public static void SetSegment(UIElement target, SegmentOption value) => target.SetValue(SegmentProperty, value);
public List <KeyWord> Segment(string sentence, SegmentOption option)
{
var dict = SingletonDict.Instance;
sentence = sentence.ToLower();
if (dict.NeedInitDict)
{
dict.TransDict = _sqlProvider.GetTransDict();
dict.AliasDict = _sqlProvider.GetAliasDict();
dict.AliasDict.AddRange(GetBoAlias());
dict.NeedInitDict = false;
}
List <WordInfo> words;
MatchOptions matchOptions = new MatchOptions();
matchOptions.IgnoreCapital = true;
if (option.MatchRule == "MinWords")
{
var parameter = new MatchParameter {
Redundancy = 2
};
words = _panGuSegmantService.DoSegment(sentence, matchOptions, parameter);
}
else
{
words = _panGuSegmantService.DoSegment(sentence, matchOptions);
}
words = words.Distinct(new WordAttributeComparer()).ToList();
var keywords = new List <KeyWord>();
if (option.Cc?.Count > 0)
{
keywords.AddRange(from item in words
let cc = item.Pos.ToString().Split(new[] { ", " }, StringSplitOptions.None).ToList()
where option.Cc.Intersect(cc).Any()
select new KeyWord
{
Term = item.Word,
Cc = option.Cc.Intersect(cc).ToList()
});
}
else
{
keywords.AddRange(words.Select(item => new KeyWord
{
Term = item.Word,
Cc = item.Pos.ToString().Split('|').ToList()
}));
}
foreach (var keyword in keywords)
{
//包含同义词
if (option.IncludeAlias)
{
foreach (var key in dict.AliasDict.Keys)
{
if (key.Equals(keyword.Term, StringComparison.OrdinalIgnoreCase))
{
keyword.Aliases.AddRange(dict.AliasDict[key]);
}
}
keyword.Aliases = keyword.Aliases.Distinct().ToList();
}
//包含翻译词
if (option.IncludeTrans)
{
foreach (var key in dict.TransDict.Keys)
{
if (key.Equals(keyword.Term, StringComparison.OrdinalIgnoreCase))
{
keyword.Translates.AddRange(dict.TransDict[key]);
}
}
keyword.Translates = keyword.Translates.Distinct().ToList();
}
}
return(keywords);
}
