Txt/Hyp term similarities

The rows are the txt words. The columns are hyp words.
Resource summary:
Acronym: AcronymLexicalResource
BasicWN: BasicWNLexicalResource
Country: CountryLexicalResource
Cyc: null
DekangLin: DekangLinLexicalResource
Google: null
InfoMap: InfoMapLexicalResource
Morpho: MorphoLexicalResource
NomBank: NomBankLexicalResource
Number: NumberLexicalResource
Ordinal: OrdinalLexicalResource
Preposition: PrepositionLexicalResource
Ravichandran: RavichandranLexicalResource
LeacockChodorowWN: null
JiangConrathWN: JiangConrathWNLexicalResource
ResnikWN: ResnikWNLexicalResource
LinWN: LinWNLexicalResource
HirstWN: null
AdaptedLeskWN: null
EdgeWN: null
StringSim: StringSimLexicalResource
WebLexicalEntailment: null
CorpusLexicalEntailment: CorpusLexicalEntailmentResource


Inference ID: PARC-7

Txt: The technician cooled the room.

Hyp: The technician did lower the temperature of the room . (yes)

The
DT		 technician
NN		 did
VBD		 lower
JJR		 the
DT		 temperature
NN		 the
DT		 room
NN		 .
.
The:DT   0.00 19.96 17.13 18.47   0.00 19.38   0.00 19.31   0.00
technician:NN 10.00   0.00 11.80   8.40 10.00   8.15 10.00   8.00 10.00
cooled:VBD 10.00 13.08   5.69   6.31 10.00   9.14 10.00 12.76 10.00
the:DT   0.00 19.96 17.13 18.47   0.00 19.38   0.00 19.31   0.00
room:NN 10.00   8.71   9.36   8.31 10.00   6.97 10.00   0.00 10.00
.:.   0.45 19.97 17.26 18.53   0.45 19.41   0.45 19.15   0.00
NO_WORD   1.00 10.00 10.00   9.00   1.00 10.00   1.00 10.00 10.00

Response: yes (CORRECT)
Justification:
Features matched (val wt name just):
 1.00 -3.08 Alignment.score
 1.00  0.11 Alignment.isGood
-1.00  0.05 Alignment.isBad
 0.00  3.00 Alignment.nbWordNotAligned
 0.00  3.00 Alignment.hypSpan
 0.00  0.67 Alignment.txtSpan
-1.00  1.00 Adjunct.addPosCxt : hyp added lower[lower-JJR]
-0.10  1.00 NullPunisher.article : the
-1.00  1.00 NullPunisher.other : temperature
-1.00  1.00 NullPunisher.other : lower
-2.00  1.00 RootEntailment.poorlyAlignedRoot : "did" aligned badly to "cooled"
 1.00 -1.67 GlickmanLexicalEntailment.score
Hand-tuned score (dot product of above): -9.7867
Threshold: -11.4590

Inference ID: PARC-8

Txt: The technician raised the temperature of the room.

Hyp: The technician did cool the room . (no)

The
DT		 technician
NN		 did
VBD		 cool
VB		 the
DT		 room
NN		 .
.
The:DT   0.00 19.96 17.13 19.73   0.00 19.31   0.00
technician:NN 10.00   0.00 11.80 13.08 10.00   8.00 10.00
raised:VBD 10.00 13.95   3.91   6.95 10.00 13.52 10.00
the:DT   0.00 19.96 17.13 19.73   0.00 19.31   0.00
temperature:NN 10.00   8.15 12.04   9.59 10.00   6.97 10.00
the:DT   0.00 19.96 17.13 19.73   0.00 19.31   0.00
room:NN 10.00   8.71   9.36 11.00 10.00   0.00 10.00
.:.   0.45 19.97 17.26 19.74   0.45 19.15   0.00
NO_WORD   1.00 10.00   1.00 10.00   1.00 10.00 10.00

Response: yes (INCORRECT)
Justification:
Features matched (val wt name just):
 1.00 -1.42 Alignment.score
 1.00  0.40 Alignment.isGood
-1.00  0.01 Alignment.isBad
 0.00  1.00 Alignment.nbWordNotAligned
 0.00  4.00 Alignment.hypSpan
 0.00  0.43 Alignment.txtSpan
-0.05  1.00 NullPunisher.aux : did
-2.00  1.00 RootEntailment.poorlyAlignedRoot : "cool" aligned badly to "raised"
-3.00  1.00 Structure.parentsMismatch : args have different parents, different relations: text "room" <-prep_of-- "temperature" vs. hyp "room" <-dobj-- "cool", which aligned to text "raised"
 1.00 -2.19 GlickmanLexicalEntailment.score
-3.00  1.00 Structure.parentsMismatch&Align.veryGood :
Hand-tuned score (dot product of above): -11.2782
Threshold: -11.4590

Inference ID: PARC-100

Txt: Bin Laden was last seen in the house.

Hyp: Osama_Bin_Laden is WH[LOCATION] . (in the house)

Osama_Bin_Laden
NNP		 is
VBZ		 WH[LOCATION]
JJ		 .
.
Bin_Laden:NNP   0.50   5.50   0.00 10.50
was:VBD 15.00   0.00 17.50 10.00
last:RB 15.00 10.23 17.50 10.00
seen:VBN 15.00   0.05 17.50 10.00
the:DT 20.00 10.96 17.50   0.00
house:NN 10.00   5.39   0.00 10.00
.:. 20.00 11.33 17.50   0.00
NO_WORD 10.00   1.00   9.00 10.00

Response: dontknow (INCORRECT)
Justification:
Features matched (val wt name just):
 1.00 -1.38 Alignment.score
 1.00  0.41 Alignment.isGood
-1.00  0.01 Alignment.isBad
 0.00  1.00 Alignment.nbWordNotAligned
 0.00  2.00 Alignment.hypSpan
 0.00  0.50 Alignment.txtSpan
 1.00  1.00 Factive.factivePassage : factive entails : seen-VBN
-0.05  1.00 NullPunisher.aux : is
-1.00  1.00 Structure.relMismatch : text "the" is det of "house" while hyp "." is punct of "WH[LOCATION]" which aligned to text "house"
 1.00 -27.63 GlickmanLexicalEntailment.score
Hand-tuned score (dot product of above): -28.6584
Threshold: -11.4590

Inference ID: PARC-101

Txt: Bin Laden was fondling an AK-47.

Hyp: WH[] was Osama_Bin_Laden holding . (an AK-47)

WH[]
NNP		 was
VBD		 Osama_Bin_Laden
VBN		 holding
VBG		 .
.
Bin_Laden:NNP   0.00   5.50 10.50 15.50 10.50
was:VBD   0.00   0.00   9.96   9.21 10.00
fondling:VBG   0.00 10.00   9.96   8.52 19.59
an:DT 20.00 11.81 20.00 19.27   0.00
AK:NNP   0.00   7.17 15.46 14.07 10.50
-47:CD   0.00 10.50 20.46 19.94 10.50
.:. 20.00 12.79 20.00 19.41   0.00
NO_WORD 10.00   1.00 10.00 10.00 10.00

Response: dontknow (INCORRECT)
Justification:
Features matched (val wt name just):
 1.00 -4.19 Alignment.score
 1.00  0.04 Alignment.isGood
-1.00  0.14 Alignment.isBad
 0.00  2.00 Alignment.nbWordNotAligned
 0.00  1.00 Alignment.hypSpan
 0.00  0.60 Alignment.txtSpan
 0.50  1.00 Adjunct.dropPosCxt : text adjunct "-47" of "AK" dropped on aligned hyp word "WH[]"
-0.05  1.00 NullPunisher.aux : was
-1.00  1.00 NullPunisher.other : holding
-2.00  1.00 RootEntailment.poorlyAlignedRoot : "Osama_Bin_Laden" aligned badly to "fondling"
 1.00 -20.75 GlickmanLexicalEntailment.score
Hand-tuned score (dot product of above): -27.5978
Threshold: -11.4590

Inference ID: PARC-102

Txt: Bin Laden was fondling an AK-47.

Hyp: Which gun was Osama_Bin_Laden holding ? (the AK-47)

Which
WDT		 gun
NN		 was
VBD		 Osama_Bin_Laden
VBN		 holding
VBG
Bin_Laden:NNP 10.50 10.50   5.50 10.50 15.50
was:VBD 11.39 14.34   0.00   9.96   9.21
fondling:VBG 20.00 12.55 10.00   9.96   8.52
an:DT   1.53 19.65 11.81 20.00 19.27
AK:NNP 11.21   8.83   7.17 15.46 14.07
-47:CD 10.50 15.50 10.50 20.46 19.94
.:.   2.67 19.17 12.79 20.00 19.41
NO_WORD 10.00 10.00 10.00 10.00 10.00

Response: dontknow (INCORRECT)
Justification:
Features matched (val wt name just):
 1.00 -6.71 Alignment.score
 1.00  0.00 Alignment.isGood
-1.00  0.67 Alignment.isBad
 0.00  3.00 Alignment.nbWordNotAligned
 0.00  1.00 Alignment.hypSpan
 0.00  0.40 Alignment.txtSpan
-1.00  1.00 NullPunisher.other : Osama_Bin_Laden
-1.00  1.00 NullPunisher.other : holding
-1.00  1.00 NullPunisher.other : gun
-3.00  1.00 Structure.parentsMismatch : args have different parents, different relations: text "an" <-det-- "AK" vs. hyp "Which" <-dobj-- "was", which aligned to text "was"
 1.00 -15.07 GlickmanLexicalEntailment.score
Hand-tuned score (dot product of above): -28.4402
Threshold: -11.4590

Inference ID: PARC-103

Txt: Some 8,000 US troops will be moved from the Japanese island of Okinawa, with other bases earmarked for closure.

Hyp: There are over 5000 U.S. troops in Okinawa currently . (yes)

There
EX		 are
VBP		 5000
NNP		 U.S.
NNP		 troops
NNS		 Okinawa
NNP		 currently
RB		 .
.
Some:DT   2.28 11.00 19.84 18.06 19.15 20.48 17.89   0.00
8,000:CD 12.76 11.15 15.50 17.60 18.89 20.46 18.56 10.50
US:NNP 12.09   5.99   9.80   0.50   8.69   9.84 12.23 10.00
troops:NNS 11.88   6.38   8.98   6.05   0.00 10.36 12.58 10.00
will:MD   1.73 10.57 19.83 17.60 19.03 20.46 17.37   0.00
be:VB 12.77   0.00 14.85 13.11 14.13 14.84 18.11 10.00
moved:VBN 11.54   1.20 13.85 11.97 13.43 13.22 17.09 10.00
the:DT   4.05 12.49 19.88 18.52 19.30 20.47 18.46   0.00
Japanese:JJ 12.93   4.03 12.32   8.01 11.06 11.45 10.31 10.50
island:NN 12.37   6.46   9.77   6.80   8.38   3.96 12.44 10.00
Okinawa:NNP 12.50   7.00 10.46   3.00   7.00   0.00 13.00 10.50
,:,   4.32 12.81 19.89 18.61 19.33 20.47 17.97   0.00
other:JJ 12.81   3.31 11.85 10.11 11.20 12.46 10.00 10.00
bases:NNS 12.57   5.75   9.71   6.64   5.10   9.45 11.89 10.00
earmarked:VBN 10.00   0.00 14.87 15.50 11.00 15.50 18.00 10.00
closure:NN 11.25   5.77   9.90   7.52   7.22   8.85 11.25 10.00
.:.   4.32 12.81 19.89 18.61 19.33 20.47 18.52   0.00
NO_WORD   1.00 10.00 10.00 10.00 10.00 10.00   9.00 10.00

Response: yes (CORRECT)
Justification:
Features matched (val wt name just):
 1.00 -3.21 Alignment.score
 1.00  0.10 Alignment.isGood
-1.00  0.06 Alignment.isBad
 0.00  3.00 Alignment.nbWordNotAligned
 0.00  3.00 Alignment.hypSpan
 0.00  0.25 Alignment.txtSpan
-1.00  1.00 Adjunct.addPosCxt : hyp added currently[currently-RB]
 0.50  1.00 Adjunct.dropPosCxt : text adjunct "8,000" of "troops" dropped on aligned hyp word "troops"
-1.00  1.00 NullPunisher.other : currently
-0.10  1.00 NullPunisher.functionWord : There
-1.00  1.00 NullPunisher.other : 5000
-1.00  1.00 Structure.relMismatch : text "troops" is nsubjpass of "moved" while hyp "troops" is prep_over of "are" which aligned to text "moved"
 1.00 -2.91 GlickmanLexicalEntailment.score
Hand-tuned score (dot product of above): -9.6770
Threshold: -11.4590

Hand-set weights Accuracy: 2/6 = 0.3333

Word similarity table built on Wed Aug 30 21:40:35 PDT 2006 using command:
java edu.stanford.nlp.rte.WordSimilarityGenerator -info /u/nlp/rte/data/byformat/align/stochastic/chris.align.xml -output /u/nlp/rte/data/byformat/wordsim/stochastic/chris.wordsim.html