Api Development

Introduction

Questions:

• how to define the m4all core service e.g. is this just model creation, or does it include tagging, comments, rating, groups, wiki etc
• how to define the mashups we'd want to create within project
• how to define apis that we and others could consume (to be part of mashups)
• how to build apis e.g. WSDL, RSS, Atom, JSON (easiest first!) and authentication system (all APIs seem to require registration and the sending of credentials in API method calls)
• how to consume apis, especially with respect to making them easy to build in mashup building software such as qedWiki
• how to build mashups e.g. qedwiki, yahoo pipes, google gagdets? etc
• how to document and promote apis
• in principle is it possible to entirely use a 3rd party service for tagging, rating and blogging per model (and potentially per micro-behaviour) e.g. is this scalable, do for instance delicious allow this level of use of their API?
  • http://del.icio.us/help/api/ (sounds like a busy m4all service would get throttled if we were getting a lot of page requests)

Resources

• Find, publish and learn about APIs: http://programmableweb.com
• Create and publish 'mashup' websites: http://services.alphaworks.ibm.com/qedwiki/
• Introduction to the process of exposing data and consuming: http://java.sys-con.com/read/325192.htm

Review of some mashups

Zoogle

http://www.zoogle.in/Default.aspx
this seems to be a tag search of as many different apis as possible. the apis are categorised by search type and the execution of the search is initiated one at a time as the user selects the target e.g. News—>BBC

Thumblicious

http://thumblicio.us/index.php
This grabs most popular links based on a tag search and generated thumbnail images of the web site where possible