Thursday, December 13, 2012

Stephan Karpischek defended his thesis on mobile barcode scanning and quality of product master data

Stephan Karpischek is, after Irena Pletikosa, the second the PhD student this week and my fifth in total successfully defending his thesis. Congratulations, Dr. Stephan Karpischek!
As his co-referee Sanjay Sarma from MIT could not attend in person, Stephan defended his thesis in a sophisticated conference room at ETH where we had high-quality connection to MIT. The quality of video and sound was amazing and prove that some travels can be indeed eliminated by these technologies.
Stephan outlined four contributions of his thesis: the implementation of the open-source mobile barcode scanning app my2cents.mobi [1], the proof of incorrect master data yielding lower usage of mobile apps [2], the introduction of a
method to measure correctness of product names and its application to public master data sources [3]. Additionally, Stephan also did
some remarkable analysis of user behavior of mobile bargaining [4] and developed one of the most renown Swiss apps [5] , Swisspeaks. However, both results did not really fit any more in his thesis "Mobile Barcode Scanning Applications for Consumers" (will be available in a couple of months here).
 In the Q and A section of his defense Stephan prove that roughly 2% of today's master data available in online data pools amazon, google, afili.net etc. are wrong, dillute the brand image and could yield financial losses. This result provides a fact-based justification of the GS1's emerging Trusted Source service [6] which aims at providing brand-owner approved master-data. He also gave a perspective of a future master-data monitoring service that brand-owners could use to measure their appearance and abuse of their barcodes by other parties on the network...
Stephan, it has been a great pleasure to working with you and I really enjoyed the inspiring discussion with you. I wish you good luck in your new job!


[1] S. Karpischek, F. Michahelles, E. Fleisch, my2cents – enabling research on consumer-product interaction. Journal of Personal and Ubiquitous Computing. Springer, 2011, [doi:10.1007/s00779-011-0426-9].
[2] S. Karpischek, F. Michahelles, my2cents - Digitizing consumer opinions and comments about retail products, Internet of Things 2010 Conference (IoT2010), 29 November - 1 December 2010, Tokyo, Japan.
[3] S. Karpischek, F. Michahelles, E. Fleisch, The Not So Unique Global Trade Identification Number - Product master data quality in publicly available sources (Extended Abstract). 14th International Conference on Electronic Commerce (ICEC'12), 7 - 8 August 2012, Singapore.
[4] S. Karpischek, D. Santani, F. Michahelles, Usage Analysis of a Mobile Bargain Finder Application. 13th International Conference on Electronic Commerce and Web Technologies (EC-Web 2012), 3 - 7 September 2012, Vienna, Austria.
[5] S. Karpischek, C. Marforio, M. Godenzi, S. Heuel, F. Michahelles, SwissPeaks - Mobile augmented reality to identify mountains, Let's Go Out: Research in Outdoor Mixed and Augmented Reality.  Workshop at the International Symposium on Mixed and Augmented Reality 2009 (ISMAR 2009), 19 October 2009, Orlando, Florida, USA.
[6] GS1 Trusted Source of Data, Project Report, GS1, July 2012 [pdf]

Tuesday, December 11, 2012

Dr. Irena Pletikosa Cvijikj, well done!

Irena Pletikosa Cvijikj has defended her PhD thesis on "Evaluation Framework for Social Media Brand Presence" (the thesis will online here in a few months from now). Irena started in 2009 with her first project about using Facebook to connect things [1]. She really got fascinated by social media and started publishing quite an impressive series of papers investigating the content users share on Facebook brandpages [2],  emotions and motivation of users to engage on such pages, effectiveness of moderator activities to increase number of fans [3,4,5], detection of trending topics on Facebook [6],  and an evaluation framework [7] summarizing her findings as best practices for practitioners (two more are under review). It is a bit ironic that Facebook has been down since her defense, I'm glad she didn't plan for a demo;)


It was again fascinating to see how single papers which appeared to be rather diverse and unrelated in the beginning, could be put together into a comprehensive and coherent thesis deriving both lessons for literature and practitioners of social media. After Mikko Lehtonen, Felix von Reischach and Erica Dubach, Irena is the fourth student receiving a PhD from our lab (the next candidate is due in a couple of days!).

I'm glad Irena will stay in our lab and start forming her own group on social media-related research within the next months.

Congralutions to your great work! Good luck, and thank you very much for the opportunity to working with you, Irena!

[1] Irena Pletikosa Cvijikj, Goekmen Cetin, Stephan Karpischek, Florian Michahelles: Influence of Facebook on Purchase Decision Making, What can the Internet of Things do for the Citizen (CIoT) Workshop at The Eighth International Conference on Pervasive Computing (Pervasive 2010), Helsinki, Finland, May 2010 [PDF].
[2] Irena Pletikosa Cvijikj, Florian Michahelles: Understanding Social Media Marketing: A Case Study on Topics, Categories and Sentiment on a Facebook Brand Page, In Proceedings of the 15th International Academic MindTrek Conference: Envisioning Future Media Environments (MindTrek '11). ACM, New York, NY, USA, 175-182. DOI: 10.1145/2181037.2181066. URL: http://doi.acm.org/10.1145/2181037.2181066. [PDF]
[3] A Case Study of the Effects of Moderator Posts within a Facebook Brand Page, Irena Pletikosa Cvijikj, Florian Michahelles, In: Datta, Anwitaman, Shulman, Stuart, Zheng, Baihua, Lin, Shou-De, Sun, Aixin, Lim, Ee-Peng (eds.) Social Informatics. Lecture Notes in Computer Science, Springer Berlin / Heidelberg, 2011, Volume 6984/2011, 161-170, DOI: 10.1007/978-3-642-24704-0_21. URL: http://dx.doi.org/10.1007/978-3-642-24704-0_21 [PDF]
[4] Irena Pletikosa Cvijikj, Erica Dubach Spiegler, Florian Michahelles: The Effect of Post Type, Category and Posting Day on User Interaction Level on Facebook, Privacy, Security, Risk and Trust (PASSAT '11), 2011 IEEE Third International Conference on and 2011 IEEE Third International Conference on Social Computing (SocialCom '11) , pp.810-813, 9-11 Oct. 2011. DOI: 10.1109/PASSAT/SocialCom.2011.21. URL: http://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=6113221&isnumber=6113084 [PDF]
[5] Irena Pletikosa Cvijikj: Usability Testing of a Facebook Brand Page, World Usability Day 2011 Slovenia, Kranj, Slovenia, 10th November 2011. [PDF]
[6] Irena Pletikosa Cvijikj, Florian Michahelles: Monitoring Trends on Facebook,Dependable, Autonomic and Secure Computing (DASC), 2011 IEEE Ninth International Conference on, pp.895-902, 12-14 Dec. 2011. DOI: 10.1109/DASC.2011.150. URL: http://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=6118886&isnumber=6118342 [PDF]
[7] Irena Pletikosa Cvijikj, Florian Michahelles: Understanding the User Generated Content and Interactions on a Facebook Brand Page, , International Journal of Social and Humanistic Computing (IJSHC) Special Issue "Contemporary Social Media Topics in Business and Public Organizations", Inderscience Publishers, 2012. [PDF]

Mobile and Ubiquitous Multimedia Conference 2012 in Ulm

Enrico Rukzio did a fabulous job in organizing MUM2012  in Ulm this year. Together with the PC-Co chairs Kaisa Väänänen-Vainio-Mattila and Mark Billinghurst I could coordinate the collection of 389 reviews from 69 PC members for the 126 paper submissions we received from 359 authors out of 32 countries.
Dieter Schmalstieg opened the conference with his keynote Augmented Reality Technology for Smartphones. He nicely showed how this technology has evolved from expensive, bulky and immature platforms towards everyday AR in today's mobile phones. Ironically, though, all the AR systems of today (LayAR, Wikitude, Swisspeaks etc.) are simply based on compass and acceleration without taking any developments about visual and natural markers into account. Dieter showed some very impressive videos how AR experience can be improved significantly by stabilizing the digital overlays using markers extracted by image processing. Futhermore, he showed systems which could also recognize 'free space', e.g. blue sky, for placing digital overlays there instead of occluding important information from the real-world. Indeed fascinating was the presentation of the recording of a skate-board stunt which was recorded life, separated from the background and replayed and introduced as an AR layer in another setting over and over again.
While I was absolutely amazed about the technical achievements of AR research I'm still struggling with recognizing meaningful applications of this technologies. In the subsequent session Dieter mentioned replacing today's habit googeling for keywords on the mobile by instant web searches based on context and situations the users is in and pointing to: AR could replace the text-field of Google search...

I believe AR is a fascinating research area and also a great tool. Combining the advancements of visual markers/sensors/localization should help to make AR a useful and intuitive experience for accessing digital services in mobile user contexts.
Throughout the three days there was interesting work presented on novel design, mobile interaction, social interaction, security, mobile augmented reality, tools, audio, field studies, and public displays (see program). As part of these sessions I was presenting our work object circles which proposes to use the Google+ Circle concept for visualizing and managing data streams of smart items as an intuitive metaphor for everyday users. The underlying principle is to use the established interface of social media to also manage and interact with items and devices [1]. In addition to that, as part of the industry track, I presented our evaluation of the Dacuda scanmouse where could prove the more effective use of this device for everyday scanning tasks compared to multi-function printers, personal scanners, and mobile phone image taking [2]. Instead of a prototype I could actually show the working product which clearly showed to me the power of demonstrations. Let's see how many visitors will buy such a mouse;)
Finally, Flavius Kehr presented our field study on evaluating capturing of driving behavior via smart-phone vs. built-in fixed unit [3]. His statistical analysis showed that the fixed unit has more accuracy, however, the measurements of the phone are still significant to provide feedback to driver about his driving behaviour. Thus, Flavius could conclude with the clear finding that mobile devices are well suited to elicit driving behavior. This opens the space for many other players, e.g. insurance companies, to provide driving feedback in addition to the built-in systems of car manufacturers.

Nigel Davies provided the second keynote of the conference, he showed his passion about public displays as part of his involved into the PD-net project. He started off with the general benefits of giving a keynote: it allows you to present results without being bound to a strict scientific methodology. Despite the ubiquitous emergence of displays in public areas all these devices still suffer major short-comings: people feel disturbed by aggressive ads, the systems are passive and don't allow for interaction. There are much smarter ways of using these expensive installations: build toolkits to allow interactivity, implement context sensitive information, support audit-trails and hand-over for strolling users across several displays. Nigel also showed how ideas and concepts may quickly become obsolete: the idea of setting bluetooth-friendly usernames appears to be ridiculous today but seemed to be completely reasonable a few years ago [4].
As scenarios for more interactive displays he proposed emergency search for missing children, emergency announcements, behavior changes (e.g. collect vouchers by by-passing) and local marketing. Public displays should be open for users he concluded. Challenged by the question of why display providers would open up their systems Nigel responded with the proposal of radically re-thinking the concept of how displays are being used today: instead of catching eyballs based on impressions they would be more integrated into users needs. Nigel drew the analogy of Google embedding ads into search results rather than catching eyeballs on the front page. Whether the ad companies are already ready for this now is hard to say, Nigel admitted, but it would be worth trying.Public displays (sometimes also referred to as digital signage) is a growing field, a research community with its own conference has already been formed: The International Symposium on Pervasive Displays, 4-5 June, 2013 - Mountain View, California.

[1] Florian Michahelles, Philip Probst: Object Circles: Modeling physical objects as social relationships, Proceedings of the 10th ACM SIGMOBILE Conference on Mobile and Ubiquitous Multimedia (MUM'12), Ulm, Germany, December 2012 [PDF].
[2] Matthias Wyss, Alexander Ilic, Florian Michahelles: The scanner at your finger tips - analysis of the effectiveness of the scan mouse device, Proceedings of the 10th ACM SIGMOBILE Conference on Mobile and Ubiquitous Multimedia (MUM'12), Ulm, Germany, December 2012 [PDF].
[3] Driving Behavior Analysis with Smartphones: Insights from a Controlled Field Study, Johannes Paefgen, Flavius Kehr, Yudan Zhai, Florian Michahelles, Proceedings of the 10th ACM SIGMOBILE Conference on Mobile and Ubiquitous Multimedia (MUM'12), Ulm, Germany, December 2012 [PDF].
[4] Nigel Davies, Adrian Friday, Peter Newman, Sarah Rutlidge, and Oliver Storz. 2009. Using bluetooth device names to support interaction in smart environments. In Proceedings of the 7th international conference on Mobile systems, applications, and services (MobiSys '09). ACM, New York, NY, USA, 151-164.

Monday, December 3, 2012

Panel Discussion: "How do users find my app?"



Thanks to Lars Erik Holmquist from Yahoo Labs I had the opportunity to host a panel on “The curse of choice – how do users find your apps” at Siggraph Asia 2012. Together with my follow colleagues from previous Research in the Large workshop Henriette Cramer, FrankBentley and Niels Henze, and with Jaden Choi from TGrape we had an interesting discussion on apps. 

Nan Zhong, master student at ETH Zurich, opened the panel with a general introduction to the head and longtail nature of markets (slides). Head markets follow the so-called Pareto-principle which builds upon the fact that shelf space is usually limited, such that the merchant has to choose wisely which products to put on displays. The general rule is that only 20% out of all products generated a 80% of total revenue, thus niche products have a hard chance to appear on the shelf. Opposite to that are longtail markets which generated most of the revenue out of a huge variety of less popular niche products. However, this is only possible in digital markets were shelf-space is virtually unlimited and transaction costs of managing all these products can be kept low. Nan investigated [1] the nature of the Google Play market based on data-set of Appaware and – surprisingly – found: in contrast to most other digital markets (Netflix, amazon…), Google play is a clear Head market with even stronger characteristics than traditional retail. Thus, Nan’s conclusion was that developers should clearly focus on developing blockbuster. Furthermore, he found that the average capacity of users to install apps is limited by average to 40 apps. This means an app has to be really good in order to make it most user’s phone. What’s more, only 28% of all users would download not more than one paid app. Thus, free apps are much more likely to be successful. Furthermore, successful apps are rated better in general, whereas niche apps are used by expert users who are more critical give lower scores. Nan closed his talk with the suggestion to develop mechanisms that allow niche to be found more easily, as current collaborative filtering techniques would rather push successful apps even further.
Inspired by these insights we discussed various strategies of how developers could foster that their apps get found. Niels proposed to focus on search engine optimization by trying to understand how Google would rank apps. Thus, building up a history of a developer account releasing several successful apps overtime might help as well as tweaking the description of apps. Frank reported about his experience [2] of professionally marketing apps as any other product by approaching influential blogs launch posts that could trigger the demand for the app. Henriette proposed [4] to build upon the sharing friends by making apps social and allowing to be propagated by digital word of mouth.
We continued with the question of the price and Jaden Choi disclosed that he sees price rather following quality of the app. Price wouldn’t really matter so much, is rather a decision between paid and free app. Nan confirmed this finding saying that the 99cent price would be rather a relict of itunes than based on rational thinking. Also his research didn’t really reveal much experimentation of developers with adapting the price of apps. Jaden added that especially in Korea in-app payment for virtual gadgets would be strong opportunity and even more successful than having a paid app.
Niels disclosed his experience of releasing apps on Sunday [3] evening as the most effective measure getting users’ attention. Based on our experience with AppAware I mentioned that creating localized clones yourself from or own could be another approach to be found more easily in various languages and to get better positions in local app markets.
The fundamental question of developing interesting and new apps in the plethora of existing apps Frank answered by reverting on established methods of ethnography and experimentation. One approach would be to go out and observe users in order to get a glance what they might need. The next step then would be to “fail early”, meaning to release in short cycles and to learn from user comments to improve quickly.
Another interesting insight to me was that more expensive devices might also allow for more expensive apps. Finally, we had some controversy discussion about whether apps would prevail or rather be replaced by HTML5 web apps. The arguments by Frank (pro apps) and Niels (pro web-apps) where along the lines of barrier to install apps and technical flexibility of accessing hardware functions on the phone.
Overall, the panel together with about 50 visitors seemed to have been a good add-on at the Siggraph Asia conference. The App Symposium will be continued as a SiggraphMobile in 2013.
Additionally, Siggraph Asia was featuring an Emerging Technologies section. Different demos could be tried out proving new developments or also art and entertainment projects. My takes were controlling and seeing through a drone by a head-mounted display, “tasting” a website's ingredients (links, pics, bandwidth) as mixed liquor, an interface for controlling dancing robots, a force-sensitive display, an actuator game, and breath-awareness through a baloon.
Additionally, the Siggraph exhibition
featured a humanoid robot demo where a user's face gets project onto a mannequin representing a proxy of the user in meeting. <

Apart from that, Singapore is a fascinating place, an island of wealth, luxury, and strange rules in the heart of Asia. I really enjoyed the visit of the conference in Singapore. 

[1] Where should You Focus: Long Tail or Superstar?, An analysis of app adoption on the Android Market, Nan Zhong, Florian Michahelles. Symposium on Apps at Siggraph Asia, Singapore, November, 2012. [pdf] [slides]
[2] Frank Bentley, Santosh Basapur:   StoryPlace.Me: the path from studying elder communication to a public location-based video service, CHI EA '12, CHI '12 Extended Abstracts on Human Factors in Computing Systems, 2012 
[3] Niels Henze, Susanne Boll: Release Your App on Sunday Eve: Finding the Best Time to Deploy Apps (poster), Adjunct proceedings of MobileHCI, 2011
[4] Henriette Cramer, Mattias Rost, Lars Erik Holmquist (2011) Performing a Check-in: Emerging Practices, Norms and ‘Conflicts’ in Location-Sharing Using Foursquare. Proc. MobileHCI’11, Stockholm, Sweden. pdf

NFC Congress 2012

 It was the third time I attended the NFC Congress at FH Hagenberg. I was happy to meet so many people still believing in the opportunities of this technology even though actual deployment is still progressing slowly. I attended the congress one day before the announcement of Apple’s iphone, but there seemed to be pretty much consensus that Apple will not be a driver of this technology either.
 It was good to see how Nokia’s development of NFC finally got integrated into Microsoft’s Proximity API of Windows 8 which markets NFC as user experience instead of technical modes and parameters. However, the importance of windows mobile among Android and iOS still remains speculation. Payment is still seen as the main driver – or currently the main show stopper – of NFC. Several talks pointed out the roadmap for NFC as starting with wireless cards being rolled out mainly in eastern and southern Europe, next would be NFC stickers on phones, and finally the full integrated into the phone.
There the challenge will be to group service around payment such as coupons, loyalty, promotion, transport in order to incentivize consumers to switch from established payment methods – in Germany mainly cash – to mobile payments.
Finally, NFC was also proposed for micro-merchants as an alternative to transaction terminals, I enjoyed the split screen application for taxi-drivers presented by the Sparkassen Verlag Overall, the time horizon of greater adoption still remains unclear. Most participants, however, shared the believe that NFC would come sooner or later, mostly driven by the introduction of wireless cards which might be substituated by NFC later. Thus, I’m already looking forward to the NFC 2013 workshop which is technically co-sponsored by IEEE Switzerland Section and will take place in Zurich, Switzerland.

Technology Transfer in Poland


Poznan was my first visit of Poland. I was invited to talk about the technology transfer practice at ETH Zurich as part of the Opencode Transfer project which deals with the development of a new technology transfer process for Polish universities in greater Poznan region.
 I was presenting on the characteristics of ETH Zurich’s spin-offs such as longer survival rates, job creation, and value generation as the main motivation for fostering entrepreneurship at universities.

Throughout the following panel discussion I learned about significant differences between Poland, Switzerland and US. Whereas ETH transfer mainly functions as a service center supporting ETH employees to commercialize scientific results, in Poland university rather see transfer offices as profit centers which should generate surpluses out of licenses, patents and shares of spin-offs. Also other participants underlined that this approach might be hardly successful when looking at the numbers of US universities such as Stanford or Univ. of Illinois, where only 4% of submitted patents finally materialize in transferred licenses.
Overall, I was fascinated by the entrepreneurial spirit and the strong ambition of technology transfer from research to practice in Poznan.   

Tuesday, November 20, 2012

GS1 Technology Day: Nano, robots, humanoid reproduction

For the first time the leaders of the GS1 EPCglobal Board of Governors met at ETH Zurich for discussing future directions of RFID and barcode technologies. As part of this meeting, we organized a technology day which featured current develops of our labs associated companies and groups (find presentations and detailed agenda here).
In addition to our own work we could also invite renown keynote speakers from ETH Zurich.
Prof. Dr. Bertram Batlogg talked about recent developments of nano-technology and shared some insights about emerging display technologies.
As part of this program we had also the chance to meet Prof. Dr. Raffaello D’Andrea who motivated his research on controlling robots by sensors by the example of the steam engine controlled by a speed limiter.
Raffaello showed the tremendous developments regarding miniaturization when he compared their most recent quadrocopter with the table-sized model of a decade ago.
However, during his robot air-show the title of his talk "Controlling the artificial" become even clearer: Raefaello showed his trust into his technologies when he took a seat in his Flying Machine Arena and his students started the show, sometimes closely above his head.

In the first demo our visitors could control the helicopter with a stick by gestures. The system ensured that the helicopters did not collide with the walls or crash down because of too hectic movements. Other demos showed the ability of balancing a standing stick on the helicopter, coordinated movements of groups, kinect gesture control and the famous ball juggling (see here). Thus, this demonstration made me confident about getting the next doener delivered by drone pretty soon....
Finally, Prof. Dr. Markus Gross talked about how to visually reproduce the real-world based on special effects in several animation movies. He reported about the ability of 3D-scanning of still faces and reproducing movements with 7 artificial muscles in contrast to the natural 35 muscles. His prophecy was that in about a five year's time horizon simulated face-movements would be indistinguishable from the real. Thus, Holywood stars would already start protecting the rights on the visual appearance of their bodies, as movements can be done by the computers later rather easily and more cost-efficient.
In the second part of his talk Prof. Gross talked about humanoid robots that would mimic actors in amusements parks. Based on 3D scans and 3D printing faces can be modeled quickly. As a commercialization of this technology he showed customized barbie puppets carrying faces of the corresponding children. Technological progress sometimes can be fascinating and scary...



Friday, November 16, 2012

White collar crime


In my lecture management information systems Michael Schermann from Lehrstuhl für Wirtschaftsinformatik of Technische Universität München presented on the incentives of in-company crime and instruments for preventing it.
He started off with the traditional explanation of crime where criminals would simply balance between their crime incentive and the opportunity of being punished. Thus, high punishment would be enough to move the balance and ban crime...
However, from practice and from recent experiments we know this does not really hold. People rather have a moral account from where they withdraw and upload credits for their activities: if you buy green, you're more likely to steal...[1].
Furthermore, the simple cost/benefit analysis of fraud is also overwritten by an inner concept relating to fraud of users. It could be shown that fraud outside one's peer-group is less accepted than one's inside peer-group [2].
As example of fraud Michael listed corruption, higher billing of services/products, and non-cash fraud, e.g. manipulation of salaries - he said 15% of salaries would be adjusted higher than they should be. This shows that the traditional concept of incentive-punishment does not hold and is rather being replaced with an inner self-concept of values: if people park illegally, they are more prone to donate to beggar - in order to compensate their wrong behavior by adjusting their moral inner account.
Interestingly, most fraud is detected by hints of others. Thus, companies have started introducing whistle blower channels in order to collect these hints in a more systematic way. Michael proposed to translate business strategy into core values, which have to be lived up by top-management. There should be a clear list of don'ts to be avoided in companies. Control in companies can focus either on input (if output is hard to measure, e.g. as in research) or on input (e.g. sales).
The practical challenge, however, is to smartly balance between burden and effect. IT can help to better manage and control performance and coherence in order to be more responsive to risks, e.g. software can help to crunch down 126k of suspect emails down to a 1,2k messages.
Michael's slides can be found here.
 
[1] Nina Mazar and Chen-Bo Zhong: Do Green Products Make Us Better People? Psychological Science February 2010 , first published on March 5, 2010 
[2] Gino, F., Ayal, S. & Ariely, D. (2009). Contagion and differentiation in unethical behavior: The effect of one bad apple on the barrel. Psychological Science, 20 (3), 393-398.
ariely: coke was gone, dollar where there -) distance to money increases crime