public int GetDistinctPunctuationsCount(FeatureParameters parameters)
{
int messageId = parameters.MessageId;
yaf_Message item = buhonlineDataProvider.ReadMessage(messageId);
return(item.Message.Where(char.IsPunctuation).Distinct().Count());
}
public bool IsFormWordsContained(FeatureParameters parameters)
{
int messageId = parameters.MessageId;
yaf_Message item = buhonlineDataProvider.ReadMessage(messageId);
return(IsWordsFromCollectionContained(item.Message, formWords));
}
public int ThreadLength(FeatureParameters parameters)
{
int messageId = parameters.MessageId;
yaf_Message item = buhonlineDataProvider.ReadMessage(messageId);
return(buhonlineDataProvider.ReadTopicMessages(item.TopicID).Count());
}
public int GetUserQuotesCount(FeatureParameters parameters)
{
int messageId = parameters.MessageId;
yaf_Message item = buhonlineDataProvider.ReadMessage(messageId);
return(FeatureRegex.QuoteExpression.Matches(item.Message).Count);
}
public bool IsSmileContained(FeatureParameters parameters)
{
int messageId = parameters.MessageId;
yaf_Message item = buhonlineDataProvider.ReadMessage(messageId);
return(smiles.Any(item.Message.Contains));
}
public bool IsTextHighlighted(FeatureParameters parameters)
{
int messageId = parameters.MessageId;
yaf_Message item = buhonlineDataProvider.ReadMessage(messageId);
return(FeatureRegex.HighlightBBTag.IsMatch(item.Message));
}
public int GetUpperLettersCount(FeatureParameters parameters)
{
int messageId = parameters.MessageId;
yaf_Message item = buhonlineDataProvider.ReadMessage(messageId);
return(item.Message.Count(char.IsUpper));
}
public int GetHtmlTagsCount(FeatureParameters parameters)
{
int messageId = parameters.MessageId;
yaf_Message item = buhonlineDataProvider.ReadMessage(messageId);
return(FeatureRegex.HtmlTags.Matches(item.Message).Count);
}
public int GetNumbersCount(FeatureParameters parameters)
{
int messageId = parameters.MessageId;
yaf_Message item = buhonlineDataProvider.ReadMessage(messageId);
return(FeatureRegex.Numbers.Split(item.Message).Length - 1);
}
public int GetQuestionMarksFeature(FeatureParameters parameters)
{
int messageId = parameters.MessageId;
yaf_Message item = buhonlineDataProvider.ReadMessage(messageId);
return(item.Message.Split('?').Length - 1);
}
public int GetPostLengthFeature(FeatureParameters parameters)
{
int messageId = parameters.MessageId;
yaf_Message item = buhonlineDataProvider.ReadMessage(messageId);
return(item.Message.Length);
}
public bool IsDateContained(FeatureParameters parameters)
{
int messageId = parameters.MessageId;
yaf_Message item = buhonlineDataProvider.ReadMessage(messageId);
return(FeatureRegex.Date.IsMatch(item.Message));
}
public int GetPoints(FeatureParameters parameters)
{
int messageId = parameters.MessageId;
yaf_Message item = buhonlineDataProvider.ReadMessage(messageId);
return(item.Points);
}
public bool IsEdited(FeatureParameters parameters)
{
int messageId = parameters.MessageId;
yaf_Message item = buhonlineDataProvider.ReadMessage(messageId);
return(item.Edited != null);
}
public bool IsThankwordsContained(FeatureParameters parameters)
{
int messageId = parameters.MessageId;
yaf_Message item = buhonlineDataProvider.ReadMessage(messageId);
HashSet <string> thankwords = Preprocessing.ThankWords;
return(FeatureRegex.Words.Split(item.Message.ToLower()).Any(thankwords.Contains));
}
public bool IsHelloContained(FeatureParameters parameters)
{
HashSet <string> helloWords = Preprocessing.HelloWords;
int messageId = parameters.MessageId;
yaf_Message item = buhonlineDataProvider.ReadMessage(messageId);
return(helloWords.Any(item.Message.ToLower().StartsWith));
}
public bool IsSameAuthorAsTopicStarter(FeatureParameters parameters)
{
int messageId = parameters.MessageId;
yaf_Message item = buhonlineDataProvider.ReadMessage(messageId);
int topicStarterAuthorID = GetTopicStarter(item.TopicID).UserID;
return(item.UserID == topicStarterAuthorID);
}
public double QuotientPostLengthToAvarageInTopic(FeatureParameters parameters)
{
int messageId = parameters.MessageId;
yaf_Message item = buhonlineDataProvider.ReadMessage(messageId);
IEnumerable <yaf_Message> messages = buhonlineDataProvider.ReadTopicMessages(item.TopicID);
return(messages.Average(m => m.Message.Length));
}
public double TimeFromTopicStarter(FeatureParameters parameters)
{
int messageId = parameters.MessageId;
yaf_Message item = buhonlineDataProvider.ReadMessage(messageId);
yaf_Message ts = buhonlineDataProvider.ReadTopicStarter(item.TopicID);
return((item.Posted - ts.Posted).TotalMinutes);
}
private static bool IsTopicStarterCited(yaf_Message item)
{
var topicStarter = GetTopicStarter(item);
return
(FeatureRegex.QuoteExpression.Matches(item.Message).Cast <Match>().Any(
m => m.Groups[1].Value.Trim().ToLower() == topicStarter.Message.Trim().ToLower()));
}
public void AddMessage(yaf_Message message)
{
if (messages.Count == keepMaxMessages)
{
messages = new Dictionary <int, yaf_Message>();
}
messages.Add(message.MessageID, message);
}
private static bool IsSmileContained(yaf_Message item)
{
var smiles =
":) ;-) :( \\:d/ :-\" =d> o:) [-o< [-x #-o =p~ :-k =; :-$ :-# [-( :-s :-({|= 8-[ :-& :^o ](*,) @}-\\-- :-n"
.Split(' ');
return(smiles.Any(item.Message.Contains));
}
private string GetQuestion(int messageId)
{
yaf_Message item = buhonlineDataProvider.ReadMessage(messageId);
int topicId = item.TopicID;
yaf_Message topicStarter = buhonlineDataProvider.ReadTopicStarter(topicId);
return(topicStarter.Message);
}
public bool IsFormulaContained(FeatureParameters parameters)
{
int messageId = parameters.MessageId;
yaf_Message item = buhonlineDataProvider.ReadMessage(messageId);
string[] parts = item.Message.Split(new[] { ']', '?' });
return(parts.Any(FeatureRegex.Formula.IsMatch));
}
public int GetMessageCitedCount(FeatureParameters parameters)
{
int messageId = parameters.MessageId;
yaf_Message item = buhonlineDataProvider.ReadMessage(messageId);
return(Preprocessing.MessageCitedCounts.ContainsKey(item.MessageID)
? Preprocessing.MessageCitedCounts[item.MessageID]
: 0);
}
public int GetAccentedWordsCount(FeatureParameters parameters)
{
int messageId = parameters.MessageId;
yaf_Message item = buhonlineDataProvider.ReadMessage(messageId);
return(GetTaggedWordsCount(item.Message, new List <string> {
"b", "s", "color", "colour"
}));
}
public double GetStopWordsToAllWordsCount(FeatureParameters parameters)
{
int messageId = parameters.MessageId;
yaf_Message item = buhonlineDataProvider.ReadMessage(messageId);
HashSet <string> stopwords = Preprocessing.StopWords;
return((double)FeatureRegex.Words.Split(item.Message.ToLower()).Count(stopwords.Contains) /
GetWordsCount(parameters));
}
private static int GetAuthorPostsCount(yaf_Message item)
{
var author = GetAuthor(item);
if (author == null)
{
return(0);
}
return(author.NumPosts);
}
private static int GetAuthorForumAge(yaf_Message item)
{
var author = GetAuthor(item);
if (author == null || author.Joined > item.Posted)
{
return(0);
}
return((int)(item.Posted - author.Joined).TotalDays);
}
private static int GetAuthorRating(yaf_Message item)
{
var author = GetAuthor(item);
if (author == null)
{
return(0);
}
return(author.Points);
}
