(via Marakana)

Privacy issues

1. I want to be in control on the data that I share
2. I am concerned about my privacy and therefore I do want to be in control of my data (or at least in as much control as possible)
3. I do not want to have anybody else making aggregations and connections concerning my profiles apart the ones that I explicitly create
4. I do not aggree that Facebook claims that everything I post/like/share on this social media platform has anything to do with my real life. Actually I am aware of my posts and likes and I was always posting keeping in mind that potentially everybody has access to this content. Facebook Timeline does not reflect my life in any sense. In order to prevent others from thinking so I at least have to be controlling all the activity going on on my timeline or delete my entire account. The latter was the better option for me.
5. Second I am a computer scientist who has some knowledge on machine learning, social media analytics and user profiling techniques and knows how to apply them. So I do know what you can do with data on Facebook and I don’t want to be subject of such action.

Friendship issues

1. 75% of my Facebook friends are not my “real” friends
2. With 75% of my Facebook friends I do not have contact
3. I can also keep in touch with my real friends without using Facebook

Focusing on more important things

1. I want to spend my time in a more productive manner rather than keeping track and correcting updates on facebook (probably made by others)
2. I do not want to use Facebook just as a tool of promoting myself (there are other social media platforms which better suite this use case such as linkedin or xing)
3. I want to spend more time maintaining my blog and less time feeding social networks with the latest gossip news
4. I am not a Social Media Expert
5. I do not want to read only filtered information that Facebook thinks is important to me (well I am aware that Google does the same but at least there is some kind of excape to search for information on other sources
6. …

This are some of my reasons for deleting my Facebook account. What do you think about this?

Die “Simple Flashlight” App für Android ist nun in der Version 3.1 im Market verfügbar. Simple Flashlight ist eine einfache Taschenlampe für dein Mobiltelefon oder deinen Tablet. Es wurde die transparente Displaybeleuchtung entfernt. Zudem gibt es jetzt die Möglichkeit, durch “Wischen” nach rechts bzw. links die Farbe des Bildschirms auf rot oder blau zu ändern.

Recent Changes (v3.1)

• Displaybeleuchtung verbessert (Transparenz entfernt)
• Swipe nach rechts => Bildschirm blau färben
• Swipe nach links => Bildschirm rot färben
• Swipe verbessert

Recent Changes (v3.0)

•  Möglichkeit die App zu starten indem man die Suchtaste gedrückt hält
•  Bildschirmbeleuchtung ist nun transparent
•  Swipe-up und Swipe-Down zum dimmen der Bildschirmhelligkeit

Downloaden und ausprobieren! https://market.android.com/details?id=at.muehlburger.android.simple.flashlight

Simple Flashlight for Android

Die “Simple Flashlight” App für Android ist nun in der Version 3.0 im Market verfügbar. Simple Flashlight ist eine einfache Taschenlampe für dein Mobiltelefon oder deinen Tablet.

Recent Changes (v3.0)

•  Möglichkeit die App zu starten indem man die Suchtaste gedrückt hält
•  Bildschirmbeleuchtung ist nun transparent
•  Swipe-up und Swipe-Down zum dimmen der Bildschirmhelligkeit

Simple Flashlight im Android Market

Downloaden und ausprobieren: https://market.android.com/details?id=at.muehlburger.android.simple.flashlight

Machine Learning has always been a topic that attracted my interest. Currently I attend the “Machine Learning” class (@ml_class) offered by Prof. Andrew Ng from Stanford University. The class is awesome. The video lectures are made of small chunks explaining every topic in detail. A lot of topics were already covered:

• Linear regression with one variable
• Linear regression with multiple variable
• One-vs-all Classification
• Regularization
• Backpropagation Algorithm
• Neural Networks
• Practical advise for applying learning algorithms
• How to develop and debug learning algorithms
• Feature and model design, setting up experiments

Other interesting topics are following. In parallel to the lectures there are homework programming exercises that have to be solved. To date programming assignments covering topics as:

• Linear regression
• Logistic regression
• Multi-class classification and Neural Networks
• Neural network learning
• Regularized linear regression and bias-variance

In order to solve the exercises you have to understand the contents and have some programming experience in GNU Octave. It is also important to have basic understanding of Linear Algebra.

Machine Learning Contests

In order to apply machine learning on real world problems you can enter a machine learning contest. There are various contests out there, some of which are

• Tuned IT
• Kaggle

I will keep you updated.

Related articles

• Apache Mahout: Scalable machine learning for everyone (lucidimagination.com)
• How to enter a data contest – machine learning for newbies like me (petewarden.typepad.com)
• AI and the Labor Market (machinedlearnings.com)
• Applications of Neural Networks (shafiqissani.wordpress.com)

Ich habe mir einmal das Datenformat angesehen, welches die Website spritpreisrechner.at für die Abfrage der Spritdaten benutzt. Es ist eigentlich ganz einfach:

Es werden per HTTP POST Request Anfragen an http://spritpreisrechner.at/espritmap-app/GasStationServlet geschickt. Wichtig ist dabei der Parameter “data” der natürlich mit Werten (Längen- und Breitengraden bzw. einen Suchstring) initialisiert werden muss.

Beispielsweise könnte man folgenden cURL Aufruf ausführen (die Daten in “data” sind url-encoded”):

curl -d "data=%5B%22Graz%22%2C%22DIE%22%2C15.414262484145%2C47.057105360725%2C15.494085024429%2C47.08048974931%5D" http://spritpreisrechner.at/espritmap-app/GasStationServlet

Der HTTP Request Header sieht konkret folgendermaßen aus:

Request URL:http://spritpreisrechner.at/espritmap-app/GasStationServlet
Request Method:POST
Status Code:200 OK
Request Headersview source
Accept:*/*
Accept-Charset:ISO-8859-1,utf-8;q=0.7,*;q=0.3
Accept-Encoding:gzip,deflate,sdch
Accept-Language:de-DE,de;q=0.8,en-US;q=0.6,en;q=0.4
Connection:keep-alive
Content-Length:101
Content-Type:application/x-www-form-urlencoded
Host:spritpreisrechner.at
Origin:http://spritpreisrechner.at
Referer:http://spritpreisrechner.at/ts/map.jsp
User-Agent:Mozilla/5.0 (Macintosh; Intel Mac OS X 10_7_1) AppleWebKit/535.1 (KHTML, like Gecko) Chrome/13.0.782.215 Safari/535.1
X-Requested-With:XMLHttpRequest
Form Dataview URL encoded
data:["","DIE",15.409674251128,47.051201316374,15.489496791403,47.074588294516]
Response Headersview source
Connection:Keep-Alive
Content-Type:text/html;charset=UTF-8
Date:Fri, 26 Aug 2011 08:27:11 GMT
Keep-Alive:timeout=4, max=150
Transfer-Encoding:chunked

Der Parameter “data” enthält einen JSON String der mit folgenden Werten befüllt wird:

data:[
	"",	            => Ort (Tankstellen werden in der Nähe dieses Ortes gesucht) Update: Vielen Dank an Andreas Stütz!: Dieser Parameter gibt an, ob auch geschlossene Tankstellen angezeigt werden sollen ("checked") oder nicht ("")
	"DIE",		    => Treibstoffart: "DIE" oder "SUP"  (Diesel oder Super 95)
	15.414258888907,    => Longitude 1 (Längengrad)
	47.057106335865,    => Latitude 1 (Breitengrad)
	15.494081429191,    => Longitude 2 (Längengrad)
	47.080490724022     => Latitude 2 (Breitengrad)
]

Longitude 1 und Latitude 1 geben Koordinaten der linken unteren Ecke, Longitude 2 und Latitude 2 die Koordinaten der oberen rechten Ecke an. Das so aufgespannte Rechteck definiert den sichtbaren Bereich auf der Google Maps Karte. Im Zentrum sollte sich sinnvollerweise die aktuelle Position befinden. Innerhalb des sichtbaren Bereiches werden die 5 billigsten Tankstellen mit den dazugehörigen Spritpreisen angezeigt. Um das Rechteck aufzuspannen kann man sich in Android folgendermaßen behelfen:

1. Aktuelle Position mittels Android LocationManager ermitteln
2. Mittels der aktuellen Position und der Methoden getLongitudeSpan() und getLatitudeSpan() die rechte obere und die linke untere Ecke berechnen und anschließend
3. Parameter “data” mit den ermittelten Werten befüllen und HTTP Post Request abschicken.

Wird der oben beschriebene HTTP Request ausgeführt, bekommt man als HTTP Response einen JSON String, der wie folgt aufgebaut ist:

[
  {
    "kredit":true,
    "self":true,
    "spritPrice":[
      {
        "amount":"1.262",
        "datAnounce":"Thu Aug 25 17:45:55 CEST 2011",
        "gasStationId":"7766",
        "errorItems":[

        ],
        "errorCode":0,
        "datValid":1314287155000,
        "spritId":"DIE"
      }
    ],
    "automat":false,
    "city":"Graz",
    "open":true,
    "distance":1.32,
    "postalCode":"8010",
    "gasStationId":7766,
    "errorItems":[

    ],
    "priceSearchDisabled":false,
    "longitude":"15.4509774",
    "payMethod":"UTA",
    "mail":"",
    "gasStationName":"Turmöl",
    "fax":"433164751524",
    "clubCard":"",
    "openingHours":[
      {
        "beginn":"06:00",
        "day":{
          "dayLabel":"Feiertag",
          "order":8,
          "errorItems":[

          ],
          "errorCode":0,
          "day":"FE"
        },
        "end":"22:00"
      },
      {
        "beginn":"05:00",
        "day":{
          "dayLabel":"Dienstag",
          "order":2,
          "errorItems":[

          ],
          "errorCode":0,
          "day":"DI"
        },
        "end":"23:00"
      },
      {
        "beginn":"05:00",
        "day":{
          "dayLabel":"Montag",
          "order":1,
          "errorItems":[

          ],
          "errorCode":0,
          "day":"MO"
        },
        "end":"23:00"
      },
      {
        "beginn":"05:00",
        "day":{
          "dayLabel":"Freitag",
          "order":5,
          "errorItems":[

          ],
          "errorCode":0,
          "day":"FR"
        },
        "end":"23:00"
      },
      {
        "beginn":"06:00",
        "day":{
          "dayLabel":"Samstag",
          "order":6,
          "errorItems":[

          ],
          "errorCode":0,
          "day":"SA"
        },
        "end":"22:00"
      },
      {
        "beginn":"06:00",
        "day":{
          "dayLabel":"Sonntag",
          "order":7,
          "errorItems":[

          ],
          "errorCode":0,
          "day":"SO"
        },
        "end":"22:00"
      },
      {
        "beginn":"05:00",
        "day":{
          "dayLabel":"Mittwoch",
          "order":3,
          "errorItems":[

          ],
          "errorCode":0,
          "day":"MI"
        },
        "end":"23:00"
      },
      {
        "beginn":"05:00",
        "day":{
          "dayLabel":"Donnerstag",
          "order":4,
          "errorItems":[

          ],
          "errorCode":0,
          "day":"DO"
        },
        "end":"23:00"
      }
    ],
    "access":"",
    "url":"",
    "serviceText":"Shop\r\nBistro\r\nWaschanlage\r\nSB-Lanzenwäsche\r\nDiesel-Schnellläufer\r\n",
    "maestro":true,
    "companionship":false,
    "address":"Conrad.-v.Hötzendorfstr. 135 ",
    "club":false,
    "errorCode":1,
    "service":false,
    "latitude":"47.0510643",
    "bar":true,
    "telephone":"43316475152",
    "gasStationInternalId":"CvH"
  }
]

Einen Parser für diesen JSON String zu schreiben, dürfte nicht allzu schwer sein. Das war auch schon alles um eine App für spritpreisrechner.at entwicklen zu können. Eigentlich ziemlich einfach. Was meint ihr? Freu mich auf Kommentare und Anregungen!

Google I/O 2011: Keynote Day 2

Android Market for Developers

Designing and Implementing Android UIs for Phones and Tablets

Building Android Apps for Google TV

Android Development Tools

Taking Android to Work

Related articles

• Google TV Gets Android Market And Android 3.1 Honeycomb (slashgear.com)
• Google TV will get Android 3.1, apps from Android Market this summer (venturebeat.com)
• What’s Up With Google TV [Google] (gizmodo.com)
• A First Look at the New Google TV (fakeiitian.com)
• Google Introduces Revamped Android Market at I/O (slashgear.com)

Google IO is currently taking place in San Francisco and for everybody who could not attend Google live streamed quite some sessions. In the following you can see the videos which are currently published:

Google I/O 2011: Android Keynote Day One

As an Android developer I am very interested in the Android sessions as you can imagine. So the following the next video first covers advaced topics of android development.

Google I/O 2011: Android Protips: Advanced Topics for Expert Android App Developers

The next video discusses business models that you as a developer can use to monetize your app. So this surely is interesting for everybody of you, isn’t it?

Google I/O 2011: Don’t just build a mobile app. Build a business.

Ever wondered about what Near Field Communication on Android is implemented? The answer gives you the next video.

Google I/O 2011: How to NFC

The next video covers the highlights of the new update release of Android Honeycomb (Version 3.1).

Google I/O 2011: Honeycomb Highlights

The last video of the first day contains a discussion with the Android Team.

Google I/O 2011: Fireside Chat with the Android Team

More videos will follow here as soon as they get published, so stay tuned!

Related articles

• Live From Google IO [Google] (gizmodo.com)
• Android Honeycomb Goes to 3.1 (mashable.com)
• Google IO 2011 (news.netapex.org)
• Google IO Brings “Music Beta” (techeblog.com)
• Google Will Remove Any Illegal, Unauthorized, Copyright-Infringing Music From Its Cloud [Google] (gizmodo.com)

Eli Pariser explains in the following TED Talk how “human information filters” get substituted by algorithmic ones, which means how recommendation engines filter information for you. Have a look at the video is is really worth watching:

Do you know other examples of web personalization or recommendations engines? Please leave me a comment at the end of this post.

Related articles

• Who Decides What You Read? Hint: It’s Not You (bigthink.com)
• TED 2011: Junk Food Algorithms and the World They Feed Us (wired.com)
• Gatekeepers Make a Lot of Sense (pravdam.com)