Abstract

Web 2.0 applications have become popular as drivers of new types of Web content, but they have also introduced a new level of interface design in Web development; they are focusing on richer interfaces, user-generated content, and better interworking of Web-based applications. The current foundations of the Web 2.0, however, are strictly imperative in nature, which makes it difficult to develop applications which are robust, interoperable, and backwards compatible. Using a declarative approach for Web 2.0 applications, this new wave of applications can be built on a more robust foundation which is more in line with the Web's style of using declarative methods whenever possible. We show a path how today's imperative Web 2.0 applications can be regarded as a testbed as well as a first implementation for a revised version of Web 2.0 technologies, which will be based on declarative markup rather than imperative code.

What is Web 2.0?

• The Web as a Platform
• Harnessing Collective Intelligence
• Being Data-Centered
• Using Web-Based Applications
• Using Light-Weight Programming Models
• Using Cross-Device Applications
• Providing Rich User Experiences

Web 2.0 Applications

• Web 2.0 applications very often leverage the network effect
  • by bringing together previously unconnected things, new value is created
  • the value thus often lies in the connections, and not in the connected things
• Formalize and exploit existing structures
  • many Web 2.0 applications thrive on representing existing structures
  • few Web 2.0 applications are design tools for creating works starting from scratch
  • Flickr, LinkedIn, FaceBook, MySpace, YouTube, Wikipedia, del.icio.us, blogs, …
• One universal observation about Web 2.0: context over content
  • context formalization, aggregation, and analysis is the hidden core of Web 2.0 applications
  • the step to a common context concept has not yet been made

Web 2.0 Technologies

• Dynamic HTML (DHTML) has been around for a long time
  • DHTML = HTML + JavaScript + DOM + CSS
  • Ajax = DHTML + XMLHttpRequest
  • newer browsers and JavaScript libraries make DHTML applications easier to write
• Rich Internet Application (RIA)
  • sometimes interpreted as an Ajax-heavy Web-based application
  • sometimes interpreted as a non-Web runtime accessing Web data sources
  • for the purpose of this talk RIAs are strictly Web-based
• Accessibility and usability problems of RIAs
  • how well do an applications work with JavaScript switched off?
  • how well do an applications work with CSS switched off?
  • how well do applications work on limited devices (PDAs, mobile browsers)

Stability of the Web 2.0 Landscape

• Interfaces to services change often and often without warning
  • building mashups by cutting&pasting code samples is fun and easy
  • maintaining mashups and fixing them when something breaks is hard and happens often
  • libraries hiding the browser incompatibilities introduce additional complexities
• The Web 2.0 landscape builds on a tightly coupled foundation
  • Web 2.0 applications are tightly coupled with their data model (through an API)
  • Web 2.0 applications are tightly coupled with JavaScript libraries
  • JavaScript libraries are tightly coupled with browser interfaces
• Tight coupling is less robust than loose coupling
  • tight coupling fails completely when components change
  • loose coupling degrades gracefully and can recover from errors more easily

Document Engineering

• API vs. Documents
  • classical distributed systems are build around APIs
  • one of the really interesting observations about the Web: No APIs!
  • document exchanges are a more loosely coupled way of building systems
  • the Web's document exchange model is called Representational State Transfer (REST)
• Real-world examples of document exchanges
  • just send us a form with all necessary details and we will process it.
  • imagine having to know all APIs of all participating business processes
  • even if all internal processes change completely, the form still works
• Loose coupling is good but hard
  • defining the right forms (i.e., document models) is challenging
  • identify commonly used concepts and define an extensible model
  • provide mustUnderstand semantics for required extensions

Reusable Web 2.0 Data

• Web 2.0 applications use custom XML formats
  • data exchange between Ajax and server is a closed scenario
  • code needs to know the XML structure for building mashups
• Standardizing Web 2.0 data formats is not likely to happen
  • applications want to have the freedom to use their own data formats
  • applications might not want to fully disclose or expose their data model
• Microformats are one possible way of encoding well-defined semantics
  • limited to simple concepts which can be easily embedded in HTML
• Semantic Web technologies are the other end of the spectrum
  • complex set of technologies for expressing arbitrary semantics
  • high barrier of entry and unlikely to succeed on a large scale

Blog Post

Blog Posts and Link Objects

• Blogs have a number of recurring features
  • they are a sequentially ordered series of blog posts
  • a blog has an owner and a permanent URI
  • posts have a date and content
  • posts very often have a number of URIs as part of their content
• Blog posts can be regarded as XLinks (blog links or blinks)
  • the complete blog (the collection of all posts) is the blog linkbase
  • each post is an XLink with a predefined structure
  • post date, post URIs, post content (HTML snippet), …
  • advanced blinks could classify referenced URIs (support, criticism, similarity, blog, …)

Blog Post Structure

Blog Post as XLink

<blink xlink:type="extended" xmlns:xlink="http://www.w3.org/1999/xlink" xlink:title="Sorry Pluto...">
 <permalink xlink:type="locator" xlink:title="Sorry Pluto..." xlink:label="permalink" xlink:href="http://docordie.blogspot.com/2006/08/sorry-pluto-and-some-thoughts-about.html"/>
 <author xlink:type="locator" xlink:title="Bob Glushko" xlink:label="author" xlink:href="http://www.ischool.berkeley.edu/˜glushko/"/>
 <blog xlink:type="locator" xlink:title="Doc Or Die" xlink:label="blog" xlink:href="http://docordie.blogspot.com/"/>
 <feed xlink:type="locator" xlink:title="’Doc Or Die’ Atom Feed" xlink:label="feed" xlink:href="http://docordie.blogspot.com/atom.xml"/>
 <posturi xlink:type="locator" xlink:label="site" xlink:href="http://en.wikipedia.org/wiki/International_Astronomical_Union"/>
 <posturi xlink:type="locator" xlink:label="site" xlink:href="http://www.modbee.com/local/story/12638580p-13341408c.html"/>
 <posturi xlink:type="locator" xlink:label="site" xlink:href="http://www.freenewmexican.com/news/48332.html"/>
 <posturi xlink:type="locator" xlink:label="site" xlink:href="http://www.miami.com/mld/miamiherald/living/15365590.htm"/>
 <post xlink:type="resource" xlink:label="post" xlink:title="Sorry Pluto..." type="text/html">
  <h2 xmlns="http://www.w3.org/1999/xhtml">Sorry Pluto...</h2>
  <!-- complete HTML snippet of blog post -->
 </post>
 <timestamp xlink:type="resource" xlink:label="timestamp" type="xs:date">2006-08-27</timestamp>
 <arc xlink:type="arc" xlink:from="permalink" xlink:to="timestamp" xlink:title="Post Time"/>
 <arc xlink:type="arc" xlink:from="site" xlink:to="post" xlink:title="Blog Post"/>
 <arc xlink:type="arc" xlink:from="permalink" xlink:to="author" xlink:title="Post Author"/>
 <arc xlink:type="arc" xlink:from="permalink" xlink:to="blog" xlink:title="Blog Home"/>
 <arc xlink:type="arc" xlink:from="blog" xlink:to="author" xlink:title="Blog Author"/>
 <arc xlink:type="arc" xlink:from="blog" xlink:to="feed" xlink:title="Blog Feed"/>
 <arc xlink:type="arc" xlink:from="author" xlink:to="blog" xlink:title="Authored Blogs"/>
</blink>

Context Publishing as Add-On

• No need for all blogs to manage their posts as XLinks
  • XLinks are just the way to formalize context across applications
• An added opportunity to make blog data available and usable and useful
  • instead of searching for blog posts, you just find them!
  • request/response interaction with the context provider (e.g., the blog search engine)
• Context must first be formalized before it can be reused
  • individual blogs provide little context (only the blog's posts)
  • blog hosting sites provide a lot of context (all hosted blogs)
  • blog search engine provide complete context (all crawled blogs)

Crawling Blogs

Using Aggregators

Native Blinks

XLink Infrastructure

• Linkbase is an XML database storing XLink documents
  • storage and access is efficient and scalable
  • queries can be based on arbitrary structural criteria
• Client-side implementation in a Firefox extension
  • for each request, a request to the linkbase is sent
  • the linkbase replies with the set of links for that URI (currently no ranking)
  • the links are transformed into XUL code
  • the XUL code is rendered and displays the links in a sidebar

XLink Mockup

H. Weinreich, H. Obendorf, and W. Lamersdorf. The Look of the Link — Concepts for the User Interface of Extended Hyperlinks, In: Proceedings of HT'01, August 2001

Image XLinks

• Images can also have links associated with them
  • annotations to an image as a whole
  • annotations to a geometric region of the image (currently only rectangles)
• Image links are just another application of the same concepts
  • the client also sends linkbase requests for all images embedded in an HTML page
  • image links are transformed to SVG code and displayed as an overlay
• Popular image annotation applications could be opened
  • Flickr supports the concept of annotating part of an image
  • this annotation is stored in Flickr and can only be viewed in Flickr
  • Flickr could provide comprehensive image publishing: image & context

Conclusions

• Declarative Web 2.0 is a goal and not a concrete technology
• Identification of frequently used concepts is hard
  • some easy targets, many hard targets
  • even for the easy targets, exact semantics may be hard to define
  • if there is not incentive to become declarative, only a few will do it
• Web architecture is more fashionable than it used to be
  • naive SOA had enough time to prove its disadvantages
  • granularity of services will increase
  • the need for composable services will increase
  • technologies providing easy and robust integration are required

Future Plans

• Atomify the Concepts
  • Atom is the next big thing: a unified format for syndication and publishing
  • linkbases are accessed using the Atom Publishing Protocol (APP)
  • links are published using Atom feeds
  • linkbase handling and management could be viewed as context syndication
• And Now for Something Completely Different …
  • linking puts things into a logical context
  • location puts things into a spatial context
  • location is used in many applications, but not as a Web-level concept
  • the location-aware Web is one facet of the declarative Web
  • add location to the Web architecture in URIs, HTTP, and HTML

References

1. D. Connolly. Gleaning Resource Descriptions from Dialects of Languages (GRDDL). World Wide Web Consortium, Candidate Recommendation CR-grddl-20070502, May 2007.
2. S. J. DeRose, E. Maler, and D. Orchard. XML Linking Language (XLink) Version 1.0. World Wide Web Consortium, Recommendation REC-xlink-20010627, June 2001.
3. R. T. Fielding. Architectural Styles and the Design of Network-based Software Architectures. PhD thesis, University of California, Irvine, Irvine, California, 2000.
4. J. Gregorio and B. de Hóra. The Atom Publishing Protocol. Internet Draft draft-ietf-atompub-protocol-15, May 2007.
5. M. Gudgin, M. Hadley, N. Mendelsohn, J.-J. Moreau, H. Frystyk Nielsen, A. Karmarkar, and Y. Lafon. SOAP Version 1.2 Part 1: Messaging Framework (Second Edition). World Wide Web Consortium, Recommendation REC-soap12-part1-20070427, April 2007.
6. A. Y. Halevy, N. Ashish, D. Bitton, M. J. Carey, D. Draper, J. Pollock, A. Rosenthal, and V. Sikka. Enterprise Information Integration: Successes, Challenges and Controversies. In: F. Özcan, editor, Proceedings of the ACM SIGMOD 2005 International Conference on Management of Data, pages 778–787, Baltimore, Maryland, June 2005. ACM Press.
7. I. Jacobs and N. Walsh. Architecture of the World Wide Web, Volume One. World Wide Web Consortium, Recommendation REC-webarch-20041215, December 2004.
8. G. Klyne and J. J. Carroll. Resource Description Framework (RDF): Concepts and Abstract Syntax. World Wide Web Consortium, Recommendation REC-rdf-concepts-20040210, February 2004.
9. R. Kumar, J. Novak, P. Raghavan, and A. Tomkins. Structure and Evolution of Blogspace. Communications of the ACM, 47(12):35–39, December 2004.
10. D. Lowe and E. Wilde. Improving Web Linking Using XLink. In: Proceedings of Open Publish 2001, Sydney, Australia, July 2001.
11. A. van Kesteren. The XMLHttpRequest Object. World Wide Web Consortium, Working Draft WD-XMLHttpRequest-20070618, June 2007.
12. E. Wilde. What are you talking about? In: Proceedings of the 2007 IEEE International Conference on Services Computing, Salt Lake City, Utah, July 2007.