All posts tagged ‘data’

Mining Flickr to Build 3D Models of the World

Microsoft’s PhotoSynth tool is jaw-droppingly awesome. But, because it’s a Microsoft project, the technology is unlikely to appear on some of your favorite non-Microsoft online apps, like Google Maps or Flickr.

However, our friends at ReadWriteWeb stumbled across a very similar tool — at least in terms of the end result — developed by the University of North Carolina in conjunction with Swiss university, ETH-Zurich.

The team has developed a method for creating 3D models by pulling in millions of photographs from Flickr and using some fancy algorithms to generate 3D models of local landmarks. Perhaps even more impressive the results can be generated using a single computer in under a day.

Project lead Jan-Michael Frahm touts the project’s efficiency saying, “our technique would be the equivalent of processing a stack of photos as high as the 828-meter Dubai Towers, using a single PC, versus the next best technique, which is the equivalent of processing a stack of photos 42 meters tall — as high as the ceiling of Notre Dame — using 62 PCs. This efficiency is essential if one is to fully utilize the billions of user-provided images continuously being uploaded to the internet.”

While the results are cool and would make an impressive addition to any number of geo-based services, more serious use cases include helping disaster workers get a better idea of where they’re headed and the extent of damage.

So far the researchers have released a movies demonstrating the technique on landmarks in both Rome (get it? built in a day…) and Berlin, and the results are impressive. For more information on how the process works, check out the UNC website.

See Also:

Share Your Mundane Details

Pie chart of ingesting liquid

The web has all sorts of data, but it’s sorely missing yours. If you request an account from Daytum, you can change that.

The site lets its users collect data about themselves and share it via beautiful charts. Track your coffee consumption, how often you exercise, or anything else that matters to you.

One of the site’s creators, Nicholas Felton creates a personal annual report each year that shows his year in data. Through gorgeous visualizations, Felton shows off the mundane. And now you can, too.

Felton gave a talk about the history of Daytum at the Future of Web Design. His slides are embedded below:

Using the site is super easy. You can drop in, add some data, and drop out. There’s also a Twitter submission system, for adding your data on the go.

The site is so simple, yet has a number of options for users. There’s lots of room for improvement, such as an API to allow some automated inclusion of data. Overall, this is a fun, easy-to-use site that’s worth checking out.

See also:

File Under: Visual Design

Dashboard Shows Our Open Data Future

Sprint data dashboard

Sprint’s beautiful dashboard is full of widgets showing a glimpse into what’s going on in the world. Among the things it tracks: sticky notes being produced, organ transplant operations, and temperature around the world.

There’s an info pane for each, explaining where the data comes from. Most appear to be based on estimates (nobody is counting each coffee cup as it’s being produced), but what if it really was all live?

If the future of the web is open APIs, this will be possible. What if we all had a dashboard like this, connecting us to the data that matters to us?

[via Daring Fireball]

See also:

File Under: Visual Design

Make Data Not Look Like Data

Just about every web page has some sort of data behind it. This blog post, for example, has a title, a date, categories, tags, and the body of the post itself, in addition to some other metadata. All of this data sits in a row in the Webmonkey database, waiting for you to call it up. When we display a blog post, you see the content–usually something new or cool in web development. You don’t see data.

The point: the technical site of creating web sites is often about showing data, but users don’t see it that way.

User interface, data, and how users see them

In his presentation Experience is the Product, Peter Merholz has a great series of slides, which I’ve summarized in the above graphic. Even though data is sometimes central to a website, it shouldn’t look like data.

Author info seen as dataUnfortunately, there are plenty of examples of websites putting the data first. This screenshot comes from a DZone article, appropriately about databases. I would guess that first name and last name are fields in their database. Separating the name into two pieces can be useful, but only behind the scenes. For example, alphabetizing by last name is possible when it’s stored separately.

The way DZone shows the author name isn’t very useful. It starts with the data, rather than the way the user expects to see the data (that is, without the field labels). Yes, this example is nit-picky, but keep an eye out and you’ll notice sites that put the data first. Maybe you’ll even notice yourself doing it, then find a way to make the data blend into the interface.

See also:

File Under: Software & Tools

Google Adds Insight to Trends

Google Insights for SearchGoogle released another data-mining tool meant to help its advertisers spend wisely. Google Insights looks a lot like Google Trends. In fact, I wonder why both tools still exist (Google has done it before: see Sites vs. Pages).

Insights does beyond Trends, with map visualization and categories. Like Trends, you can download the data for your own number crunching.

Here’s an example of how Insights lets you dig a little deeper. Say I wanted to find out when and where users have been searching for Sam Adams, the beer (and yes, the colonial U.S. statesman):

Sam Adams searches

Well, there’s another Sam Adams I haven’t mentioned. I happen to know about him because he’s the mayor-elect of my home city, Portland, Oregon. Check out how blue Oregon is in the map, and how the graph ticked up dramatically around our May 20 election.

With Insights, I can restrict to just the beer by selecting the Food and Drink category. With that done, the graph and the map make a little more sense:

Sam Adams searches in the food and drink category

One feature from Trends that is lacking in Insights is the news view. This would have been useful to someone who doesn’t know about my city’s new mayor. In Trends, the graph of search term popularity is labeled with news stories that might explain the move up or down. Google Finance also has this feature, but it’s hit and miss correlating stocks with press releases.

User searching seems more rational than investing. For example, this Google trends graph explains the surge in search for “obama” in June with a link to news.

Trends search for Obama

Insights doesn’t.

Trends search for Obama

The other Insights features make up for it. Use it to determine which locales would be most open to your product/service or which presidential candidate has more buzz in contentious Michigan.

This is normally the spot where we would beg for an API, to automate our research, or build something even cooler on top of this data. Well, we might not have to. From what I can see, the download option acts as a de facto API. The same arguments sent to the web version can be sent to a CSV (Comma Separated File) version.

But that might not be the most exciting part of Google Insights. Notice in its logo there’s a “for search” tacked onto the end of the name. Also, the URL is /insights/search instead of just /insights. That suggests we might be seeing Google Insights for Something Else down the line. Any guesses?

See also: