Extracting Query Facets From Search Results Date : 2013/08/20 Source : SIGIR’13 Authors : Weize...

32
Extracting Query Facets From Search Results Date : 2013/08/20 Source : SIGIR’13 Authors : Weize Kong and James Allan Advisor : Dr.Jia-ling, Koh Speaker : Wei, Chang 1

description

What is query facet ? Definition : query facet a set of coordinate terms ( terms that share a semantic relationship by being grouped under a relationship ) 3 a query facet (Mars rovers)

Transcript of Extracting Query Facets From Search Results Date : 2013/08/20 Source : SIGIR’13 Authors : Weize...

Page 1: Extracting Query Facets From Search Results Date : 2013/08/20 Source : SIGIR’13 Authors : Weize Kong and James Allan Advisor : Dr.Jia-ling, Koh Speaker.

1

Extracting Query Facets From Search ResultsDate : 2013/08/20Source : SIGIR’13Authors : Weize Kong and James AllanAdvisor : Dr.Jia-ling, KohSpeaker : Wei, Chang

Page 2: Extracting Query Facets From Search Results Date : 2013/08/20 Source : SIGIR’13 Authors : Weize Kong and James Allan Advisor : Dr.Jia-ling, Koh Speaker.

2

OUTLINE Introduction Approach Experiment Conclusion

Page 3: Extracting Query Facets From Search Results Date : 2013/08/20 Source : SIGIR’13 Authors : Weize Kong and James Allan Advisor : Dr.Jia-ling, Koh Speaker.

3

What is query facet ? Definition : query facet

a set of coordinate terms( terms that share a semantic relationship by being grouped under a relationship )

a query facet(Mars rovers)

Page 4: Extracting Query Facets From Search Results Date : 2013/08/20 Source : SIGIR’13 Authors : Weize Kong and James Allan Advisor : Dr.Jia-ling, Koh Speaker.

4

WHAT CAN WE DO WITH QUERY FACETS ?

• Flight type• Domestic• International

• Travel Class• First • Business• Economy

Page 5: Extracting Query Facets From Search Results Date : 2013/08/20 Source : SIGIR’13 Authors : Weize Kong and James Allan Advisor : Dr.Jia-ling, Koh Speaker.

5

GOAL Extract query facets from the top-k web

search results D={, , … , }

Page 6: Extracting Query Facets From Search Results Date : 2013/08/20 Source : SIGIR’13 Authors : Weize Kong and James Allan Advisor : Dr.Jia-ling, Koh Speaker.

6

OUTLINE Introduction Approach

Step 1 : Extracting candidate lists Step 2 : Finding query facets from candidate lists

Experiment Conclusion

Page 7: Extracting Query Facets From Search Results Date : 2013/08/20 Source : SIGIR’13 Authors : Weize Kong and James Allan Advisor : Dr.Jia-ling, Koh Speaker.

7

PATTERN-BASED SEMANTIC CLASS EXTRACTION Reference from : Z. Dou, S. Hu, Y. Luo, R. Song, and J.-R.

Wen. Finding dimensions for queries.

For example : There are many Mars rovers, such as Curiosity, Opportunity,

and Spirit. <ul> <li>first class</li> <li>business class</li> <li>economy class</li> </ul>

Page 8: Extracting Query Facets From Search Results Date : 2013/08/20 Source : SIGIR’13 Authors : Weize Kong and James Allan Advisor : Dr.Jia-ling, Koh Speaker.

8

CANDIDATE LISTS

The candidate lists are usually noisy, and could be non-relevant to the issued query.

To address this problem, we use a supervised method.

• All the list items are normalized by converting text to lowercase and removing non-alphanumeric characters.

• Then, we remove stopwords and duplicate items in each lists.• Finally, we discard all lists that contain fewer than two item or more than

200 items.

Page 9: Extracting Query Facets From Search Results Date : 2013/08/20 Source : SIGIR’13 Authors : Weize Kong and James Allan Advisor : Dr.Jia-ling, Koh Speaker.

9

NOTE : WHAT IS SUPERVISED METHOD

Quiz 1 Quiz 2 Quiz 3 Final Exam

John A B+ B- BEric A+ A A+ APeter B+ A- A+ A+Steve A+ A+ B- B+Mark C A+ B+ BLarry B+ B+ B+ A

LA-99 (Training Data)

LA-100 Quiz 1 Quiz 2 Quiz 3 Final

ExamDavid A- B+ A- ?James B A A ?

EXAMPLE :

Page 10: Extracting Query Facets From Search Results Date : 2013/08/20 Source : SIGIR’13 Authors : Weize Kong and James Allan Advisor : Dr.Jia-ling, Koh Speaker.

10

NOTE : WHAT IS SUPERVISED LEARNING

TrainingTraining

data (with features)

Model

New Data Model Prediction

Page 11: Extracting Query Facets From Search Results Date : 2013/08/20 Source : SIGIR’13 Authors : Weize Kong and James Allan Advisor : Dr.Jia-ling, Koh Speaker.

11

OUTLINE Introduction Approach

Step 1 : Extracting candidate lists Step 2 : Finding query facets from candidate lists

Experiment Conclusion

Page 12: Extracting Query Facets From Search Results Date : 2013/08/20 Source : SIGIR’13 Authors : Weize Kong and James Allan Advisor : Dr.Jia-ling, Koh Speaker.

12

PROBLEM DEFINITION

Whether a list item is a facet term Whether a pair of list items is in one query

facet

Page 13: Extracting Query Facets From Search Results Date : 2013/08/20 Source : SIGIR’13 Authors : Weize Kong and James Allan Advisor : Dr.Jia-ling, Koh Speaker.

13

FEATURES

Page 14: Extracting Query Facets From Search Results Date : 2013/08/20 Source : SIGIR’13 Authors : Weize Kong and James Allan Advisor : Dr.Jia-ling, Koh Speaker.

14

GRAPH

Page 15: Extracting Query Facets From Search Results Date : 2013/08/20 Source : SIGIR’13 Authors : Weize Kong and James Allan Advisor : Dr.Jia-ling, Koh Speaker.

15

LOGISTIC-BASED CONDITIONAL PROBABILITY DISTRIBUTIONS

Page 16: Extracting Query Facets From Search Results Date : 2013/08/20 Source : SIGIR’13 Authors : Weize Kong and James Allan Advisor : Dr.Jia-ling, Koh Speaker.

16

PARAMETER ESTIMATION

Maximizing the log-likelihood using gradient descent.

Page 17: Extracting Query Facets From Search Results Date : 2013/08/20 Source : SIGIR’13 Authors : Weize Kong and James Allan Advisor : Dr.Jia-ling, Koh Speaker.

17

INFERENCE The training is finished. The graphical model does not enforce the

labeling to produce strict partitioning for facet terms. For example, when=1, =1, we may have = 0.

Page 18: Extracting Query Facets From Search Results Date : 2013/08/20 Source : SIGIR’13 Authors : Weize Kong and James Allan Advisor : Dr.Jia-ling, Koh Speaker.

18

REPHRASE THE OPTIMIZATION PROBLEM

This optimization problem is NP-hard, which can be provedby a reduction from the Multiway Cut problem. Therefore, we propose two algorithms, QF-I and QF-J, to approximate the results.

The optimization target becomes , where is the set of all possible query facet sets that can be generated from L with the strict partitioning constraint.

Page 19: Extracting Query Facets From Search Results Date : 2013/08/20 Source : SIGIR’13 Authors : Weize Kong and James Allan Advisor : Dr.Jia-ling, Koh Speaker.

19

QF-I1. Select list items with as facet terms.2.

Page 20: Extracting Query Facets From Search Results Date : 2013/08/20 Source : SIGIR’13 Authors : Weize Kong and James Allan Advisor : Dr.Jia-ling, Koh Speaker.

20

QF-J

Page 21: Extracting Query Facets From Search Results Date : 2013/08/20 Source : SIGIR’13 Authors : Weize Kong and James Allan Advisor : Dr.Jia-ling, Koh Speaker.

21

RANKING QUERY FACETS

score for a query facet :

score for a facet term :

Page 22: Extracting Query Facets From Search Results Date : 2013/08/20 Source : SIGIR’13 Authors : Weize Kong and James Allan Advisor : Dr.Jia-ling, Koh Speaker.

22

OUTLINE Introduction Approach

Step 1 : Extracting candidate lists Step 2 : Finding query facets from candidate lists

Experiment Evaluation Experiment Result

Conclusion

Page 23: Extracting Query Facets From Search Results Date : 2013/08/20 Source : SIGIR’13 Authors : Weize Kong and James Allan Advisor : Dr.Jia-ling, Koh Speaker.

23

DATA

Using Top 10 query facets generated by different models.

Page 24: Extracting Query Facets From Search Results Date : 2013/08/20 Source : SIGIR’13 Authors : Weize Kong and James Allan Advisor : Dr.Jia-ling, Koh Speaker.

24

EVALUATION METRICS Using “∗” to distinguish between system

generated results and human labeled results, which we used as ground truth.

Page 25: Extracting Query Facets From Search Results Date : 2013/08/20 Source : SIGIR’13 Authors : Weize Kong and James Allan Advisor : Dr.Jia-ling, Koh Speaker.

25

CLUSTERING QUALITY

Page 26: Extracting Query Facets From Search Results Date : 2013/08/20 Source : SIGIR’13 Authors : Weize Kong and James Allan Advisor : Dr.Jia-ling, Koh Speaker.

26

OVERALL QUALITY

fp-nDCG is weighted by rp-nDCG is weighted by

Page 27: Extracting Query Facets From Search Results Date : 2013/08/20 Source : SIGIR’13 Authors : Weize Kong and James Allan Advisor : Dr.Jia-ling, Koh Speaker.

27

OUTLINE Introduction Approach

Step 1 : Extracting candidate lists Step 2 : Finding query facets from candidate lists

Experiment Evaluation Experiment Result

Conclusion

Page 28: Extracting Query Facets From Search Results Date : 2013/08/20 Source : SIGIR’13 Authors : Weize Kong and James Allan Advisor : Dr.Jia-ling, Koh Speaker.

28

FACET TERMS

Page 29: Extracting Query Facets From Search Results Date : 2013/08/20 Source : SIGIR’13 Authors : Weize Kong and James Allan Advisor : Dr.Jia-ling, Koh Speaker.

29

CLUSTERING FACET TERMS

Page 30: Extracting Query Facets From Search Results Date : 2013/08/20 Source : SIGIR’13 Authors : Weize Kong and James Allan Advisor : Dr.Jia-ling, Koh Speaker.

30

OVERALL

Page 31: Extracting Query Facets From Search Results Date : 2013/08/20 Source : SIGIR’13 Authors : Weize Kong and James Allan Advisor : Dr.Jia-ling, Koh Speaker.

31

OUTLINE Introduction Approach

Step 1 : Extracting candidate lists Step 2 : Finding query facets from candidate lists

Experiment Evaluation Experiment Result

Conclusion

Page 32: Extracting Query Facets From Search Results Date : 2013/08/20 Source : SIGIR’13 Authors : Weize Kong and James Allan Advisor : Dr.Jia-ling, Koh Speaker.

32

CONCLUSION We developed a supervised method based on a

graphical model to recognize query facets from the noisy facet candidate lists extracted from the top ranked search results.

We proposed two algorithms for approximate inference on the graphical model.

We designed a new evaluation metric for this task to combine recall and precision of facet terms with grouping quality.

Experimental results showed that the supervised method significantly outperforms other unsupervised methods, suggesting that query facet extraction can be effectively learned.