private TagFreqItem <NS>[] LoadDat(string path)
{
if (!File.Exists(path))
{
return(null);
}
var bytes = File.ReadAllBytes(path);
int index = 0;
int size = BitConverter.ToInt32(bytes, index);
index += 4;
var valueArr = new TagFreqItem <NS> [size];
for (int i = 0; i < size; i++)
{
var currSize = BitConverter.ToInt32(bytes, index);
index += 4;
var tfi = new TagFreqItem <NS>();
for (int j = 0; j < currSize; j++)
{
var tag = BitConverter.ToInt32(bytes, index);
index += 4;
var freq = BitConverter.ToInt32(bytes, index);
index += 4;
tfi.labelMap.Add((NS)tag, freq);
}
valueArr[i] = tfi;
}
return(valueArr);
}
public override TagFreqItem <NR>[] OnLoadValue(string path)
{
var valueArr = LoadDat(path + ".value.dat");
if (valueArr != null)
{
return(valueArr);
}
var valueList = new List <TagFreqItem <NR> >();
try
{
foreach (var line in File.ReadLines(path))
{
var tuple = TagFreqItem <NR> .Create(line);
var tfi = new TagFreqItem <NR>();
foreach (var p in tuple.Item2)
{
//tfi.AddLabel((NR)Enum.Parse(typeof(NR), p.Key), p.Value);
tfi.labelMap[(NR)Enum.Parse(typeof(NR), p.Key)] = p.Value;
}
valueList.Add(tfi);
}
}
catch (Exception e)
{
return(null);
}
return(valueList.ToArray());
}
private static List <TagFreqItem <NS> > RoleTag(List <Vertex> vertices, WordNet wordnet_all)
{
var tagList = new List <TagFreqItem <NS> >();
for (int i = 0; i < vertices.Count; i++)
{
var vertex = vertices[i];
var nature = vertex.GetNature();
if (Nature.ns == nature && vertex.attr.totalFreq <= 1000)
{
if (vertex.realWord.Length < 3) // 二字地名
{
tagList.Add(new TagFreqItem <NS>(NS.H, NS.G));
}
else
{
tagList.Add(new TagFreqItem <NS>(NS.G));
}
continue;
}
var tfi = PlaceDictionary.dict.Get(vertex.word); // 使用等效词
if (tfi == null)
{
tfi = new TagFreqItem <NS>(NS.Z, PlaceDictionary.trans_tr_dict.GetFreq(NS.Z));
}
tagList.Add(tfi);
}
return(tagList);
}
/// <summary>
/// 找出给定顶点列表中的顶点的关联词性标签，以及对应在机构词典中的《标签，频率》pair。
/// </summary>
/// <param name="vertices"></param>
/// <param name="wordNetAll"></param>
/// <returns></returns>
public static List <TagFreqItem <NT> > RoleTag(List <Vertex> vertices, WordNet wordNetAll)
{
var tagList = new List <TagFreqItem <NT> >();
for (int i = 0; i < vertices.Count; i++) // 遍历顶点
{
var vertex = vertices[i]; // 当前顶点
// 找出当前词条的所有关联词性，并作为
var nature = vertex.GetNature(); // 当前顶点（词条）的词性
switch (nature)
{
case Nature.nrf: // 音译人名
if (vertex.attr.totalFreq <= 1000)
{
tagList.Add(new TagFreqItem <NT>(NT.F, 1000));
continue;
}
break;
case Nature.ni: // 机构相关名称
case Nature.nic:
case Nature.nis:
case Nature.nit:
var tfi = new TagFreqItem <NT>(NT.K, 1000); //
tfi.AddLabel(NT.D, 1000);
tagList.Add(tfi);
continue;
case Nature.m:
tagList.Add(new TagFreqItem <NT>(NT.M, 1000));
continue;
}
var tagItem = OrgDictionary.dictionary.Get(vertex.word); // 此处使用等效词，更加精准
if (tagItem == null)
{
tagItem = new TagFreqItem <NT>(NT.Z, OrgDictionary.transformMatrixDictionary.GetFreq(NT.Z));
}
tagList.Add(tagItem);
}
return(tagList);
}
private TagFreqItem <NR>[] LoadDat(string path)
{
//var bytes = File.ReadAllBytes(path);
var ba = ByteArray.Create(path);
if (ba == null)
{
return(null);
}
//if (bytes == null || bytes.Length < 5) return null;
//int index = 0;
//int size = ByteUtil.Bytes2Int(bytes, index);
//index += 4;
int size = ba.NextInt();
var valueArr = new TagFreqItem <NR> [size];
for (int i = 0; i < size; i++)
{
//var currSize = ByteUtil.Bytes2Int(bytes, index);
//index += 4;
var currSize = ba.NextInt();
var tfi = new TagFreqItem <NR>();
for (int j = 0; j < currSize; j++)
{
//var enumVal = ByteUtil.Bytes2Int(bytes, index);
//index += 4;
//var freq = ByteUtil.Bytes2Int(bytes, index);
//index += 4;
var enumVal = ba.NextInt();
var freq = ba.NextInt();
tfi.AddLabel((NR)enumVal, freq);
}
valueArr[i] = tfi;
}
return(valueArr);
}
/// <summary>
/// 角色观察
/// </summary>
/// <param name="wordSegs"></param>
/// <returns></returns>
public static List <TagFreqItem <NR> > RoleObserve(List <Vertex> wordSegs)
{
var tagList = new List <TagFreqItem <NR> >()
{
new TagFreqItem <NR>(NR.A, NR.K)
}; // 始 ## 始 A K
var dict = ChsPersonNameDict.dictionary;
for (int i = 1; i < wordSegs.Count; i++) // 跳过起始辅助节点
{
var vertex = wordSegs[i];
var nritem = dict.Get(vertex.realWord); // 获取词条（节点）的字符串值对应的《标签，频率》pair
if (nritem == null) // 如果没有字符串对应的TagFreqItem，那就由顶点对应的词性来帮助分析
{
switch (vertex.GuessNature())
{
case Nature.nr: // 如果词性是人名，
if (vertex.attr.totalFreq <= 1000 && vertex.realWord.Length == 2)
{
nritem = new TagFreqItem <NR>(NR.X, NR.G);
}
else
{
nritem = new TagFreqItem <NR>(NR.A, ChsPersonNameDict.transformMatrixDictionary.GetFreq(NR.A));
}
break;
case Nature.nnt: // 职务职称
nritem = new TagFreqItem <NR>(NR.G, NR.K);
break;
default:
nritem = new TagFreqItem <NR>(NR.A, ChsPersonNameDict.transformMatrixDictionary.GetFreq(NR.A));
break;
}
} // 如果人名词典中存在当前顶点这样的词条，那直接添加对应的TagFreqItem到列表中
tagList.Add(nritem);
}
return(tagList);
}
public override TagFreqItem <NT>[] OnLoadValue(string path)
{
var valueArr = LoadDat(path + ".value.dat");
if (valueArr != null)
{
return(valueArr);
}
var valueList = new List <TagFreqItem <NT> >();
try
{
foreach (var line in File.ReadLines(path))
{
if (string.IsNullOrWhiteSpace(line))
{
continue;
}
var tuple = TagFreqItem <NT> .Create(line);
var tfi = new TagFreqItem <NT>();
foreach (var p in tuple.Item2)
{
tfi.labelMap.Add((NT)Enum.Parse(typeof(NT), p.Key), p.Value);
}
valueList.Add(tfi);
}
}
catch (Exception e)
{
// log load error
}
return(valueList.ToArray());
}