我应该如何表示有效搜索和比较字符串的数据

Question

I have two array with length 300. They look like this (JSON representation):

[
    [
        ["word1",0.000199],
        ["word2",0.000102],
          ...
        ["word15",0.000102]
    ],
      ...
    [
        ["anotherword1",0.0032199],
        ["anotherword2",0.032302],
          ...
        ["anotherword15",0.0320102]
    ]
]

And I have this bruteforce algorithm:

for(var i = 0; i < 300; i++)
    {
        for(var j = 0; j < 15; j++)
        {
            for(var ii = i + 1; ii < 300; ii++)
            {
                for(var jj = 0; jj < 15; jj++)
                {
                    for(var jjj = 0; jjj < 15; jjj++)
                    {
                        if(new_keywords[i][j][0] === new_keywords[ii][jj][0] && new_keywords[ii][jj][0] === state_keywords[i][jjj][0])
                        {
                            console.log(0);
                        }
                    }
                }
            }
        }
}

I need to search for same words in those arrays and if words are the same, then I sum values and divide sum by 3 and replace that value in state_keywords array. So for each word which is more then once in array I have means of its values.

Now... my approach is very bad because I have now about 300 mil iterations and that is crazy. I need some better implementation of my array in JavaScript. Something like lexikographical tree or kd-tree or something.

Thank you.

EDIT:

Here is http://jsfiddle.net/dD7yB/1/ with example.

EDIT2:

I'm sorry if I'm not clear enough. So what exaclty I'm doing:

• I have array state_keywords . Indexes are from 0 to 299 and they representing a themes ...
• Each theme may be represented by 15 words and every time new_keywords array arrives, they may be different.
• When new_keywords array arrive I need to check every word in that array if it is in state_keywords array on same theme index.
• If it is: add probabilities up and divide by 2.
• If it is not: add new word into state_keyword array BUT if they are more than 15 words for one theme (which now are) I need to store just first 15 sorted by probabilities.

And this I need to do as effectively as possbile, because I need to do this every second so it must be FAST.

EDIT3:

Now I use this code:

var i, j, jj, l;
for(i = 0; i < 300; i++)
{
    for(j = 0; j < 15; j++)
    {
        l = new_keywords[i].length;
        for(jj = 0; jj < l; jj++)
        {
            if(state_keywords[i][j][0] === new_keywords[i][jj][0])
            {  
                state_keywords[i][j][1] = (state_keywords[i][j][1] + new_keywords[i][jj][1]) / 2;
            }
        }
    }
}

which is much faster then the previous one.

Answer 1

Why don't you make those arrays into objects with the strings as keys to the values? Then you can just just look up the words directly and get the values?

var wordlists = [
    {
        "word1":0.000199,
        "word2":0.000102,
          ...
        "word15":0.000102
    },
      ...
    {
        "anotherword1":0.0032199,
        "anotherword2":0.032302,
          ...
        "anotherword15":0.0320102
    }
]

and then lookup with

wordlists[0]["word2"]  //0.000102