Advances in Neural Semantic Parsing

Data Recombination for
Neural Semantic Parsing
Robin Jia and Percy Liang
Stanford University
Semantic Parsing
What states border Texas ?
2
Semantic Parser
And(State, NextTo(StateId(Texas)))
[New Mexico, Oklahoma, Arkansas, Louisiana]
Executor
Zelle and Mooney (1996), Zettlemoyer and Collins (2005, 2007), Liang et al.
(2011), Artzi and Zettlemoyer (2013), Berant et al. (2013).
Outline
 
•
Neural Semantic Parser
•
Data Recombination
•
Discussion
3
Outline
•
Neural Semantic Parser
•
Data Recombination
•
Discussion
4
Semantic Parsing
•
Traditional semantic parsers are engineering-
heavy
•
Can we get good performance with a domain-
general model?
5
Semantic Parsing
•
Geoquery
•
Input: 
what is the population of iowa ?
•
Output: 
_answer ( A , ( _population ( B , A ) ,
_const ( B , _stateid ( iowa ) ) ) )
•
ATIS
•
Input: 
list all flights from chicago to milwaukee
•
Output: 
( _lambda $0 e ( _and ( _flight $0 ) ( _from
$0 chicago:_ci ) ( _to $0 milwaukee:_ci ) ) )
•
Overnight
•
Input: 
what restaurants have takeout
•
Output: 
( call list ( call filter ( call getProperty
( call singleton en.restaurant ) ( string ! type ) )
( string takeout ) ) )
6
Zelle and Mooney (1996), Dahl et al. (1994), Wang et al. (2015).
Sequence-to-sequence Models
•
Machine Translation
•
Input: 
the blue house
•
Output: 
la maison bleue
•
Syntactic Parsing
•
Input: 
the dog barked
•
Output: 
(S (NP DT NN )
NP
 (VP VBD)
VP
 )
S
•
And many more…
7
Sutskever et al. (2014), Bahdanau et al. (2014), Vinyals et al. (2015).
Semantic Parsing
 
•
Treat semantic parsing as a sequence-to-
sequence task
•
Use domain-general attention-based neural
model
8
Sutskever et al. (2014), Bahdanau et al. (2014), Luong et al. (2015).
Neural Semantic Parser
9
what     states   border    texas
Neural Semantic Parser
10
what     states   border    texas
Neural Semantic Parser
11
what     states   border    texas
Neural Semantic Parser
12
what     states   border    texas
Neural Semantic Parser
13
what     states   border    texas
Neural Semantic Parser
14
what     states   border    texas
Neural Semantic Parser
15
what     states   border    texas
(
)
And
City
State
iowa
utah
...
Neural Semantic Parser
16
what     states   border    texas
(
)
And
City
State
iowa
utah
...
Neural Semantic Parser
17
what     states   border    texas
And
Neural Semantic Parser
18
what     states   border    texas
And
Rare Entities
 
city('alabama','al','birmingham',284413).
city('alabama','al','mobile',200452).
city('alabama','al','montgomery',177857).
city('alabama','al','huntsville',142513).
city('alabama','al','tuscaloosa',75143).
city('alaska','ak','anchorage',174431).
city('arizona','az','phoenix',789704).
city('arizona','az','tucson',330537).
city('arizona','az','mesa',152453).
city('arizona','az','tempe',106919).
city('arizona','az','glendale',96988).
city('arizona','az','scottsdale',88622).
city('arkansas','ar','little rock',158915).
city('arkansas','ar','fort smith',71384).
city('arkansas','ar','north little rock',64388).
city('california','ca','los angeles',2966850).
city('california','ca','san diego',875538).
city('california','ca','san
francisco',678974).
city('california','ca','san jose',629442).
city('california','ca','long beach',361334).
city('california','ca','oakland',339337).
city('california','ca','sacramento',275741).
city('california','ca','anaheim',219311).
city('california','ca','fresno',218202).
city('california','ca','santa ana',203713).
city('california','ca','riverside',170876).
city('california','ca','huntington
beach',170505).
city('california','ca','stockton',149779).
city('california','ca','glendale',139060).
city('california','ca','fremont',131945).
…
19
Rare Entities
 
•
Rare entities pose a problem!
•
If we see an entity name we didn’t see during
training, how can we even generate the right
logical form?
•
Solution: 
Attention-based copying
20
See also Gu et al. (2016), Gulcehre et al. (2016)
Neural Semantic Parser
21
what     states   border    texas
(
)
And
City
State
iowa
utah
...
    And (
  State ,
  NextTo (
    StateId (
 
what
states
border
texas
Neural Semantic Parser
22
what     states   border    texas
(
)
And
City
State
iowa
utah
...
    And (
  State ,
  NextTo (
    StateId (
what
states
border
texas
Implementation Details
 
•
Training: maximize loglikelihood of correct
logical form with SGD
•
Test time
•
Decode with beam search
•
Add missing parentheses
•
Prune logical forms that result in execution error
23
Results (Take 1)
24
* Not directly comparable; used a seed lexicon for predicates.
 
How can we do better?
Outline
•
Neural Semantic Parser
•
Data Recombination
•
Discussion
25
Structural Regularities
26
 
•
Given
•
what are the major cities in iowa ?
And(City, Major, LocatedIn(StateId(iowa)))
•
We know how to parse
•
what are the major cities in 
texas 
?
 And(City, Major, LocatedIn(StateId(
texas
)))
 
Structural Regularities
•
How do we build a neural model that respects
compositional structural regularities?
27
Recombinant Examples
Dataset:
                    what are the major cities in             ?
And(City, Major, LocatedIn(StateId(      )))
                    what are states that border            ?
                 
And(State, NextTo(StateId(      )))
28
 
iowa
iowa
 
texas
texas
Recombinant Examples
29
what are the major cities in iowa ?
And(City, Major, LocatedIn(StateId(iowa)))
Step 1: Apply high-precision alignment rules
Recombinant Examples
30
what are the major cities in iowa ?
And(City, Major, LocatedIn(
StateId
(iowa)))
Step 1: Apply high-precision alignment rules
Step 2: Infer types of aligned fragments
Recombinant Examples
31
R
OOT
 → (
what are the major cities in 
S
TATE
 
?
,
      And(City, Major, LocatedIn(
S
TATE
))
)
S
TATE
 → (
iowa
,
 
StateId(iowa)
)
Step 1: Apply high-precision alignment rules
Step 2: Infer types of aligned fragments
Step 3: Generate grammar rules by abstracting
aligned fragments
Recombinant Examples
32
R
OOT
 → (
what are states that border 
S
TATE
 
?
,
      And(State, NextTo(
S
TATE
))
)
S
TATE
 → (
texas
,
 
StateId(texas)
)
Step 1: Apply high-precision alignment rules
Step 2: Infer types of aligned fragments
Step 3: Generate grammar rules by abstracting
aligned fragments
(Repeat for other examples)
Recombinant Examples
33
what are the major cities in 
State 
?
And(City, Major, LocatedIn(
State
))
Step 1: Apply high-precision alignment rules
Step 2: Infer types of aligned fragments
Step 3: Generate grammar rules by abstracting
aligned fragments
Step 4: Combine rules from different examples to
form new “recombinant” examples
Recombinant Examples
34
what are the major cities in texas ?
And(City, Major, LocatedIn(StateId(texas)))
Step 1: Apply high-precision alignment rules
Step 2: Infer types of aligned fragments
Step 3: Generate grammar rules by abstracting
aligned fragments
Step 4: Combine rules from different examples to
form new “recombinant” examples
Data Recombination
35
Step 1: 
Generate recombinant examples 
using training
data and prior knowledge about domain
Data Recombination
36
Model
 
Step 2: 
Train on these examples 
to increase model’s
awareness of task structure
Training the Model
•
At each epoch, sample new recombinant
examples from grammar
•
Choose production rules uniformly at random
•
Train on recombinant examples plus original
training examples in 1:1 ratio
37
Results (Take 2)
38
* Not directly comparable; used a seed lexicon for predicates.
Structural Regularities, Part 2
39
 
•
Given
•
what are the major cities in iowa ?
And(City, Major, LocatedIn(StateId(iowa)))
•
what are states that border texas ?
And(State, NextTo(StateId(texas)))
•
We know how to parse
•
what are the major cities in
        
states that border texas 
?
 
And(City, Major, LocatedIn(
    
And(State, NextTo(StateId(texas)))
))
 
Abstracting Whole Phrases
40
What are states that border texas ?
And(State, NextTo(StateId(texas)))
Step 1: Apply high-precision alignment rules
Abstracting Whole Phrases
41
What are states that border texas ?
And(
State
, NextTo(StateId(texas)))
Step 1: Apply high-precision alignment rules
Step 2: Infer types of aligned fragments
Abstracting Whole Phrases
42
What are states that border texas ?
And(
State
, NextTo(StateId(texas)))
Step 1: Apply high-precision alignment rules
Step 2: Infer types of aligned fragments
New Grammar Rule
•
State → (
states that border texas
,
               
And(State, NextTo(StateId(texas)))
)
Results (Take 3)
43
* Not directly comparable; used a seed lexicon for predicates.
Composition of Strategies
 
•
Grammar induction strategies are functions on
grammars
•
Process each rule in grammar independently to
generate new grammar rules
•
Initial grammar
•
Root → (
what are the major cities in iowa ?
,
    And(City, Major, LocatedIn(StateId(iowa)))
)
•
Root → (
what are states that border texas ?
,
       And(State, NextTo(StateId(texas)))
)
44
Composition of Strategies
45
Initial Gramar
 
AbsWholePhrases Grammar
Composition of Strategies
46
S
TATE
 →
      
(
states that border texas 
,
    And(State, NextTo(StateId(texas)))
)
Composition of Strategies
47
Initial Gramar
AbsWholePhrases Grammar
Composed Grammar
Results (Take 4)
48
* Not directly comparable; used a seed lexicon for predicates.
What else can we do?
 
•
What happens if we artificially make longer
examples?
49
Concatenation
50
•
Given
•
what are the major cities in iowa ?
And(City, Major, LocatedIn(StateId(iowa)))
•
what are states that border texas ?
And(State, NextTo(StateId(texas)))
•
Create new example
•
what are the major cities in iowa ? </s> what are
states that border texas ?
 
And(City, Major, LocatedIn(StateId(iowa)))
 </s> And(State, NextTo(StateId(texas)))
Concatenation
•
Can cast as grammar induction
•
Compose with previous grammar induction
strategies
51
Final Results
52
* Not directly comparable; used a seed lexicon for predicates.
Outline
•
Neural Semantic Parser
•
Data Recombination
•
Discussion
53
Data Recombination
54
Model
 
Step 1: 
Generate recombinant examples 
using training
data and prior knowledge about domain
Data Recombination
55
Data Augmentation
•
Apply local transformations to a single example
to generate more examples
56
Krizhevsky et al. (2012).
Data Augmentation
•
Apply local transformations to a single example
to generate more examples
57
Krizhevsky et al. (2012).
Horizontal Reflection
Data Augmentation
•
Apply local transformations to a single example
to generate more examples
58
Krizhevsky et al. (2012).
Cropping
Data Recombination
•
Get a richer generative model by combining
information from 
multiple examples
59
                    what are the major cities in             ?
And(City, Major, LocatedIn(StateId(      )))
                    what are states that border            ?
                 
And(State, NextTo(StateId(      )))
 
iowa
iowa
 
texas
texas
Empirical Distribution
60
Data Augmentation
61
Data Recombination
62
Thank you!
63
Model
 
Code, data, and experiments available on
https://worksheets.codalab.org/worksheets/0x50757a37779b485f89012e4ba03b6f4f/
Composition of Strategies
64
S
TATE
 →
      
(
states that border texas 
,
    And(State, NextTo(StateId(texas)))
)
New Grammar Rule
•
S
TATE
 → (
states that border 
S
TATE
I
D
,
               
And(State, NextTo(
S
TATE
I
D
))
)
An Artificial Experiment
 
•
Make up some simple artificial data
•
friends of relatives of alice
Friend(Relative(alice))
•
colleagues of brothers of bob
Colleague(Brother(bob))
•
Generate recombinant examples to train on
•
Same length
: only swap entities
•
friends of relatives of 
bob
Friend(Relative(
bob
))
•
Longer
: nest whole phrases
•
friends of relatives of 
colleagues of brothers of 
alice
Friend(Relative(
Colleague(Brother(
alice
))
))
65
An Artificial Experiment
•
Test only on “short” examples like those in the
original training data
•
siblings of roommates of eve
Sibling(Roommate(eve))
66
An Artificial Experiment
67

								

								
								

										

												
Slide Note

												

										

								


								
						

							
						


							

								

									
										
Embed
									
									
										
Share
									

									
									
									

								

								
							


							

							 
 							 
 							 
						  


							
							

								
Delve into the realm of neural semantic parsing with a focus on data recombination techniques, traditional parsers, and the shift towards domain-general models. Explore the application of sequence-to-sequence models and attention-based neural frameworks in semantic parsing tasks. Discover the evolving landscape of semantic parsing with innovative methodologies and models proposed by researchers such as Zelle, Mooney, Sutskever, and Bahdanau.


								
    
																		
  •   Neural Semantic Parsing
    • 
																	
  •    Data Recombination
    • 
																	
  •    Sequence-to-Sequence Models
    • 
																	
  •    Attention-Based Models
    • 
																	
  •    Semantic Parsing
    • 
																

								

								

									

										noy_p
										

											 noy_p Follow
											

										

									

									
Uploaded on Nov 25, 2024 | 0 Views

								

								


								

									
Download Presentation

									
								

								
Please find below an Image/Link to download the presentation.

								
The content on the website is provided AS IS for your information and personal use only. It may not be sold, licensed, or shared on other websites without obtaining consent from the author.If you encounter any issues during the download, it is possible that the publisher has removed the file from their server.


									
									

									


										 
										 
										 
										 

									

									

									

								
You are allowed to download the files provided on this website for personal or commercial use, subject to the condition that they are used lawfully. All files are the property of their respective owners.

								

									 
									
Download Presentation
								

								
The content on the website is provided AS IS for your information and personal use only. It may not be sold, licensed, or shared on other websites without obtaining consent from the author.


								

									
 E N D 

								


							
Presentation Transcript

							

							
  1. Data Recombination for Neural Semantic Parsing Robin Jia and Percy Liang Stanford University
    2. 
            
            
            
  2. Semantic Parsing What states border Texas ? Semantic Parser And(State, NextTo(StateId(Texas))) Executor [New Mexico, Oklahoma, Arkansas, Louisiana] Zelle and Mooney (1996), Zettlemoyer and Collins (2005, 2007), Liang et al.  (2011), Artzi and Zettlemoyer (2013), Berant et al. (2013). 2
    3. 
            
            
            
  3. Outline     Neural Semantic Parser     Data Recombination     Discussion 3
    4. 
            
            
            
  4. Outline     Neural Semantic Parser     Data Recombination     Discussion 4
    5. 
            
            
            
  5. Semantic Parsing     Traditional semantic parsers are engineering- heavy      Can we get good performance with a domain- general model? 5
    6. 
            
            
            
  6. Semantic Parsing     Geoquery     Input: what is the population of iowa ?     Output: _answer ( A , ( _population ( B , A ) ,  _const ( B , _stateid ( iowa ) ) ) )     ATIS     Input: list all flights from chicago to milwaukee     Output: ( _lambda $0 e ( _and ( _flight $0 ) ( _from  $0 chicago:_ci ) ( _to $0 milwaukee:_ci ) ) )     Overnight     Input: what restaurants have takeout     Output: ( call list ( call filter ( call getProperty  ( call singleton en.restaurant ) ( string ! type ) )  ( string takeout ) ) ) Zelle and Mooney (1996), Dahl et al. (1994), Wang et al. (2015). 6
    7. 
            
            
            
  7. Sequence-to-sequence Models     Machine Translation     Input: the blue house     Output: la maison bleue     Syntactic Parsing     Input: the dog barked     Output: (S (NP DT NN )NP (VP VBD)VP )S     And many more    Sutskever et al. (2014), Bahdanau et al. (2014), Vinyals et al. (2015). 7
    8. 
            
            
            
  8. Semantic Parsing     Treat semantic parsing as a sequence-to- sequence task     Use domain-general attention-based neural  model Sutskever et al. (2014), Bahdanau et al. (2014), Luong et al. (2015). 8
    9. 
            
            
            
  9. Neural Semantic Parser ? ? ? ? ?    4    0    1    2    3 what     states   border    texas ? ? ? ? ?    5    4    1    2    3 9
    10. 
            
            
            
  10. Neural Semantic Parser ? ? ? ?    4    1    2    3 what     states   border    texas ? ? ? ?    4    1    2    3 10
    11. 
            
            
            
  11. Neural Semantic Parser what     states   border    texas ?1 ?2 ?3 ?4 11
    12. 
            
            
            
  12. Neural Semantic Parser ? ? ? ? ? ?    4    4    0    1    2    3 what     states   border    texas ?1 ? ? ? ? ? ?    5    4    1    1    2    3 12
    13. 
            
            
            
  13. Neural Semantic Parser what     states   border    texas ?1 ?2 ?3 ?4 ?1 13
    14. 
            
            
            
  14. Neural Semantic Parser what     states   border    texas ?1 ?2 ?3 ?4 ?1 ?1 14
    15. 
            
            
            
  15. Neural Semantic Parser what     states   border    texas ?1 ?2 ?3 ?4 ( ) And City State iowa utah ... ?1 ?1 ?1= 15
    16. 
            
            
            
  16. Neural Semantic Parser what     states   border    texas ?1 ?2 ?3 ?4 ( ) And City State iowa utah ... ?1 ?1 ?1= 16
    17. 
            
            
            
  17. Neural Semantic Parser what     states   border    texas ?1 ?2 ?3 ?4 ?1 ?1 ?1= And ?2 17
    18. 
            
            
            
  18. Neural Semantic Parser what     states   border    texas ?1 ?2 ?3 ?4 ?2 ? = And 18
    19. 
            
            
            
  19. Rare Entities city('alabama','al','birmingham',284413). city('alabama','al','mobile',200452). city('alabama','al','montgomery',177857). city('alabama','al','huntsville',142513). city('alabama','al','tuscaloosa',75143). city('alaska','ak','anchorage',174431). city('arizona','az','phoenix',789704). city('arizona','az','tucson',330537). city('arizona','az','mesa',152453). city('arizona','az','tempe',106919). city('arizona','az','glendale',96988). city('arizona','az','scottsdale',88622). city('arkansas','ar','little rock',158915). city('arkansas','ar','fort smith',71384). city('arkansas','ar','north little rock',64388). city('california','ca','los angeles',2966850). city('california','ca','san diego',875538). city('california','ca','san  francisco',678974). city('california','ca','san jose',629442). city('california','ca','long beach',361334). city('california','ca','oakland',339337). city('california','ca','sacramento',275741). city('california','ca','anaheim',219311). city('california','ca','fresno',218202). city('california','ca','santa ana',203713). city('california','ca','riverside',170876). city('california','ca','huntington  beach',170505). city('california','ca','stockton',149779). city('california','ca','glendale',139060). city('california','ca','fremont',131945).     19
    20. 
            
            
            
  20. Rare Entities     Rare entities pose a problem!     If we see an entity name we didn   t see during  training, how can we even generate the right  logical form?     Solution: Attention-based copying See also Gu et al. (2016), Gulcehre et al. (2016)  20
    21. 
            
            
            
  21. Neural Semantic Parser what     states   border    texas ?1 ?2 ?3 ?4 ( ) And City State iowa utah ... ? = And ( what states border texas   State ,    NextTo (     StateId ( ?? ?? ??= 21
    22. 
            
            
            
  22. Neural Semantic Parser what     states   border    texas ?1 ?2 ?3 ?4 ( ) And City State iowa utah ... ? = And ( what states border texas   State ,    NextTo (     StateId ( ?1 ?1 ?1= 22
    23. 
            
            
            
  23. Implementation Details     Training: maximize loglikelihood of correct  logical form with SGD     Test time     Decode with beam search     Add missing parentheses     Prune logical forms that result in execution error 23
    24. 
            
            
            
  24. Results (Take 1) System Geoquery ATIS 84.6 Overnight Zettlemoyer and Collins (2007) Kwiatkowski et al. (2010) Liang et al. (2011)* Kwiatkowski et al. (2013) Zhao and Huang (2015) Wang et al. (2015) Our RNN Model 88.9 91.1 89.0 88.9 84.2 58.8 75.8 85.0 76.3 How can we do better? * Not directly comparable; used a seed lexicon for predicates. 24
    25. 
            
            
            
  25. Outline     Neural Semantic Parser     Data Recombination     Discussion 25
    26. 
            
            
            
  26. Structural Regularities     Given     what are the major cities in iowa ?  And(City, Major, LocatedIn(StateId(iowa)))     We know how to parse     what are the major cities in texas ?   And(City, Major, LocatedIn(StateId(texas))) 26
    27. 
            
            
            
  27. Structural Regularities     How do we build a neural model that respects  compositional structural regularities? Initial data Recombinant Examples Prior Knowledge 27
    28. 
            
            
            
  28. Recombinant Examples Dataset: iowa iowa                     what are the major cities in             ? And(City, Major, LocatedIn(StateId(      ))) texas texas                     what are states that border            ? And(State, NextTo(StateId(      ))) 28
    29. 
            
            
            
  29. Recombinant Examples what are the major cities in iowa ?  And(City, Major, LocatedIn(StateId(iowa))) Step 1: Apply high-precision alignment rules 29
    30. 
            
            
            
  30. Recombinant Examples what are the major cities in iowa ?  And(City, Major, LocatedIn(StateId(iowa))) Step 1: Apply high-precision alignment rules Step 2: Infer types of aligned fragments 30
    31. 
            
            
            
  31. Recombinant Examples ROOT    (what are the major cities in STATE?,       And(City, Major, LocatedIn(STATE))) STATE    (iowa,StateId(iowa)) Step 1: Apply high-precision alignment rules Step 2: Infer types of aligned fragments Step 3: Generate grammar rules by abstracting  aligned fragments 31
    32. 
            
            
            
  32. Recombinant Examples ROOT    (what are states that border STATE?,       And(State, NextTo(STATE))) STATE    (texas,StateId(texas)) Step 1: Apply high-precision alignment rules Step 2: Infer types of aligned fragments Step 3: Generate grammar rules by abstracting  aligned fragments (Repeat for other examples) 32
    33. 
            
            
            
  33. Recombinant Examples what are the major cities in STATE?  And(City, Major, LocatedIn(STATE)) Step 1: Apply high-precision alignment rules Step 2: Infer types of aligned fragments Step 3: Generate grammar rules by abstracting  aligned fragments Step 4: Combine rules from different examples to  form new    recombinant    examples 33
    34. 
            
            
            
  34. Recombinant Examples what are the major cities in texas ?  And(City, Major, LocatedIn(StateId(texas))) Step 1: Apply high-precision alignment rules Step 2: Infer types of aligned fragments Step 3: Generate grammar rules by abstracting  aligned fragments Step 4: Combine rules from different examples to  form new    recombinant    examples 34
    35. 
            
            
            
  35. Data Recombination Initial data Recombinant Examples Prior Knowledge Step 1: Generate recombinant examples using training  data and prior knowledge about domain 35
    36. 
            
            
            
  36. Data Recombination Initial data Recombinant Examples Model Prior Knowledge Step 2: Train on these examples to increase model   s  awareness of task structure 36
    37. 
            
            
            
  37. Training the Model     At each epoch, sample new recombinant  examples from grammar     Choose production rules uniformly at random     Train on recombinant examples plus original  training examples in 1:1 ratio 37
    38. 
            
            
            
  38. Results (Take 2) System Geoquery ATIS 84.6 Overnight Zettlemoyer and Collins (2007) Kwiatkowski et al. (2010) Liang et al. (2011)* Kwiatkowski et al. (2013) Zhao and Huang (2015) Wang et al. (2015) Our RNN Model + Abstracting Entities 88.9 91.1 89.0 88.9 84.2 58.8 75.8 75.3 85.0 85.4 76.3 79.9 * Not directly comparable; used a seed lexicon for predicates. 38
    39. 
            
            
            
  39. Structural Regularities, Part 2     Given     what are the major cities in iowa ?  And(City, Major, LocatedIn(StateId(iowa)))     what are states that border texas ? And(State, NextTo(StateId(texas)))     We know how to parse     what are the major cities in  states that border texas ?  And(City, Major, LocatedIn( And(State, NextTo(StateId(texas))))) 39
    40. 
            
            
            
  40. Abstracting Whole Phrases What are states that border texas ? And(State, NextTo(StateId(texas))) Step 1: Apply high-precision alignment rules 40
    41. 
            
            
            
  41. Abstracting Whole Phrases What are states that border texas ? And(State, NextTo(StateId(texas))) Step 1: Apply high-precision alignment rules Step 2: Infer types of aligned fragments 41
    42. 
            
            
            
  42. Abstracting Whole Phrases What are states that border texas ? And(State, NextTo(StateId(texas))) Step 1: Apply high-precision alignment rules Step 2: Infer types of aligned fragments New Grammar Rule     STATE    (states that border texas, And(State, NextTo(StateId(texas)))) 42
    43. 
            
            
            
  43. Results (Take 3) System Geoquery ATIS 84.6 Overnight Zettlemoyer and Collins (2007) Kwiatkowski et al. (2010) Liang et al. (2011)* Kwiatkowski et al. (2013) Zhao and Huang (2015) Wang et al. (2015) Our RNN Model + AbsEntities + AbsWholePhrases 88.9 91.1 89.0 88.9 84.2 58.8 75.8 75.3 75.9 85.0 85.4 87.5 76.3 79.9 --- * Not directly comparable; used a seed lexicon for predicates. 43
    44. 
            
            
            
  44. Composition of Strategies     Grammar induction strategies are functions on  grammars     Process each rule in grammar independently to  generate new grammar rules     Initial grammar     ROOT    (what are the major cities in iowa ?,     And(City, Major, LocatedIn(StateId(iowa))))     ROOT    (what are states that border texas ?,        And(State, NextTo(StateId(texas)))) 44
    45. 
            
            
            
  45. Composition of Strategies Initial Gramar Abstract Whole Phrases AbsWholePhrases Grammar 45
    46. 
            
            
            
  46. Composition of Strategies STATE    (states that border texas ,      And(State, NextTo(StateId(texas)))) 46
    47. 
            
            
            
  47. Composition of Strategies Initial Gramar Abstract Whole Phrases AbsWholePhrases Grammar Abstract Entities Composed Grammar 47
    48. 
            
            
            
  48. Results (Take 4) System Geoquery ATIS 84.6 Overnight Zettlemoyer and Collins (2007) Kwiatkowski et al. (2010) Liang et al. (2011)* Kwiatkowski et al. (2013) Zhao and Huang (2015) Wang et al. (2015) Our RNN Model + AbsEntities + AbsWholePhrases + AbsWholePhrases, AbsEntities 88.9 91.1 89.0 88.9 84.2 58.8 75.8 75.3 75.9 75.3 85.0 85.4 87.5 88.9 76.3 79.9 --- --- * Not directly comparable; used a seed lexicon for predicates. 48
    49. 
            
            
            
  49. What else can we do?     What happens if we artificially make longer  examples? 49
    50. 
            
            
            
  50. Concatenation     Given     what are the major cities in iowa ?  And(City, Major, LocatedIn(StateId(iowa)))     what are states that border texas ? And(State, NextTo(StateId(texas)))     Create new example     what are the major cities in iowa ? </s> what are  states that border texas ? And(City, Major, LocatedIn(StateId(iowa)))  </s> And(State, NextTo(StateId(texas))) 50
    51. 
            
            
            
						

						

					

					
					
					

						
Related

						

						


							
							
							
							

						

							

									

      

        
          

            

              LR Parsing and State Merging Techniques
            

            

              
LR Parsing and State Merging Techniques

              
soren

            

          

        
      


     

      

        
          

            

              Artificial Neural Networks From Scratch
            

            

              
Artificial Neural Networks From Scratch

              
inara

            

          

        
      


     

      

        
          

            

              Semantic Memory Models in Cognitive Psychology
            

            

              
Semantic Memory Models in Cognitive Psychology

              
shepherd

            

          

        
      


     

      

        
          

            

              Core Area of Statutory Bank Branch Audit and Review of Advances by Hitesh Pomal
            

            

              
Core Area of Statutory Bank Branch Audit and Review of Advances by Hitesh Pomal

              
issa

            

          

        
      


     

      

        
          

            

              Semantic Roles in Sentence Structure
            

            

              
Semantic Roles in Sentence Structure

              
wisteria

            

          

        
      


     

      

        
          

            

              Neural Quantum States and Symmetries in Quantum Mechanics
            

            

              
Neural Quantum States and Symmetries in Quantum Mechanics

              
adelinda

            

          

        
      


     

      

        
          

            

              Semantic Analysis of Clinical Narratives Using Complex Knowledge Graphs
            

            

              
Semantic Analysis of Clinical Narratives Using Complex Knowledge Graphs

              
linden

            

          

        
      


     

      

        
          

            

              Introduction to Neural Networks in IBM SPSS Modeler 14.2
            

            

              
Introduction to Neural Networks in IBM SPSS Modeler 14.2

              
anouk

            

          

        
      


     

      

        
          

            

              Dynamic Oracle Training in Constituency Parsing
            

            

              
Dynamic Oracle Training in Constituency Parsing

              
barack

            

          

        
      


     

      

        
          

            

              Shift-Reduce Parsing Example in Mr. Lupoli's F2012
            

            

              
Shift-Reduce Parsing Example in Mr. Lupoli's F2012

              
ayda

            

          

        
      


     

      

        
          

            

              Semantic Web Technologies and Knowledge Representation Overview
            

            

              
Semantic Web Technologies and Knowledge Representation Overview

              
aris_eel

            

          

        
      


     

      

        
          

            

              Multimodal Semantic Indexing for Image Retrieval at IIIT Hyderabad
            

            

              
Multimodal Semantic Indexing for Image Retrieval at IIIT Hyderabad

              
kailan

            

          

        
      


     

      

        
          

            

              Advanced Parsing Techniques for NLP Evaluation
            

            

              
Advanced Parsing Techniques for NLP Evaluation

              
sazm

            

          

        
      


     

      

        
          

            

              Sentence Comprehension and Memory in Psycholinguistics
            

            

              
Sentence Comprehension and Memory in Psycholinguistics

              
kayteel

            

          

        
      


     

      

        
          

            

              Top-Down Parsing in Context-Free Syntax
            

            

              
Top-Down Parsing in Context-Free Syntax

              
celi_661

            

          

        
      


     

      

        
          

            

              Revisiting Semantic Feature Analysis: A Classic Therapy Technique
            

            

              
Revisiting Semantic Feature Analysis: A Classic Therapy Technique

              
lakenzien

            

          

        
      


     

      

        
          

            

              Exploring Fast & Accurate Parsing With Learning to Prune
            

            

              
Exploring Fast & Accurate Parsing With Learning to Prune

              
ada_be

            

          

        
      


     

      

        
          

            

              Introduction to NLP Parsing Techniques and Algorithms
            

            

              
Introduction to NLP Parsing Techniques and Algorithms

              
weirich_a

            

          

        
      


     

      

        
          

            

              Enhancing Name and Address Parsing for Data Standardization
            

            

              
Enhancing Name and Address Parsing for Data Standardization

              
threlkeld_b

            

          

        
      


     

      

        
          

            

              Ensembling Diverse Approaches to Question Answering
            

            

              
Ensembling Diverse Approaches to Question Answering

              
tobo_yo

            

          

        
      


     

      

        
          

            

              Annotating Semantic Issues in Translation for Students
            

            

              
Annotating Semantic Issues in Translation for Students

              
jart449

            

          

        
      


     

      

        
          

            

              Semantic Classification of Prepositions in Bulgarian WordNet: A Comprehensive Overview
            

            

              
Semantic Classification of Prepositions in Bulgarian WordNet: A Comprehensive Overview

              
juanjes

            

          

        
      


     							

						

					

					

				


				

			

		
		

			

				
				

					
More Related Content

					

						

														

					

				

				
			

		












 





  
  
  



giItT1WQy@!-/#giItT1WQy@!-/#giItT1WQy@!-/#giItT1WQy@!-/#giItT1WQy@!-/#giItT1WQy@!-/#giItT1WQy@!-/#giItT1WQy@!-/#giItT1WQy@!-/#