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);
}
}
