Biocentric information and the role of taxonomy and taxonomists
Philosophy
Software
Flexibility
Content organization and finding content
Working with distributed and local content, linkouts
Quality control
Options for using the star approach (collections, star*sites, star*nets)
Sustainability
Names and classification schemes
CU*STAR
Biopedia
XID and matrix keys

Biocentric organization and the role of taxonomy and taxonomists

Most of biology is made up of biocentric information - information about named entities. Since the time of Linnaeus, this information has been very effectively organized by biological taxonomy - suites of rules and guidelines about how to develop and apply names so that they are stable and unambiguous, and the placement of those names within concept hierarchies called classifications.Taxonomists are the custodians of taxonomic knowledge and so are the key players in organizing and indexing information about organisms. In an informatics context, they oversee the most important array of biological meta-data. One of their tasks must be to transfer their organization and indexing skills, as well as their knowledge of organisms, into a world where generic skills can be put to use by the internet, and their knowledge accessed through the internet.

Star*sites aim to help this process. We are embedding the organizational and indexing power of taxonomy within the internet. In so doing, we seek to add an array of internet services for biologists.

Philosophy

Our philosophy is to help develop contemporary knowledge environments for biology that are inclusive and flexible. Star*sites are intended to be very scaleable - and managers have a range of options for the scope of a site - from assembling a small collection within a pre-existing site to building an extensive network of sites. We also adopt a principle of inclusiveness - rather than exclusiveness. The software has been designed to work with a diversity of different data types - although has begun with images and descriptions. Sites can adopt different classification schemes as part of their organization. Inclusiveness also extends to sharing of content - such that any star*site can 'sunch' with other sites or collections, and expand their own indices to include content at remote sites. Inclusiveness embraces the ability to combine local and remote or distributed data. The software also is built to be compliant with an 'inclusive' philosophy. The software is modular and we welcome developers to add to the range of items in the software library. The appearance of the site and its functionality are controlled by variables, and site managers will be able to use the variables to control the appearance, appearance and content of the sites. We believe that the result will make the system very flexible and very scaleable, and embues it with the capacity to grow top a very large size and to evolve to meet changing needs and demands.

There are some aspects which are not flexible at this time. Because of the sharing philosophy, we presume that builders of star*assets will be placing their content in the public domain. Star*central also needs to be made aware of certain dimensions of individual sites in order to be able to facilitate content sharing.

Software

Star software is modular. The software is written in PHP, and so can work within a public domain PHP - mySQL - Apache environment. A module library is available through the toolbox at star*central. Users can pick up modules, or create new ones. Just talk to us about how to get going with this.

Flexibility

As the star*software develops, site managers will have the ability to control the front page, the 'chrome' or 'livery' so that the appearance can be moulded to meet the needs of the site. There will be choice over which functions (modules) will be included within a site, and interested parties will be able to expand the arrange of functions that are available. Each site will be able to define its default classification and what part of the classification the site opens at. The array of outlinks will be flexible and users will also have the choice of using bio*pedia. In addition, although installing a star*site will create a default directory structure, site managers will be able to reorganize this to meet their local needs. As far as possible we seek to define as many aspects of the system as variables and to place the options as a buffet-style menu for both site developers and site visitors. The data are presented in 'dynamic fact sheets', and the composition and layout of those sheets will also be under the control of the site manager. Some elements will not be optional, such as the dialogue with star*central which is required to maintain oversight of all of the activities among individual sites...

Content organization, and finding content

The contents of star*sites are referred to as assets. Assets are placed within collections. Collections can be placed inside other collections. Currently, star*sites are set up to deal with images, but will be expanded to deal with other assets. Tools to create collections or to add and edit assets are available through our toolbox.

Visitors to star*sites can find assets using a number of pathways. There are search ('Look-for') functions at various levels within the system and these are appropriate for visitors wo know precisely what they are looking for. Search functions are quick and direct. Secondly, visitors can browse and search within classification schemes. Site managers and users can chose the classification scheme to use. This allows visitors to navigate towards assets within an informed phylogenetic context. It allows for the search process to be broadened or narrowed. In addition, visitors can use the hierarchical collections to navigate from a broad or imprecise search objective to an increasingly precise and targeted approach. This places within a contextual environment. Site managers will be able to edit and expand the hierarchical collection structure to provide expert guidance to assets.

Working with local and distributed content, linkouts

The inclusiveness philosophy dictates that star*sites work with local and distributed data. Access to distributed data is empowered by two means. Firstly, we provide tools to allow the easy creation of other star*sites, and tools can be used to synch the indices of diferent sites so that any one site becomes a portal to some or all of the assets located elsewhere within the star family. Secondly, we use a system of linkout inspired by NCBI. The linkouts will be assembled within a database, and star site managers, and perhaps even users, will be able to decide which linkouts are used in a given site at a given time. The linkouts are contextualised in two ways. Linkouts are mapped against the classification scheme, so that they only appear within specified clades or specified subsets of taxa (e.g. only when visiting pages that deal with mammals or possibly mammals of Australia). Linkouts are also tagged with metadata from an open-ended controlled vocabulary. This limits the linkouts to ones appropriate to the audience or the needs of users (the linkouts might, for example, be limited to those appropaiet to K-12 students or to those relating to molecular data).

Quality control

Star*site are intended to be expert biuological knowledge environments, providing resources for - among others - research biologists, educators and students, decision makers, and those who seek to inform or even disinform us. It is of importance that the quality (accuracy and completeness) of the information within a star*site be explicit. We adopt two strategies to help protect data quality. Firstly, star*sites will have the option of adding comments boxes to every element of the databases, so that visitors can contribute comments and these can be subsequently viewed by visitors. We advocate that visibility of comments be controlled by the site manager. Secondly, we are building a system that will permit three layers of quality 'judgement' to be applied to content. The lowest level allows anyone to add anything and no quality control is imposed. At the second level, content is vetted through a secretariat for each site, and according to agreed guidelines, the secretariat adds a seal of approval to the content. Finally, we will provide tools which will permit panels of experts to evaluate, comment, and edit content to the highest standards. Each web site will have software switches which will hide content at lower levels of quality and/or show the different levels of quality in different ways.

Options for using the star approach (collections, star*sites, star*nets)

Someone wishing to use the star*approach will have an array of choices. At the minimalist level, they can identify the owner of a collection of a star*site and ask to add assets to that collection. This is made possible by through tools (internet services) accessed at our toolbox through a web browser. At the next level, a owner may wish to create a collection within an existing star*site, and again does this using internet services tools that can be accessed at our toolbox through a web browser. At the next level, an owner may wish to create a new star*site, and again does this by downloading software that is accessed at our toolbox through a web browser. The new star*site manager will need to set upo a server running Apache, mySQL and PHP but our toolbox will provide help there too. Finally, a group of experts may chose to create a network of co-operating star*sites. These are referred to as star*nets, and the first of these is plankton*net. Members of star*nets have a dedicated synching function which allows each site to carry the contents of the indices of all other members. each site therefore becomes a portal to the content assem,bles by larger community. Coupled with our robust sustainability model, this should enable large teams to build large-scale knowledge environments on the internet.

Sustainability

Web knowledge resources face considerable problems with sustainability. Many sites have appeared, inspired by enthusiasm but often with a short-term supply of support. When the enthusiasm diminishes or the support runs out, the site stals, and begins a process of degradation. star*central offers a starchiving service - in which not only theindices of star*sites but also the content can be archived. This ensures that the content remains accessible even after individual sites stop growing. Compliant with out philosophy, any star*site will be empowered with the starchiving function - thereby protecting the sustainability model against the collapse of a single star*central.

Names and classifications

For the star*sites to realize their vision, the taxonomic systems that we call upon must be populated with names. star*sites work closely with the uBio project. This project has been populating a structure called NameBank with objective taxonomic information (names of organisms) and also maintains a second database in which these are placed within alternative classifications. At the time of writing, NameBank has access to about 1,700,000 names, and a further 300,000 are waiting to be added. NameBank includes fairly comprehensive generic coverage of all living organisms (several groups of insects, some Cnidaria and a few other groups have not yet been covered). This ensures that the compilation of names can be used to index assets relating to any kind of organism. NameBank through ClassificationBank places names within vying classification schemes. NameBank adds considerable nomenclatural information, maps occurrences and uses of names, identifies synonyms and alternative names such as colloquial names of mis-spellings that may have found their way into the literature, internet or databases, and finally helps navigates through the problems created by homonyms (names spelled in the same way but referring to different entities). Star*sites begin with the CU*STAR classification (being a comprehensive and unified classification of all life), but managers of star*sites can switch to other classifications. The CU*STAR classification is based on Patterson, D. J. 1999. The diversity of eukaryotes. American Naturalist 154: S96-124. (pdf).There will also be editing tools that will allow users to build their own classifications. As star*sites grow, new names that are needed to index assets that are to be fed into the uBio Namebank structure. Please do let us know if you would like to add names. We will provide you with access to names adding tools.

The classification system is an important part of the star*sites for many reasons:
it is of intrisnic value because it reveals perceptions of relationships
it provides access to nomenclatural and other ancillary data at related sites
it provides a pre-formed framework around which assets, whether local or distributed, can be organized
it provides services for taxa for which we have no assets - such as outlinks
it allows for the reconciliation of synonynms, misspellings and other alternative names
it identifies homonyms to be identified and flagged to avoid errors in navigation to information
it provides a framework around which other services can be built - such as a 'expert*ease' service which identifies expert taxonomists who know about the taxa in question
it allows clades or other collections of names to be annotated with generic notes
it provides the basis of a names registry system
etc.

CU*STAR

The CU*STAR classification offers a unified classification to ease navigation from one taxonomic region to another. The classification has been placed in a flexible and editable environment so that users can build their own classifications. We seek to keep CU*STAR up to date with names from NameBank. Different regions of CU*STAR are monitored by taxonomic experts - we refer to them as custodians.

As most star*sites can be expected to include descriptions of taxa, we are creating a species star*site called biopedia. Biopedia is intended to be a communal resource where users can place basic information about organisms (initially descriptions, but later descriptions together with images, key literature, information about experts, and so on). Biopedia will be connected to NameBank to pick up associated nomenclatural data and information on synonyms. It is likely that biopedia will include links to other data sources, but one model for growth involves atomizing the data to create a super*matrix that will facilitate the discovery of identities of taxa using matrix identification tools

X:ID and matrix keys

Matrix keys are a very effective means of identifying organisms. Information about the organism (characters and character states) can be entered into a matrix. Organisms can then be found by a process of elimination (which organisms live in Massachusetts, in marine environments, are one foot long, with hard shell, have a spikey tail, and legs underneath). Examples of software through which matrix keys can be built are Delta (more for experts), Lucid, oir X:ID. The latter will be developed so that the entries on characters and taxa can be linked to appropriate data in both NameBank and star*sites. This increases the integration of data resources with tools for users. In addition, the development of keys will feed 'atomised' information into a resource referred to as the Super*matrix. Subsequent keys will have the option of using the matrix information available within super*matrix.