Keys to the Flowering Plants of Australia

This section describes the rationale and approach behind the Keys to the Flowering Plants of Australia project and its underlying platforms, KeyBase and IdentifyLife

Why two systems?

The two main types of analytical identification platform - pathway (often called dichotomous) and matrix-based (often called interactive) keys - both have advantages and disadvantages:

Pathway key advantages:
  • Relatively easy and quick to create
  • Often more accurate than matrix-based keys
  • Often more jargon-laden than good matrix-based keys
  • Good starting point: many legacy keys already exist
  • Good learning tools - users quickly get to know the critical features in a group

Pathway key disadvantages
  • The unanswerable couplet problem - identification stops if you come to a couplet you can't answer

Matrix-based key advantages
  • Descriptive data compiled for the key has many other uses
  • Keys are flexible (the unanswerable couplet problem is not a problem)
  • Often more approachable - easier to use plain language rather than jargon

Matrix-based key disadvantages
  • Data compilation is laborious
  • Very sensitive to Type I errors (falsely discarding correct taxa due to errors in the underlying data)
  • Not so good as learning tools - keys tend to be "black boxes" that give the answer but little knowledge of critical features in the group)

It's clear that both types of identification tools have good and bad features. This project seeks to create parallel keys, in both KeyBase and IdentifyLife, so users can have the best of both worlds and use the most appropriate platform for their circumstance.

What's so good about KeyBase?

Most dichotomous keys on the web are displayed as pretty much straight text, much as you'd find on a printed page. Sometimes, people put hyperlinks on the couplet numbers, and think this is pretty cool. KeyBase goes much further, and turns plain old dichotomous keys into interactive web applications. With KeyBase, we can:

  • link keys together to create a seamless key to all species of flowering plants in Australia
  • display keys coupleted or indented, or use the KeyBase Player, which allows users to interact with keys in a more sophisticated way
  • illustrate keys to make them easier to use for naive users
  • glossarise keys, automatically connecting technical words in the couplets to online glossary entries
  • geographically filter keys (for example, create an on-the-fly key to just the plants that occur in your local area)
  • create custom on-the-fly keys to any set of taxa
  • access KeyBase via web services (for example, you will be able to insert keys derived from KeyBase into your own Flora project)

Please note that KeyBase is still under development. While it's been designed to do all these things, it currently only does the first three of these dot-points. Watch the KeyBase space for further developments.

What's so good about IdentifyLife?

Most matrix-based (interactive) identification systems, such as Lucid and DELTA, are desk-top applications, with very limited capacity for collaborative key building and maintenance. IdentifyLife is a new, web-based system that allows groups of people to collaboratively build and maintain the descriptive data that underlies their keys. Unlike the desk-top-based applications, which effectively build small, self-contained "island" of data disconnected from other key projects, IdentifyLife maintains a "sea" of descriptve data and allows individual key projects to share resources and data. This potentially makes IdentifyLife a more flexible and efficient means of building keys, as new projects can readily draw on data provided for previous projects. By storing its data in the cloud, using a very generalised underlying data model, and allowing groups of collaborators to create, manage and maintain their data, IdentifyLife is more future-proof than existing systems.

What's more, data captured in IdentifyLife can be used for a variety of purposes, such as building natural-language descriptions or painting characters on phylogenies, as well as for the core purpose of deploying identification keys. This potential repurposing makes the capture of data more effective than building a one-off key.

Why manage descriptive data in KeyBase and IdentifyLife instead of in other Flora projects?

Traditionally, Flora projects have created and published a bundled package of content - identification keys, nomenclatural information, descriptions, distribution maps etc - for a whole flora or for a taxonomic group. The package is only made available when complete; this may take many years.

The vision of the Keys to the Flowering Plants of Australia project is to separate the identification keys from the other parts of the package, and to use specialised repositories that each do one thing - the management and deployment of dichotomous keys and of matrix-based keys respectively - really well. Separating the keys from the rest should allow us to deploy to our users identification keys - arguably one of the most important parts of a Flora - while the other content is being created. This will have advantage for our users - it will be useful to be able to identify a plant even if you then have to trawl the web for other information about it, than not to be able to identify it at all.

How will the Keys to Flowering Plants of Australia project proceed?


How can I help?

Please go back to the main page for this project to see how you can help contribute

-- KevinThiele - 2012-12-25