Better way to search an array

Programming Questions

ikillbombs

Better way to search an array

in Programming Questions • 3 years ago

hi,

i want to make an arraysearch faster. The problem is that I want the array to search itself and when a word resembles the same word in a different array, to save the positions.
Now i'm doing it like this:

[code]
import java.util.ArrayList;

XMLElement xml;

void draw() {
loadXML();
noLoop();
}

void loadXML() {

xml = new XMLElement(this, "aol.xml");
// XMLElement aolFile = xml.getChild(0);
int num_aol = xml.getChildCount();
int rows = 1;
String[][] myArray = new String[num_aol][rows];
String[][] myResult = new String[num_aol][rows];

for (int i=0; i<num_aol; i++) {
    XMLElement ClickURL = xml.getChild(i);
    XMLElement thro = ClickURL.getChild(0);

    //String Query = xml.getString(1);
    int AnonID = int(ClickURL.getChild(0).getContent());
    String Query = ClickURL.getChild(1).getContent();
    String QueryTime = ClickURL.getChild(2).getContent();
    String ItemRank = ClickURL.getChild(3).getContent();

    String[] splitQuery = splitTokens(Query);
    int stringAdd = splitQuery.length;
    myArray[i] = (String[]) expand(myArray[i], stringAdd);
    for (int j = 0; j < splitQuery.length; j++) {

      myArray[i][j] = splitQuery[j];
     // println(myArray[i][j]);
    }
}
// search part
for (int i = 0; i < num_aol; i++)
    for (int j = 0; j < myArray[i].length-1; j++)
      for (int l = 0; l < num_aol; l++)
        for (int k = 0; k < myArray[l].length-1; k++) {

      if ( myArray[l][k] == myArray[i][j] ){
        if(i == l){

        }
        else{
          println("value 1 = "+ myArray[i][j]+ " value 2 = " + myArray[l][k]);
          }
        }
        }
}
[/code]

Is there a faster way than this?

Replies(7)

kooogy

Re: Better way to search an array

3 years ago

yes, almost certainly. those quadruple nested loops are just wrong.

but you need to post your data so we can run the code the way you are.

ikillbombs

Re: Better way to search an array

3 years ago

XML file only

http://paste2.org/p/1014564

complete

http://rapidshare.com/files/422721828/AOL_XML.zip

kooogy

Re: Better way to search an array

3 years ago

are you sure you want to match every word in 1000+ sentences with every other word in the same 1000+ sentences - that's a lot of potential matches, especially with words like 'and' and 'http' - i keep getting Out Of memory exceptions. am working on it...

and you won't match anything comparing strings with =, you need to use .equals() as they are classes.

and generally, if you're comparing things with themselves you can do the following

for (x = 0 ; x < size ; x++) {
for (y = x + 1 ; y < size ; y++) {
// compare x and y
}
}

that way you are only comparing the Xs in the diagram below - matching(2,3) is the same as matching(3,2), no need to do both.

123
1.XX
2..X
3...

kooogy

Re: Re: Better way to search an array

3 years ago

perhaps a better solution would be a hashmap<string, arraylist> full of positions, keyed on word. then you'd have to

for every sentence {
for every word in the sentence {
    if word is not in hash {
      create new hash(word)
    }
    add position to hash(word)
}
}

then it's trivial to get a list of all words, a count of occurances and all their positions

ikillbombs

Re: Better way to search an array

3 years ago

hi, thanks for your help. It's definantly allot quicker! So it must be something like this? I

   for (int i = 0; i < num_aol; i++)
    for (int j = 0; j < myArray[i].length-1; j++) {
      hm.put(myArray[i][j], "s "+i+" w "+j);
    }

for (int i = 0; i < num_aol; i++)
    for (int j = 0; j < myArray[i].length-1; j++) {
      if( hm.containsKey(myArray[i][j]) ) {

        Object Result = hm.get(myArray[i][j]);
        String[] ResultVal = split(Result.toString(), ' ');
      println (myArray[int(ResultVal[1])][int(ResultVal[2])]);


      }

I'm trying to figure out how to store the data, but i cannot think of a good way. But how can i request the information correctly?

kooogy

Re: Re: Better way to search an array

3 years ago

beware that doing this:

hm.put(myArray[i][j], "s "+i+" w "+j);

will replace the previous references to the word with this new one. replace, not add. read up on hashmaps.

i was thinking more of a hashmap where the word was the key and the value was a List of occurrences. kinda fiddly. easier if you are using newer processing and use generics.

also, for the love of god, use {}s around all your blocks, even if the block is only one line long. SO much clearer and easier to read, especially if the forum code is going to chomp indenting or you mess up the pasting (there's a missing close } in that last block, two opens, one close)

ikillbombs

Re: Better way to search an array

3 years ago

I think this is the best way:

for (int i = 0; i < num_aol; i++)
    for (int j = 0; j < myArray[i].length-1; j++) {
      hm.put(myArray[i][j], "s "+i+" w "+j);
    }

for (int i = 0; i < num_aol; i++)
    for (int j = 0; j < myArray[i].length-1; j++) {
      if( hm.containsKey(myArray[i][j]) ) {

        Object Result = hm.get(myArray[i][j]);
        String[] ResultVal = split(Result.toString(), ' ');
      println (myArray[int(ResultVal[1])][int(ResultVal[2])]);


      }