CareerCup

Using Reservoir Sampling, you could basically read from the stream of search queries and add them with probablity = k/k+i, where k is 1 million and i denotes the ith hindi query after the kth hindi query. You are ignoring the non-hindi queries as you read from the stream. When you have read all the queries the k queries you have are randomly selected hindi queries.

How does one determine if a query is in Hindi? Assuming we know how to do that, lookup Reservoir Sampling.

The library function getLanguage(Query) is given

The library function getLanguage(Query) is given

Sachin, did you even lookup what Reservoir Sampling is?

Why should [only] Sachin look up what Reservoir Sampling is? Talk about compulsions :D

int i=0;
int k = 1 Million;
String[] hinQ = new String[K];
while(reader has not reached eof)
{
String querry = reader.NextLine();
if(getLanuage(querry) == 'Hindi')
{
i++;
if(i < K)hinQ[i]=querry;
else if(random(i)<K)
{
hinQ[random(K)]= querry;
}
}
}
// Also note that if memory space is not enough to hold 1 million querry, you can copy the same file to different location and limit your K and i (where to start and where to end except the last)