I used a word list under MIT licence from here: https://github.com/jeremy-rifkin/Wordlist Each input is split by spaces (so I get words).
For each word, the scoring is: If the word is valid and new: +10 if it starts with the right letter, +1 for each additional occurence of a letter. If the word is valid but already used: -1 point If the word is invalid: -5 points (from the top of my head, exact numbers might be slightly different). To keep track of what words were guessed already, I put the words into a Set<string>.