public UserTopicsWalking(QuestionList questionList, ICollection<string> categories, TopicsStatistics topicStatistics)
{
this.topicStatistics = topicStatistics;
UserQuestions = questionList.GetAllQuestions().Where(q => categories.Contains(q.Category))
.GroupBy(q => q.AuthorEmail, (userEmail, hisQuestions) => new { userEmail, hisQuestions })
.Where(u => u.hisQuestions.Count() >= 2)
.ToDictionary(u => u.userEmail, u => u.hisQuestions);
foreach (var questions in UserQuestions.Values)
{
foreach (var question in questions)
{
var topic = topicStatistics.GetTopicByQuestionId(question.Id, Threshold);
if (topic == null) continue;
if (topicDistribution.ContainsKey(topic.Item1))
{
topicDistribution[topic.Item1]++;
}
else
{
topicDistribution[topic.Item1] = 1;
}
}
}
userQuestionsCount = topicDistribution.Sum(it => it.Value);
topicMoves = GetTopicSequenceFrequences(2);
}
public static void ModifyTyposCorpus(QuestionList ql)
{
var detector = new SpellChecker(TrigramIndex.CreateFrom(ql));
Console.WriteLine("I am Modifying");
var start = DateTime.Now;
foreach (var question in ql.GetAllQuestions())
{
question.Text = String.Join(" ", question.Text.SplitInWordsAndStripHTML().Select(detector.Fix));
question.Title = String.Join(" ", question.Title.SplitInWordsAndStripHTML().Select(detector.Fix));
}
Console.WriteLine("Questions modified in {0}", (DateTime.Now - start).TotalSeconds);
start = DateTime.Now;
foreach (var answer in ql.GetAllAnswers())
{
answer.Text = String.Join(" ", answer.Text.SplitInWordsAndStripHTML().Select(detector.Fix));
}
Console.WriteLine("Answers modified in {0}", (DateTime.Now - start).TotalSeconds);
File.WriteAllLines(Program.QuestionsNoTyposFileName, ql.GetAllQuestions().Select(Question.FormatStringWrite));
File.WriteAllLines(Program.AnswersNoTyposFileName, ql.GetAllAnswers().Select(Answer.FormatStringWrite));
}
public void TestId()
{
var ql = new QuestionList(QuestionsFileName, AnswersFileName);
var hasIdenticId = false;
foreach (var question in ql.GetAllQuestions())
{
foreach (var answer in ql.GetAllAnswers())
{
hasIdenticId = true;
if (answer.Id == question.Id)
Console.WriteLine("BAD ID!!!!!!!!! " + answer.Id);
}
//Console.WriteLine(question.Id);
}
Assert.AreEqual(true, hasIdenticId);
}
public UserStatistics(QuestionList questionList)
: base(questionList)
{
var parser = new MailUserPageParser(Program.MailUsersDirectory);
var questionUsers = questionList.GetAllQuestions().Select(q => q.AuthorEmail);
var answerUsers = questionList.GetAllAnswers().Select(a => a.AuthorEmail);
var questionListUsers = new HashSet<string>(questionUsers.Union(answerUsers));
users = parser.GetObjects().Where(u => questionListUsers.Contains(u.Email)).ToList();
TopicStatistics = new TopicsStatistics(questionList);
}
private static IEnumerable<string> GetAllVertices(QuestionList ql)
{
return ql.GetAllQuestions().Select(q => q.AuthorEmail).Concat(ql.GetAllAnswers().Select(q => q.AuthorEmail)).Distinct();
}
private static IEnumerable<Tuple<string, string, double>> GetAllEdges(QuestionList ql)
{
var edges = new Dictionary<Tuple<string, string>, double>();
foreach (var question in ql.GetAllQuestions())
{
foreach (var answer in question.GetAnswers())
{
var from = answer.AuthorEmail;
var to = question.AuthorEmail;
edges.UpdateOrAdd(Tuple.Create(from,to), v => v + 1d, 1d);
}
}
return edges.Select(q => Tuple.Create(q.Key.Item1,q.Key.Item2, q.Value));
}