public static void TrainingIncreaseTarget()
{
TraningDataset.InitIncreaseStock();
var PreviewId = "";
var PreviewRoot = new HTMLEngine.MyRootHtmlNode();
foreach (var increase in TraningDataset.IncreaseStockList)
{
if (PreviewId.Equals(increase.id))
{
var htmlfile = Program.DocBase + @"\FDDC_announcements_round1_train_20180518\round1_train_20180518\定增\html\" + increase.id + ".html";
PreviewRoot = HTMLEngine.Anlayze(htmlfile);
}
TableAnlayzeTool.PutTrainingItem(PreviewRoot, increase.PublishTarget);
}
var Rank = new List <int>();
Rank = TableAnlayzeTool.TrainingTitleResult.Values.ToList();
Rank.Sort();
Rank.Reverse();
var Top10 = Rank[9];
foreach (var title in TableAnlayzeTool.TrainingTitleResult)
{
if (title.Value >= Top10)
{
Console.WriteLine(title.Key + ":" + title.Value);
}
}
}
private static void Traning()
{
Training = new StreamWriter("Training.log");
TraningDataset.InitContract();
TraningDataset.InitStockChange();
TraningDataset.InitIncreaseStock();
ContractTraning.Train();
Training.Close();
}
private static void Traning()
{
TraningDataset.InitContract();
TraningDataset.InitStockChange();
TraningDataset.InitReorganization();
Training = new StreamWriter("Training.log");
//ContractTraning.Train();
//ReOrganizationTraning.Train();
Training.Close();
}
public static void AnlayzeEntitySurroundWords()
{
var ContractPath_TRAIN = Program.DocBase + @"\FDDC_announcements_round1_train_20180518\重大合同";
Surround JiaFangSurround = new Surround();
Surround YiFangSurround = new Surround();
Surround ProjectNameSurround = new Surround();
Surround ContractNameSurround = new Surround();
LeadingWord JiaFangNameLeadingWord = new LeadingWord();
LeadingWord YiFangNameLeadingWord = new LeadingWord();
LeadingWord ProjectNameLeadingWord = new LeadingWord();
LeadingWord ContractNameLeadingWord = new LeadingWord();
foreach (var filename in System.IO.Directory.GetFiles(ContractPath_TRAIN + @"\html\"))
{
var fi = new System.IO.FileInfo(filename);
var Id = fi.Name.Replace(".html", String.Empty);
if (TraningDataset.GetContractById(Id).Count == 0)
{
continue;
}
var contract = TraningDataset.GetContractById(Id).First();
var doc = new AnnouceDocument(filename);
//if (!string.IsNullOrEmpty(contract.JiaFang)) JiaFangSurround.AnlayzeEntitySurroundWords(doc, contract.JiaFang);
//if (!string.IsNullOrEmpty(contract.YiFang)) YiFangSurround.AnlayzeEntitySurroundWords(doc, contract.YiFang);
//if (!string.IsNullOrEmpty(contract.ProjectName)) ProjectNameSurround.AnlayzeEntitySurroundWords(doc, contract.ProjectName);
//if (!string.IsNullOrEmpty(contract.ContractName)) ContractNameSurround.AnlayzeEntitySurroundWords(doc, contract.ContractName);
if (!string.IsNullOrEmpty(contract.JiaFang))
{
JiaFangNameLeadingWord.AnlayzeLeadingWord(doc, contract.JiaFang);
}
if (!string.IsNullOrEmpty(contract.YiFang))
{
YiFangNameLeadingWord.AnlayzeLeadingWord(doc, contract.YiFang);
}
if (!string.IsNullOrEmpty(contract.ProjectName))
{
ProjectNameLeadingWord.AnlayzeLeadingWord(doc, contract.ProjectName);
}
if (!string.IsNullOrEmpty(contract.ContractName))
{
ContractNameLeadingWord.AnlayzeLeadingWord(doc, contract.ContractName);
}
}
//JiaFangSurround.GetTop(10);
//YiFangSurround.GetTop(10);
//ProjectNameSurround.GetTop(10);
//ContractNameSurround.GetTop(10);
JiaFangLeadingDict = JiaFangNameLeadingWord.GetTop(5);
YiFangLeadingDict = YiFangNameLeadingWord.GetTop(5);
ProjectNameLeadingDict = ProjectNameLeadingWord.GetTop(5);
ContractNameLeadingDict = ContractNameLeadingWord.GetTop(5);
}
static void Main(string[] args)
{
//生成PDF的TXT文件的批处理命令
//PDFToTXT.GetBatchFile();
//初始化
Encoding.RegisterProvider(CodePagesEncodingProvider.Instance);
BussinessLogic.LoadCompanyName(@"Resources\FDDC_announcements_company_name_20180531.json");
TraningDataset.InitContract();
TraningDataset.InitStockChange();
TraningDataset.InitIncreaseStock();
ContractTraning.TraningMaxLenth();
ContractTraning.EntityWordPerperty();
Training.Close();
UT();
Extract();
Logger.Close();
Score.Close();
}
public static void RunWordAnlayze()
{
var root = HTMLEngine.Anlayze(Program.DocBase + @"\FDDC_announcements_round1_train_20180518\round1_train_20180518\重大合同\html\1044779.html");
var Contract = TraningDataset.GetContractById("1044779")[0];
EntityWordAnlayzeTool.AnlayzeEntitySurroundWords(root, Contract.ProjectName);
root = HTMLEngine.Anlayze(Program.DocBase + @"\FDDC_announcements_round1_train_20180518\round1_train_20180518\重大合同\html\1450.html");
Contract = TraningDataset.GetContractById("1450")[0];
EntityWordAnlayzeTool.AnlayzeEntitySurroundWords(root, Contract.ProjectName);
root = HTMLEngine.Anlayze(Program.DocBase + @"\FDDC_announcements_round1_train_20180518\round1_train_20180518\重大合同\html\1042224.html");
Contract = TraningDataset.GetContractById("1042224")[0];
EntityWordAnlayzeTool.AnlayzeEntitySurroundWords(root, Contract.ProjectName);
root = HTMLEngine.Anlayze(Program.DocBase + @"\FDDC_announcements_round1_train_20180518\round1_train_20180518\重大合同\html\917362.html");
Contract = TraningDataset.GetContractById("917362")[0];
EntityWordAnlayzeTool.AnlayzeEntitySurroundWords(root, Contract.ProjectName);
}
public static void AnlayzeEntitySurroundWords()
{
var ContractPath_TRAIN = Program.DocBase + @"\FDDC_announcements_round1_train_20180518\round1_train_20180518\重大合同";
Console.WriteLine("前导词：甲方");
foreach (var filename in System.IO.Directory.GetFiles(ContractPath_TRAIN + @"\html\"))
{
var fi = new System.IO.FileInfo(filename);
var Id = fi.Name.Replace(".html", "");
if (TraningDataset.GetContractById(Id).Count == 0)
{
continue;
}
var contract = TraningDataset.GetContractById(Id).First();
if (contract.JiaFang == "")
{
continue;
}
var root = HTMLEngine.Anlayze(filename);
EntityWordAnlayzeTool.AnlayzeEntitySurroundWords(root, contract.JiaFang);
}
}
/// <summary>
/// 分析实体周边词语
/// </summary>
public static void AnlayzeEntitySurroundWords()
{
var ContractPath_TRAIN = Program.DocBase + @"\FDDC_announcements_round1_train_20180518\重大合同";
Surround JiaFangSurround = new Surround();
Surround YiFangSurround = new Surround();
Surround ProjectNameSurround = new Surround();
Surround ContractNameSurround = new Surround();
LeadingColonWord JiaFangLeadingColonWord = new LeadingColonWord();
LeadingColonWord YiFangLeadingColonWord = new LeadingColonWord();
LeadingColonWord ProjectNameLeadingColonWord = new LeadingColonWord();
LeadingColonWord ContractNameLeadingColonWord = new LeadingColonWord();
foreach (var filename in System.IO.Directory.GetFiles(ContractPath_TRAIN + @"\html\"))
{
var fi = new System.IO.FileInfo(filename);
var Id = fi.Name.Replace(".html", String.Empty);
if (TraningDataset.GetContractById(Id).Count == 0)
{
continue;
}
var contract = TraningDataset.GetContractById(Id).First();
var doc = new Contract();
doc.Init(filename);
if (!string.IsNullOrEmpty(contract.JiaFang))
{
JiaFangSurround.AnlayzeEntitySurroundWords(doc, contract.JiaFang);
}
if (!string.IsNullOrEmpty(contract.YiFang))
{
YiFangSurround.AnlayzeEntitySurroundWords(doc, contract.YiFang);
}
if (!string.IsNullOrEmpty(contract.ProjectName))
{
ProjectNameSurround.AnlayzeEntitySurroundWords(doc, contract.ProjectName);
}
if (!string.IsNullOrEmpty(contract.ContractName))
{
ContractNameSurround.AnlayzeEntitySurroundWords(doc, contract.ContractName);
}
if (!string.IsNullOrEmpty(contract.JiaFang))
{
JiaFangLeadingColonWord.AnlayzeLeadingWord(doc, contract.JiaFang);
}
if (!string.IsNullOrEmpty(contract.YiFang))
{
YiFangLeadingColonWord.AnlayzeLeadingWord(doc, contract.YiFang);
}
if (!string.IsNullOrEmpty(contract.ProjectName))
{
ProjectNameLeadingColonWord.AnlayzeLeadingWord(doc, contract.ProjectName);
}
if (!string.IsNullOrEmpty(contract.ContractName))
{
ContractNameLeadingColonWord.AnlayzeLeadingWord(doc, contract.ContractName);
}
}
JiaFangSurround.WriteToLog(Program.Training);
Program.Training.WriteLine("甲方：冒号前导词");
JiaFangLeadingColonWord.WriteToLog(Program.Training);
JiaFangLeadingDict = Utility.ConvertRankToCIDict(Utility.FindTop(5, JiaFangLeadingColonWord.LeadingWordDict));
YiFangSurround.WriteToLog(Program.Training);
Program.Training.WriteLine("乙方：冒号前导词");
YiFangLeadingColonWord.WriteToLog(Program.Training);
YiFangLeadingDict = Utility.ConvertRankToCIDict(Utility.FindTop(5, YiFangLeadingColonWord.LeadingWordDict));
ProjectNameSurround.WriteToLog(Program.Training);
Program.Training.WriteLine("工程名：冒号前导词");
ProjectNameLeadingColonWord.WriteToLog(Program.Training);
ProjectNameLeadingDict = Utility.ConvertRankToCIDict(Utility.FindTop(5, ProjectNameLeadingColonWord.LeadingWordDict));
ContractNameSurround.WriteToLog(Program.Training);
Program.Training.WriteLine("合同名：冒号前导词");
ContractNameLeadingColonWord.WriteToLog(Program.Training);
ContractNameLeadingDict = Utility.ConvertRankToCIDict(Utility.FindTop(5, ContractNameLeadingColonWord.LeadingWordDict));
}
/// <summary>
/// 使用LTP方式分析实体周边词语
/// </summary>
public static void AnlayzeEntitySurroundWordsLTP()
{
var ContractPath_TRAIN = Program.DocBase + @"\FDDC_announcements_round1_train_20180518\重大合同";
var JiaFangDP = new LTPTrainingDP();
var JiaFangSRL = new LTPTrainingSRL();
var YiFnagDP = new LTPTrainingDP();
var YiFnagSRL = new LTPTrainingSRL();
var ContractNameDP = new LTPTrainingDP();
var ContractNameSRL = new LTPTrainingSRL();
var ProjectNameDP = new LTPTrainingDP();
var ProjectNameSRL = new LTPTrainingSRL();
foreach (var filename in System.IO.Directory.GetFiles(ContractPath_TRAIN + @"\html\"))
{
var fi = new System.IO.FileInfo(filename);
var Id = fi.Name.Replace(".html", String.Empty);
if (TraningDataset.GetContractById(Id).Count == 0)
{
continue;
}
var contract = TraningDataset.GetContractById(Id).First();
var c = new Contract();
c.Init(filename);
if (!string.IsNullOrEmpty(contract.JiaFang))
{
JiaFangDP.Training(c.Dplist, contract.JiaFang);
JiaFangSRL.Training(c.Srllist, contract.JiaFang);
}
if (!string.IsNullOrEmpty(contract.YiFang))
{
YiFnagDP.Training(c.Dplist, contract.YiFang);
YiFnagSRL.Training(c.Srllist, contract.YiFang);
}
if (!string.IsNullOrEmpty(contract.ContractName))
{
ContractNameDP.Training(c.Dplist, contract.ContractName);
ContractNameSRL.Training(c.Srllist, contract.ContractName);
}
if (!string.IsNullOrEmpty(contract.ProjectName))
{
ProjectNameDP.Training(c.Dplist, contract.ProjectName);
ProjectNameSRL.Training(c.Srllist, contract.ProjectName);
}
}
Program.Training.WriteLine("甲方附近词语分析（DP）：");
JiaFangDP.WriteToLog(Program.Training);
Program.Training.WriteLine("甲方附近词语分析（SRL）：");
JiaFangSRL.WriteToLog(Program.Training);
Program.Training.WriteLine("乙方附近词语分析（DP）：");
YiFnagDP.WriteToLog(Program.Training);
Program.Training.WriteLine("乙方附近词语分析（SRL）：");
YiFnagSRL.WriteToLog(Program.Training);
Program.Training.WriteLine("合同名附近词语分析（DP）：");
ContractNameDP.WriteToLog(Program.Training);
Program.Training.WriteLine("合同名附近词语分析（SRL）：");
ContractNameSRL.WriteToLog(Program.Training);
Program.Training.WriteLine("工程名附近词语分析（DP）：");
ProjectNameDP.WriteToLog(Program.Training);
Program.Training.WriteLine("工程名附近词语分析（SRL）：");
ProjectNameSRL.WriteToLog(Program.Training);
}