hypertree decompositions g. gottlob technical university of vienna, austria n. leone and f....
TRANSCRIPT
![Page 1: Hypertree Decompositions G. Gottlob Technical University of Vienna, Austria N. Leone and F. Scarcello University of Calabria, Italy For papers and further](https://reader036.vdocument.in/reader036/viewer/2022062515/56649cf05503460f949bf3ec/html5/thumbnails/1.jpg)
Hypertree Decompositions
G. Gottlob Technical University of Vienna, Austria
N. Leone and F. ScarcelloUniversity of Calabria, Italy
For papers and further material see:http://ulisse.deis.unical.it/~frank/Hypertrees/
![Page 2: Hypertree Decompositions G. Gottlob Technical University of Vienna, Austria N. Leone and F. Scarcello University of Calabria, Italy For papers and further](https://reader036.vdocument.in/reader036/viewer/2022062515/56649cf05503460f949bf3ec/html5/thumbnails/2.jpg)
Three Problems:
HOM: The homomorphism problem
CSP: Constraint satisfaction problem
BCQ: Boolean conjunctive query evaluation
Important problems in different areas.All these problems are hypergraph based.
But actually: HOM = BCQ = CSP
![Page 3: Hypertree Decompositions G. Gottlob Technical University of Vienna, Austria N. Leone and F. Scarcello University of Calabria, Italy For papers and further](https://reader036.vdocument.in/reader036/viewer/2022062515/56649cf05503460f949bf3ec/html5/thumbnails/3.jpg)
The Homomorphism Problem
),...,,,(
),...,,,(
21
21
k
k
SSSVB
RRRUA
Given two relational structures
Decide whether there exists a homomorphism h from A to B
ii ShR
i
VUh
)(
,thatsuch
:
xx
x
![Page 4: Hypertree Decompositions G. Gottlob Technical University of Vienna, Austria N. Leone and F. Scarcello University of Calabria, Italy For papers and further](https://reader036.vdocument.in/reader036/viewer/2022062515/56649cf05503460f949bf3ec/html5/thumbnails/4.jpg)
HOM is NP-complete
Membership: Obvious, guess h.
Hardness: Transformation from 3COL.
(well-known)
3
21
45
6
1 2
1
32
3
3 4
2 5
4 5
3 6
red
greenredred
redgreen
green
green
blue
blueblueblue
AB
Graph 3-colourable iff HOM(A,B ) yes-instance.
![Page 5: Hypertree Decompositions G. Gottlob Technical University of Vienna, Austria N. Leone and F. Scarcello University of Calabria, Italy For papers and further](https://reader036.vdocument.in/reader036/viewer/2022062515/56649cf05503460f949bf3ec/html5/thumbnails/5.jpg)
HOM is NP-complete
Membership: Obvious, guess h.
Hardness: Transformation from 3COL.
(well-known, independently proved in various contexts)
3
21
45
6
1 2
1
32
3
3 4
2 5
4 5
3 6
red
greenredred
redgreen
green
green
blue
blueblueblue
AB
hh
Graph 3-colourable iff HOM(A,B ) yes-instance.
![Page 6: Hypertree Decompositions G. Gottlob Technical University of Vienna, Austria N. Leone and F. Scarcello University of Calabria, Italy For papers and further](https://reader036.vdocument.in/reader036/viewer/2022062515/56649cf05503460f949bf3ec/html5/thumbnails/6.jpg)
Conjunctive Queries, CSPsDatabase schema (scopes): Enrolled (Pers#, Course, Reg-Date) Teaches (Pers#, Course, Assigned) Parent (Pers1, Pers2)
Is there any teacher having a child enrolled in her course?ans Enrolled(S,C,R) Teaches(P,C,A)
Parent(P,S)
![Page 7: Hypertree Decompositions G. Gottlob Technical University of Vienna, Austria N. Leone and F. Scarcello University of Calabria, Italy For papers and further](https://reader036.vdocument.in/reader036/viewer/2022062515/56649cf05503460f949bf3ec/html5/thumbnails/7.jpg)
Conjunctive Queries, CSPs (2)
Database schema (scopes): Enrolled (Pers#, Course, Reg-Date) Teaches (Pers#, Course, Assigned) Parent (Pers1, Pers2)
Is there any teacher whose child attend some course?ans Enrolled(S,C’,R) Teaches(P,C,A)
Parent(P,S)
![Page 8: Hypertree Decompositions G. Gottlob Technical University of Vienna, Austria N. Leone and F. Scarcello University of Calabria, Italy For papers and further](https://reader036.vdocument.in/reader036/viewer/2022062515/56649cf05503460f949bf3ec/html5/thumbnails/8.jpg)
BCQ = HOMView query Q (=scopes) as a relational structure
Universe: Variables of the queryRelations: sets of query-atoms for each database relation.
The database D is itself a relational structure.The Boolean conjunctive query (CSP) is equivalen to the HOM instance HOM(Q,D ).
Vive-versa, every HOM instance can be reformulated as a Boolean Conjunctive Query (CSP).
This talk will mainly concentrate on BCQ
![Page 9: Hypertree Decompositions G. Gottlob Technical University of Vienna, Austria N. Leone and F. Scarcello University of Calabria, Italy For papers and further](https://reader036.vdocument.in/reader036/viewer/2022062515/56649cf05503460f949bf3ec/html5/thumbnails/9.jpg)
Queries, CSPs, and Hypergraphs
ans Enrolled(S,C,R) Teaches(P,C,A) Parent(P,S)
S
C
AR
P
![Page 10: Hypertree Decompositions G. Gottlob Technical University of Vienna, Austria N. Leone and F. Scarcello University of Calabria, Italy For papers and further](https://reader036.vdocument.in/reader036/viewer/2022062515/56649cf05503460f949bf3ec/html5/thumbnails/10.jpg)
Queries, CSPs, and Hypergraphs
ans Enrolled(S,C’,R) Teaches(P,C,A) Parent(P,S)
S
C’
AR
P
C
![Page 11: Hypertree Decompositions G. Gottlob Technical University of Vienna, Austria N. Leone and F. Scarcello University of Calabria, Italy For papers and further](https://reader036.vdocument.in/reader036/viewer/2022062515/56649cf05503460f949bf3ec/html5/thumbnails/11.jpg)
Boolean Conjunctive Queries
The problem BCQ ( = constraint satisiability)
Instance: < DB, Q> (= <Relations, Scope>)
Question: Has Q a nonempty result over DB?
Combined Complexity(Vardi ’82)
![Page 12: Hypertree Decompositions G. Gottlob Technical University of Vienna, Austria N. Leone and F. Scarcello University of Calabria, Italy For papers and further](https://reader036.vdocument.in/reader036/viewer/2022062515/56649cf05503460f949bf3ec/html5/thumbnails/12.jpg)
Problems Equivalent to BCQ
Conjunctive Query Containment
Query of Tuple Problem
Constraint Satisfaction in AI
Clause Subsumption in Theorem Proving
)()( 2121
dbQdbQdbQQ
? )( dbQt
? s.t. DC
BCQ CSP
![Page 13: Hypertree Decompositions G. Gottlob Technical University of Vienna, Austria N. Leone and F. Scarcello University of Calabria, Italy For papers and further](https://reader036.vdocument.in/reader036/viewer/2022062515/56649cf05503460f949bf3ec/html5/thumbnails/13.jpg)
Example of CSP: Crossword Puzzle
![Page 14: Hypertree Decompositions G. Gottlob Technical University of Vienna, Austria N. Leone and F. Scarcello University of Calabria, Italy For papers and further](https://reader036.vdocument.in/reader036/viewer/2022062515/56649cf05503460f949bf3ec/html5/thumbnails/14.jpg)
Complexity of BCQNP-complete in the general case (Chandra and Merlin ’77)NP-hard even for fixed database
Polynomial if Q has an acyclic hypergraph(Yannakakis ’81)LOGCFL-complete (in NC2) (G.L.S. ’98)
Interest in larger tractable classes of CQS
![Page 15: Hypertree Decompositions G. Gottlob Technical University of Vienna, Austria N. Leone and F. Scarcello University of Calabria, Italy For papers and further](https://reader036.vdocument.in/reader036/viewer/2022062515/56649cf05503460f949bf3ec/html5/thumbnails/15.jpg)
Teaches(P,C,A)
Parent(P,S)
Acyclic queries or CSPs
ans Enrolled(S,C’,R) Teaches(P,C,A) Parent(P,S)
S
C’
AR
P
C
Enrolled(S,C’,R)
Join Tree
![Page 16: Hypertree Decompositions G. Gottlob Technical University of Vienna, Austria N. Leone and F. Scarcello University of Calabria, Italy For papers and further](https://reader036.vdocument.in/reader036/viewer/2022062515/56649cf05503460f949bf3ec/html5/thumbnails/16.jpg)
Theorem [GLS99]: Answering acyclic BCQs is LOGCFL-complete
LOGCFL: class of problems/languages that are logspace-reducible to a CFL
NPPACNCNCACSACLOGCFLNLAC 2110
Characterization of LOGCFL [Ruzzo80]:
LOGCFL = Class of all problems solvable with a logspace ATM with polynomial tree-size
LOGCFL
![Page 17: Hypertree Decompositions G. Gottlob Technical University of Vienna, Austria N. Leone and F. Scarcello University of Calabria, Italy For papers and further](https://reader036.vdocument.in/reader036/viewer/2022062515/56649cf05503460f949bf3ec/html5/thumbnails/17.jpg)
![Page 18: Hypertree Decompositions G. Gottlob Technical University of Vienna, Austria N. Leone and F. Scarcello University of Calabria, Italy For papers and further](https://reader036.vdocument.in/reader036/viewer/2022062515/56649cf05503460f949bf3ec/html5/thumbnails/18.jpg)
Is this query hard?
),','(),',()',',,,(
)','()','()',',(),(
),(),',()',',',,(),,',,(
FXBqFXBpYXYXJj
ZYhZXgZFFfZYe
ZXdZCCcFCYYSbFCXXSaans
n size of the databasem number of atoms in the query
• Classical methods worst-case complexity:O(n
m)
m = 11 !
• Despite its apparence, this query is nearly acyclic
It can be evaluated in O(m·n 2· logn)
![Page 19: Hypertree Decompositions G. Gottlob Technical University of Vienna, Austria N. Leone and F. Scarcello University of Calabria, Italy For papers and further](https://reader036.vdocument.in/reader036/viewer/2022062515/56649cf05503460f949bf3ec/html5/thumbnails/19.jpg)
),','(),',()',',,,(
)','()','()',',(),(
),(),',()',',',,(),,',,(
FXBqFXBpYXYXJj
ZYhZXgZFFfZYe
ZXdZCCcFCYYSbFCXXSaans
It can be evaluated in O(m·n 2· logn)
S
X X’ C F
Y Y’ C’ F’
Z Z’J
B B’
![Page 20: Hypertree Decompositions G. Gottlob Technical University of Vienna, Austria N. Leone and F. Scarcello University of Calabria, Italy For papers and further](https://reader036.vdocument.in/reader036/viewer/2022062515/56649cf05503460f949bf3ec/html5/thumbnails/20.jpg)
Nearly Acyclic Queries & CSPs
Bounded Treewidth (tw) a measure of the cyclicity of graphs for queries: tw(Q) = tw(G(Q))
For fixed k: checking tw(Q) k Computing a tree decomposition
linear time(Bodlaender’96)
Answering BCQ of treewidth k:O(nk log n) (Chekuri & Rajaraman’97, Kolaitis & Vardi,
98)
LOGCFL-complete (G.L.S.’98)
![Page 21: Hypertree Decompositions G. Gottlob Technical University of Vienna, Austria N. Leone and F. Scarcello University of Calabria, Italy For papers and further](https://reader036.vdocument.in/reader036/viewer/2022062515/56649cf05503460f949bf3ec/html5/thumbnails/21.jpg)
Primal graphs of Queries
ans Enrolled(S,C,R) Teaches(P,C,A) Parent(P,S)
S
C
AR
P
C
AR
S P
Hypergraph H(Q) Primal graph G(Q)
![Page 22: Hypertree Decompositions G. Gottlob Technical University of Vienna, Austria N. Leone and F. Scarcello University of Calabria, Italy For papers and further](https://reader036.vdocument.in/reader036/viewer/2022062515/56649cf05503460f949bf3ec/html5/thumbnails/22.jpg)
Example: a cyclic graph
g
q
ab
f
c
d
p hl
nm
ok
e
i
j
![Page 23: Hypertree Decompositions G. Gottlob Technical University of Vienna, Austria N. Leone and F. Scarcello University of Calabria, Italy For papers and further](https://reader036.vdocument.in/reader036/viewer/2022062515/56649cf05503460f949bf3ec/html5/thumbnails/23.jpg)
A tree decomposition of width 2
ah
ahq
hij abc
hkl
hkp klo
mno
bcdcefag
![Page 24: Hypertree Decompositions G. Gottlob Technical University of Vienna, Austria N. Leone and F. Scarcello University of Calabria, Italy For papers and further](https://reader036.vdocument.in/reader036/viewer/2022062515/56649cf05503460f949bf3ec/html5/thumbnails/24.jpg)
Connectedness condition for h
ah
ahq
hij abc
hkl
hkp klo
mno
bcdcefag
![Page 25: Hypertree Decompositions G. Gottlob Technical University of Vienna, Austria N. Leone and F. Scarcello University of Calabria, Italy For papers and further](https://reader036.vdocument.in/reader036/viewer/2022062515/56649cf05503460f949bf3ec/html5/thumbnails/25.jpg)
Game characterization of Treewidth
A robber and k cops play the game on a graph
The cops have to capture the robber
Each cop controls a vertex of the graph
Each cop, at any time, can fly to any vertex of the graph
The robber tries to elude her capture, by running arbitrarily fast on the vertices of the graph,but on those vertices controlled by cops
![Page 26: Hypertree Decompositions G. Gottlob Technical University of Vienna, Austria N. Leone and F. Scarcello University of Calabria, Italy For papers and further](https://reader036.vdocument.in/reader036/viewer/2022062515/56649cf05503460f949bf3ec/html5/thumbnails/26.jpg)
Playing the game
g
q
ab
f
c
d
p hl
nm
ok
e
i
j
![Page 27: Hypertree Decompositions G. Gottlob Technical University of Vienna, Austria N. Leone and F. Scarcello University of Calabria, Italy For papers and further](https://reader036.vdocument.in/reader036/viewer/2022062515/56649cf05503460f949bf3ec/html5/thumbnails/27.jpg)
Playing the game
g
q
ab
f
c
d
p hl
nm
ok
e
i
j
![Page 28: Hypertree Decompositions G. Gottlob Technical University of Vienna, Austria N. Leone and F. Scarcello University of Calabria, Italy For papers and further](https://reader036.vdocument.in/reader036/viewer/2022062515/56649cf05503460f949bf3ec/html5/thumbnails/28.jpg)
Playing the game
g
q
ab
f
c
d
p hl
nm
ok
e
i
j
![Page 29: Hypertree Decompositions G. Gottlob Technical University of Vienna, Austria N. Leone and F. Scarcello University of Calabria, Italy For papers and further](https://reader036.vdocument.in/reader036/viewer/2022062515/56649cf05503460f949bf3ec/html5/thumbnails/29.jpg)
Playing the game
g
q
ab
f
c
d
p hl
nm
ok
e
i
j
![Page 30: Hypertree Decompositions G. Gottlob Technical University of Vienna, Austria N. Leone and F. Scarcello University of Calabria, Italy For papers and further](https://reader036.vdocument.in/reader036/viewer/2022062515/56649cf05503460f949bf3ec/html5/thumbnails/30.jpg)
Logical characterization of Treewidth
Logic Querying Basic FO ,L
vars.1most at with : k LL 1k
)disallowed , ,( ,on based FO : Logic L
(Kolaitis & Vardi ’98)1kL ]TW[k
![Page 31: Hypertree Decompositions G. Gottlob Technical University of Vienna, Austria N. Leone and F. Scarcello University of Calabria, Italy For papers and further](https://reader036.vdocument.in/reader036/viewer/2022062515/56649cf05503460f949bf3ec/html5/thumbnails/31.jpg)
Hypergraphs vs Graphs (1)
S
C’
AR
P
C C
AR
S P
C’
An acyclic hypergraph Its cyclic primal graph
![Page 32: Hypertree Decompositions G. Gottlob Technical University of Vienna, Austria N. Leone and F. Scarcello University of Calabria, Italy For papers and further](https://reader036.vdocument.in/reader036/viewer/2022062515/56649cf05503460f949bf3ec/html5/thumbnails/32.jpg)
Hypergraphs vs Graphs (1)
S
C’
AR
P
C C
AR
S P
C’
There are two cliques.We cannot know where they come from
![Page 33: Hypertree Decompositions G. Gottlob Technical University of Vienna, Austria N. Leone and F. Scarcello University of Calabria, Italy For papers and further](https://reader036.vdocument.in/reader036/viewer/2022062515/56649cf05503460f949bf3ec/html5/thumbnails/33.jpg)
Drawbacks of treewidth
Acyclic queries may have unbounded TW!
Example:q p1(X1, X2,…, Xn) … pm(X1, X2,…, Xn)
is acyclic, obviously polynomial, but has treewidth n-1
![Page 34: Hypertree Decompositions G. Gottlob Technical University of Vienna, Austria N. Leone and F. Scarcello University of Calabria, Italy For papers and further](https://reader036.vdocument.in/reader036/viewer/2022062515/56649cf05503460f949bf3ec/html5/thumbnails/34.jpg)
Beyond treewidth
Bounded Degree of Cyclicity
Bounded Query width
(Gyssens & Paredaens ’84)
(Chekuri & Rajaraman ’97)
Group together query atoms (hyperedges) instead of variables
![Page 35: Hypertree Decompositions G. Gottlob Technical University of Vienna, Austria N. Leone and F. Scarcello University of Calabria, Italy For papers and further](https://reader036.vdocument.in/reader036/viewer/2022062515/56649cf05503460f949bf3ec/html5/thumbnails/35.jpg)
Query Decompositionq p1(X1, X2,…, Xn) … pm(X1, X2,…, Xn)
p1(X1, X2,…, Xn)
p1(X1, X2,…, Xn) p2(X1, X2,…, Xn) pm(X1, X2,…, Xn)
Query width = 1
• Every atom appears in some node• Connectedness conditions for variables and atoms
![Page 36: Hypertree Decompositions G. Gottlob Technical University of Vienna, Austria N. Leone and F. Scarcello University of Calabria, Italy For papers and further](https://reader036.vdocument.in/reader036/viewer/2022062515/56649cf05503460f949bf3ec/html5/thumbnails/36.jpg)
Decomposition of cyclic queriesq s(Y,Z,U) g(X,Y) t(Z,X) s(Z,W,X) t(Y,Z)
g(X,Y), t(Y,Z)
t(Z,X)
s(Z,W,X)
s(Y,Z,U)
Query width = 2
BCQ is polynomial for queries of bounded query width, if a query decomposition is given
![Page 37: Hypertree Decompositions G. Gottlob Technical University of Vienna, Austria N. Leone and F. Scarcello University of Calabria, Italy For papers and further](https://reader036.vdocument.in/reader036/viewer/2022062515/56649cf05503460f949bf3ec/html5/thumbnails/37.jpg)
Open Problems by Chekuri & Rajaraman ‘97
Are the following problems solvable in polynomial time for fixed k ?
Decide whether Q has query width at most k
Compute a query decomposition of Q of width k
![Page 38: Hypertree Decompositions G. Gottlob Technical University of Vienna, Austria N. Leone and F. Scarcello University of Calabria, Italy For papers and further](https://reader036.vdocument.in/reader036/viewer/2022062515/56649cf05503460f949bf3ec/html5/thumbnails/38.jpg)
A negative answer (G.L.S. ’99)
Theorem: Deciding whether a query has query width at most k isNP-complete
Proof: Very involved reduction fromEXACT COVERING BY 3-SETS
![Page 39: Hypertree Decompositions G. Gottlob Technical University of Vienna, Austria N. Leone and F. Scarcello University of Calabria, Italy For papers and further](https://reader036.vdocument.in/reader036/viewer/2022062515/56649cf05503460f949bf3ec/html5/thumbnails/39.jpg)
Important Observation
NP-hardness id due to an overly strong conditionin the definition of query decomposition
p(X,Y,Z), c(T,W)
d(X,T)
a(X,U,W), b(Y,V,W)
c(Y,T)
p(X,Y,Z), q(U,V,Z)
Forbidden !
![Page 40: Hypertree Decompositions G. Gottlob Technical University of Vienna, Austria N. Leone and F. Scarcello University of Calabria, Italy For papers and further](https://reader036.vdocument.in/reader036/viewer/2022062515/56649cf05503460f949bf3ec/html5/thumbnails/40.jpg)
Important ObservationBut the reuse of p(X,Y,Z) is harmless here:
we could added an atom p(X,Y,Z’) without changing the query
p(X,Y,Z), c(T,W)
d(X,T)
a(X,U,W), b(Y,V,W)
c(Y,T)
p(X,Y,Z), q(U,V,Z)
p(X,Y,Z’), c(T,W)
p(X,Y,Z), q(U,V,Z)
![Page 41: Hypertree Decompositions G. Gottlob Technical University of Vienna, Austria N. Leone and F. Scarcello University of Calabria, Italy For papers and further](https://reader036.vdocument.in/reader036/viewer/2022062515/56649cf05503460f949bf3ec/html5/thumbnails/41.jpg)
Hypertree Decompositions
More liberal than query decomposition
Query atoms can be used “partially”as long as the full atom appearssomewhere else
![Page 42: Hypertree Decompositions G. Gottlob Technical University of Vienna, Austria N. Leone and F. Scarcello University of Calabria, Italy For papers and further](https://reader036.vdocument.in/reader036/viewer/2022062515/56649cf05503460f949bf3ec/html5/thumbnails/42.jpg)
Grouping and Reusing Atoms
We use p(X,Y,Z) partially p(X,Y,_), c(T,W)
d(X,T)
a(X,U,W), b(Y,V,W)
c(Y,T)
p(X,Y,Z), q(U,V,Z)
p(X,Y,_),p(X,Y,_), c(T,W)
We group atoms p(X,Y,Z), q(U,V,Z)
![Page 43: Hypertree Decompositions G. Gottlob Technical University of Vienna, Austria N. Leone and F. Scarcello University of Calabria, Italy For papers and further](https://reader036.vdocument.in/reader036/viewer/2022062515/56649cf05503460f949bf3ec/html5/thumbnails/43.jpg)
Reusing atoms
We use p(X,Y,Z) partially p(X,Y,_), c(T,W)
d(X,T)
a(X,U,W), b(Y,V,W)
c(Y,T)
p(X,Y,Z), q(U,V,Z)
p(X,Y,_),p(X,Y,_), c(T,W)
![Page 44: Hypertree Decompositions G. Gottlob Technical University of Vienna, Austria N. Leone and F. Scarcello University of Calabria, Italy For papers and further](https://reader036.vdocument.in/reader036/viewer/2022062515/56649cf05503460f949bf3ec/html5/thumbnails/44.jpg)
a(S,X,X’,C,F), b(S,Y,Y’,C’,F’)
j(J,X,Y,X’,Y’)
j(_,X,Y,_,_), c(C,C’,Z) j(_,_,_,X’,Y’), f(F,F’,Z’)
d(X,Z) e(Y,Z) h(Y’,Z’)g(X’,Z’), f(F,_,Z’)
p(B,X’,F) q(B’,X’,F)
),','(),',()',',,,(
)','()','()',',(),(
),(),',()',',',,(),,',,(
FXBqFXBpYXYXJj
ZYhZXgZFFfZYe
ZXdZCCcFCYYSbFCXXSaans
![Page 45: Hypertree Decompositions G. Gottlob Technical University of Vienna, Austria N. Leone and F. Scarcello University of Calabria, Italy For papers and further](https://reader036.vdocument.in/reader036/viewer/2022062515/56649cf05503460f949bf3ec/html5/thumbnails/45.jpg)
Connectedness Condition
a(S,X,X’,C,F), b(S,Y,Y’,C’,F’)
j(J,X,Y,X’,Y’)
j(_,X,Y,_,_), c(C,C’,Z) j(_,_,_,X’,Y’), f(F,F’,Z’)
d(X,Z) e(Y,Z) h(Y’,Z’)g(X’,Z’), f(F,_,Z’)
p(B,X’,F) q(B’,X’,F)
![Page 46: Hypertree Decompositions G. Gottlob Technical University of Vienna, Austria N. Leone and F. Scarcello University of Calabria, Italy For papers and further](https://reader036.vdocument.in/reader036/viewer/2022062515/56649cf05503460f949bf3ec/html5/thumbnails/46.jpg)
Special Condition
a(S,X,X’,C,F), b(S,Y,Y’,C’,F’)
j(J,X,Y,X’,Y’)
j(_,X,Y,_,_), c(C,C’,Z) j(_,_,_,X’,Y’), f(F,F’,Z’)
d(X,Z) e(Y,Z) h(Y’,Z’)g(X’,Z’), f(F,_,Z’)
p(B,X’,F) q(B’,X’,F)
Each variable that disappeared at some vertex v
Does not reappear inthe subtrees rootedat v
J X Y
![Page 47: Hypertree Decompositions G. Gottlob Technical University of Vienna, Austria N. Leone and F. Scarcello University of Calabria, Italy For papers and further](https://reader036.vdocument.in/reader036/viewer/2022062515/56649cf05503460f949bf3ec/html5/thumbnails/47.jpg)
Special Condition
a(S,X,X’,C,F), b(S,Y,Y’,C’,F’)
j(J,X,Y,X’,Y’)
j(_,X,Y,_,_), c(C,C’,Z) j(_,_,_,X’,Y’), f(F,F’,Z’)
d(X,Z) e(Y,Z) h(Y’,Z’)g(X’,Z’), f(F,_,Z’)
p(B,X’,F) q(B’,X’,F)
Each variable that disappeared at some vertex v
Does not appear inthe subtrees rootedat v
J X Y
![Page 48: Hypertree Decompositions G. Gottlob Technical University of Vienna, Austria N. Leone and F. Scarcello University of Calabria, Italy For papers and further](https://reader036.vdocument.in/reader036/viewer/2022062515/56649cf05503460f949bf3ec/html5/thumbnails/48.jpg)
Positive Results onHypertree Decompositions
For each query Q, hw(Q) qw(Q)
In some cases, hw(Q) < qw(Q)
For fixed k, deciding whether hw(Q) k is in polynomial time (LOGCFL)
Computing hypertree decompositions is feasible in polynomial time (for fixed k)
![Page 49: Hypertree Decompositions G. Gottlob Technical University of Vienna, Austria N. Leone and F. Scarcello University of Calabria, Italy For papers and further](https://reader036.vdocument.in/reader036/viewer/2022062515/56649cf05503460f949bf3ec/html5/thumbnails/49.jpg)
Evaluating queries having bounded hypertree widthk fixed
Given:a database db
a query Q over db such that hw(Q) ka width k hypertree decomposition of Q
Deciding whether Q(db) is not empty is in O(n k+1 log n) and complete for LOGCFL
Computing Q(db) is feasible in output-polynomial time
![Page 50: Hypertree Decompositions G. Gottlob Technical University of Vienna, Austria N. Leone and F. Scarcello University of Calabria, Italy For papers and further](https://reader036.vdocument.in/reader036/viewer/2022062515/56649cf05503460f949bf3ec/html5/thumbnails/50.jpg)
Comparison results
Hypertree Decomposition
Hinge Decomposition+
Tree ClusteringCycle Hypercutset
Tree Clusteringw* treewidth
Cycle Cutset
HingeDecomposition
Biconnected Components
![Page 51: Hypertree Decompositions G. Gottlob Technical University of Vienna, Austria N. Leone and F. Scarcello University of Calabria, Italy For papers and further](https://reader036.vdocument.in/reader036/viewer/2022062515/56649cf05503460f949bf3ec/html5/thumbnails/51.jpg)
Characterizations ofHypertree width
Is hypertree width A natural concept? A natural generalization of
hypergraph acyclicity?
Are there nice characterizations In terms of logic? In terms of games?
Yes !
![Page 52: Hypertree Decompositions G. Gottlob Technical University of Vienna, Austria N. Leone and F. Scarcello University of Calabria, Italy For papers and further](https://reader036.vdocument.in/reader036/viewer/2022062515/56649cf05503460f949bf3ec/html5/thumbnails/52.jpg)
Characterizations ofHypertree width
Logical characterization:Loosely guarded logic
Game characterization:The robber and marshals game
![Page 53: Hypertree Decompositions G. Gottlob Technical University of Vienna, Austria N. Leone and F. Scarcello University of Calabria, Italy For papers and further](https://reader036.vdocument.in/reader036/viewer/2022062515/56649cf05503460f949bf3ec/html5/thumbnails/53.jpg)
Guarded Formulas )( gX
Guard atom: ))( var(gfree
)( 21
k
gggXk-guarded Formulas (loosely guarded):
k-guard
GF(FO), GFk(FO) are well-studied fragments of FO (Van Benthem’97, Gradel’99)
![Page 54: Hypertree Decompositions G. Gottlob Technical University of Vienna, Austria N. Leone and F. Scarcello University of Calabria, Italy For papers and further](https://reader036.vdocument.in/reader036/viewer/2022062515/56649cf05503460f949bf3ec/html5/thumbnails/54.jpg)
Logical Characterization of HW
Theorem: )(GF HWkk
L
From this general result, we also get anice logical characterization of acyclic queries:
)GF( ACYCLIC HW1
LCorollary:
![Page 55: Hypertree Decompositions G. Gottlob Technical University of Vienna, Austria N. Leone and F. Scarcello University of Calabria, Italy For papers and further](https://reader036.vdocument.in/reader036/viewer/2022062515/56649cf05503460f949bf3ec/html5/thumbnails/55.jpg)
An Example)),(),,(),,(),,(( .,,,,, WTsUZYrTYXqZYXpWUTZYX
p(X,Y,Z)
q(X,Y,T)
s(T,W)
r(Y,Z,U)
Is acyclic:
Indeed, there exists an equivalent guarded formula:
),,(( .,, ZYXpZYX )),(.),,((q . WTsWTYXT
)),,(r . UZYUGuard
Guarded subformula
![Page 56: Hypertree Decompositions G. Gottlob Technical University of Vienna, Austria N. Leone and F. Scarcello University of Calabria, Italy For papers and further](https://reader036.vdocument.in/reader036/viewer/2022062515/56649cf05503460f949bf3ec/html5/thumbnails/56.jpg)
Game characterization:Robber and Marshals
A robber and k marshals play the game on a hypergraph
The marshals have to capture the robber
The robber tries to elude her capture, by running arbitrarily fast on the vertices of the hypergraph
![Page 57: Hypertree Decompositions G. Gottlob Technical University of Vienna, Austria N. Leone and F. Scarcello University of Calabria, Italy For papers and further](https://reader036.vdocument.in/reader036/viewer/2022062515/56649cf05503460f949bf3ec/html5/thumbnails/57.jpg)
Robbers and Marshals: the rules
Each marshal stays on an edge of the hypergraph and controls all of its vertices at once
The robber can go from a vertex to another vertex running along the edges, but she cannot pass through vertices controlled by some marshal
The marshals win the game if they are able to monotonically shrink the moving space of the robber, and thus eventually capture her
Consequently, the robber wins if she can go back to some vertex previously controlled by marshals
![Page 58: Hypertree Decompositions G. Gottlob Technical University of Vienna, Austria N. Leone and F. Scarcello University of Calabria, Italy For papers and further](https://reader036.vdocument.in/reader036/viewer/2022062515/56649cf05503460f949bf3ec/html5/thumbnails/58.jpg)
Step 0: the empty hypergraph
VP R
S
X Y
ZT U
W
![Page 59: Hypertree Decompositions G. Gottlob Technical University of Vienna, Austria N. Leone and F. Scarcello University of Calabria, Italy For papers and further](https://reader036.vdocument.in/reader036/viewer/2022062515/56649cf05503460f949bf3ec/html5/thumbnails/59.jpg)
V
Step 1: first move of the marshals
VP R
S
X Y
ZT U
W
![Page 60: Hypertree Decompositions G. Gottlob Technical University of Vienna, Austria N. Leone and F. Scarcello University of Calabria, Italy For papers and further](https://reader036.vdocument.in/reader036/viewer/2022062515/56649cf05503460f949bf3ec/html5/thumbnails/60.jpg)
Step 1: first move of the marshals
VP R
S
X Y
ZT U
W
![Page 61: Hypertree Decompositions G. Gottlob Technical University of Vienna, Austria N. Leone and F. Scarcello University of Calabria, Italy For papers and further](https://reader036.vdocument.in/reader036/viewer/2022062515/56649cf05503460f949bf3ec/html5/thumbnails/61.jpg)
Step 2a: shrinking the space
VP R
S
X Y
ZT U
W
![Page 62: Hypertree Decompositions G. Gottlob Technical University of Vienna, Austria N. Leone and F. Scarcello University of Calabria, Italy For papers and further](https://reader036.vdocument.in/reader036/viewer/2022062515/56649cf05503460f949bf3ec/html5/thumbnails/62.jpg)
Step 2a: shrinking the space
VP R
S
X Y
ZT U
W
![Page 63: Hypertree Decompositions G. Gottlob Technical University of Vienna, Austria N. Leone and F. Scarcello University of Calabria, Italy For papers and further](https://reader036.vdocument.in/reader036/viewer/2022062515/56649cf05503460f949bf3ec/html5/thumbnails/63.jpg)
Step 2a: shrinking the space
VP R
S
X Y
ZT U
W
![Page 64: Hypertree Decompositions G. Gottlob Technical University of Vienna, Austria N. Leone and F. Scarcello University of Calabria, Italy For papers and further](https://reader036.vdocument.in/reader036/viewer/2022062515/56649cf05503460f949bf3ec/html5/thumbnails/64.jpg)
The capture
VP R
S
X Y
ZT U
W
![Page 65: Hypertree Decompositions G. Gottlob Technical University of Vienna, Austria N. Leone and F. Scarcello University of Calabria, Italy For papers and further](https://reader036.vdocument.in/reader036/viewer/2022062515/56649cf05503460f949bf3ec/html5/thumbnails/65.jpg)
V
A different robber’s choice
VP R
S
X Y
ZT U
W
![Page 66: Hypertree Decompositions G. Gottlob Technical University of Vienna, Austria N. Leone and F. Scarcello University of Calabria, Italy For papers and further](https://reader036.vdocument.in/reader036/viewer/2022062515/56649cf05503460f949bf3ec/html5/thumbnails/66.jpg)
V
Step 2b: the capture
VP R
S
X Y
ZT U
W
V
![Page 67: Hypertree Decompositions G. Gottlob Technical University of Vienna, Austria N. Leone and F. Scarcello University of Calabria, Italy For papers and further](https://reader036.vdocument.in/reader036/viewer/2022062515/56649cf05503460f949bf3ec/html5/thumbnails/67.jpg)
R&M Game and Hypertree Width
Let H be a hypergraph.Theorem: H has hypertree width k if and only if k marshals have a winning strategy on H.Corollary: H is acyclic if and only if one marshal has a winning strategy on H.
Winning strategies on H correspond to hypertree decompositions of H and vice versa.
![Page 68: Hypertree Decompositions G. Gottlob Technical University of Vienna, Austria N. Leone and F. Scarcello University of Calabria, Italy For papers and further](https://reader036.vdocument.in/reader036/viewer/2022062515/56649cf05503460f949bf3ec/html5/thumbnails/68.jpg)
VVP R
S
X Y
ZT UW
Strategies and Decompositions
),,(),,(),(
),(),,(),,,(),,,(
ZXWdVPRfYXg
ZYeZUTcPUYSbRTXSaans
![Page 69: Hypertree Decompositions G. Gottlob Technical University of Vienna, Austria N. Leone and F. Scarcello University of Calabria, Italy For papers and further](https://reader036.vdocument.in/reader036/viewer/2022062515/56649cf05503460f949bf3ec/html5/thumbnails/69.jpg)
VVP R
S
X Y
ZT UW
a(S,X,T,R), b(S,Y,U,P)
First choice of the two marshals
![Page 70: Hypertree Decompositions G. Gottlob Technical University of Vienna, Austria N. Leone and F. Scarcello University of Calabria, Italy For papers and further](https://reader036.vdocument.in/reader036/viewer/2022062515/56649cf05503460f949bf3ec/html5/thumbnails/70.jpg)
VVP R
S
X Y
ZT UW
a(S,X,T,R), b(S,Y,U,P)
A possible choice for the robber
![Page 71: Hypertree Decompositions G. Gottlob Technical University of Vienna, Austria N. Leone and F. Scarcello University of Calabria, Italy For papers and further](https://reader036.vdocument.in/reader036/viewer/2022062515/56649cf05503460f949bf3ec/html5/thumbnails/71.jpg)
VP R
S
X Y
ZT UW
a(S,X,T,R), b(S,Y,U,P)
f(R,P,V)
V
The capture
![Page 72: Hypertree Decompositions G. Gottlob Technical University of Vienna, Austria N. Leone and F. Scarcello University of Calabria, Italy For papers and further](https://reader036.vdocument.in/reader036/viewer/2022062515/56649cf05503460f949bf3ec/html5/thumbnails/72.jpg)
VVP R
S
X Y
ZT UW
a(S,X,T,R), b(S,Y,U,P)
f(R,P,V)
The second choice for the robber
![Page 73: Hypertree Decompositions G. Gottlob Technical University of Vienna, Austria N. Leone and F. Scarcello University of Calabria, Italy For papers and further](https://reader036.vdocument.in/reader036/viewer/2022062515/56649cf05503460f949bf3ec/html5/thumbnails/73.jpg)
VVP R
S
X Y
ZT UW
a(S,X,T,R), b(S,Y,U,P)
f(R,P,V) g(X,Y), c(T,Z,U)
The marshals corner the robber
![Page 74: Hypertree Decompositions G. Gottlob Technical University of Vienna, Austria N. Leone and F. Scarcello University of Calabria, Italy For papers and further](https://reader036.vdocument.in/reader036/viewer/2022062515/56649cf05503460f949bf3ec/html5/thumbnails/74.jpg)
VVP R
S
X Y
ZT UW
a(S,X,T,R), b(S,Y,U,P)
f(R,P,V) g(X,Y), c(T,Z,U)
g(X,Y), d(W,X,Z)
The capture
![Page 75: Hypertree Decompositions G. Gottlob Technical University of Vienna, Austria N. Leone and F. Scarcello University of Calabria, Italy For papers and further](https://reader036.vdocument.in/reader036/viewer/2022062515/56649cf05503460f949bf3ec/html5/thumbnails/75.jpg)
Conclusions and open questions
There are some interesting open questions: Special condition
Hypertree decomposition is a very natural notion Issues of fixed-parameter tractability Nonmonotonic capturing vs monotonic
capturing
For papers and further material see:http://ulisse.deis.unical.it/~frank/Hypertrees/