public void ClassifyArticle()
{
Tags = new List <string>();
List <string> sentences = SplitSentences();
// analyse article using Bag of Words technique
TopicsList topicsList = new TopicsList("XF40Demo.Resources.NewsBoW.csv");
AnalyseSentences(sentences, topicsList);
// analyse article again to identify false positives
TopicsList falseTopicsList = new TopicsList("XF40Demo.Resources.NewsFalseBoW.csv");
AnalyseSentences(sentences, falseTopicsList);
// subtract false positives from topics list
foreach (Topic falseTopic in falseTopicsList.Topics)
{
Topic topic = topicsList.Topics.Find(x => x.Name.Equals(falseTopic.Name));
topic.Count -= falseTopic.Count;
}
// select topic + tags
Topic tempTopic = topicsList.Topics.OrderByDescending(o => o.Count).First();
Topic = (tempTopic.Count < 2 || string.Equals(Title, "week in review", StringComparison.OrdinalIgnoreCase)) ? "Unclassified" : tempTopic.Name;
foreach (Topic topic in topicsList.Topics.OrderByDescending(o => o.Count).Take(4))
{
if (topic.Count > 0)
{
Tags.Add(topic.Name);
}
}
}
public void AnalyseSentences(List <string> sentences, TopicsList topicsList)
{
foreach (string sentence in sentences)
{
Parallel.ForEach(topicsList.Topics, topic =>
{
foreach (string term in topic.Terms)
{
if (sentence.Contains(term.ToLower()))
{
topic.Count++;
}
}
});
}
}
