84 lines
2.5 KiB
Java
84 lines
2.5 KiB
Java
|
/**
|
||
|
|
* The WordNode class represents one or more consecutive words from the training data, and
|
||
|
|
* every one word that came after that word in the training data, as well as the number of
|
||
|
|
* times each word appeared.
|
||
|
|
*
|
||
|
|
* Example:
|
||
|
|
* The dog -
|
||
|
|
* eats, 5
|
||
|
|
* is, 2
|
||
|
|
* likes, 1
|
||
|
|
* chases, 3
|
||
|
|
*
|
||
|
|
* That example uses two words as the key. Any number of words can technically be used, but
|
||
|
|
* anything over two will likely result in it directly plagairizing the training data.
|
||
|
|
*/
|
||
|
|
|
||
|
|
import java.util.ArrayList;
|
||
|
|
|
||
|
|
public class WordNode {
|
||
|
|
private String key;
|
||
|
|
private ArrayList<String> wordValues;
|
||
|
|
private ArrayList<Integer> intValues;
|
||
|
|
|
||
|
|
/**
|
||
|
|
* Constructor method used to create a WordNode with an initial key value pair.
|
||
|
|
*/
|
||
|
|
public WordNode(String k, String v) {
|
||
|
|
key = k;
|
||
|
|
wordValues = new ArrayList<>();
|
||
|
|
intValues = new ArrayList<>();
|
||
|
|
wordValues.add(v);
|
||
|
|
intValues.add(1);
|
||
|
|
}
|
||
|
|
|
||
|
|
/**
|
||
|
|
* Method used to add a word that follows this WordNode.
|
||
|
|
*/
|
||
|
|
public void add(String str) {
|
||
|
|
str = str.trim();
|
||
|
|
|
||
|
|
// If str has already been found, increase the corresponding int by 1.
|
||
|
|
if(wordValues.contains(str)) {
|
||
|
|
int index = wordValues.indexOf(str);
|
||
|
|
intValues.set(index, intValues.get(index) + 1);
|
||
|
|
} else { // If str has not been found. Add it and a 1 to appropriate ArrayLists.
|
||
|
|
wordValues.add(str);
|
||
|
|
intValues.add(1);
|
||
|
|
}
|
||
|
|
}
|
||
|
|
|
||
|
|
/**
|
||
|
|
* Returns a word from wordValues, using the number of times each word
|
||
|
|
* appeared to determine the probability of returning that word.
|
||
|
|
*/
|
||
|
|
public String generateWord() {
|
||
|
|
|
||
|
|
/* Total represents the sum of every element in intValues, which is
|
||
|
|
the number of times the key appears in the training data. This
|
||
|
|
is used to determine the range for the random number generator. */
|
||
|
|
int total = 0;
|
||
|
|
for(int x : intValues) {
|
||
|
|
total += x;
|
||
|
|
}
|
||
|
|
|
||
|
|
// random number 0 through (total - 1)
|
||
|
|
int random = (int) (Math.random() * total);
|
||
|
|
|
||
|
|
int index = 0;
|
||
|
|
int count = intValues.get(index);
|
||
|
|
for(int i = 0; i < random; i++) {
|
||
|
|
count --;
|
||
|
|
if(count == 0) {
|
||
|
|
index++;
|
||
|
|
count = intValues.get(index);
|
||
|
|
}
|
||
|
|
}
|
||
|
|
|
||
|
|
return wordValues.get(index);
|
||
|
|
}
|
||
|
|
|
||
|
|
public String getKey() { return key; }
|
||
|
|
public ArrayList<String> getWordVals() { return wordValues; }
|
||
|
|
public ArrayList<Integer> getNumVals() { return intValues; }
|
||
|
|
}
|