Learning Ensembles of First-Order Clauses for Recall-Precision Curves A Case Study in Biomedical Information Extraction

1
Learning Ensembles ofFirst-Order Clauses for
Recall-Precision CurvesA Case Study
inBiomedical Information Extraction

• Mark Goadrich, Louis Oliphant and Jude Shavlik
• Department of Computer Sciences
• University of Wisconsin Madison USA
• 6 Sept 2004

2
Talk Outline

• Link Learning and ILP
• Our Gleaner Approach
• Aleph Ensembles
• Biomedical Information Extraction
• Evaluation and Results
• Future Work

3
ILP Domains

• Object Learning
• Trains, Carcinogenesis
• Link Learning
• Binary predicates

4
Link Learning

• Large skew toward negatives
• 500 relational objects
• 5000 positive links means 245,000 negative links
• Difficult to measure success
• Always negative classifier is 98 accurate
• ROC curves look overly optimistic
• Enormous quantity of data
• 4,285,199,774 web pages indexed by Google
• PubMed includes over 15 million citations

5
Our Approach

• Develop fast ensemble algorithms focused on
  recall and precision evaluation
• Key Ideas of Gleaner
• Keep wide range of clauses
• Create separate theories for different recall
  ranges
• Evaluation
• Area Under Recall Precision Curve (AURPC)
• Time Number of clauses considered

6
Gleaner - Background

• Focus evaluation on positive examples
• Recall
• Precision
• Rapid Random Restart (Zelezny et al ILP 2002)
• Stochastic selection of starting clause
• Time-limited local heuristic search
• We store variety of clauses (based on recall)

7
Gleaner - Learning

• Create B Bins
• Generate Clauses
• Record Best
• Repeat for K seeds

Precision
Recall
8
Gleaner - Combining

• Combine K clauses per bin
• If at least L of K clauses match, call example
  positive
• How to choose L ?
• L1 then high recall, low  precision
• LK then low  recall, high precision
• Our method
• Choose L such that ensemble recall matches bin b
• Bin bs precision should be higher than any
  clause in it
• We should now have set of high precision rule
  sets spanning space of recall levels

9
How to use Gleaner

• Generate Curve
• User Selects Recall Bin
• Return ClassificationsWith Precision Confidence

Precision
Recall 0.50 Precision 0.70
Recall
10
Aleph Ensembles

• We compare to ensembles of theories
• Algorithm (Dutra et al ILP 2002)
• Use K different initial seeds
• Learn K theories containing C clauses
• Rank examples by the number of theories
• Need to balance C for high performance
• Small C leads to low recall
• Large C leads to converging theories

11
Aleph Ensembles (100 theories)
12
Biomedical Information Extraction

• Given Medical Journal abstracts tagged
  with protein localization relations
• Do Construct system to extract protein
  localization phrases from unseen text
• NPL3 encodes a nuclear protein with an RNA
  recognition motif and similarities to a family of
  proteins involved in RNA metabolism.

13
Biomedical Information Extraction

• Hand-labeled dataset (Ray Craven 01)
• 7,245 sentences from 871 abstracts
• Examples are phrase-phrase combinations
• 1,810 positive 279,154 negative
• 1.6 GB of background knowledge
• Structural, Statistical, Lexical and Ontological
• In total, 200 distinct background predicates

14
Evaluation Metrics
1.0

• Two dimensions
• Area Under Recall-Precision Curve (AURPC)
• All curves standardized to cover full recall
  range
• Averaged AURPC over 5 folds
• Number of clauses considered
• Rough estimate of time
• Both are stop anytime parallel algorithms

Precision
Recall
1.0
15
AURPC Interpolation

• Convex interpolation in RP space?
• Precision interpolation is counterintuitive
• Example 1000 positive 9000 negative

TP FP TP Rate FP Rate Recall Prec
500 500 0.50 0.06 0.50 0.50

1000 9000 1.00 1.00 1.00 0.10
750
4750
0.75
0.53
0.75
0.14
Example Counts
RP Curves
ROC Curves
16
AURPC Interpolation
17
Experimental Methodology

• Performed five-fold cross-validation
• Variation of parameters
• Gleaner (20 recall bins)
• seeds 25, 50, 75, 100
• clauses 1K, 10K, 25K, 50K, 100K, 250K, 500K
• Ensembles (0.75 minacc, 35,000 nodes)
• theories 10, 25, 50, 75, 100
• clauses per theory 1, 5, 10, 15, 20, 25, 50

18
Results Testfold 5 at 1,000,000 clauses
Gleaner
Ensembles
19
Results Gleaner vs Aleph Ensembles
20
Conclusions

• Gleaner
• Focuses on recall and precision
• Keeps wide spectrum of clauses
• Good results in few cpu cycles
• Aleph ensembles
• Early stopping helpful
• Require more cpu cycles
• AURPC
• Useful metric for comparison
• Interpolation unintuitive

21
Future Work

• Improve Gleaner performance over time
• Explore alternate clause combinations
• Better understanding of AURPC
• Search for clauses that optimize AURPC
• Examine more ILP link-learning datasets
• Use Gleaner with other ML algorithms

22
Take-Home Message

• Definition of Gleaner
• One who gathers grain left behind by reapers
• Gleaner and ILP
• Many clauses constructed and evaluated in ILP
  hypothesis search
• We need to make better use of those that arent
  the highest scoring ones
• Thanks, Questions?

23
Acknowledgements

• USA NLM Grant 5T15LM007359-02
• USA NLM Grant 1R01LM07050-01
• USA DARPA Grant F30602-01-2-0571
• USA Air Force Grant F30602-01-2-0571
• Condor Group
• David Page
• Vitor Santos Costa, Ines Dutra
• Soumya Ray, Marios Skounakis, Mark Craven
• Dataset available at (URL in proceedings)
• ftp//ftp.cs.wisc.edu/machine-learning/shavlik-gro
  up/datasets/IE-protein-location

24
Deleted Scenes

• Aleph Learning
• Clause Weighting
• Sample Gleaner Recall-Precision Curve
• Sample Extraction Clause
• Gleaner Algorithm
• Director Commentary
• on off

25
Aleph - Learning

• Aleph learns theories of clauses (Srinivasan, v4,
  2003)
• Pick a positive seed example and saturate
• Use heuristic search to find best clause
• Pick new seed from uncovered positivesand repeat
  until threshold of positives covered
• Theory produces one recall-precision point
• Learning complete theories is time-consuming
• Can produce ranking with theory ensembles

26
Clause Weighting

• Single Theory Ensemble
• rank by how many clauses cover examples
• Weight clauses using tuneset statistics
• CN2 (average precision of matching clauses)
• Lowest False Positive Rate Score
• Cumulative
• F1 score Recall
• Precision Diversity

27
Clause Weighting
28
Further Results
29
Biomedical Information Extraction

• NPL3 encodes a nuclear protein with

marked location
alphanumeric
30
Sample Extraction Clause

• P Protein, L Location, S Sentence
• 29 Recall 34 Precision on testset 1

contains alphanumeric
contains marked location
contains no between halfX verb
contains alphanumeric
31
Gleaner Algorithm

• Create B equal-sized recall bins
• For K different seeds
• Generate rules using Rapid Random Restart
• Record best rule (precision x recall)found for
  each bin
• For each recall bin B
• Find threshold L of K clauses such thatrecall of
  at least L of K clauses match examples recall
  for this bin
• Find recall and precision on testset using each
  bins at least L of K decision process