static string GetProjectName(MyRootHtmlNode root) { var Extractor = new EntityProperty(); //这些关键字后面 Extractor.LeadingWordList = new string[] { "项目名称:", "工程名称:", "中标项目:", "合同标的:", "工程内容:" }; Extractor.Extract(root); foreach (var item in Extractor.CandidateWord) { var ProjectName = item.Trim(); if (EntityWordAnlayzeTool.TrimEnglish(ProjectName).Length > ContractTraning.MaxContractNameLength) { continue; } Program.Logger.WriteLine("项目名称候补词(关键字):[" + item + "]"); return(ProjectName); } var MarkFeature = new EntityProperty.struMarkFeature(); MarkFeature.MarkStartWith = "“"; MarkFeature.MarkEndWith = "”"; MarkFeature.InnerEndWith = "标段"; var MarkFeatureConfirm = new EntityProperty.struMarkFeature(); MarkFeatureConfirm.MarkStartWith = "“"; MarkFeatureConfirm.MarkEndWith = "”"; MarkFeatureConfirm.InnerEndWith = "标"; Extractor.MarkFeature = new EntityProperty.struMarkFeature[] { MarkFeature, MarkFeatureConfirm }; Extractor.Extract(root); foreach (var item in Extractor.CandidateWord) { var ProjectName = item.Trim(); if (EntityWordAnlayzeTool.TrimEnglish(ProjectName).Length > ContractTraning.MaxContractNameLength) { continue; } Program.Logger.WriteLine("工程名称候补词(《XXX》):[" + item + "]"); return(ProjectName); } var list = BussinessLogic.GetProjectName(root); if (list.Count > 0) { return(list[0]); } return(""); }
static string GetProjectName(MyRootHtmlNode root) { var Extractor = new ExtractProperty(); //这些关键字后面 Extractor.LeadingWordList = new string[] { "项目名称:" }; Extractor.Extract(root); foreach (var item in Extractor.CandidateWord) { Program.Logger.WriteLine("项目名称候补词(关键字):[" + item + "]"); return(item.Replace(" ", "")); } var MarkFeature = new ExtractProperty.struMarkFeature(); MarkFeature.MarkStartWith = "“"; MarkFeature.MarkEndWith = "”"; MarkFeature.InnerEndWith = "标段"; var MarkFeatureConfirm = new ExtractProperty.struMarkFeature(); MarkFeatureConfirm.MarkStartWith = "“"; MarkFeatureConfirm.MarkEndWith = "”"; MarkFeatureConfirm.InnerEndWith = "标"; Extractor.MarkFeature = new ExtractProperty.struMarkFeature[] { MarkFeature, MarkFeatureConfirm }; Extractor.Extract(root); foreach (var item in Extractor.CandidateWord) { Program.Logger.WriteLine("工程名称候补词(《XXX》):[" + item + "]"); return(item); } var list = BussinessLogic.GetProjectName(root); if (list.Count > 0) { return(list[0]); } return(""); }