主觀題自動(dòng)評(píng)分技術(shù)研究.pdf

1、華中科技大學(xué)碩士學(xué)位論文主觀題自動(dòng)評(píng)分技術(shù)研究姓名：郭扉扉申請(qǐng)學(xué)位級(jí)別：碩士專業(yè)：工業(yè)工程指導(dǎo)教師：尹文生20110112華中科技大學(xué)碩士學(xué)位論文華中科技大學(xué)碩士學(xué)位論文IIAbstractAslesDickenssaidinhisnovelTheTaleofTwoCities“Itwasthebestoftimesitwasthewstoftimes”.Lifeofusisfloodedwithalargeamountofinfmat

2、ionsohowtochoosevaluableinfmationisbecomingmemeimptantfeveryoneatanytimewhenwefacetheInter.Thusmemepeoplepayattentiontotheresearchonartificialintelligencedataminingnaturallanguageunderstingetc.Naturallanguageprocessingte

3、chnologycombinedcomputersciencetechnologyartificialintelligencetechnologywasuniversallyappliedininfmationretrievalfieldatfirst.Baseoninfmationretrievaltechnologysomekeyissuesaboutnaturallanguageunderstingwillbestudieddis

4、cussedinthispaperappliedtosubjectiveautomatedassessmentfieldtoobtainsomedesireresults.Thedevelopmentapplicationofknowledgeretrievaltechnologyisintroducedfirstlythenthepresentresearchstatusofsubjectiveautomatedassessmentt

5、echnologyisdescribedafterwards.Theanalysessummarizationofthedevelopmentientationofnaturallanguageprocessingtechnologymakeagoodfoundationfsubsequentresearch.Thepaperincludesthreeaspects.ThefirstaspectisaboutChinesewdsegme

6、ntation.Adictionarybasedonallacterindexstructureisproposed.AHashtableisfirstlyestablishedfallcommonlyusedactersineachphrasesothedictionarycansupptnonfirstacterqueryfuzzyquery.Theconstructionmaintenanceofthedictionaryissi

7、mpleeasytorealize.TakingWdSenseCodeasdataresourcetoconstructallacterindexingdictionarythetestindicatesthattheChinesewdsegmentationalgithmbasedonthisdictionarycanseparatephrasescrectly.ThesecondaspectisaboutLatentSemantic

8、Analysis(LSA)itsapplication.SomeconcretetestsshowthatthismethodLSAhasthefunctionofextractingrepresentingtheimplicitsemanticofahighdermatrixintheassociationoftermswithdocumentsbyusingalowdermatrix.LSAwksbyusingSingularVal

9、ueDecomposition(SVD)whichisawayinLinearAlgebrabyusingstatisticalcomputationappliedtoalargecpusoftext.UnftunatelyperfmancedatashowstheeffectisnotdesirablewhenLSAisappliedtosubjectiveautomatedassessment.Itisbecausethereisn