August 9, 2015
Penny Pincher is a fast template-based gesture recognizer, developed by Eugene Taranta and Joseph LaViola (full paper reference below). The algorithm is well-suited for mobile applications since it is both fast and accurate and, as shown in the evaluation by the authors, outperforms other recognizers. My Github project provides a Swift implementation of Penny Pincher and shows its usage in a simple example project. Also, the framework contains a UIGestureRecognizer subclass that integrates well into the existing gesture recognition framework of iOS.
Here’s the full reference for the paper:
Eugene M. Taranta, II and Joseph J. LaViola, Jr.. 2015. Penny pincher: a blazing fast, highly accurate $-family recognizer. In Proceedings of the 41st Graphics Interface Conference (GI ’15). Canadian Information Processing Society, Toronto, Ont., Canada, Canada, 195-202.
July 4, 2014
AccessRank is a Swift implementation of the AccessRank algorithm by Fitchett and Cockburn (full reference below) for iOS and Mac OS apps. The algorithm predicts which list items users might select or visit next by taking multiple sources of input into account. For instance, you could use AccessRank to generate a list of predictions for:
- which documents a user is most likely to open next.
- which commands in an auto-complete menu might be triggered next.
- which fonts in a font-chooser widget might be selected next.
- basically every part in a user interface where reuse or revisitation of things is involved…
To improve on other common methods such as recency-based and frequency-based predictions, AccessRank adds Markov weights, time weighting, and other parameters for calculating a final score for each item, while the algorithm tries to maximize both prediction accuracy and list stability. Prediction accuracy is important since top items are easier and faster to access than items in bottom sections; list stability is important since automatic reordering of items can impede usability when users try to reselect an item based on an already learned location. You can configure the algorithm depending on whether you prefer more prediction accuracy or more list stability.
Once AccessRank has calculated predictions, you can use the resulting list to enhance your user interface in various ways. For example, you could display the most likely next items in an additional list as suggestions, or you could visually highlight relevant objects to give users cues where they might want to go next.
Here’s the full reference for the AccessRank algorithm by Fitchett and Cockburn:
Stephen Fitchett and Andy Cockburn. 2012. AccessRank: predicting what users will do next. In Proceedings of the SIGCHI Conference on Human Factors in Computing Systems (CHI ’12). ACM, New York, NY, USA, 2239-2242.
June 9, 2014
BayesianTouchCriterion is an Objective-C implementation of the Bayesian Target Criterion (BTC) developed by Xiaojun Bi and Shumin Zhai. Given a touch point and some touch targets, it improves the accuracy of selections by finding the touch target with the shortest Bayesian Touch Distance (BTD). This element is the one that the user intended to select. The results of the evaluation of Bi and Zhai show that selection based on BTC is much more precise than using the visual boundary of an element. (Visual boundary checking is how most touch frameworks determine if the finger touches a target.)
The image above shows how randomly placed touch points (the smaller, semi-transparent circles) select their corresponding targets (the nine big circles). In the user interface of a mobile app, for example, BTC could be used to more accurately determine which menu item the user intended to select. The point of using BTC instead of visual boundary checking is: The intended element is not necessarily the element where the user’s touch lies exactly within the visual boundary of the element (e.g., the touch point might be somewhere slightly outside of the visual boundary). BTC has the advantage that it does not require more detailed information about the touch point (finger posture, touch area…). The only required input parameters of the formula are touch location, target center, and target diameter.
If you are interested in the full mathematical details and derivations, here’s the full reference:
Xiaojun Bi and Shumin Zhai. Bayesian touch: A statistical criterion of target selection with finger touch. In Proceedings of the 26th Annual ACM Symposium on User Interface Software and Technology, UIST ’13, pages 51–60, New York, NY, USA, 2013. ACM.
The paper is freely available on Research at Google.
February 5, 2014
Plumber is a tiny Objective-C library for creating curved connection lines between nodes (for flowcharts, for example). The constructed path connects the bounding boxes such that the distance between two sides is minimized. I’ve created it since a quick Google search did not reveal anything simple for iOS. The curve calculation algorithm is a direct port of the function used in the Graffle Example of the vector drawing library Raphaël.
November 19, 2013
Nowadays, developers often work in several different development environments and text editors. For example, you may need Eclipse for regular Java development, Android Studio for Android development, Xcode for iOS development, Visual Studio for C#, Sublime Text for web development, etc.. Every tool, however, has different keyboard shortcuts. Since it is hard to remember all shortcuts, there’s a constant loss of productivity when switching tools. If you don’t want to edit all shortcut sets in every tool, you can instead use CodingKeys as an “abstraction layer”.
CodingKeys lets you define unified shortcuts, which are dynamically re-mapped to existing shortcuts of other applications when you switch tools. All mappings can be conveniently edited in a single configuration file. The config file also gives you a nice overview over all shortcuts and grows as you add new apps to your coding toolbox.
November 30, 2012
DollarP-ObjC is an Objective-C port of the $P gesture recognizer to be used in iOS applications.
What is $P?
From the $P website:
The $P Point-Cloud Recognizer is a 2-D gesture recognizer designed for rapid prototyping of gesture-based user interfaces. In machine learning terms, $P is an instance-based nearest-neighbor classifier with a Euclidean scoring function, i.e., a geometric template matcher. $P is the latest in the dollar family of recognizers that includes $1 for unistrokes and $N for multistrokes. Although about half of $P’s code is from $1, unlike both $1 and $N, $P does not represent gestures as ordered series of points (i.e., strokes), but as unordered point-clouds. By representing gestures as point-clouds, $P can handle both unistrokes and multistrokes equivalently and without the combinatoric overhead of $N. When comparing two point-clouds, $P solves the classic assignment problem between two bipartite graphs using an approximation of the Hungarian algorithm. The $P recognizer is distributed under the New BSD License agreement.
May 24, 2012
How it works
TUIO is based on the OSC protocol and usually transferred via UDP. Tuio.js uses node.js and Websockets (Socket.IO) to push OSC/TUIO messages to the browser. TUIO.js converts the messages to events that applications can register with.
August 24, 2011
Recently, the facebook API has been voted »Worst API« in a developer survey. Bad news first: It really is the worst API (or at least the worst API that we have worked with). Good news: It would be quite straightforward for facebook to drastically improve the overall app development experience by working on at least one specific area (see end of this article).
March 9, 2011
Very good presentation by Dan North about »Deliberate Discovery: Code Like You Mean It«. »Ignorance is multivariate« and »We’re ignorant about our ignorance« – yes, based on past project experience, I guess I can confirm that! ;) He also mentions an interesting book called »A mind of its own: How Your Brain Distorts and Deceives«…
February 8, 2011