The Snoopy Concept
Fighting Heterogeneity in Semistructured and
 Collaborative Information Systems by using
             Recommendations



    Wolfgang Gassler, Eva Zangerle, Günther Specht
          Databases and Information Systems
            University of Innsbruck, Austria


                CTS 2011, Philadelphia
Knowledge is structured
                          2
Type: Desktop
Type: iMac                Location: Room 3S01
Location: Room 3S01       Manufacturer: HP
User: Wolfgang            Shipment Date: 11/01/01
Color: white              Serial: #123456
IP-Address: 192.168.0.5   Hostname: wg.uibk.ac.at
                          IP: 192.168.0.10




    Hardware Inventory Management
                                                    3
Type: Desktop
        Type: iMac                Location: Room 3S01
        Location: Room 3S01       Manufacturer: HP
        User: Wolfgang            Shipment Date: 11/01/01
        Color: white              Serial: #123456
        IP-Address: 192.168.0.5   Hostname: wg.uibk.ac.at
                                  IP: 192.168.0.10



Steve                                                 Bill




            Hardware Inventory Management
                                                             4
Traditional Relational DB System
Steve:                    Bill:
Type: iMac                Type: Desktop
Location: Room 3S01       Location: Room 3S01
User: Wolfgang            Manufacturer: HP
Color: white              Shipment Date: 11/01/01
IP-Address: 192.168.0.5   Serial: #123456
                          Hostname: wg.uibk.ac.at
                          IP: 192.168.0.10




                                                    5
Traditional Relational DB System
Steve:                    Bill:
Type: iMac                Type: Desktop
Location: Room 3S01       Location: Room 3S01
User: Wolfgang            Manufacturer: HP
Color: white              Shipment Date: 11/01/01
IP-Address: 192.168.0.5   Serial: #123456
                          Hostname: wg.uibk.ac.at
                          IP: 192.168.0.10




                                                    6
Traditional Relational DB System
Steve:                    Bill:
Type: iMac                Type: Desktop
Location: Room 3S01       Location: Room 3S01
User: Wolfgang            Manufacturer: HP
Color: white              Shipment Date: 11/01/01
IP-Address: 192.168.0.5   Serial: #123456
                          Hostname: wg.uibk.ac.at
                          IP: 192.168.0.10



                                50%
                            information
                                lost
                                                    7
Wiki System
●
    Free format
●
    No schema
●
    100% information




                                8
Wiki System
●
    Free format
●
    No schema
●
    100% information


BUT:
●
    No structure
●
    No aligned information
●
    Poor search facilities (complex queries)
                                               9
Semantic Web / Semantic Wiki
●
    Free format
●
    No schema
    100% information




                       lo
●




                         ca
●
    Typed Links




                          edt
                             In
                                Pennsylvania



                                               10
Semantic Web / Semantic Wikis
●
    Free format
●
    No schema
    100% information




                              lo
●




                                ca
●
    Typed Links




                                   t
                                  ed
                                     In
BUT:
                                       Pennsylvania
●
    No aligned information
●
    User is not guided or motivated
                                                      11
Features
                 Structure   Extensibility   Homogeneity
Relational DB    yes         no              yes
Wikis            no          yes             no
Semantic Wikis   yes         yes             no




                                                           12
Features
               Structure   Extensibility   Homogeneity
Relational DB  yes         no              yes
Wikis          no          yes             no
Semantic Wikis yes         yes             no
Snoopy Concept yes         yes             yes




                                                         13
The Snoopy Concept
●   Similar to RDF (subject, property, value)
    <CTS2011> <Location> <Philadelphia>
    <CTS2011> <Venue> <Sheraton>

●   Recommendations based on already inserted
    data (self-learning system)
    ●   Encourage user to enter more data
    ●   Encourage re-usage of properties and values
    ●   -> Decrease proliferation of schemata
                                                      14
SnoopyDB Prototype




●   Recommends Structure
●   Avoids Synonyms
●   Exploits user's extensive and valuable
    knowledge („snoops“ information)
                                             15
SnoopyDB Prototype




●   Encourages Semantic Refinements
●   Validations & Recommendations of Types   16
Evaluation
●   24 test users
    ●   2/3 computer scientists / students
    ●   1/3 standard computer users
●   Two Snoopy Systems
    ●   Without any guidance, recommendations
    ●   All guidance and recommendation features
●   Task
    ●   Insert City, University, Band, Car Model in non-
        guided and guided system
                                                           17
Evaluation (2)




                 18
Evaluation (3)




                 19
Conclusion – Snoopy Concept
●   Avoid proliferation of structures (self-learning)
●   Avoid synonyms in the system
    ●   49% properties reused
●   Exploit user's extensive and valuable
    knowledge („snooping“ as much as possible)
    ●   Increase the quantity of information in the system
    ●   31% more data
    ●   Semantic refinements by resolving homonyms
    ●   Increase the quality of information in the system
                                                             20

The Snoopy Concept Fighting Heterogeneity in Semistructured and Collaborative Information Systems by using Recommendations

  • 1.
    The Snoopy Concept FightingHeterogeneity in Semistructured and Collaborative Information Systems by using Recommendations Wolfgang Gassler, Eva Zangerle, Günther Specht Databases and Information Systems University of Innsbruck, Austria CTS 2011, Philadelphia
  • 2.
  • 3.
    Type: Desktop Type: iMac Location: Room 3S01 Location: Room 3S01 Manufacturer: HP User: Wolfgang Shipment Date: 11/01/01 Color: white Serial: #123456 IP-Address: 192.168.0.5 Hostname: wg.uibk.ac.at IP: 192.168.0.10 Hardware Inventory Management 3
  • 4.
    Type: Desktop Type: iMac Location: Room 3S01 Location: Room 3S01 Manufacturer: HP User: Wolfgang Shipment Date: 11/01/01 Color: white Serial: #123456 IP-Address: 192.168.0.5 Hostname: wg.uibk.ac.at IP: 192.168.0.10 Steve Bill Hardware Inventory Management 4
  • 5.
    Traditional Relational DBSystem Steve: Bill: Type: iMac Type: Desktop Location: Room 3S01 Location: Room 3S01 User: Wolfgang Manufacturer: HP Color: white Shipment Date: 11/01/01 IP-Address: 192.168.0.5 Serial: #123456 Hostname: wg.uibk.ac.at IP: 192.168.0.10 5
  • 6.
    Traditional Relational DBSystem Steve: Bill: Type: iMac Type: Desktop Location: Room 3S01 Location: Room 3S01 User: Wolfgang Manufacturer: HP Color: white Shipment Date: 11/01/01 IP-Address: 192.168.0.5 Serial: #123456 Hostname: wg.uibk.ac.at IP: 192.168.0.10 6
  • 7.
    Traditional Relational DBSystem Steve: Bill: Type: iMac Type: Desktop Location: Room 3S01 Location: Room 3S01 User: Wolfgang Manufacturer: HP Color: white Shipment Date: 11/01/01 IP-Address: 192.168.0.5 Serial: #123456 Hostname: wg.uibk.ac.at IP: 192.168.0.10 50% information lost 7
  • 8.
    Wiki System ● Free format ● No schema ● 100% information 8
  • 9.
    Wiki System ● Free format ● No schema ● 100% information BUT: ● No structure ● No aligned information ● Poor search facilities (complex queries) 9
  • 10.
    Semantic Web /Semantic Wiki ● Free format ● No schema 100% information lo ● ca ● Typed Links edt In Pennsylvania 10
  • 11.
    Semantic Web /Semantic Wikis ● Free format ● No schema 100% information lo ● ca ● Typed Links t ed In BUT: Pennsylvania ● No aligned information ● User is not guided or motivated 11
  • 12.
    Features Structure Extensibility Homogeneity Relational DB yes no yes Wikis no yes no Semantic Wikis yes yes no 12
  • 13.
    Features Structure Extensibility Homogeneity Relational DB yes no yes Wikis no yes no Semantic Wikis yes yes no Snoopy Concept yes yes yes 13
  • 14.
    The Snoopy Concept ● Similar to RDF (subject, property, value) <CTS2011> <Location> <Philadelphia> <CTS2011> <Venue> <Sheraton> ● Recommendations based on already inserted data (self-learning system) ● Encourage user to enter more data ● Encourage re-usage of properties and values ● -> Decrease proliferation of schemata 14
  • 15.
    SnoopyDB Prototype ● Recommends Structure ● Avoids Synonyms ● Exploits user's extensive and valuable knowledge („snoops“ information) 15
  • 16.
    SnoopyDB Prototype ● Encourages Semantic Refinements ● Validations & Recommendations of Types 16
  • 17.
    Evaluation ● 24 test users ● 2/3 computer scientists / students ● 1/3 standard computer users ● Two Snoopy Systems ● Without any guidance, recommendations ● All guidance and recommendation features ● Task ● Insert City, University, Band, Car Model in non- guided and guided system 17
  • 18.
  • 19.
  • 20.
    Conclusion – SnoopyConcept ● Avoid proliferation of structures (self-learning) ● Avoid synonyms in the system ● 49% properties reused ● Exploit user's extensive and valuable knowledge („snooping“ as much as possible) ● Increase the quantity of information in the system ● 31% more data ● Semantic refinements by resolving homonyms ● Increase the quality of information in the system 20