergence and Serena Williams

May 17, 2019 No Comments

Retail Zipline, a startup aiming to improve communication between retail stores and corporate decision makers, announced today that it has raised $ 9.6 million in Series A funding.

CEO Melissa Wong previously worked in corporate communications for Old Navy, where she said she saw “such a disconnect between what was decided in headquarters and what was decided in stores.” For example, management might decide on a big marketing push to sell any remaining Mother’s Day-related items after the holiday has passed, but then “the stores wouldn’t do it.”

“The stores would say there were too many messages, they didn’t see the memo, they didn’t know it was a priority,” Wong said.

So she founded Retail Zipline with CTO Jeremy Baker, with the goal of building better communication tools for retailers. Baker said that while they looked at existing chat and task management software for inspiration, those tools were “mostly built for people sitting at a desk all day,” rather than workers who are “on the floor, dealing with customers.”

Retail Zipline’s features include messaging and task management — plus a centralized library of documents and multimedia and a survey tool to track results and feedback from stores.

Retail Zipline screenshot

To illustrate how the software is actually being used, Baker outlined a scenario where an athletic shoe company is launching “a huge initiative,” with a big-name athlete signed on to promote the latest pair of shoes.

“In a traditional environment, someone might FedEx over a package to [the store], someone might send an email down, ‘Hey, look for a package on this day,’ someone else from the marketing team might say, ‘Hey guys, we’re doing a shoe launch,’ ” he said. “All of this in these disparate systems, where people have to piece together the story. It’s kind of like a murder mystery.”

Baker said that Retail Zipline, on the other hand, provides a single place to find all the needed materials and tasks “tied together with a bow, instead of a store manager spending 10-plus hours in the back room trying to piece this thing together, or even worse not seeing it.”

The company’s customers include Casper, LEGO and Lush Cosmetics. Wong said Retail Zipline works “with anyone that has a retail location” — ranging from Gap, Inc. with thousands of stores, to Toms Shoes with 10.

The funding was led by Emergence, with Santi Subotovsky and Kara Egan from Emergence both joining the startup’s board of directors. Serena Williams’ new firm Serena Ventures also participated.

“As someone with an incredibly active life, I understand the need to be dynamic, and capable of quickly adapting to shifting priorities, but I’m also aware of the stress a fast-paced work environment can impose,” Williams said in a statement. “Retail Zipline is tackling this issue head-on in retail – a notoriously stressful industry – by pioneering products that help store associates get organized, communicate efficiently, and deliver amazing customer experiences.”


Startups – TechCrunch