6 : (FrameNet) [2] [3] [4] [5] [6] (ChineseFrameNetCFN) % 1520 ; [7] 3151 ( ) ; FrameNet [8] atract(

Size: px

Start display at page:

Download "6 : (FrameNet) [2] [3] [4] [5] [6] (ChineseFrameNetCFN) % 1520 ; [7] 3151 ( ) ; FrameNet [8] atract("

育衾唐
7 years ago
Views:

1 JOURNAL OFCHINESEINFORMATIONPROCESSING Vol.25No.6 Nov.2011 : (2011) ( ) : (ChineseFrameNetCFN) Filmore FrameNet 25 F %52.70%61.62%; : ; ; :TP391 :A ResearchonChineseFrameNetConstructionandApplicationtechnologies LIU Kaiying (SchoolofComputerandInformationTechnologyShanxiUniversityTaiyuanShanxi030006China) Abstract:ChineseFrameNetisancomputationallexicalsemanticdatabasewhichisbasedontheframesemanticsby FilmorereferringtotheFrameNetbytheCaliforniaUniversityatBerkeleyandsupportedbyChineserealcopus. ThisarticleintroducestheCFN basictheory:framesemanticsandtheenglishframenetproject.thenitanalyzes theconstructingtechnologyintroducestheongoingresearchesonautomaticsemanticannotationbasedoncfn.the experimentalresultsonal25framesdatafortheprecisiontherecalandf1-valuereached74.16%52.70% 61.62% respectively.atlastitintroducesthesituationofsomeresearchesbasedoncfn. Keywords:ChineseFramenet;semanticrolelabeling;framesemanticdependencygraph (Frame Ele- 1 ments) (FrameSemantics) Filmore (Heating-instrument) [1] (Frame) (MotivatingContext) : : (Cook) (Food) (Container) : ( ); 863 (2006AA01Z142); (07CYY022) : (1931 )

2 6 : (FrameNet) [2] [3] [4] [5] [6] (ChineseFrameNetCFN) % 1520 ; [7] 3151 ( ) ; FrameNet [8] atract( ) cast( ) catapult( ) drag() (CauseMotion)9 : (Agent) (Area) (Cause) (Goal) (Initial State) (Path) (Result) (Source) (Theme);11 : (Degree) (Depictive) (Distance) (Explanation) (Handle) (Instrument) (Manner) (Means) (Place) (Subregion) (Time) 8 ;(3) ;(4) 1589 FrameNet (American NationalCorpusANC) : (Multiply- CFN [9] (1)CFN :(1) ;(2) 1 a)cfn FN AnnotatedSub-Corpus ) : v 2 CFN

3 CFN : <1> vvvvv v v CFN CFN v v v <1> ( ) <2> (Ent) (At) (Val1) (Val2) (Inis)(Finis) (Dif) (Val ran) (cir) (Cor) (Dur) (Cor1) (Cor2) 獉獉 (Manr) (Path) (Place) (Speed) (Time) 獉獉 : : : :[ ] : :[ ] : v v v v v v v vvv vv :[ ][ ] v v <2> v v v v v d) ( ) b) : (2) ; CFN : CCL 1 F2000 獉獉 2 獉獉 1 ( ) 10~20 2 ( 2 1 ) CFN : <at-np-subj > <tgt=[ ; ] > <diff-np-obj > tgt [ ] ;at ( 1)np subj c) FN CFN ;

4 6 : % F1-71.9% F1- PropBank : 50 6 Arg0-Arg5 [13] : PropBank (3) (24) np.dlc(1);np.subj s(1) np.obj(11);dj.obj(4);dni(2); ini(3)np.ext(2);np.subj (1); vp.obj(1) (3) sp.obj(2);pp[].adva(1) (2) dj.comp(1);ap.adva(1) (6) tp.adva (4);tp.subj s(1);tp.adva s(1) (1) pp[].adva(1) (1) np.obj(1) [18] CFN CFN 80.1% CFN 69.9% [19-21] (CRF) : 3 (CRF) [10] [11] PropBank [12] PropBank Arg2Arg5 PropBank 2 [14] 2 PropBank [15]PropBank (28) np.subj(9);np.ext(11);ini(4); np.head (1);pp[].ext (1); CRF [16] PropBank [17] IOB Senseval-3 FrameNet

5 mp mr mf std(mf) CRF++ Source- forge(htp://crfpp.sourceforge.net/) 63.04% 60.20% 61.58% CRF CRFL2 C=1 CRF 71.67% 55.27% 62.41% SVM 25 4 ; 61.36% 54.59% 57.78% fold ME 32-fold 3 F % 61.03% 62.61% CRF % 54.84% 62.45% CRFSVM mp mr mf std(mf) 72.60% 49.50% 58.86% % 56.44% 59.48% % 50.73% 60.19% % 57.25% 60.42% % 68.08% 71.68% % 58.07% 59.23% ME Stanford 71.70% 45.46% 55.64% [12] ME 5 mp 1) ;2) ;3) ;mr ; mf F1-;mA 43 2-fold ;std(mf) 3 F1- mf std(mf)= 槡 Var(mF ) (CRF) (SVM) CRF :(1) CRF (ME) ;(2) 25 ; (mf)71.68%; (ma)84.08%; (3) ( ) (mf) CRF 63.26% (SVM) (ME) 4 CFN (CRF) 4.1 4

6 6 : 51 ( ) 4.3 [26] CFN :(1) (Lexical UnitDetection)(2) (Unknown FrameDetection)(3) (FrameDisambigu- ( ation) [22] ) fold % [23] HumSent T-CRF % % [27] [24-25] % ; 1 :

7 ed.multilingualframenetsincomputationallexicog- raphy.methodsand Applications.Berlin/New York: MoutondeGruyter2009: [4] BoasHansC.BilingualFrameNetDictionariesfor Ma- chineranslation[c]//proceedingsofthethirdinterna- : tionalconferenceonlanguageresourcesandevalua- tion.eds.gonzlezm.rodrguezandpazsurezc. Araujo.Vol.IV.2002: [5] OharaKyoko HiroseSeikoFujietal.TheJapa- neseframenetproject:a [28] Preliminary Report[C]// ProceedingsofPacific AssociationforComputational Linguistics(PACLING 03)2003: [6] SalomoMaria M M.FrameNetBrasil:Umtrabalho emprogresso[j].calidoscpio20097:3. [7] YouLPLiuK Y.BuildingChineseFrameNetData base[c]//proceedings of 2005 IEEE International OwL Conference on Natural Language Processing and KnowledgeEngineering (IEEE NLP-KE)2005:301- : 306. eclipse3.3 Jena2.3 [8]. [D] [9]. (CFN) [R]. CFN [10] : (LOC)72%Sun HLJurafsky D.Shalow semanticparsingof Chinese [C]//Hirschberg JB ed. Proceedings of (TIME)73% (MEANS)62% NAACL-HLT2004.Boston:ACL2004: [11] XueNWPalmer M.Automaticsemanticrolelabe- lingforchineseverbs[c]//bramer M ed.proceed- 863 ingsofthenineteenthinternationaljointconference on ArtificialInteligence.Edinburgh:IJCAI2005: [12] XueNW.LabelingChinesepredicateswithsemantic roles[j].computationallinguistics200834(2): FrameNet [13]. [J] (3):3-15. [14] CohnTBlunsom P.Semanticrolelabelingwithtree FrameNet conditionalrandom fields[c]//knight KNg HT OflazerKeds.ProceedingsofCoNLL2005.Ann ColinF.Baker FrameNet Arbor:ACL2005: [15]. [C]// 2006: [1] Filmore.Frame semantics[c]//linguisticsin the MorningCalm.1982: [2] Baker CF Filmore CJ Lowe JB. The Berkeley FrameNetproject[C]//ProceedingsoftheCO LING- ACL 98.Montreal:ACLPress1998: [3] SubiratsCarlos.SpanishFramenet:Aframe-seman- ticanalysisofthespanishlexicon[c]//hans Boas [16] YuJDFanXPang Wetal.Semanticrolelabeling basedonconditionalrandom fields[j].journalof Southeast University(English Edition) (3): [17]. [C]// 2007: ( 62)

8 encyparsing[c]//proceedingsofthe2011 Confer- enceonempiricalmethodsinnaturallanguagepro- cessing.edinburghscotlanduk : [26] WanxiangCheTingLiu.Jointly Modeling WSDand SRL with MarkovLogic[C]//Proceedingsofthe23rd InternationalConferenceon ComputationalLinguis- tics (Coling2010).BeijingChina : [27] WanxiangCheTingLiuShengLi.A New Chinese NaturalLanguageUnderstanding ArchitectureBased on Multilayer Search Mechanism [C ]//Third SIGHAN Workshopon Chinese Language Process- ing [28]. [C]// 櫚櫚櫚櫚櫚櫚櫚櫚櫚櫚櫚櫚櫚櫚櫚櫚櫚櫚櫚櫚櫚櫚櫚櫚櫚櫚櫚櫚櫚櫚櫚櫚櫚櫚櫚櫚櫚櫚櫚櫚櫚櫚櫚櫚櫚櫚櫚櫚櫚 ( 52) [18]. Conference on Computational Linguistics (Coling [C]// 2010): : [19]. [J] [D] [20]. [25]. [J]. (JournalofSoftware) (4): [21] Jihong LIRuibo WANGWeilin WANG Bo GU GuochenLI.AutomaticLabelingofSemanticRoleon ChineseFrameNetUsingConditionalRandom Fields [C]//2009 IEEE/WIC/ACM International Confer- enceon WebInteligenceandInteligentAgentTech- nology-workshops. [22]. [28]. [J] (3): [24]. 24(1): [J] (4): [26]. [J] (6): [27] Ru LiShuanghongLi. The Semantic Computing Model of Sentence Similarity Based on Chinese FrameNet[C]//Web Inteligence/IAT Workshops 2009: [J] (2): [23] RuLiHaijingLiuShuanghongLi.ChineseFrameI- dentificationusing T-CRF Model[C]//International

GATE(General Architecturefor Text Engi- neering) 1 Java [1] GATE (2) GATE GATE UIMA (Unstructured Information Manage- mentarchitecture) 2 [2]

GATE(General Architecturefor Text Engi- neering) 1 Java [1] GATE (2) GATE GATE UIMA (Unstructured Information Manage- mentarchitecture) 2 [2] 25 6 201111 JOURNAL OFCHINESEINFORMATIONPROCESSING Vol.25No.6 Nov.2011 :1003-0077(2011)06-0053-10 ( 150001) : LTP(LanguageTechnologyPlatform) CoNLL2009 2006 2010 400 20116 : ; :TP391 :A LanguageTechnologyPlatform