/// <summary> /// 获得合同名 /// </summary> /// <returns></returns> string GetContractName() { var e = new EntityProperty(); e.PropertyName = "合同名称"; e.PropertyType = EntityProperty.enmType.NER; e.MaxLength = ContractTraning.ContractES.MaxLength; e.MinLength = ContractTraning.ContractES.MinLength; e.LeadingColonKeyWordList = new string[] { "合同名称:" }; e.QuotationTrailingWordList = new string[] { "协议书", "合同书", "确认书", "合同", "协议" }; e.QuotationTrailingWordList_IsSkipBracket = true; //暂时只能选True var KeyList = new List <ExtractPropertyByDP.DPKeyWord>(); KeyList.Add(new ExtractPropertyByDP.DPKeyWord() { StartWord = new string[] { "签署", "签订" }, //通过SRL训练获得 StartDPValue = new string[] { LTPTrainingDP.核心关系, LTPTrainingDP.定中关系, LTPTrainingDP.并列关系 }, EndWord = new string[] { "补充协议", "合同书", "合同", "协议书", "协议", }, EndDPValue = new string[] { LTPTrainingDP.核心关系, LTPTrainingDP.定中关系, LTPTrainingDP.并列关系, LTPTrainingDP.动宾关系, LTPTrainingDP.主谓关系 } }); e.DpKeyWordList = KeyList; var StartArray = new string[] { "签署了", "签订了" }; //通过语境训练获得 var EndArray = new string[] { "合同" }; e.ExternalStartEndStringFeature = Utility.GetStartEndStringArray(StartArray, EndArray); e.ExternalStartEndStringFeatureCandidatePreprocess = (x) => { return(x + "合同"); }; e.MaxLengthCheckPreprocess = str => { return(Utility.TrimEnglish(str)); }; //最高级别的置信度,特殊处理器 e.LeadingColonKeyWordCandidatePreprocess = str => { var c = Normalizer.ClearTrailing(TrimEndJianCheng(str)); return(c); }; e.CandidatePreprocess = str => { var c = Normalizer.ClearTrailing(TrimEndJianCheng(str)); var RightQMarkIdx = c.IndexOf("”"); if (!(RightQMarkIdx != -1 && RightQMarkIdx != c.Length - 1)) { //对于"XXX"合同,有右边引号,但不是最后的时候,不用做 c = c.TrimStart("“".ToCharArray()); } c = c.TrimStart("《".ToCharArray()); c = c.TrimEnd("》".ToCharArray()).TrimEnd("”".ToCharArray()); return(c); }; e.ExcludeContainsWordList = new string[] { "日常经营重大合同" }; //下面这个列表的根据不足,正确做法是【尚未签署】 e.ExcludeEqualsWordList = new string[] { "若干项重大合同", "中标合同", "正式合同", "合同", "重大合同", "项目合同", "终止协议", "经营合同", "特别重大合同", "相关项目合同" }; e.Extract(this); //冒号优先 return(e.EvaluateCI()); }
/// <summary> /// 获得合同名 /// </summary> /// <returns></returns> string GetContractName() { var e = new EntityProperty(); e.PropertyName = "合同名称"; e.PropertyType = EntityProperty.enmType.NER; e.MaxLength = 200; e.MinLength = 4; e.LeadingColonKeyWordList = new string[] { "合同名称:" }; e.QuotationTrailingWordList = new string[] { "商务合同补充协议", "承包合同补充协议", "补充协议", "经营合同补充协议", "协议书", "合同书", "确认书", "合同", "协议" }; e.QuotationTrailingWordList_IsSkipBracket = false; var StartArray = new string[] { "签署了", "签订了" }; //通过语境训练获得 var EndArray = new string[] { "合同" }; e.ExternalStartEndStringFeature = Utility.GetStartEndStringArray(StartArray, EndArray); e.ExternalStartEndStringFeatureCandidatePreprocess = (x) => { return(x + "合同"); }; e.MaxLengthCheckPreprocess = str => { return(Utility.TrimEnglish(str)); }; //最高级别的置信度,特殊处理器 e.LeadingColonKeyWordCandidatePreprocess = str => { var c = Normalizer.ClearTrailing(TrimEndJianCheng(str)); return(c); }; e.CandidatePreprocess = str => { var c = Normalizer.ClearTrailing(TrimEndJianCheng(str)); var RightQMarkIdx = c.IndexOf("”"); if (!(RightQMarkIdx != -1 && RightQMarkIdx != c.Length - 1)) { //对于"XXX"合同,有右边引号,但不是最后的时候,不用做 c = c.TrimStart("“".ToCharArray()); } c = c.TrimStart("《".ToCharArray()); c = c.TrimEnd("》".ToCharArray()).TrimEnd("”".ToCharArray()); return(c); }; e.ExcludeContainsWordList = new string[] { "日常经营重大合同" }; //下面这个列表的根据不足,正确做法是【尚未签署】 e.ExcludeEqualsWordList = new string[] { "若干项重大合同", "中标合同", "正式合同", "合同", "重大合同", "项目合同", "终止协议", "经营合同", "特别重大合同", "相关项目合同" }; e.Extract(this); //冒号优先 var contractname = e.EvaluateCI(); return(contractname); }
/// <summary> /// 获得合同名 /// </summary> /// <returns></returns> string GetContractName() { var e = new EntityProperty(); e.PropertyName = "合同名称"; e.PropertyType = EntityProperty.enmType.NER; e.MaxLength = ContractTraning.MaxContractNameLength; e.MinLength = 5; /* 训练模式下 * e.LeadingColonKeyWordList = ContractTraning.ContractNameLeadingDict * .Where((x) => { return x.Value >= 40; }) //阈值40%以上 * .Select((x) => { return x.Key + ":"; }).ToArray(); */ e.LeadingColonKeyWordList = new string[] { "合同名称:" }; e.QuotationTrailingWordList = new string[] { "协议书", "合同书", "确认书", "合同", "协议" }; e.QuotationTrailingWordList_IsSkipBracket = true; //暂时只能选True var KeyList = new List <ExtractPropertyByDP.DPKeyWord>(); KeyList.Add(new ExtractPropertyByDP.DPKeyWord() { StartWord = new string[] { "签署", "签订" }, //通过SRL训练获得 StartDPValue = new string[] { LTPTrainingDP.核心关系, LTPTrainingDP.定中关系, LTPTrainingDP.并列关系 }, EndWord = new string[] { "补充协议", "合同书", "合同", "协议书", "协议", }, EndDPValue = new string[] { LTPTrainingDP.核心关系, LTPTrainingDP.定中关系, LTPTrainingDP.并列关系, LTPTrainingDP.动宾关系, LTPTrainingDP.主谓关系 } }); e.DpKeyWordList = KeyList; var StartArray = new string[] { "签署了", "签订了" }; //通过语境训练获得 var EndArray = new string[] { "合同" }; e.ExternalStartEndStringFeature = Utility.GetStartEndStringArray(StartArray, EndArray); e.ExternalStartEndStringFeatureCandidatePreprocess = (x) => { return(x + "合同"); }; e.MaxLengthCheckPreprocess = str => { return(EntityWordAnlayzeTool.TrimEnglish(str)); }; //最高级别的置信度,特殊处理器 e.LeadingColonKeyWordCandidatePreprocess = str => { var c = Normalizer.ClearTrailing(TrimJianCheng(str)); return(c); }; e.CandidatePreprocess = str => { var c = Normalizer.ClearTrailing(TrimJianCheng(str)); var RightQMarkIdx = c.IndexOf("”"); if (!(RightQMarkIdx != -1 && RightQMarkIdx != c.Length - 1)) { //对于"XXX"合同,有右边引号,但不是最后的时候,不用做 c = c.TrimStart("“".ToCharArray()); } c = c.TrimStart("《".ToCharArray()); c = c.TrimEnd("》".ToCharArray()).TrimEnd("”".ToCharArray()); return(c); }; e.ExcludeContainsWordList = new string[] { "日常经营重大合同" }; //下面这个列表的根据不足 e.ExcludeEqualsWordList = new string[] { "合同", "重大合同", "项目合同", "终止协议", "经营合同", "特别重大合同", "相关项目合同" }; e.Extract(this); //是否所有的候选词里面包括(测试集无法使用) var contractlist = TraningDataset.ContractList.Where((x) => { return(x.id == this.Id); }); if (contractlist.Count() > 0) { var contract = contractlist.First(); var contractname = contract.ContractName; if (!String.IsNullOrEmpty(contractname)) { e.CheckIsCandidateContainsTarget(contractname); } } //置信度 e.Confidence = ContractTraning.ContractES.GetStardardCI(); return(e.EvaluateCI()); }
/// <summary> /// 获得甲方 /// </summary> /// <returns></returns> public string GetJiaFang() { //最高置信度的抽取 EntityProperty e = new EntityProperty(); e.ExcludeContainsWordList = new string[] { "招标代理" }; e.LeadingColonKeyWordList = new string[] { "甲方:", "合同买方:", "发包人:", "发包单位:", "发包方:", "发包机构:", "发包人名称:", "招标人:", "招标单位:", "招标方:", "招标机构:", "招标人名称:", "业主:", "业主单位:", "业主方:", "业主机构:", "业主名称:", "采购单位:", "采购单位名称:", "采购人:", "采购人名称:", "采购方:", "采购方名称:" }; e.CandidatePreprocess = (x => { x = Normalizer.ClearTrailing(x); return(CompanyNameLogic.AfterProcessFullName(x).secFullName); }); e.MaxLength = ContractTraning.JiaFangES.MaxLength; e.MaxLengthCheckPreprocess = Utility.TrimEnglish; e.MinLength = 3; e.Extract(this); //这里不直接做Distinct,出现频次越高,则可信度越高 //多个甲方的时候,可能意味着没有甲方! if (e.LeadingColonKeyWordCandidate.Distinct().Count() > 1) { foreach (var candidate in e.LeadingColonKeyWordCandidate) { Program.Logger.WriteLine("发现多个甲方:" + candidate); } } if (e.LeadingColonKeyWordCandidate.Count > 0) { return(e.LeadingColonKeyWordCandidate[0]); } //招标 var Extractor = new ExtractPropertyByHTML(); var CandidateWord = new List <String>(); var StartArray = new string[] { "招标单位", "业主", "收到", "接到" }; var EndArray = new string[] { "发来", "发出", "的中标" }; Extractor.StartEndFeature = Utility.GetStartEndStringArray(StartArray, EndArray); Extractor.Extract(root); foreach (var item in Extractor.CandidateWord) { var JiaFang = CompanyNameLogic.AfterProcessFullName(item.Value.Trim()); if (JiaFang.secFullName.Contains("招标代理")) { continue; //特殊业务规则 } JiaFang.secFullName = JiaFang.secFullName.Replace("业主", String.Empty).Trim(); JiaFang.secFullName = JiaFang.secFullName.Replace("招标单位", String.Empty).Trim(); if (Utility.TrimEnglish(JiaFang.secFullName).Length > ContractTraning.JiaFangES.MaxLength) { continue; } if (JiaFang.secFullName.Length < 3) { continue; //使用实际长度排除全英文的情况 } if (!Program.IsMultiThreadMode) { Program.Logger.WriteLine("甲方候补词(招标):[" + JiaFang.secFullName + "]"); } CandidateWord.Add(JiaFang.secFullName); } //合同 Extractor = new ExtractPropertyByHTML(); StartArray = new string[] { "与", "与业主" }; EndArray = new string[] { "签署", "签订" }; Extractor.StartEndFeature = Utility.GetStartEndStringArray(StartArray, EndArray); Extractor.Extract(root); foreach (var item in Extractor.CandidateWord) { var JiaFang = CompanyNameLogic.AfterProcessFullName(item.Value.Trim()); JiaFang.secFullName = JiaFang.secFullName.Replace("业主", String.Empty).Trim(); if (JiaFang.secFullName.Contains("招标代理")) { continue; //特殊业务规则 } if (Utility.TrimEnglish(JiaFang.secFullName).Length > ContractTraning.JiaFangES.MaxLength) { continue; } if (JiaFang.secFullName.Length < 3) { continue; //使用实际长度排除全英文的情况 } if (!Program.IsMultiThreadMode) { Program.Logger.WriteLine("甲方候补词(合同):[" + JiaFang.secFullName + "]"); } CandidateWord.Add(JiaFang.secFullName); } return(CompanyNameLogic.MostLikeCompanyName(CandidateWord)); }
/// <summary> /// 获得甲方 /// </summary> /// <returns></returns> string GetJiaFang(String YiFang) { //最高置信度的抽取 EntityProperty e = new EntityProperty(); e.ExcludeContainsWordList = new string[] { "招标代理" }; e.LeadingColonKeyWordList = new string[] { "甲方:", "合同买方:", "发包人:", "发包单位:", "发包方:", "发包机构:", "发包人名称:", "招标人:", "招标单位:", "招标方:", "招标机构:", "招标人名称:", "项目招标人:", "业主:", "业主单位:", "业主方:", "业主机构:", "业主名称:", "采购单位:", "采购单位名称:", "采购人:", "采购人名称:", "采购方:", "采购方名称:" }; e.CandidatePreprocess = (x => { x = Normalizer.ClearTrailing(x); return(CompanyNameLogic.AfterProcessFullName(x).secFullName); }); e.MaxLength = 32; e.MaxLengthCheckPreprocess = Utility.TrimEnglish; e.MinLength = 3; e.Extract(this); //这里不直接做Distinct,出现频次越高,则可信度越高 //多个甲方的时候,可能意味着没有甲方! if (e.LeadingColonKeyWordCandidate.Distinct().Count() > 1) { foreach (var candidate in e.LeadingColonKeyWordCandidate) { Program.Logger.WriteLine("发现多个甲方:" + candidate); } } if (e.LeadingColonKeyWordCandidate.Count > 0) { return(e.LeadingColonKeyWordCandidate[0]); } var ner = SearchJiaFang(); var NerJia = String.Empty; if (!String.IsNullOrEmpty(ner)) { foreach (var cn in companynamelist) { if (cn.secShortName == ner) { ner = cn.secFullName; } } if (String.IsNullOrEmpty(YiFang)) { NerJia = ner; } if (!YiFang.Equals(ner)) { NerJia = ner; } } //招标 var Extractor = new ExtractPropertyByHTML(); var CandidateWord = new List <String>(); var StartArray = new string[] { "招标单位", "业主", "收到", "接到" }; var EndArray = new string[] { "发来", "发出", "的中标" }; Extractor.StartEndFeature = Utility.GetStartEndStringArray(StartArray, EndArray); Extractor.Extract(root); foreach (var item in Extractor.CandidateWord) { var JiaFang = CompanyNameLogic.AfterProcessFullName(item.Value.Trim()); if (JiaFang.secFullName.Contains("招标代理")) { continue; //特殊业务规则 } JiaFang.secFullName = JiaFang.secFullName.Replace("业主", String.Empty).Trim(); JiaFang.secFullName = JiaFang.secFullName.Replace("招标单位", String.Empty).Trim(); if (Utility.TrimEnglish(JiaFang.secFullName).Length > 32) { continue; } if (JiaFang.secFullName.Length < 3) { continue; //使用实际长度排除全英文的情况 } if (!Program.IsMultiThreadMode) { Program.Logger.WriteLine("甲方候补词(招标):[" + JiaFang.secFullName + "]"); } CandidateWord.Add(JiaFang.secFullName); } //合同 Extractor = new ExtractPropertyByHTML(); StartArray = new string[] { "与", "与业主" }; EndArray = new string[] { "签署", "签订" }; Extractor.StartEndFeature = Utility.GetStartEndStringArray(StartArray, EndArray); Extractor.Extract(root); foreach (var item in Extractor.CandidateWord) { var JiaFang = CompanyNameLogic.AfterProcessFullName(item.Value.Trim()); JiaFang.secFullName = JiaFang.secFullName.Replace("业主", String.Empty).Trim(); if (JiaFang.secFullName.Contains("招标代理")) { continue; //特殊业务规则 } if (Utility.TrimEnglish(JiaFang.secFullName).Length > 32) { continue; } if (JiaFang.secFullName.Length < 3) { continue; //使用实际长度排除全英文的情况 } if (!Program.IsMultiThreadMode) { Program.Logger.WriteLine("甲方候补词(合同):[" + JiaFang.secFullName + "]"); } CandidateWord.Add(JiaFang.secFullName); } if (!String.IsNullOrEmpty(NerJia)) { //原则上,有NER中提取的甲方,则使用甲方 foreach (var c in CandidateWord) { //但是,这里有可能是正确的解答,例如 //NER:(集团)有限公司 实际上应该是 XXXX(集团)有限公司 if (c.EndsWith(NerJia)) { return(c); } } return(NerJia); } else { return(CompanyNameLogic.MostLikeCompanyName(CandidateWord)); } }