cwmexample.java

一个数据挖掘软件ALPHAMINERR的整个过程的JAVA版源代码

     // -----------------------------------------------------------------------
     //  Physical Model
     // -----------------------------------------------------------------------

     // DEFINING TABLES AND COLUMNS
     // Create the product table:
     Table prodTable = relationalPackage.getTable().createTable();
     prodTable.setName("Product");

     // Create the columns:
     Column relUPC = relationalPackage.getColumn().createColumn();
     relUPC.setName("UPC");
     relUPC.setOwner(prodTable);

     Column relProdDesc = relationalPackage.getColumn().createColumn();
     relProdDesc.setName("Description");
     relProdDesc.setOwner(prodTable);

     Column relBrand = relationalPackage.getColumn().createColumn();
     relBrand.setName("Brand");
     relBrand.setOwner(prodTable);

     Column relType = relationalPackage.getColumn().createColumn();
     relType.setName("Type");
     relType.setOwner(prodTable);

     Column relCat = relationalPackage.getColumn().createColumn();
     relCat.setName("Category");
     relCat.setOwner(prodTable);

     Column relColor = relationalPackage.getColumn().createColumn();
     relColor.setName("Color");
     relColor.setOwner(prodTable);

     // Create the geography table:
     Table geogTable = relationalPackage.getTable().createTable();
     geogTable.setName("Geography");

     // Create the columns:
     Column relState = relationalPackage.getColumn().createColumn();
     relState.setName("State");
     relState.setOwner(geogTable);

     Column relGeogDesc = relationalPackage.getColumn().createColumn();
     relGeogDesc.setName("Description");
     relGeogDesc.setOwner(geogTable);

     Column relTerr = relationalPackage.getColumn().createColumn();
     relTerr.setName("Territory");
     relTerr.setOwner(geogTable);

     Column relRegion = relationalPackage.getColumn().createColumn();
     relRegion.setName("Region");
     relRegion.setOwner(geogTable);

     // Create the time table:
     Table timeTable = relationalPackage.getTable().createTable();
     timeTable.setName("Time");

     // Create the columns:
     Column relWeek = relationalPackage.getColumn().createColumn();
     relWeek.setName("Week");
     relWeek.setOwner(timeTable);

     Column relStartDate = relationalPackage.getColumn().createColumn();
     relStartDate.setName("Start Date");
     relStartDate.setOwner(timeTable);

     Column relEndDate = relationalPackage.getColumn().createColumn();
     relEndDate.setName("End Date");
     relEndDate.setOwner(timeTable);

     Column relMonth = relationalPackage.getColumn().createColumn();
     relMonth.setName("Month");
     relMonth.setOwner(timeTable);

     Column relQuarter = relationalPackage.getColumn().createColumn();
     relQuarter.setName("Quarter");
     relQuarter.setOwner(timeTable);

     Column relYear = relationalPackage.getColumn().createColumn();
     relYear.setName("Year");
     relYear.setOwner(timeTable);

     // Create the sales fact table:
     Table salesFactTable = relationalPackage.getTable().createTable();
     salesFactTable.setName("Sales Fact Table");

     // Create the columns:
     Column relSalesTimeId = relationalPackage.getColumn().createColumn();
     relSalesTimeId.setName("Time");
     relSalesTimeId.setOwner(salesFactTable);

     Column relSalesGeogId = relationalPackage.getColumn().createColumn();
     relSalesGeogId.setName("Geography");
     relSalesGeogId.setOwner(salesFactTable);

     Column relSalesProdId = relationalPackage.getColumn().createColumn();
     relSalesProdId.setName("Product");
     relSalesProdId.setOwner(salesFactTable);

     Column relSales = relationalPackage.getColumn().createColumn();
     relSales.setName("Sales");
     relSales.setOwner(salesFactTable);

     Column relCost = relationalPackage.getColumn().createColumn();
     relCost.setName("Cost");
     relCost.setOwner(salesFactTable);

     // Create the population fact table:
     Table popFactTable = relationalPackage.getTable().createTable();
     popFactTable.setName("Population Fact Table");

     // Create the columns:
     Column relpopTimeId = relationalPackage.getColumn().createColumn();
     relpopTimeId.setName("Time");
     relpopTimeId.setOwner(popFactTable);

     Column relpopGeogId = relationalPackage.getColumn().createColumn();
     relpopGeogId.setName("Geography");
     relpopGeogId.setOwner(popFactTable);

     Column relTotalRevenue = relationalPackage.getColumn().createColumn();
     relTotalRevenue.setName("Total Revenue");
     relTotalRevenue.setOwner(popFactTable);

     // Add keys to product table:
     PrimaryKey prodPrimaryKey = relationalPackage.getPrimaryKey().createPrimaryKey();
     prodTable.addOwnedElement(prodPrimaryKey);
     prodPrimaryKey.addFeature(relUPC);

     // Add keys to sales fact table:
     PrimaryKey salesPrimaryKey = relationalPackage.getPrimaryKey().createPrimaryKey();
     salesFactTable.addOwnedElement(salesPrimaryKey);
     salesPrimaryKey.addFeature(relSalesTimeId);
     salesPrimaryKey.addFeature(relSalesGeogId);
     salesPrimaryKey.addFeature(relSalesProdId);

     // Add foreign keys:
     ForeignKey salesToProd = relationalPackage.getForeignKey().createForeignKey();
     salesFactTable.addOwnedElement(salesToProd);
     salesToProd.addFeature(relSalesProdId);
     salesToProd.setUniqueKey(prodPrimaryKey);


     // -----------------------------------------------------------------------
     //  Physical Deployment Models
     // -----------------------------------------------------------------------

     // CREATE DEPLOYMENT GROUP
     DeploymentGroup deploy1 = olapPackage.getDeploymentGroup().createDeploymentGroup();
     dimensionalSchema.addOwnedElement(deploy1);


     // -----------------------------------------------------------------------
     //  Mapping between the Models
     // -----------------------------------------------------------------------

     // MAPPING THE LOGICAL MODEL
     TransformationMap standardProdHierTM = transformationPackage.getTransformationMap().createTransformationMap();

     ClassifierMap standardProdHierCM = transformationPackage.getClassifierMap().createClassifierMap();
     standardProdHierTM.addOwnedElement(standardProdHierCM);

     // Create the featuremaps and assign target and sources:
     FeatureMap stdProdHierFMId = transformationPackage.getFeatureMap().createFeatureMap();
     standardProdHierCM.addFeatureMap(stdProdHierFMId);
     stdProdHierFMId.addTarget(stdId);
     stdProdHierFMId.addSource(catHLRId);
     stdProdHierFMId.addSource(typeHLRId);
     stdProdHierFMId.addSource(brandHLRId);
     stdProdHierFMId.addSource(upcHLRId);

     FeatureMap stdProdHierFMDesc = transformationPackage.getFeatureMap().createFeatureMap();
     standardProdHierCM.addFeatureMap(stdProdHierFMDesc);
     stdProdHierFMDesc.addTarget(stdLongDescription);
     stdProdHierFMDesc.addSource(catHLRLongDescription);
     stdProdHierFMDesc.addSource(typeHLRLongDescription);
     stdProdHierFMDesc.addSource(brandHLRLongDescription);
     stdProdHierFMDesc.addSource(upcHLRLongDescription);

     FeatureMap stdProdHierFMParent = transformationPackage.getFeatureMap().createFeatureMap();
     standardProdHierCM.addFeatureMap(stdProdHierFMParent);
     stdProdHierFMParent.addTarget(stdParent);
     stdProdHierFMParent.addSource(typeHLRparent);
     stdProdHierFMParent.addSource(brandHLRparent);
     stdProdHierFMParent.addSource(upcHLRparent);

     // ...
   }

   /**
    * Demonstrates the CWM use for warehousing. <p>
    *
    * From [1], example of Chapter 7 "Web-Enabled Data Warehouse Model".
    *
    * @throws Exception cannot run tests
    */
   public void createWebEnabledDataWarehouseModel() throws Exception {

     // Create CWM factory and get required packages:
     CWMCompletePackage cwmFactory = new CWMCompletePackage();
     OlapPackage olapPkg = cwmFactory.getOlap();
     javax.olap.serversidemetadata.SchemaClass schemaClass = olapPkg.getSchema();
     javax.olap.serversidemetadata.Schema schema = schemaClass.createSchema();
     schema.setName( "WebDW" );


     // Create Time Dimensions:
     DimensionClass dimensionClass = olapPkg.getDimension();
     Dimension timeDimension = dimensionClass.createDimension();
     timeDimension.setName( "Time" );
     timeDimension.setIsTime( true );

     // Add the Time Dimension to the OLAP Schema:
     Collection<Dimension> dimCol = schema.getDimension();
     dimCol.add( timeDimension );
     timeDimension.setSchema( schema );

     // Get neccessary class factories:
     CorePackage corePkg = cwmFactory.getCore();
     KeysIndexesPackage keysIndexesPkg = cwmFactory.getKeysIndexes();

     AttributeClass attributeClass = corePkg.getAttribute();
     DataTypeClass dataTypeClass = corePkg.getDataType();
     UniqueKeyClass uniqueKeyClass = keysIndexesPkg.getUniqueKey();
     StereotypeClass stereotypeClass = corePkg.getStereotype();
     TaggedValueClass taggedValueClass = corePkg.getTaggedValue();

     // Create the primary key of the Time Dimension using the
     // "surrogate key" pattern. Note that the data types we
     // create here will be reused by other attributes.

     DataType integerType = dataTypeClass.createDataType();
     integerType.setName( "Integer" );

     Attribute keyAttribute = attributeClass.createAttribute();
     keyAttribute.setName( "TimeID" );
     keyAttribute.setType( integerType );
     keyAttribute.setNamespace( timeDimension );
     List<Attribute> featureList = timeDimension.getFeature();
     featureList.add( keyAttribute );

     UniqueKey uniqueKey = uniqueKeyClass.createUniqueKey();
     uniqueKey.setName( "TimeUK" );
     List<Attribute> keyFeatureList = uniqueKey.getFeature();
     keyFeatureList.add( keyAttribute );
     uniqueKey.setNamespace( timeDimension );
     Collection<ModelElement> ownedElements = timeDimension.getOwnedElement();
     ownedElements.add( uniqueKey );

     // Create the "surrogate key" stereotype, as required by
     // the "surrogate key" pattern.
     // Note that this stereotype will be reused by other
     // dimension keys.

     Stereotype surrogateKeyStereotype = stereotypeClass.createStereotype();
     surrogateKeyStereotype.setName( "SurrogateKey" );
     Collection<UniqueKey> extendedElements = surrogateKeyStereotype.getExtendedElement();
     extendedElements.add( uniqueKey );

     TaggedValue surrogateKeyTaggedValue = taggedValueClass.createTaggedValue();
     surrogateKeyTaggedValue.setTag( "encoding" );
     surrogateKeyTaggedValue.setValue( "serial integer" );
     surrogateKeyTaggedValue.setStereotype( surrogateKeyStereotype );

     Collection<TaggedValue> requiredTags = surrogateKeyStereotype.getRequiredTag();
     requiredTags.add( surrogateKeyTaggedValue );

     // Create the non-key attributes and add them to the Time Dimension.
     // Note that the data types created here will be re-used by the
     // other attributes.

     DataType booleanType = dataTypeClass.createDataType();
     booleanType.setName( "Boolean" );
     DataType stringType = dataTypeClass.createDataType();
     stringType.setName( "String" );

     Attribute nonKeyAttribute = attributeClass.createAttribute();

     nonKeyAttribute.setName( "DateType" );
     nonKeyAttribute.setType( stringType );
     nonKeyAttribute.setNamespace( timeDimension );
     featureList.add( nonKeyAttribute );

     nonKeyAttribute = attributeClass.createAttribute();
     nonKeyAttribute.setName( "CalendarType" );
     nonKeyAttribute.setType( stringType );
     nonKeyAttribute.setNamespace( timeDimension );
     featureList.add( nonKeyAttribute );

     nonKeyAttribute = attributeClass.createAttribute();
     nonKeyAttribute.setName( "DayOfWeek" );
     nonKeyAttribute.setType( integerType );
     nonKeyAttribute.setNamespace( timeDimension );
     featureList.add( nonKeyAttribute );

     nonKeyAttribute = attributeClass.createAttribute();
     nonKeyAttribute.setName( "DayNumberInWeek" );
     nonKeyAttribute.setType( integerType );
     nonKeyAttribute.setNamespace( timeDimension );
     featureList.add( nonKeyAttribute );

     nonKeyAttribute = attributeClass.createAttribute();
     nonKeyAttribute.setName( "DayNumberInMonth" );
     nonKeyAttribute.setType( integerType );
     nonKeyAttribute.setNamespace( timeDimension );
     featureList.add( nonKeyAttribute );

     // ...

     nonKeyAttribute = attributeClass.createAttribute();
     nonKeyAttribute.setName( "Workday" );
     nonKeyAttribute.setType( booleanType );
     nonKeyAttribute.setNamespace( timeDimension );
     featureList.add( nonKeyAttribute );

     // ...

     nonKeyAttribute = attributeClass.createAttribute();
     nonKeyAttribute.setName( "Second" );
     nonKeyAttribute.setType( integerType );
     nonKeyAttribute.setNamespace( timeDimension );
     featureList.add( nonKeyAttribute );

     nonKeyAttribute = attributeClass.createAttribute();
     nonKeyAttribute.setName( "Timespan" );