Twenital: Determines gender from a twitter handle

Twitter never stores the gender of their users. It makes the life of analytic engines very difficult to find that what is the ratio of men vs women on twitter.
Here I have created quick and dirty mash-up to get  gender of the twitter user.

How It Works?
Thanks to face.com API, they have recently released face recognition API. I am simply using their face recognition API to do this.

Here is AJAX code

function getusergender(imageurl) {
var url = 'http://api.face.com/faces/detect.json?api_key=1ce92a4c5e60f04c37bedf86c2d19387&urls=' + imageurl + '&callback=?';
$.getJSON(url, function (data) {
if (data.status == "success") {
if (data.photos[0].tags.length == 0) {
$("#resdiv").html('<div class="errormsg">hmmm..Let him/her put better profile picture. </b>TIP: put high resolution profile picture, facing front.</div>');
}
else if (data.photos[0].tags.length > 1) {
showerror('This user is multifaceted, Let him/her put better profile picture..');
}
else {
if (data.photos[0].tags[0].attributes.gender.value) {
var gender = data.photos[0].tags[0].attributes.gender.value;
if (data.photos[0].tags[0].attributes.gender.confidence > 40) {
if (gender == 'male') {
$("#resdiv").text('100% Male');
}
else {
$("#resdiv").text('Gorgeous Female');
}

}
else {
if (gender == 'male') {
$("#resdiv").text('Male');
}
else {
$("#resdiv").text('Female');
}
}
}
else {
showerror('hmmm... not sure!!');
}

}
}
else {
if (data.usage.used > 197) {
showerror('Sorry!! Application Overloaded!!');
}
else {
showerror('Processing failed!! Can not determine gender!!');
}
}
});

}

   DEMO  

Offline Google Maps in C#.net

Sometime back I wrote an application for windows mobile in c#.net which can give you turn-by-turn navigation on phone completely offline. Map and direction info will be pulled by a peer Desktop application which will export all these information in an XML file. This XML file will have these 2 Items

1. Static Map Images (covering the track at zoomed & top level))
2. Direction steps and description.

Static map images can be downloaded from Google using there static map API. Direction steps and descriptions can only be downloaded via a AJAX API. Since I had only AJAX way of getting this data, I created a webpage which can do all necessary DOM operation using JavaScript.  Once the DOM is populated I can easily grab the data from C# API (Web browser control + GetElementByID APIs).

Downloading the static Images from Google is not enough, you need to map Latitude and Longitude on the static map. Which is done like this

public PixelPoint GetCordinateOnStaticImage(double latitude, double longitude, 
double centerLat, 
double centerLang, 
int zoom,
double width,
double height)
{

       long val = 1 << ((21 - zoom));
       double target_y = LatToY(latitude); 
       double target_x = LongToX(longitude);
       double delta_x = (((target_x - LongToX(centerLang))) / (val));
       double delta_y  =(((target_y - LatToY(centerLat)) )/ (val));
       double marker_x = (width/2) + delta_x;
       double marker_y = (height/2) + delta_y;
       PixelPoint p = new PixelPoint(marker_x,marker_y);
       return p;
}

You can see Full source code here. Originally Idea is taken from here.


Direction
Google returns the poly-line of the direction path. Some of the points on this polyline will be identified as step, step has direction description, e.g "at blah blah circle take left".
This project solve another problem of pulling the right direction description from the given location.

SourceCode
Complete source-code of this project is available on Google code here.

Demo
Here is a demo video.

Create quick and dirty YQL Mash-up in 10 seconds (using MashupBuilder)

Sometime, you might have experienced that  you have to wait till whole website is downloaded just to see a stock-quote, game score, or league points table(etc.). Mostly these kind of websites are poorly managed, and they are very slow or contains too many annoying ads. I have created a Mash-up-builder that will help you in creating YQL based mash-up, very quickly. These mash-ups will be capable of showing just a small portion of the original web page. Mash-up will be dynamic and content will be pulled from original page every-time you will visit the mash-up. Additionally mash-up will be hosted on my website, so you need not to bother about hosting, though you can copy HTML code anytime if you want to host on your own.(these mashups are pure JavaScript, no server side coding).

Advantage is getting a faster & neat web page.

It was very easy to create mash-up-builder. I used YQL's html table.

How it works?
As you see that YQL's HTML table gives you output if you provide the URL and xpath of the content. Similarly Mash-up builder take 2 parameters as input, a URL of the web page and an XPATH of the content. Once these 2 things are ready, I cleverly used JSONP to render the content on page.

I recommend you to see the source {view-source} to understand the working of the mash-up.


Warning
Since mash-up builder injects HTML content directly from external source, its  prone to CrossSite Scripting attack. My website is not hosting any private content , so to me its fine because no security can be breached, do not try to replicate this functionality on your website if your website is hosting some private content.

  LIVE DEMO  

Feel free to share on twitter, Facebook & other social networking sites. 

IPL Tables made sortable

Ever since I visited the official IPL website, I hated those static points table.

So I used YQL to grab that data and place in very sortable fashion here.

To create this I am using,
YQL 
Google visualization API  (optionally one can use YUI)
Jquery tabs

How to write your own spell checker ? C#.net

Hi There,
Today I wrote a readable & usable implementation of Norvig's Spell corrector.
Norvi's implementation of spell check is awesome! On his site he has link to many other implementation in various languages, including C#.

Most of the implementation are focused on minimum number of lines , so they are not readable. I have written my own implementation which you can use on your own risk.

Here is the source code

You can also check-out entire code using this svn command

            svn checkout https://jugad.googlecode.com/svn/trunk/C%23/SpellCheck SpellCheck

How Algorithm works?



1.Initially Algorithms builds a dictionary* data-structure of words, from a huge English text. (so algorithm does not uses dictionary# file as such). This dictionary* contains words-to-frequency-of-occurrence mapping. lets say it as nWords.

2.When you will fetch a word(lets say 'w') for spell correction,Then the algorithm builds a list of all possible words that can be formed from w by doing following operation

1. Deletion (delete 1 character, for all possible combination)
2. Transposition (flip 1 character, for all possible combination)
3. Alteration (modify 1 character, for all possible combination)
4. Insertion (insert 1 character, for all possible combination)

Lets say name of this list is 'edits' , note that all words in this list are just 1 step away from original word.
3. If edits has 1 or more words that exists in nWords, then it returns the word from nWords with maximum frequency, as a result.
4. If No words in 'edits' are present in nWords, then it creates another edits using edits to get second level word (word which are 2 step away from original input word)
5.Repeats from step-3 (or give up if your are already 2/3 steps apart from original word 'w')
Read more Here.

How To Use This code ?
To use this code instantiate class spell by passing huge-string of English text (from where it will build dictionary)
Then call function correct() to get the corrected word as result.

Note:This code is just to demonstrate the algorithm, modify constructor to build dictionary directly from file. You can also avoid calling sort operation in function 'correct'. Thoroughly understand the code before using it. Notify me if you find any critical problem.


*dictionary data-structure of programming
#real dictionary