Title: Exploration%20of%20a%20Heuristic%20Approach%20to%20Threshold%20Learning%20in%20Adaptive%20Filtering
1Exploration of a Heuristic Approach to Threshold
Learning in Adaptive Filtering
- Chengxiang Zhai, Peter Jansen, David A. Evans
- Claritech Corporation
- (Now Clairvoyance Corporation)
c.zhai, p.jansen, dae_at_clairvoyancecorp.com
Clairvoyance Corporation
2Threshold Learning in Adaptive Filtering
no
doc vector
Utility Evaluation
Scoring
Thresholding
yes
profile vector
threshold
Vector Learning
Threshold Learning
Feedback Information
3Delivery-ratio Initial Threshold Setting
- Select a threshold corresponding to a given ratio
of acceptance/delivery - No relevance judgments needed
Delivery-ratio 0.2 1/5
Profile vector
D1 25.5 D2 23.2 ... Dk 11.3
Scoring
Ref corpus
4Beta-Gamma Threshold Learning
- Optimize utility function
- Exploit relevance judgments
5Research Questions
- Learning and initial inaccuracies Can learning
compensate for initial inaccuracies? - Exploitation vs. exploration Does exploration
(lowering threshold) pay off in the long run?
score
ideal adaptive
ideal fixed
actual adaptive
actual fixed
time
6Experiment Design
- Data collection AP88 Topics 1-50
- Utility function LF1 3R - 2R-
- IDF statistics WSJ87
- Fixed profile vector (no vector learning,?0)
- Threshold updated at every accepted doc
- Varying two parameters fix others
- initial threshold level (ratio 0.002, 0.001,
0.0005, 0.00025, 0.000125) - learning convergence speed (gamma 0.001, 0.01,
0.1,0.5)
7Learning Effect 1 Correction of Inappropriate
Initial Threshold Setting
bad initial threshold without updating
bad initial threshold with updating
8Learning Effect 2 Early Exploration Pays Off
9Learning Effect 3 Regular Exploration Pays Off
Later
10Tradeoff between Exploration and Exploitation
under-explore
over-explore
11Conclusions Future Work
- Conclusions
- Learning helps correct inappropriate initial
threshold (threshold too low) - Exploration (deliberately lowering the threshold)
pays off in the long run - For a given length of document stream, the
optimal performance is achieved at the right
tradeoff between exploration and exploitation - Future work
- Formalize exploitation-exploration tradeoff
(reinforcement learning?)