1 / 17

Modeling Interactive Web Sources for Information Mediation

Modeling Interactive Web Sources for Information Mediation. Bertram Ludäscher, Amarnath Gupta San Diego Supercomputer Center, UCSD. Information Mediation Framework/Motivation Modeling Interactive Sources with Interaction Diagrams Computing Derived Capabilities Conclusions & Future Work.

eadoin
Télécharger la présentation

Modeling Interactive Web Sources for Information Mediation

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Modeling Interactive Web Sources for Information Mediation Bertram Ludäscher, Amarnath Gupta San Diego Supercomputer Center, UCSD Information Mediation Framework/Motivation Modeling Interactive Sources with Interaction Diagrams Computing Derived Capabilities Conclusions & Future Work

  2. Mediation of Information Using XML (MIX) BBQ UI Mediator XMAS View Def. XML-queries Mediator XML-answers Wrapper Wrapper Wrapper WWW RDB OODB MIX Mediator Architecture Modeling Web Sources for Information Mediation

  3. Motivation • Wrappers export schema, capabilities, and data of sources • Specifics of Web sources (e.g., vs RDB/OODB sources): • limited query capabilities, i.e., • restricted i/o binding patterns • need for user interaction • complex navigations • data extraction • accessibility • ... Modeling Web Sources for Information Mediation

  4. Example: ATM/bank locator • "Find someATM/bank locations close to a location given by street, city, state, user interaction." • involves • input and output attributes • forms, navigations, user interaction • data extraction Modeling Web Sources for Information Mediation

  5. BBQ UI Mediator XMAS View Def. user/source interaction XML-queries Mediator XML-answers Wrapper Wrapper Wrapper WWW RDB OODB "Non-Wrappable" User Interactions Modified MIX Mediator Architecture Modeling Web Sources for Information Mediation

  6. Modeling Interactive Sources... • Model input/output behavior of (HTML) elements as select-project queries: • Given values for input attributes a, • select tuples satisfying y(a)from source relation R, • project on output attributes b : • pb(sy(a)(R)) Modeling Web Sources for Information Mediation

  7. Modeling Interactive Sources... • Example (HTML Link): • clicking on an airport code apc produces the airport address with zip code: • pzip(sy(apc)(R)) • where • y(apc):= (apc=$apc) and • $apcrepresents the actual parameter (here: link label) • (or: answer(ZIP) APC=$APC, R(APC,ZIP,...) ) Modeling Web Sources for Information Mediation

  8. ... Modeling Interactive Sources ... • Example (HTML Forms): • Given a street and city name, extract the zip code: • pzip(sy(street,city)(R)) • where y(street,city):= (street=$scity=$c) • By default, forms are modeled as conjunctions of equalities. • (similar for menus and other elements) Modeling Web Sources for Information Mediation

  9. ... Modeling Interactive Sources ... • Example (Non-Wrappable Elements): • interactions requiring explicit user interaction with the source, and where the result depends on a1, ..., an are denoted by • !ui(a1, ..., an) • The user interaction is modeled by a new, internal attribute x: • py(sy(a1, ..., an, x)(R)) Modeling Web Sources for Information Mediation

  10. ... Modeling Interactive Sources... • Use select-project queries to model i/o behavior of individual user interaction elements (links, forms, menus, maps, ...) • Missing: how to "glue" together ... • navigations through a Web site and • data extraction • ... in order to model source capabilities of complex interactions • model a Web source using interaction diagrams: • nodessource page(s) and exported data • edgestransitions and required interactions Modeling Web Sources for Information Mediation

  11. i[,u\v] t: xy Interaction Diagrams • diagram: labeled graph over given attributes • node source page(s) with same i/o behavior • node label exported attributes (tuple/set/list): (a_1,...,a_n), {(a_1,...,a_n)}, [(a_1,...,a_n)] • edge transition, edge label required interaction: [a1,... ] {b1, ... } • interaction i{href, form, menu, !ui, ...} •  transition condition • t's result depends on attributes u but not v Modeling Web Sources for Information Mediation

  12. Example Interaction Diagram • transition path t1.t3.t5: • fill in form with street, city, menu with state, • fill in form s.t. radius>0, • follow all bankid links • derivable query template q(t1.t3.t5) = (+street,+city,+state,+radius,-bname,-bstreet, -bcity) Modeling Web Sources for Information Mediation

  13. i[,u\v] t: xy Deriving Query Capabilities: Single Transitions • input atts • in(t) := (atts(x) atts(i)  u) \ v • output atts • out(t) := atts(y) • interaction requirements • act(i) := i   Modeling Web Sources for Information Mediation

  14. Deriving Query Capabilities:Transition Paths t1.t2.t (t = t3. ... .tn) • in(t1.t2.t) := in(t1)  in(t2.t) \ propagate(con(t1.t2))) • out(t1.t2.t) := out(t2.t) • act(t1.t2.t) := act(t1) con(t1.t2))  act(t2.t) • where • con(t1.t2)) connects out(t1) with in(t2) (e.g., ) • propagate(...) are attributes whose values are passed along • (e.g., out(t1)  in(t2)) • denotes serial conjunction Modeling Web Sources for Information Mediation

  15. Use of Interaction Diagrams • modeling tool • complex query capabilities (binding patterns) along paths t = t1. ... .tn  qt(in(t), out(t)) • sequence of interaction requirements execution plans • subsumption and equivalence of binding patterns  check for supported queries • distinction between wrappable vs. unwrappable queries using the number of !ui nodes in paths Modeling Web Sources for Information Mediation

  16. Interacting with the Mediator • Diagram-enabled wrapper can do local query optimizations • Models of wrapper-mediator interaction: • Wrapper exports complete interaction diagram, mediator plans query • Mediator hands complete subquery to wrapper, wrapper optimizes query • Wrapper gives mediator partial plans, mediator computes overall query plan • An interesting model • Touting wrapper: identifies missing element in a query that would make an infeasible query feasible Modeling Web Sources for Information Mediation

  17. Conclusions & Future Work • Ongoing: • Stand-alone Web site modeling tool (SiteModel: P. Nguyen) which exports an interaction diagram • Definition & analysis of query capabilities of complex sources • Future Work: • support for query evaluation at the MIX mediator and user/source interaction • use interactions diagrams for semi-automatic wrapper generation Modeling Web Sources for Information Mediation

More Related