Abstract

Based on the proposals submitted last week, individual AJAXLink project parts will be assigned today. The majority of today's class is reserved for the individual project parts to start working on the interfaces with all related parts. Today's result should be sufficient input for everybody to prepare a detailed project outline implementation strategy, due in two weeks.

Friday Proposals

Tuesday Suggestions

Igor

• Turn basic AJAXLink code into a firefox extension
  • basic AJAXLink code takes document and XLinks and updates the document
  • explore the possibilities and limitations of bookmarklets (lightweight extensions)
• Required extension behavior
  • fetch XLinks from linkbase and add them to a document
  • display document XLinks in some additional user interface element
• Optional extension behavior
  • advanced filters for linkbase access (get only a subset of XLinks)
  • generate XLinks based on user interactions (linkbase write access)

CREAM Metadata and Annotations

S. Handschuh, S. Staab, Authoring and Annotation of Web Pages in CREAM, Proceedings of WWW 2002, May 2002

Jeff & Michael

• Basic AJAXLink functionality
  • starting with embedded XLinks and mapping these to AJAX code
  • using some kind of AJAX UI library will make life much easier
• Stepwise refinement of the functionality
  • handle additional media types (text, images)
  • handle out-of-line links as well
  • additional fragment identifiers (such as XPointers for XML)
  • decide what to do with overlapping links (hard to code)
• Many interactions with Libby
  • design questions of rendering links and handling interactions
  • comprehensive list of all kinds of complex XLink structures

John

• Explore the possibilities of annotating Sylvia
  • Sylvia content is relevant for all students
  • annotations (notes and links) create the iSchool student linkbase
• Challenges are introduced by Sylvia's application model
  • one Sylvia syllabus is rendered in various presentation formats
  • annotations should be to a syllabus, not to some specific format
  • XLink links resources, so it should link the syllabus XML
• Explore how AJAXLink can be reused/integrated in Sylvia
  • AJAXLink should be declarative Web 2.0
  • embedding annotations in syllabus display would be typical Web 2.0
  • which requirements can be derived from this use case?

Libby

• XLink is a simple but powerful language
  • links can have 0-n resources and 0-n locators
  • links can have various arcs with different semantics
  • links can link resources of various media types
• Create a comprehensive list of all possible combinations
  • the list will not be exhaustive, but should be informative
  • decide on which combinations are more important than others
  • create user interface designs for the relevant combinations
• Looking at Web 2.0 applications is a good way to start
  • if AJAXLink is declarative Web 2.0, many should be covered
  • if applications are not covered, why?
  • AJAXLink should have a good ratio of coverage vs. complexity

Lois

• Google scholar creates connections between multiple resources
  • resources are cited by resources and cite resources
  • we need a way to extract this information (API or small-scale crawling)
• AJAXLink code could be used to visualize these links
  • retrieve and visualize Google Scholar links using AJAX
  • as a first step, this could be an improved interface for Google Scholar
• How reasonable is it to link to links?
  • Google Scholar links are authoritative information (read-only)
  • users would like to attach these links to resources
  • how well do XLink and AJAXLink handle this use case?

Ryan

• Multimedia links connecting images, video, and audio
  • most of these cannot be implemented using HTML (images maybe)
  • there should be a concept for a fallback strategy for HTML only
• Libby's list of supported links and media types needs to be extended
  • adding the additional media types and arcs and behaviors which are supported
  • working out a consistent user interface design across implementations
  • we need a good concept for link display and graceful degradation
• Interfaces to the more basic AJAXLink modules
  • XLinks are usually handled by the standard code (Jeff & Michael)
  • additional XLinks may be fetched from linkbases (Igor)

Yiming

• Relational linkbase for storing XLinks
  • the relevant parts of an XLink are easy to find
  • XLinks may contain additional information (attributes or resources)
• Storing XML in relational databases always is a challenge
  • store the complete XLink serialization as text
  • extract the elements and their attributes
  • if no updates are allowed, this can be implemented efficiently
• Protocol design should be REST-based
• Implementing the linkbase can be done incrementally
  • a read-only linkbase maps queries to SQL and returns the results
  • write access (maybe by Igor) requires an extended protocol
  • management features can be done completely separate from the rest

Project Timeline

• 2007-02-20: detailed outline and implementation strategy
• 2007-03-06: progress reports, technical alignment
• 2007-03-20: prototype implementations, interoperability testing starts
• 2007-04-03: stabilized first prototypes, proposals for improvement
• 2007-04-17: prototype implementations, interoperability testing starts
• 2007-05-01: finalized implementation, project reports, wrap-up