Progress Report090806
Download as PPTX, PDF0 likes520 views
The document describes an experiment using a pair method to determine if two XML files match. It involves: 1. Analyzing two XML files to extract features for each file. 2. Combining the feature representations and using them to train a libSVM classifier on other file pairs to predict if they match or not. 3. Checking the classifier's predictions - if it predicts a pair matches, output the attribute number of the files.
1 of 9
Download to read offline
More Related Content
What's hot (15)
Ad
Viewers also liked (8)
Ad
Similar to Progress Report090806 (20)
Progress Report090806
- 1. ExperimentResult for Pair Method2009/08/06Chi-I Kuan1
- 2. Examinationflow2Combine Files as Training DataXMLFeature AnalyzerTraining Xml File1representationXMLFeature AnalyzerTraining Xml File2representationlibSVMClassifierTesting Xml File1XMLFeature AnalyzerCombine Filesas Training DatarepresentationModelXMLFeature AnalyzerTesting Xml File2representationDecideMapping
- 3. Training and Testing DataCombine two analyzed xml files as input.We has 6 bookstores’ xml file , each time we choose 2 xml file as testing data and use the remaining 4 xml files as training data.3
- 4. XMLAnalyzerIsDateIsTimeIsFileIsHyperLinkIsOnlyTextIsNotTextIsMixTextHasOnlyDigitHasMeasurementHasUpperWordLastIsDateLastIsTimeLastIsFileLastIsHyperLinkLastIsOnlyTextLastIsNotTextLastIsMixTextLastHasOnlyDigitLastHasMeasurementLastHasUpperWord4Calculate xml files’ featureRepresentationFor each xml file, takemost frequently used feature to represent all instances’ feature.5Many instancesOne representation instance
- 5. CombineCombine two xml analyzed files as a pair.6
- 6. ClassifierUse libSVM to predict whether they are matching or not.7
- 7. Decide MappingCheck classifier’s prediction result , if the vector is prediction “Match” ,then output the vector’s attribute number as result. 8output