Search code examples
javaperformancehashmapspelling

Comparing one data structure against another resulting in run time of over 50 mins


I'm writing code which reads in a text file (each line a tweet) and goes through each tweet comparing it against a list of English words to see if the word is misspelled.

So the list of English words is read in from a text file as well, this is then stored in a List. When I run the code for this alone, it operates in less than one second. When I run the code for storing each word in the tweet file (without checking for spelling) for the 1,000,000 tweets, it stores each word and its frequency in a HashMap<String, Integer> in around 20-30sec.

But when I add the line to check if the word is spelled correctly, it causes a ridiculous run time increase, to the point where I could almost watch a movie before it finished running.

The simple aspect of invoking isSpelledCorrectly(X) (which just invokes list.contains(x), which has a worst case run-time of O(n)), yet it seems quite confounding that it causes the code to go from a 30 sec runtime to a 50 min runtime?

Code:

Spelling:

static List<String> spellCheck = new ArrayList<String>();

public AssignTwo() throws IOException{
    spellCheck = initCorrectSpelling("C:\\Users\\Gregs\\InfoRetrieval\\src\\english-words");
}

public static List<String> initCorrectSpelling(String filename) throws IOException { //store correct spelling of words in list
    Scanner scanner = new Scanner(new FileInputStream(filename));
    try{
        while(scanner.hasNextLine()){
            String next = scanner.nextLine();
            spellCheck.add(next);
        }
    }
    finally{
        scanner.close();
    }
    return spellCheck;
}

public static boolean isSpelledCorrectly(String word){  //check if any given word is spelled correctly by seeing if it is 
    boolean output = false;                             //contained within the spellCheck list
    if(spellCheck.contains(word)) output = true;                    
    return output;
}

Code storing Tweets:

public static HashMap<String, Integer> misSpell;
public AssignOne() throws IOException {         //read in file from path, test functions
    index("C:\\Users\\Gregs\\InfoRetrieval\\src\\tweets");
}

public static void index(String filename)  throws IOException {
    misSpell = new HashMap<String, Integer>();
    Scanner scanner = new Scanner(new FileInputStream(filename));
    try{
        while(scanner.hasNextLine()){
            String line = scanner.nextLine();       
            String[] lineArr = line.split(" ");
            for(int i=3; i<lineArr.length; i++){
                int count=1;
                lineArr[i] = lineArr[i].replaceAll("[^a-zA-Z0-9]", "");
                //if(!AssignTwo.isSpelledCorrectly(lineArr[i].toLowerCase())){  //with this line commented out, runtime <30sec, with line >50mins
                    if(misSpell.containsKey(lineArr[i].toLowerCase())){             
                        count = 1 + misSpell.get(lineArr[i].toLowerCase());             
                    }
                    misSpell.put(lineArr[i].toLowerCase(), count);
                //}
            }
        }
    } 
    finally{
        scanner.close();
    }
}

Any suggestion on where to improve code or how to make the comparisons more efficient? Is there a faster data structure for correctly spelled words?


Solution

  • List.contains() is O(N), N being the number of words in the dictionary.

    Use a HashSet, where contains() is O(1).

    Using A buffered reader would also speed things up. And avoiding to call toLowerCase() three times on each word would too.