Avirup rup Sil Silv lviu iu Cuce cerza zan T emp mple e Univer versi sity, , Philadel ladelphi phia Micr croso soft Re Rese sear arch ch avi@tem empl ple.edu e.edu si silviu@m iu@mic icros osoft.c .com
Introduction to the T emporal Slot Filling T ask Our Approach Gathering Training Data from Wikipedia Relationship Classifier Date Classifier Experiments Conclusion and Future Work
“ Bill Clinton, the forty-second president of the US, was the first to pay down principle..” Output of Relation Extraction systems [Etzion , 00] : tzioni et. al, , 05, Agich ichstein tein & & Grava avano, President_of(Bill Clinton, United States) Limitation: Does not capture temporal validity of the relationship ▪ President_of(Bill Clinton, USA) is true during time-frame 1993-2001
In Input ut: A binary relation ▪ Example: spouse(Brad Pitt, Jennifer Aniston) A document supporting the relation Outp tput ut: A 4-tuple timestamp [T1, T2, T3, T4] ▪ [2000-07-29,nil, nil, 2005-10-02] A sentence supporting the temporal validity of the relation ▪ “ Pitt married Jennifer Aniston on July 29, 2000… the couple divorced five years later in October 2, 2005. ”
T ext Analysis Conference (TAC): T emporal Slot Filling track has the following relation types: 1. Spouse Brad Pitt: Jennifer Aniston 2. Title Barack Obama: President 3. Employee Of Carol Bartz: Yahoo! Inc. 4. Cities of Residence Arturo Gatti: Montreal 5. States/Provinces of Residence Michael Vick: Virginia 6. Countries of Residence Josh Fattal: Iran 7. T op Employees/Members Microsoft: Steve Ballmer Query y Entit ity Slot Filler er
Introduction to the T emporal Slot Filling T ask Our Approach Gathering Training Data from Wikipedia Relationship Classifier Date Classifier Experiments Conclusion and Future Work
No training data available We build our own training data from Wikipedia sentences For every relation: ▪ Extract Slot-Filler Names from Infoboxes from all Wikipedia pages ▪ Apply MSR Entity Linker to resolve entity disambiguation and coreferences ▪ Collect sets of contiguous sentences containing the slot-filler names ▪ Build a language model by bootstrapping [Ag Agic icht htein in & & Gravano no, , Sp Spouse: Katie Holmes 00] textual patterns supporting the relations
Wikipe ipedia dia Sentence nces: s: No training data available On October 6, 2005, Cruise and We build our own training data from Holmes announced they were expecting a child.. Wikipedia sentences … On November 18, 2006, Holmes and Cruise were married at the For every relation: 15th-century Odescalchi Castle in ▪ Extract Slot-Filler Names from Infoboxes from all Bracciano, Italy… Wikipedia pages On June 29, 2012, it was announced that Holmes had filed for divorce ▪ Apply MSR Entity Linker to resolve entity from Cruise after five and a half disambiguation and coreferences years of marriage. ▪ Collect sets of contiguous sentences containing the slot-filler names ▪ Build a language model by bootstrapping [Ag Agic icht htein in & & Gravano no, , 00] textual patterns supporting the relations
No training data available Patterns Extracted: • DATE: X and Y were expecting a We build our own training data from child Wikipedia sentences • DATE: X and Y were married For every relation: • DATE: X had filed for divorce from Y ▪ Extract Slot-Filler Names from Infoboxes from all Wikipedia pages • … ▪ Apply MSR Entity Linker to resolve entity disambiguation and coreferences X==Query Entity Y== Slot Filler ▪ Collect sets of contiguous sentences containing We extract up to 5-grams. the slot-filler names ▪ Build a language model by bootstrapping [Ag Agic icht htein in & & no, 00] textual patterns supporting the relations Gravano
We run Stanford SUTime [Chang & Manning, 12] to resolve date surface forms Raw Input Document ument: <DOC id="AFP_ENG_20090626.0737" type="story" > <HEADLINE>Distraught Madonna 'can't stop crying' over Jackson</HEADLINE> <DATELINE>Los Angeles, June 25, 2009 (AFP)</DATELINE> <TEXT><P>Pop diva Madonna revealed she was left in tears over the death of Michael Jackson on Thursday, saying the music world had lost ..</P> </TEXT> </DOC> Docum ument ent normaliz malized ed with Timestamps: stamps: <DOC id="AFP_ENG_20090626.0737" type="story" > <HEADLINE>Distraught Madonna 'can't stop crying' over Jackson</HEADLINE> <DATELINE>Los Angeles, June 25, 2009 (AFP)</DATELINE> <TEXT><P>Pop diva Madonna revealed she was left in tears over the death of <TIMEX3 t0=“2009 -06- 25”> Thursday </TIMEX3> Michael Jackson on Thursday, saying the music world had lost ..</P> </TEXT> </DOC>
Training: Example: ▪ Query Entity (X): T om Cruise; Slot Filler (Y): Katie Holmes ▪ Sentence 1: “ On November 18, 2006, Holmes and Cruise were married in Bracciano, Italy... ” ▪ Sentence 2: “ In 2003, Cruise starred in the historical drama The Last Samurai.. ” Features es X and Y were Y, who o died in in were married d in .. .. X X X’s wife Y Y, who o died married Label married DATE LOC married d in DATE Sentence 1 0 1 .. 0 0 0 1 +1 +1 1 Sentence 0 0 0 .. 0 0 0 0 -1 Spouse se: Katie Holmes 2 Classifier: Boosted Decision Trees [Burges, 2010]
T esting: TAC TSF Eval Docume ment <DOC id="NYT_ENG_20101121.0120" type="story" > Example: <HEADLINE>NORRIS CHURCH MAILER, ARTIST AND WRITER, DIES AT 61</HEADLINE> ▪ Query Entity: Norris Church <TEXT> <P>Norman Mailer, whom Norris married in ▪ Slot Filler: Norman Mailer 1980, was an attentive father..</P> <P>Norman Mailer, who died in 2007 at 84, who dreamed up Church because he..</P> <P>Norris gave birth to John Buffalo in 1978 and spent..</P>
T esting: TAC TSF Eval Docume ment <DOC id="NYT_ENG_20101121.0120" type="story" > Example: <HEADLINE>NORRIS CHURCH MAILER, ARTIST AND WRITER, DIES AT 61</HEADLINE> ▪ Query Entity: Norris Church <TEXT> <P> Y , whom X married in _ DATE , was an attentive ▪ Slot Filler: Norman Mailer father..</P> <P> Y , who died in _DATE at 84, who dreamed up X because he..</P> <P> X gave birth to John Buffalo in _DATE TE and spent..</P>
T esting: TAC TSF Eval Docume ment <DOC id="NYT_ENG_20101121.0120" type="story" > Example: <HEADLINE>NORRIS CHURCH MAILER, ARTIST AND WRITER, DIES AT 61</HEADLINE> ▪ Query Entity: Norris Church <TEXT> <P> Y , whom X married in _ DATE , was an attentive ▪ Slot Filler: Norman Mailer father..</P> <P> Y , who died in _DATE at 84, who dreamed up X because he..</P> <P> X gave birth to John Buffalo in _DATE TE and spent..</P> Features X and Y were Y, who o died in in were married d in .. .. X married d in X’s wife Y, who o married married DATE LOC DATE Y died Sentence 0 0 0 .. .. 1 0 0 1 1 Sentence 0 1 0 .. 0 0 1 0 2 Sentence 0 0 0 .. 0 0 0 0 3
T esting: TAC TSF Eval Docume ment <DOC id="NYT_ENG_20101121.0120" type="story" > Example: <HEADLINE>NORRIS CHURCH MAILER, ARTIST AND WRITER, DIES AT 61</HEADLINE> ▪ Query Entity: Norris Church <TEXT> <P> Y , whom X married in _ DATE , was an attentive ▪ Slot Filler: Norman Mailer father..</P> <P> Y , who died in _DATE at 84, who dreamed up X because he..</P> <P> X gave birth to John Buffalo in _DATE TE and spent..</P> Features X and Y were Y, who o died in in were married d in .. .. X married d in X’s wife Y, who o married married DATE LOC DATE Y died Sentence 0 0 0 .. 1 0 0 1 1 Sentence 0 1 0 .. .. 0 0 1 0 2 Sentence 0 0 0 .. 0 0 0 0 3
T esting: TAC TSF Eval Docume ment <DOC id="NYT_ENG_20101121.0120" type="story" > Example: <HEADLINE>NORRIS CHURCH MAILER, ARTIST AND WRITER, DIES AT 61</HEADLINE> ▪ Query Entity: Norris Church <TEXT> <P> Y , whom X married in _ DATE , was an attentive ▪ Slot Filler: Norman Mailer father..</P> <P> Y , who died in _DATE at 84, who dreamed up X because he..</P> <P> X gave birth to John Buffalo in _DATE TE and spent..</P> Features X and Y were Y, who o died in in were married d in .. .. X married d in X’s wife Y, who o married married DATE LOC DATE Y died Sentence 0 0 0 .. 1 0 0 1 1 Sentence 0 1 0 .. 0 0 1 0 2 Sentence 0 0 0 .. .. 0 0 0 0 3
Recommend
More recommend
