Upload
others
View
0
Download
0
Embed Size (px)
Citation preview
1
1
Research in the Logic Group
Michael GeneserethLogic Group
Stanford University
2
Computational Logic
above(yellow,blue)on(red,yellow)
∀x.∀y.(on(x,y) ⇒ above(x,y))
∃y.on(blue,y)
¬on(yellow,blue)
on(green,yellow)∨ on(green,blue)
2
3
Automated Theorem Proving
Group Axioms�
Theorem
(x × y) × z = x × (y × z)x × e = xe × x = x
x × x −1 = e
x −1 × x = e
4
Constraint Satisfaction Systems
3
5
Database Tables
Queriesquery(X,Z) :- parent(X,Y) & parent(Y,Z)
Constraintsillegal :- parent(X,X)illegal :- parent(X,Y) & parent(Y,X)
Deductive Database Systems
parentart bobart beabea coe
parent(art, bob)parent(art, bea)parent(bob,coe)
6
Sample Applications Logical Spreadsheets
Data Integration
Web of Data
Computational Law
General Game Playing
4
Logical Spreadsheets
8
Huge Success individual users companies conglomerates
Good Features Automatic computation of values Ease of specification using simple math formulas
Computerized Spreadsheets
5
9
Functional formulas
Unidirectional Update
B1 B2 B3
Limitations of Traditional Spreadsheets
2 3
B3 = B1 + B2
5
10
Logical Spreadsheets
Extension Relational constraints on the values of cells
Good Features Automatic computation of values Ease of set-up (using logical formulas)
Challenges Automatic update Temporary inconsistencies User feedback
6
11
yescale3nobobe2noarte1
timeroomprojectorownerevent
eveningafternoonmorning
g300g200g100
classroomnog300
classroomnog200
theateryesg100
SeatingProjectorRoom
12
Constraints
If a projector is needed, an event cannot bescheduled in a room without a projector.
illegal :- event(E,O,yes,R,T) & room(R,no,S)
Bob is not permitted to schedule events in g100.
illegal :- event(E,O,P,g100,T) & owner(E,bob)
7
13
yescale3nobobe2
morningg100noarte1TimeRoomProjectorOwnerEvent
eveningafternoonmorning
g300g200g100
classroomnog300
classroomnog200
theateryesg100
SeatingProjectorRoom
14
yescale3nobobe2
morningg100noarte1TimeRoomProjectorOwnerEvent
eveningafternoon
e1morningg300g200g100
classroomnog300
classroomnog200
theateryesg100
SeatingProjectorRoom
8
15
yescale3nobobe2
morningg100noarte1TimeRoomProjectorOwnerEvent
eveninge2afternoon
e1morningg300g200g100
classroomnog300
classroomnog200
theateryesg100
SeatingProjectorRoom
16
yescale3afternoong200nobobe2morningg100noarte1
TimeRoomProjectorOwnerEvent
eveninge2afternoon
e1morningg300g200g100
classroomnog300
classroomnog200
theateryesg100
SeatingProjectorRoom
9
17
yescale3afternoong200nobobe2morningg100noarte1
TimeRoomProjectorOwnerEvent
eveninge2afternoon
e3e1morningg300g200g100
classroomnog300
classroomnog200
theateryesg100
SeatingProjectorRoom
18
morningg300yescale3afternoong200nobobe2morningg100noarte1
TimeRoomProjectorOwnerEvent
eveninge2afternoon
e3e1morningg300g200g100
classroomnog300
classroomnog200
theateryesg100
SeatingProjectorRoom
10
19
morningg300yescale3afternoong200nobobe2morningg100noarte1
TimeRoomProjectorOwnerEvent
e4eveninge2afternoon
e3e1morningg300g200g100
classroomnog300
classroomnog200
theateryesg100
SeatingProjectorRoom
20
eveningg300e4morningg300yescale3afternoong200nobobe2morningg100noarte1
TimeRoomProjectorOwnerEvent
e4eveninge2afternoon
e3e1morningg300g200g100
classroomnog300
classroomnog200
theateryesg100
SeatingProjectorRoom
11
21
Research Topics
Paraconsistent Logic (Eric Kao) reasoning with inconsistent data and constraints
Differential Logic (Mike Kassoff) computing differences from differences analogy with differential calculus
Dynamic Logic (Ashwin Deshpande) constraints on transitions, not just state privacy, security, responsibility workflow management
22
Collaborative Spreadsheets
12
Data Integration
24
Structured Data Sources
Public Sources Company Directories Weather Reports Product Catalogs Airline Schedules Product Reviews Drug Studies
Enterprise Sources Personnel Records Orders Equipment Databases Inventories Room Schedules
13
25
Consumers of Structured Data
Different users Pilots and mechanics and airline marketers Passengers and travel agents FAA and NTSB
Different uses often require different datasets Airline personnel want to see their own flights Passengers want to see all flights for their trips
26
Data Integration
Answer
Data Broker
Manufacturer 1Manufacturer 2
Marketplace Data
Query
Product analysis
SatisfactionRatings
Supplier 1Supplier 2
Supplier 3Supplier 4
14
27
Complications
Distribution Data at different sites Handled through magic of networking
Format Heterogeneity e.g. relational databases, XML, tab-delimited text Handled through hand-coded translators
*Conceptual Heterogeneity Differences in schema and vocabulary
28
20ironsaucepanc04
carter
30ironskilletc0340aluminumsaucepanc0250aluminumskilletc01
pricematerialtypeid
Source 1 - Carter
15
29
saucepanm04
kind
saucepanm03skilletm02skilletm01valueid
Source 2 - Mirkwood
20m04
price
40m0350m0260m01
valueidcoating
yesm02yesm01
valueid
30
30msrps03tefloncoatings03
stainlessmaterials03skillettypes03
superchefmakers0350msrpr01
marvel
aluminummaterialr01skillettyper01
renfrewmakerr01valueattributeid
Source 3 - Marvel
16
31
Source 4 - NHMA
nonstick
yesteflonnocoppernoceramic
valueid
americausa
country
europeukeuropefranceamericacanada
areaid
francesuperchef
company
canadarenfrewukmirkwoodusacarter
nationid
32
Consumer 1 - Xanadu
50skilletrenfrewr0120saucepancarterc04
xanadu
30skilletcarterc0340saucepancarterc0250skilletcarterc01
pricetypemakerid
All cookware products manufactured in the UnitedStates.
17
33
Consumer 2 - Yankee
40frypanrenfrewr0116potcarterc04
yankee
24frypancarterc0332potcarterc0240frypancarterc01
pricetypemakerid
All cookware products manufactured in the UnitedStates with European types and prices in euros.
34
Consumer 3 - Zebulon
zebulon
30skilletsuperchefs0350skilletmirkwoodm0260skilletmirkwoodm01
pricetypemakerid
All skillets manufactured in Europe that are madefrom non-corrosible materials.
18
35
Schemas
Consumer Relations: xanadu(X,Y,Z,W) yankee(X,Y,Z,W) values in European units alproduct(X) ceproduct(X) feproduct(X) ssproduct(X) ncproduct(X)
Source Relations: carter(X,Y,Z,W) kind(X,Y) nonstick(X,Y) marvel(X,Y,Z) coating(X,Y) company(X,Y)
price(X,Y) country(X,Y)
36
Relational LogicSafe, Horn Rules grandparent(X,Z) :- parent(X,Y), parent(Y,Z)
Existential Rules parent(X,f(X,Z)) :- grandparent(X,Z)
Disjunctive Rules father(X,Y) | mother(X,Y) :- parent(X,Y)
Recursive Rules ancestor(X,Y) :- parent(X,Y) ancestor(X,Z) :- parent(X,Y), ancestor(Y,Z)
19
37
Direct Mapping
Schema Schema Schema
Schema Schema Schema
38
Sometimes called Global-As-View Integration
Source-Based Integration
20
39
Relationships Among Sources
Replicated data Cached data Materialized views (as in data warehouses)
Heterogeneity (different schemas or vocabularies) values in euros versus values in dollars French instead of English different numbers of tables or attributes
Real World Constraints Physical laws Governmental laws Business rules
40
Collaborative Data Management
When sources are independent, they can be updatedindependently.
In the face of interrelationships among sources,individuals performing updates must collaborate(explicitly or implicitly) to ensure correct updates.
Collaborative Data Management must replaceindependent data management.
21
41
Data Manager
Data Manager
Database 2
Database 5
Database 1 Database 3
Database 4 Database 6
Metadata
Data
Constraints
42
Update Integration
Data Broker
Manufacturer 1Manufacturer 2
Marketplace Data
Update
Product analysis
SatisfactionRatings
Supplier 1Supplier 2
Supplier 3Supplier 4
22
Web of Data
44
Future - Web of Data
World Wide Web Collaborative Document Management System Interlinked Documents Keyword Search, Lists of documents as answers Individual source update, source ownership
Web of Data Collaborative Data Management System Objects/relations and constraints Query not Search, Answers instead of Documents Single Entry Principle, data ownership
23
45
Showcase - Digital DepartmentGoal - Enterprise Data Management for Stanford
Components Multiple Departments University Databases
Areas Room reservations Event Management (Equipment, Food, Mail lists) Office Assignments Curriculum (prerequisites, requirements, policies) Programs (Undergraduate, Masters, Doctoral)
46
Showcase - Digital GovernmentGoal - Enterprise Data Management for governments
Components Federal Agencies States Local Governments
Output Government people and organizations Bills, votes, laws
Input Taxes, licenses, reports, court proceedings, results Online Turbotax for everything
24
47