Presentation is loading. Please wait.

Presentation is loading. Please wait.

Query Processing over Incomplete Autonomous Web Databases MS Thesis Defense by Hemal Khatri Committee Members: Prof. Subbarao Kambhampati (chair) Prof.

Similar presentations


Presentation on theme: "Query Processing over Incomplete Autonomous Web Databases MS Thesis Defense by Hemal Khatri Committee Members: Prof. Subbarao Kambhampati (chair) Prof."— Presentation transcript:

1 Query Processing over Incomplete Autonomous Web Databases MS Thesis Defense by Hemal Khatri Committee Members: Prof. Subbarao Kambhampati (chair) Prof. Chitta Baral Prof. Yi Chen Prof. Huan Liu

2 Introduction to Web databases n Many websites allow user query through a form based interface and are supported by backend databases n Consider used cars selling websites such as Cars.com, Yahoo! autos, etc

3 Incompleteness in Web databases n Web databases are often input by lay individuals without any curation. For e.g. Cars.com, Yahoo! Autos n Web databases are being populated using automated information extraction techniques which are inherently imperfect n The local schema of data sources may not support certain attributes supported by the global schema n Incomplete/Uncertain tuple: A tuple in which one or more of its attributes have a missing value Website# of attributes # of tuples incomplete tuples body styleengine autotrader.com 132512733.67%3.6%8.1% carsdirect.com 143256498.74%55.7%55.8%

4 Problem Statement n Many entities corresponding to tuples with missing values might be relevant to the user query n Current query processing techniques return answers that exactly satisfy the user query –Such techniques return results with high precision but low recall n Relevant Uncertain tuple: A tuple which does not exactly satisfy the query predicates but the entity represented by that tuple might be relevant to the query n How to support query processing over incomplete autonomous databases in order to retrieve ranked uncertain results? nullAccord2003sedan Q:Make=Honda

5 Challenges Involved n How to predict missing values in autonomous databases? n As autonomous databases are accessible only through form-based interfaces, how to retrieve relevant uncertain answers? –How to keep query processing cost manageable in retrieving uncertain tuples? n How to rank the retrieved uncertain answers?

6 Related Work n Probabilistic databases –Incomplete databases are similar to probabilistic databases once we assess the probabilities for missing values –TRIO: uncertainty with lineage –ConQuer: handling inconsistency over databases Assume probability distributions are given for uncertain or inconsistent attributes –We assess probability distribution for missing attribute and use it to rank rewritten queries to retrieve relevant answers since the probabilities cannot be stored in databases –Our query rewriting framework is general and can be used by these systems if the databases are autonomous n Handling Missing Values –EM algorithm, Bayes Net, Association rules

7 Possible Approaches For a query Q:body style = convt 1.Certain Answers Only (CAO): Return certain answers only as in traditional databases 2. All Uncertain Answers (AUA): Null matches any concrete value, hence return all answers having body style=convt along with answers having body style as null 3. Relevant Uncertain Answers (RUA): Ranking answers by predicting values of missing attribute Low Recall Low Precision, infeasible Costly, infeasible

8 Outline n Introduction n QPIAD: Query Processing over Incomplete Autonomous Databases n Data Integration over Incomplete Autonomous Databases n Other Contributions n Conclusion

9 QPIAD System Architecture

10 RRUA: Generating Rewritten Queries n Restricted Relevant Uncertain Answers (RRUA) approach only retrieves only relevant incomplete tuples instead of retrieving all tuples as in AUA and RUA Consider a query Q:Body style=convt MakeModelYearPriceBody style Audia4200420000convt BMWz4200317000convt Porscheboxster200013000convt …..…… Rewritten queries are based on the determining set from AFD for Body style: Model ~~> Body style:0.9 Q 1 :model=‘a4’ Q 2 :model=‘z4’ Q 3 :model=‘boxster’ Determining Attribute set(dtrSet) Base Result Set:RS(Q)

11 Learning Attribute Correlations n AFD: VIN ~~> Model where VIN is an Approximate Key(AKey) with high confidence n VIN will not be useful for query rewriting and feature selection since it will not be able to retrieve additional new tuples

12 RRUA: Ranking Rewritten Queries n All queries may not be equally good in retrieving relevant answers –“z4” model cars are more likely to be convertibles than a car with “a4” model n When database or network resources are limited, the mediator can choose to issue the top K queries to get the most relevant uncertain answers

13 Learning Value Distributions n Used to rank queries based on the determining set of attributes from the AFD for query attribute n We use Naïve Bayes Classifier with m- estimates with AFD as a feature selection step n Rank of a rewritten query Q i = P(A m =v m |t i ), where t i ε П dtrSet(Am) (RS(Q)) –Q 1 :model=‘a4’, R(Q 1 ) = P(bodystyle=convt|model=a4) = 0.4 –Q 2 :model=‘z4’, R(Q 2 ) = P(bodystyle=convt|model=z4)= 1.0 –Q 3 :model=‘boxster’, R(Q 3 ) = P(bodystyle=convt|model=boxster)=0.7 R(Q 2 ) > R(Q 3 ) > R(Q 1 ) n Relevant uncertain answers are ranked based on the rank of the rewritten query that retrieved it

14 Combining AFDs and Classifiers n More than one AFD may exist for some attributes n Experimented with several approaches: –Only best-AFD having highest confidence –All attributes ignoring AFDs –Hybrid One-AFD –Ensemble of classifiers

15 Empirical Evaluation of QPIAD n Test Databases: AutoTrader database containing 100K tuples and Census database from UCI Repository containing 50K tuples n Oracular study: To evaluate the effectiveness of our system against a ground truth, we artificially insert missing values in 10% of the tuples within these databases

16 RRUA vs AUA vs RUA

17 Precision over Top K Tuples

18 Ranking the Rewritten Queries Cars database Census database

19 Robustness of QPIAD

20 User Relevance Issues with QPIAD n When the query processor presents incomplete tuples, it becomes a recommender system For a query Q:year=2000 n How to convince users into believing the system results? MakeModelYearPriceMileage HondaCivicnull1500018000 Explanation We have determined that this car’s year is 60% likely to 2000 based on price=15000 and mileage=18000

21 Outline n Introduction n QPIAD: Query Processing over Incomplete Autonomous Databases n Data Integration over Incomplete Autonomous Databases n Other Contributions n Conclusion

22 Leveraging Correlations between Data Sources Mediator:GS(Make,Model,Year,Price,Mileage,Bodystyle) Q:Body style=coupe

23 Correlated Source and Maximum Correlated Source n Consider four sources with schema: –S1(Make,Model,Year,Price) –S2(Engine,Drive,Bodystyle), AFD: {Engine, Drive} -> Body style confidence 0.7 –S3(Make,Model,Body style) AFD: Model -> Body style confidence 0.8 –S4(Make,Price,Body style) AFD: {Make, Price} -> Body Style confidence 0.6 –Mediator global schema GS(Make,Model,Year,Price, Bodystyle, Engine, Drive) n S3 and S4 are correlated sources with S1 on Body style attribute n S3 is the maximum correlated source for S1 on Body style attribute

24 Retrieving Relevant Uncertain Answers from CarsDirect.com Consider a query Q:body style = coupe(GS) n Cars.com has an AFD: Model ~~> Body style(0.9) n Cars.com is the maximum correlated source for CarsDirect.com which doesn’t support Body style but supports Model attribute MakeModelYearPriceBody style HondaAccord200319000coupe FordMustang200429100coupe AcuraLegend199712000coupe BMW325200328000coupe Q 1 :model=Accord Q 2 :model=Mustang Q 3 :model=Legend Q 4 :model=325

25 Empirical Evaluation of using Correlation between Data Sources n We consider a mediator performing data integration over three sources: Cars.com, Yahoo! Autos and CarsDirect.com n Yahoo! Autos and CarsDirect.com do not allow querying on body style but when the tuples are retrieved we can check the body style attribute to determine if the tuple retrieved has the body style specified in the query n Evaluation using attribute correlations and value distributions learned from Cars.com for 5 test queries on body style attribute

26 Retrieving Relevant Answers using Correlations from Cars.com

27 Handling Joins over Incomplete Autonomous databases n Mediator performing data integration across two sources: –Source S1 is incomplete –Source S2 is complete SourceLocal Schema S1Cars(Make,Model,Year,Price) S2Review(Model,Ratings) Mediator View UsedCars(Make,Model,Year,Price,Ratings) :- Cars(Make,Model,Year,Price), Review(Model, Ratings)

28 Issues in Handling Joins n Performing joins over probabilistic databases will lead to a disjunction in join results n Consider joining uncertain tuples from the two sources: MakeModelYearPrice Hondanull [0.6 Civic] [0.4 Accord] 200318000 ModelRatings Civic5 Accord4 MakeModelYearPriceRatings HondaCivic2003180005 HondaAccord2003180004 or 0.6 0.4 Approximation

29 Handling Join Queries n Q:σ Make=Honda (UsedCars) n Assume AFDs: {Make,Year} ~~> Model, Model ~~> Make MakeModel(FK)YearPrice HondaOdyssey200010000 HondaAccord200420000 Hondanull200015000 nullAccord200218000 ToyotaCamry200316000 Model(PK)Ratings Civic5 Corolla4 Accord4 Altima3 Camry5 Odyssey3 HondaOdyssey2000100003 HondaAccord2004200004 nullAccord2002180004 Hondanull2000150005 1.0 0.6 Q 1 : Model=Odyssey:R(Q 1 )=1 Q 2 : Model=Accord:R(Q 2 )=1 0.6 Civic 0.4 Accord Queries on source S2 to join Q 3 :Model=Odyssey:R(Q 3 )=1 Q 4 :Model=Accord:R(Q 4 )=1 Q 5 :Model=Civic:R(Q 5 )=0.6

30 Experimental Results Joins

31 Outline n Introduction n QPIAD: Query Processing over Incomplete Autonomous Databases n Data Integration over Incomplete Autonomous Databases n Other Contributions n Conclusion

32 QUIC: Querying under Imprecision and Incompleteness Consider a query Q:model like Civic(Cars) n User might be interested in similar cars like “Accord”, ”Camry”, etc n Ranking results in presence of both similar and incomplete tuples IdMakeModelYearBody style 1HondaCivic2000Sedan 2HondaAccord2004Coupe 3ToyotaCamry2001Sedan 4Hondanull2004Coupe 5Hondanull2000Sedan 6HondaCivic2004Coupe 7BMW3series2001convt 8Toyotanull1999sedan

33 Other Contributions[*Collaboration with Garrett Wolf] n Handling multi-attribute selection queries for incomplete databases* n QUIC system for query processing under imprecision and incompleteness n Online learning of value distribution based on base result set to avoid sample biases

34 Conclusion n Thesis proposed a framework for query processing over incomplete autonomous web databases: –QPIAD: Query processing over incomplete autonomous databases –QPIAD: Data Integration over multiple incomplete data sources n Results of empirical evaluation on real world databases show that our system returns relevant answers with high precision while keeping the query processing cost manageable

35 Thank You!! Questions??


Download ppt "Query Processing over Incomplete Autonomous Web Databases MS Thesis Defense by Hemal Khatri Committee Members: Prof. Subbarao Kambhampati (chair) Prof."

Similar presentations


Ads by Google