MS Defense

Modeling Individual Nodes In Dynamic Link Prediction

Maksym Morawski

2:00pm Thursday, 25 April 2013, ITE325b, UMBC

The question of how to predict which links will form in a graph, given the graph’s history, is an open research problem in computer science. There are many different approaches to the link prediction problem, one of which involves building a set of features for pairs of nodes and using supervised learning to build a model that predicts when these pairs of nodes will link. Typically, this model is learned over the entire graph. In this thesis, I investigate building this model over each individual node in an attempt to learn the particular ways in which that node behaves before making predictions about it. In addition, research into link prediction to date lacks intelligent ways of utilizing the graph over large timespans. To address this, I introduce a variety of ways to include temporality into the link prediction process by introducing new ways of using existing features.

Committee: Dr. Marie desJardins (Chair), Dr. Tim Oates, Dr. Tim Finin

Update (24.04.2013): A standalone eBook Bundle option is now available as well. The bundle contains two eBooks — for iBooks, Kindle, Nook and most other eBook readers and devices.

It’s that time of the year again: We’ve started working on a brand new Smashing Book! The book will be as ambitious as it is beautiful — a unique, extraordinary artefact, filled with valuable insight for Web designers and developers.

The Smashing Book #4 consists of two separate hardcover books, neatly packaged into one slipcase. The covers are designed by illustrator Anna Shuvalova.

The Smashing Book #4: New Perspectives on Web Design will consist of two hardcover books, “Coding” and “Design,” each with approximately eight chapters, both neatly packaged into one slipcase. New Perspectives will have practical tips on current practices and common design problems. We strongly believe in print and in the benefits of tangible books. And this time, we’d like to invite you to join us as we embark on the journey of publishing!

The Beauties And The Box

Excited as we are? Good! The release of this book is expected to be in September or October 2013. However, by pre-ordering your copy now, you will support our expansion of the Smashing Book series. And if you’re quick enough, you can grab a couple of extra (and exclusive) goodies.

New Perspectives On Coding And Design

The book will feature valuable insight into large-scale projects, adaptive interfaces, customer support, user psychology and typography. We will also uncover smart front-end strategies, obscure back-end techniques and find out what it takes to improve performance for faster and more robust websites and apps.

Table Of Contents

We’ve assembled a remarkable line-up of authors for the book. All of them are well-respected designers and developers who will share their practical insights, design strategies and hands-on tools as they explore new perspectives in Web design today. (Please note that chapter titles might still change!)

New Perspectives On Coding:

New Perspectives On Design:

Technical Details

• Two books, approximately 300 pages each, 16.5 × 24.0 cm (6.5 × 9.5 inches).
• Quality hardcovers, with stitched binding and ribbon page markers.
• A unique slipcase with a special surprise — just for you.
• Excellent for both new and experienced Web designers and developers.
• To be delivered from both the US and Germany.
• Free worldwide shipping, with a 100-day money-back guarantee.
• All eBook formats are included in both packages (PDF, EPUB, Kindle).

The Smashing Book #4 consists of two separate hardcover books, neatly packaged into one slipcase. The covers are designed by illustrator Anna Shuvalova.

Pre-Order Your Copy Today!

We want to create the best printed book experience that we’ve crafted so far — a truly smashing book, with content, design and packaging that you will love. Please note that the Smashing Book #4 is also available as a part of the Smashing Book #4 + eBook Library bundle and the Smashing Fan bundle.

© Smashing Editorial for Smashing Magazine, 2013.

The Mobile First philosophy has radically changed how professionals approach Web design and become the way companies as diverse as Facebook and IBM build their products.

The Mobile First approach is to start designing for mobile devices — which typically have less screen size and less capabilities than desktops — and progressively enhance the product; so that desktops get an enhanced site experience rather than mobiles getting a pared down one.

We grabbed the opportunity to interview Luke Wroblewski, who first defined the Mobile First concept back in 2009, about how he used the principle to create Polar at his latest startup, Input Factory Inc. where he is co-founder and CEO.

How did you get started building apps and what kept you fascinated?

The first mobile apps I worked on were during my time at Yahoo! I joined the Yahoo! Search team back in 2005 and a bit later was heading up a small “tiger team” focused on ideas for products that were 3 to 5 years out. At that time I was fascinated with where the web was going and in particular with mobile.

We started out building some experiences for newer Nokia devices, as Nokia was the big player back then. Soon after though, Apple announced the app store for iOS and we jumped into iOS applications as well. At the time we were experimenting with services that connected mobile apps to networked TVs and more traditional computers like laptops and desktops. It was a really great opportunity to explore what was coming and we came up with a bunch of concepts that I’m still passionate about today.

I think that’s what keeps me interested in this space. You can see the future: more connected devices of every shape and size; interactions between those devices; more real time access to useful information, services, and people –it’s all coming. But these things don’t just appear out of thin air. They take years of effort, trial and error to make real. So I keep at it because I keep seeing a more and more exciting future ahead.

When developing a product, how do you identify a gap in such a saturated market?

I don’t. If a market is saturated I think that’s a great sign it’s interesting on a number of levels. For me, its much more important to focus on problems I can understand and actually do something about. When you have deep experience in an area, you can often see a future other people can’t.

For instance, I’ve dug really deep into web form and mobile design. I even wrote two popular books on these topics. So I feel like I’ve increased my ability to see problems in these areas. And when I look across the Internet I see lots of people eager to share their opinions and get the opinions of others. But the solutions out there are just really bad.

You’ve got surveys that basically consist of multiple pages of form elements: checkboxes, radio buttons, text fields, and so on. Because they’re so painful to complete, companies are paying people to take these surveys and even then the participation rates are really low. I look at this and think: we can do better.

It doesn’t really matter if there are a lot of companies out there with apps for making surveys or soliciting feedback. If you see a problem and think you can do a better job solving it, that’s the on-ramp. That may sound overly confident but I think you need confidence to get out there and start doing your own thing. You have to believe in it or no one else will.

Why is the Mobile First approach a better way to do things?

Well the reasons have been stacking up over the years. But when I outlined the idea originally I pointed to three main reasons: growth, focus, and innovation.

Growth is pretty obvious these days. More than 2 million modern mobile devices enter the world each day. Compare that to the 371,000 children born per day and you can quickly see how these numbers add up. All these devices connected to networks is a huge opportunity and many companies are now feeling it in their stats as mobile begins to take over other kinds of devices in usage.

Focus comes from the natural constraints of mobile. These devices need to be portable, so their screens are small, they connect to networks anywhere and everywhere with varying success, and they get used in very diverse environments often full of distractions. These constraints push you toward more focused, simplified solutions. You can only fit so much on the screen, people often have to wait for it, and they’re unlikely to give you their full attention. So make it easy to understand and use and focus on the important things first. Mobile is a great forcing function for simplicity.

But mobile isn’t just about constraining yourself; quite the opposite. There are lots of things that make mobile experiences more powerful and engaging. Not least of which is the fact that mobile devices can be used all the time and just abut anywhere. That not only creates new uses but also means people can be connected throughout the entire day.

If that weren’t enough, due to the capabilities of mobile devices, we have new ways of creating experiences. Thanks to local detection technology, we know where people are down to 10 meters. Thanks to integrated cameras and microphones we can take in visual and audio input, process it, change it, and share it. Thanks to motion sensors we can tell where a device is in three-dimensional space, and the list goes on.

It’s easy to dismiss these capabilities as technology for technology’s sake. Instead think of them as new techniques or paints on your design palette. With them, you can paint a totally unique user experience that allows you to innovate and move beyond existing solutions that came about before these technologies were available to us.

How do you know your design is hitting the mark, especially when aiming for speed and simplicity?

Well, you can test for both. In fact, that’s exactly what we did for our app, Polar. We designed Polar for mobile first and foremost so speed and simplicity were, of course, top of my mind. Earlier I mentioned that we thought we could make collecting and sharing opinions fast, easy, and fun, almost the exact opposite of what most survey tools are like online today.

Polar is our first attempt to do that. The most important interactions in Polar are collecting and sharing opinions and, as a result, we spent a lot of time trying to get those interactions right. To make sure they were fast and easy, we used one-handed, timed tests. Our goal was to allow anyone to vote on 10 polls or create a new poll in under 60 seconds.

If you design for the extremes, the middle usually works itself out. To quote Dan Formosa at Smart Design: When they designed garden shears, they tested them with people who had arthritis. If this “extreme” case could use the garden shears, then anyone could. That’s the same approach we take with timed, one-thumb use. If you can make it work for that extreme it will work for everyone.

Is there anything you can do from a design perspective to make sure that people who download the app actually use it?

Sure. Let them actually use it. In all seriousness, so many apps start off the process by wanting to tell you all about themselves and having you tell them all about you. Fill in this form, give us your phone number, take this intro tour, and so on. All this instead of just letting you get in and start using the app you’re spending all your time memorizing which gestures the app has and connecting to Facebook.

No actually, you’re skipping past all that trying to actually use the app. So my approach is just get to the good stuff. Let me say, however, that I know this approach is controversial. There are a number of examples out there that show forcing registration up front increases your sign-up numbers. Which means you increased the number of people who filled out a form. But they’ve never used your service. They might not even know what it is, so how valuable are these users?

I’m biased toward people who’ve seen and used the app, then decided to sign-up as a result. That means they liked what they saw enough to take the next step. The total number of sign-ups may be lower but the number of qualified users may end up being higher. At least that’s our approach!

Have you heard horror stories of people screwing up their signup process in their products?

The best one I saw recently was published by Greg Nudleman. It was an app for finding nearby restrooms made by Charmin. The intro process was so labor intensive that Greg very appropriately titled his write up: Let Them Pee!

You’ve use the term “gradual engagement” a lot in the past when you talk about sign up process, can you elaborate on what that is?

Sure. Gradual engagement is an alternative to the sign-up form issue I just described. Through gradual engagement, we can communicate what apps do and why people should care by allowing them to actually interact with them in gradual ways.

For example, Polar is all about sharing and collecting opinions. So we allow anyone opening the app for the first time to vote on the polls they see. 88% of people who download the app do just that. We hold on to all their votes locally so if they ever take the next step on the “walkway” (when you want to leave a comment or create a poll of your own), all your votes carry over to your account. This is what I mean by gradually getting people to understand and use your service. It’s about creating a clear and welcome “pathway” vs. putting up walls.

There has been much debate recently over whether improved device capabilities will render mobile-first obsolete, where do you stand on this?

I certainly hope all our devices keep getting better and that we develop new ways of interacting with information and with each other. So I’m not building a moat around mobile or anything. That said, the idea of having a connected device with you anywhere and everything is really powerful.

For proof, just look at Flurry’s recent analysis of user sessions and activity across all phone and tablet sizes. The clear winner for both was the “medium” sized (3.5”-5”) phone. I think this is a testament to the value of a portable computer that you can turn to at any moment for answers, conversations, and frankly just about anything. That kind of mobility and its importance does not show any signs of letting up in the near future. So I’m still really bullish on mobile.

We’d like to thank Luke for taking the time to answer these questions.

Do you take a Mobile First approach? What stops you from using an app or mobile site? Let us know in the comments.

Featured image/thumbnail, mobile internet image via Shutterstock.

The Ultimate Productivity eBook Bundle – only $29!