Bieber, Catanio & Zhang, NJIT © Ubiquitous Metainformation and the W Y W W Y W I Principle Michael Bieber*, Joe Catanio*, Li Zhang** *Information Systems Department **Computer Science Department College of Computing Sciences New Jersey Institute of Technology November 2003 This talk ties together much of our current research.. It also gives a vision of where the WWW is heading.
Bieber, Catanio & Zhang, NJIT © Thinking Exercise Close your eyes. Think about your favorite Web site, or one that you rely on for information... Focus on an element you’re interested in –a text string, icon, element within a table or an animation...
Bieber, Catanio & Zhang, NJIT © The W Y W W Y W I Principle What you want, when you want it Wanting to point to something and say: –Tell me more about this! –What is this? –How can I use this? What do I need to know to use it? –Can I modify this? –How does this differ from similar ones? –What is the next step? This is all metainformation & people should get it!
Bieber, Catanio & Zhang, NJIT © Ubiquitous Metainformation Goal: Metainformation widespread in everyday systems How: provide tools for developers Relationship Analysis –systematically determining metainformation Metainformation Engine –automatically generating metainformation WYWWYWI –widespread accepted design philosophy
Bieber, Catanio & Zhang, NJIT © Outline Examples Metainformation (what to provide) Relationship Analysis (how to find metainformation) Metainformation Engine (how to automate it) Virtual Documents (many real world documents) Related Work WYWWYWI (what it will take)
Bieber, Catanio & Zhang, NJIT © Two Examples Purchasing System Digital Library screen shot of our prototype later...
Bieber, Catanio & Zhang, NJIT © Sample Screen from Purchasing System: All text with no links...
Bieber, Catanio & Zhang, NJIT © But we could want metainformation about almost any element...
Bieber, Catanio & Zhang, NJIT © V {vendor} Vendor Details {Vendor IS} Vendor Reliability {Vendor IS} Vendor Agreements {Vendor IS} Other Possible Vendors {Purchasing Data Warehouse} Your Purchasing History {Purchasing IS} All Screens with this Vendor {CASE Workbench}
Bieber, Catanio & Zhang, NJIT © Author and Document Owner {Metadata Repository} Define this concept {Agricultural Thesaurus}
Bieber, Catanio & Zhang, NJIT © Metainformation The full context about and around an element Provides a fuller understanding metadata element relationships
Bieber, Catanio & Zhang, NJIT © Metainformation metadata (about selected element) content relationships (based on display content) structural relationships (based on element type or “class”) annotation relationships (user-declared, knowledge-sharing) metainformation-based navigation (user-directed) Roberto Galnares’ dissertation
Bieber, Catanio & Zhang, NJIT © Author and Document Owner {Metadata Repository} Define this concept {Agricultural Thesaurus}
Bieber, Catanio & Zhang, NJIT © Author and Document Owner {Metadata Repository} metadata content relationship annotation structural relationships Define this concept {Agricultural Thesaurus} content relationship
Bieber, Catanio & Zhang, NJIT © V {vendor} Vendor Details {Vendor IS} Vendor Reliability {Vendor IS} Vendor Agreements {Vendor IS} Other Possible Vendors {Purchasing Data Warehouse} Your Purchasing History {Purchasing IS} All Screens with this Vendor {CASE Workbench} all are structural relationships
Bieber, Catanio & Zhang, NJIT © Outline Examples Metainformation (what to provide) Relationship Analysis (how to find metainformation) Metainformation Engine (how to automate it) Virtual Documents (many real world documents) Related Work WYWWYWI (what it will take)
Bieber, Catanio & Zhang, NJIT © Relationship Analysis (RA) What metainformation could we provide? RA: a systematic methodology to determine relationships (& metadata and new destination elements) –New systems analysis technique –Fills a major hole in software engineering –Analysts gain deeper understanding of a system –Yields richer analyses and designs –Relationships become links Joe Catanio’s dissertation
Bieber, Catanio & Zhang, NJIT © Relationship Analysis (RA), cont. approach: brainstorming with domain experts for existing systems: –pick elements from screen shots for new systems: –pick entities from use cases Ask questions from RA taxonomy
Bieber, Catanio & Zhang, NJIT © RA Taxonomy based on Guilford’s Structure of Intellect theory [1950] –describing intellect and creativity refined by Rao & Turoff’s Hypertext Morphology [1991] – for systems analysis
Bieber, Catanio & Zhang, NJIT © RA Taxonomy
Bieber, Catanio & Zhang, NJIT © RA Taxonomy
Bieber, Catanio & Zhang, NJIT © RA: Brainstorming Questions RA Template
Bieber, Catanio & Zhang, NJIT © Outline Examples Metainformation (what to provide) Relationship Analysis (how to find metainformation) Metainformation Engine (how to automate it) Virtual Documents (many real world documents) Related Work WYWWYWI (what it will take)
Bieber, Catanio & Zhang, NJIT © Metainformation Engine “Just in time” metainformation –required for virtual documents (e.g., query results) Automatically: –generates link anchors –generates links to services providing metainformation: metadata, content, structural, annotation relationships –incorporates metainformation-based navigation Provides lightweight systems integration through linking to everyday systems Roberto Galnares’ dissertation
Bieber, Catanio & Zhang, NJIT © V {vendor} Vendor Details {Vendor IS} Vendor Reliability {Vendor IS} Vendor Agreements {Vendor IS} Other Possible Vendors {Purchasing Data Warehouse} Your Purchasing History {Purchasing IS} All Screens with this Vendor {CASE Workbench}
Bieber, Catanio & Zhang, NJIT © To Integrate: (1) wrapper: parses screens to identify elements (2) provide metadata/structural rel’ship rules (3) identify glossaries for content relationships User’s Web Browser Vendor IS Wrapper Purchasing D.W. Wrapper Purchasing IS Wrapper CASE Workbench Wrapper Service Wrapper (i) Vendor Information System Purchasing Data Warehouse Purchasing Information System CASE Workbench Service (i) ME Relationship Engine ME Broker ME Desktop Metainformation Engine ME Lexical Analysis existing system or Web service uses Java, XML, Xpath, etc.
Bieber, Catanio & Zhang, NJIT © User’s Web Browser Vendor IS Wrapper Purchasing D.W. Wrapper Purchasing IS Wrapper CASE Workbench Wrapper Service Wrapper (i) Vendor Information System Purchasing Data Warehouse Purchasing Information System CASE Workbench Service (i) ME Relationship Engine ME Broker ME Desktop Metainformation Engine ME Lexical Analysis To Integrate: (1) wrapper: parses screens to identify elements (2) provide metadata/structural rel’ship rules (3) identify glossaries for content relationships existing system or Web service uses Java, XML, Xpath, etc.
Bieber, Catanio & Zhang, NJIT © V {vendor} Vendor Details {Vendor IS} Vendor Reliability {Vendor IS} Vendor Agreements {Vendor IS} Other Possible Vendors {Purchasing Data Warehouse} Your Purchasing History {Purchasing IS} All Screens with this Vendor {CASE Workbench} Relationship Rules element type (“vendor”) link display label (“Vendor Details”) relationship metadata for filtering links –semantic relationship type (“elaboration”) –relationship keywords destination system (“Vendor Info System”) exact command(s) for destination system (“retrieve_full(ID, details)”) conditions –user types and tasks, expertise required, access restrictions
Bieber, Catanio & Zhang, NJIT © Relationship Rules Mechanism for implementing access to: –Metadata –Structural relationships –Content relationships –Annotation relationships –Metainformation navigation
Bieber, Catanio & Zhang, NJIT © Metadatum Rule element type ( “vendor”) metadatum link display label (“Vendor Name”) metadatum’s metadata for filtering links –semantic relationship type (“name”) –metadatum keywords destination system (“Vendor Info System”) exact command(s) for destination system (“ select(vendor_table, vendor_ID, vendor_name) ”) conditions –user types and tasks, expertise required, access restrictions
Bieber, Catanio & Zhang, NJIT © Annotation Rule (display comments) element type ( “generic_element”) link display label (“ view comments on this”, element_type) ”) relationship metadata for filtering links –semantic relationship type (“annotation”) destination system (“Annotation System”) exact command(s) for destination system (“display_annotations(element_ID)”) conditions check_condition(“Core Annotation Service”, existence_check(“annotations”, element_ID)) = true)
Bieber, Catanio & Zhang, NJIT © Prototype: Digital Library Integration
Bieber, Catanio & Zhang, NJIT © Interesting Issues Information overload! –Must filter and rank order list of links –Too many anchors –Requires good user interface design Semantics –Systems/services should use same element types Unique persistent identifiers –For every screen, document, element of interest
Bieber, Catanio & Zhang, NJIT © Outline Examples Metainformation (what to provide) Relationship Analysis (how to find metainformation) Metainformation Engine (how to automate it) Virtual Documents (many real world documents) Related Work WYWWYWI (what it will take)
Bieber, Catanio & Zhang, NJIT © Virtual Documents from user interaction, queries, customizations Metainformation must be added “just in time” Example –do a decision support analysis (“# vehicles needed”) –add comments to calculation results –bookmark screen (“make it a favorite”) –close screen –follow bookmark later (“system regenerates screen”) –system must re-locate comment anchors “just in time”
Bieber, Catanio & Zhang, NJIT © Virtual Documents Re-generate virtual documents –without re-entering parameters –then wrapper parses to add metainfo anchors Re-identify elements –Location can shift –content can change (e.g., stock price) Re-locate anchors Li Zhang’s dissertation
Bieber, Catanio & Zhang, NJIT © Outline Examples Metainformation (what to provide) Relationship Analysis (how to find metainformation) Metainformation Engine (how to automate it) Virtual Documents (many real world documents) Related Work WYWWYWI (what it will take)
Bieber, Catanio & Zhang, NJIT © Related Work Web Services Semantic Web Microsoft SmartTags/NBC-Interactive’s QuickClick, etc. Link Services and Hypermedia Engines
Bieber, Catanio & Zhang, NJIT © Web Services The Metainformation Engine (ME) is a kind of Web service –a metainformation/linking service ME can integrate with existing Web services –include Web services in set of links –supplement Web services with links
Bieber, Catanio & Zhang, NJIT © Semantic Web Advocates the use of ontologies (groups of related terms) to define concepts and their relationships Goal: machine-readable semantic description / processing Recognizes the need to accommodate diverse knowledge representations and conflicting definitions. The ME could use ontologies to: –standardize element types –Find related elements and provide links to their metainformation
Bieber, Catanio & Zhang, NJIT © Microsoft SmartTags / NBC-Interactive’s QuickClick, etc. Set of links generated automatically –Based on known keywords –Can add additional links manually No structural linking
Bieber, Catanio & Zhang, NJIT © Microsoft Smart Tag critiques Users find too many links annoying Content providers do not want their documents altered Manipulation: Companies can pay to have competitors name recognized and linked to their sites Provides content based links (word recognition) not structural linkages (element recognition)
Bieber, Catanio & Zhang, NJIT © Link Services and Hypermedia Engines SFX (links to referenced articles) Link Services –Chimera, Microcosm Distributed Link Service, DHM: manual linking or automated through keyword search requiring modifications for integration Hypermedia Engines with minimal modifications –Microcosm Universal Viewer, Freckles: manual linking –OO-Navigator: SmallTalk only Web Database Applications, e.g., e-shopping –database queries only, normally single links
Bieber, Catanio & Zhang, NJIT © Outline Examples Metainformation (what to provide) Relationship Analysis (how to find metainformation) Metainformation Engine (how to automate it) Virtual Documents (many real world documents) Related Work WYWWYWI (what it will take)
Bieber, Catanio & Zhang, NJIT © What you want, when you want it What will it take? WYWWYWI mindset for developers & public –Allow metainformation (user-directed) navigation –a design philosophy for developers –demanded by the general public –this requires exposure! Developer Tools Ubiquitous Access
Bieber, Catanio & Zhang, NJIT © What you want, when you want it What will it take? WYWWYWI mindset for developers & public Developer Tools –Relationship Analysis –Metainformation Engine –Wrappers for everyday systems –Annotation/knowledge-sharing services (linking, comments, guided tours, etc.) Ubiquitous Access
Bieber, Catanio & Zhang, NJIT © What you want, when you want it: What will it take? WYWWYWI mindset for developers & public Developer Tools Ubiquitous Access –Repositories of relationship rules –Thesauri and glossaries
Bieber, Catanio & Zhang, NJIT © Interesting Issues Access privileges to others’ metainformation Right to third party authoring/linking Quality of metainformation –Rating metainformation –Bogus metainformation (such as advertisements)
Bieber, Catanio & Zhang, NJIT © Metainformation –broader conceptualization Relationship Analysis (how to find metainformation) Metainformation Engine (how to automate it) –Lightweight systems integration through linking Virtual Documents –Re-generation, re-identification, re-location WYWWYWI: a design philosophy –What you want, when you want it Research Contributions Thank you! Questions, please?