예제 #1
0
    /// <summary>
    /// 获得工程名
    /// </summary>
    /// <returns></returns>
    string GetProjectName()
    {
        var e = new EntityProperty();

        e.PropertyName            = "工程名称";
        e.LeadingColonKeyWordList = new string[] { "项目名称:", "工程名称:", "中标项目:", "合同标的:", "工程内容:" };
        e.LeadingColonKeyWordCandidatePreprocess  = TrimEndJianCheng;
        e.QuotationTrailingWordList_IsSkipBracket = true;
        e.QuotationTrailingWordList = new string[] {
            "标段施工项目", "标段土建工程", "标段施工总承包", "标段的工程", "标段工程", "标段施工总价承包",
            "标段施工总承包工程", "标段施工工程", "标段土建工程建设项目", "标段站前工程", "标段工程(施工)",
            "工程施工工程", "项目施工工程", "施工工程",
            "工程项目", "工程标段", "标段的施工项目", "标段项目", "标段施工",
            "招标采购项目", "招标活动", "采购活动", "招标项目",
            "项目", "采购", "总承包",
            "工程", "标段", "标",
        };
        e.Extract(this);
        var prj = e.EvaluateCI();

        if (!String.IsNullOrEmpty(prj))
        {
            return(prj);
        }

        //var Stardard = TraningDataset.ContractList.Where(x => x.Id == this.Id).ToList();
        //if (Stardard.Count == 1)
        //{
        //Console.WriteLine("标准答案:" + Stardard[0].ProjectName);
        //}

        //var ProjectNameList = ProjectNameLogic.GetProjectNameByCutWord(this);
        //var ProjectNameListNER = ProjectNameLogic.GetProjectNameByNer(this);

        var StartArray = new string[] { "公司为", "参与了", "确定为" };
        var EndArray   = new string[] { "的中标单位", "的公开招投标", "的中标人", "候选人" };

        e.ExternalStartEndStringFeature = Utility.GetStartEndStringArray(StartArray, EndArray);
        e.Extract(this);
        prj = e.EvaluateCI();
        if (!String.IsNullOrEmpty(prj))
        {
            if (ExtractPropertyByHTML.FindWordCnt(prj + "项目", root).Count >= 1)
            {
                return(prj + "项目");
            }
            return(prj);
        }

        foreach (var item in quotationList)
        {
            if (item.Value.Contains("推荐的中标候选人公示"))
            {
                prj = Utility.GetStringBefore(item.Value, "推荐的中标候选人公示");
                return(prj);
            }
        }
        return(string.Empty);
    }
예제 #2
0
파일: Contract.cs 프로젝트: lxxwin/FDDC
    /// <summary>
    /// 获得工程名
    /// </summary>
    /// <returns></returns>
    string GetProjectName()
    {
        var e = new EntityProperty();

        e.PropertyName            = "工程名称";
        e.LeadingColonKeyWordList = new string[] { "项目名称:", "工程名称:", "中标项目:", "合同标的:", "工程内容:" };
        e.LeadingColonKeyWordCandidatePreprocess = TrimEndJianCheng;
        e.QuotationTrailingWordList = new string[] { "工程", "标段", "标", "招标活动", "项目", "采购" };
        var StartArray = new string[] { "公司为", "参与了", "确定为" };
        var EndArray   = new string[] { "的中标单位", "的公开招投标", "的中标人", "候选人" };

        e.ExternalStartEndStringFeature = Utility.GetStartEndStringArray(StartArray, EndArray);
        e.Extract(this);
        var prj = e.EvaluateCI();

        if (!String.IsNullOrEmpty(prj))
        {
            return(prj);
        }
        foreach (var item in quotationList)
        {
            if (item.Value.Contains("推荐的中标候选人公示"))
            {
                return(Utility.GetStringBefore(item.Value, "推荐的中标候选人公示"));
            }
        }
        return(string.Empty);
    }
예제 #3
0
파일: Contract.cs 프로젝트: lxxwin/FDDC
    /// <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());
    }
예제 #4
0
    /// <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);
    }
예제 #5
0
파일: Contract.cs 프로젝트: PeiYangLiu/FDDC
    /// <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());
    }