private Dictionary<string, HashSet<int>> GetDefaultTrigramIndex(QuestionList questionList)
{
var getDataFunction = new Func<Tuple<string, HashSet<int>>[]>(
() => CalculateTrigramIndex().Select(pair => Tuple.Create(pair.Key, pair.Value)).ToArray());
return DataActualityChecker.Check
(
new Lazy<Tuple<string, HashSet<int>>[]>(getDataFunction),
t => t.Item1 + "\x2" + String.Join("\x2", t.Item2),
s =>
{
var q = s.Split('\x2');
return Tuple.Create(q[0], new HashSet<int>(q.Skip(1).Select(int.Parse)));
},
new FileDependencies(
String.Format("TrigramIndex_{0}.txt", questionList.GetHashCode()),
Program.QuestionsFileName,
Program.AnswersFileName)
)
.ToDictionary(item => item.Item1, item => item.Item2);
}
public static string[] GetFrequentWords(QuestionList questionList)
{
var getDataFunction = new Func<string[]>(
() =>
{
var statistics = new Statistics.Statistics(questionList);
return statistics.WordFrequencyDistribution(new EmptyStemmer())
.Where(item => item.Value > 10)
.Select(item => item.Key)
.ToArray();
});
return DataActualityChecker.Check
(
new Lazy<string[]>(getDataFunction),
t => t,
s => s,
new FileDependencies(String.Format("FrequentWords_{0}.txt", questionList.GetHashCode()),
Program.QuestionsFileName,
Program.AnswersFileName)
).ToArray();
}
