The Dragonfiles system

This is the beta trial release of the Dragonfiles project. Dragonfiles is intended to provide a flexible database structure and interface which can be used to prepare searchable information sets for any taxonomic group. Dragonflies are the first.

This is a functional, data-accumulating, version of the project. Later releases will build further on the growing dataset and will provide additional capabilities to query the information in the database. We anticipate changes in the interface in response to user comments.

Development of biological database resources are big projects. Having looked at several efforts, including a number of very valuable literature accumulation projects in the Odonata, it is clear to us that an open-source type structure is the only way to accumulate the necessary eyes to find the errors and clean the tables. And the research community benefits from the resource they build.

The Dragonfiles database system is intended to provide a resource to link together available information on all dragonfly species. To effect this we have created a structure to provide links to the detailed data, stored on hundreds of sites world wide. Dragonfiles provides one leg of a tripod - the Tree of Life Odonata page provides a general introduction to the Order and its biology and DragonID (working title) will provide the capacity to identify dragonflies to species ... which will immediately open a link to all the information that has been indexed through the Dragonfiles database system.

The Dragonfiles system is dependent on an international web collaboration of many Odonatologists around the world. It is not a 'wiki' in that entries are checked and cross-checked by other experts. All 'facts' on the dragonfiles system are to be supported by references to their primary source (in this release we aren't there yet). Where such sources are known to exist in an accessible digitised form on the web URL links will (in time) provide near immediate access to the detailed information.

The intention is to provide information, and to back that information with an authoritative source. Core taxonomic data (the labelling system) is maintained on the Dragonfiles site (and, it is intended, on mirrors) but the name list on the Dragonfiles database is used to provide links to a diversity of data stored on systems world wide. Species names are based on the species list maintained by Jan van Tol of Naturalis (the National Natural History Museum) Leiden Holland.

What the dragonfiles system can do:

As with all 'knowledge systems' this project is only as good as the data entered. Once the system matures and data is entered you should be able to retrieve taxonomic, ecological, biogeographical and conservation data on every recognised dragonfly species. You should also be able to find keys etc applicable to the various regions in which the species occurs. Links will exist to any specialised sites dealing with the species as well as links to representative photographs etc on the web. For the moment coverage is patchy as people have to volunteer their knowledge.

Searches on the front page are of two sorts: the Genus species pair on the left searches for recognised genera/species; the 'Last resort search' on the right searches also for synonyms, subgenera and subspecies and accepts part names.
For the moment both forms of searching are fast (much less than 1/10s on the server ... but web transmission time can be somewhat longer).

You can search the reference list by author names or by terms in the title, keywords or abstract (where these latter have been entered) (please be aware however, that not all authors are as logical as you would be in choosing the words in a title). Searching in general is presently limited in scope but 'in principle' paths to increase this capability are mapped out. Full access to the power and flexibility of the database needs however, to await the control of some obvious security 'holes'. You can also check the species entries that use any given reference. This list is dynamically generated according to where the reference has been used in the Dragonfiles database. It is not produced statically by scanning the reference article.

The links to electronically available material will (in what is becoming a surprising number of cases) allow you to access the original work wherever you may be. There are active programmes to digitise older and rarer work to make them available to users world wide. This includes one associated with the Dragonfiles project. Please bear with us as this is both a time consuming, and unfunded, task. Finding digitised literature is also not a straightforward activity ... so feel very free to inform us of the URLs of any unlinked digitised literature that you come across.

The usefulness of the system will depend on people entering data - we can't afford to do it all and experts can do things quickly and easily that might take people without their specialised knowledge a long time.

References are the core of the system. References have been obtained from many sources and there are style differences between various formulations. We are starting with some 5000 references and anticipate having at least 15000 in the not too distant future. Ideally every Odonata reference should end up here. You should be able to transfer primary information into the format you desire without too much effort. Some references have additional information in [ ... ]. This includes such cases as records of different pagination in different sources ... and in the case of some references (e.g. some by Selys) both pagination for separates and for associated journal articles. Where part works occur the individual parts should have their own year of publication and pagination data. This may make it difficult for people who wish to cite the whole work, however it prevents reference links from being split up and duplicated. Where year of publication is suspect alternate dates should be in [ ... ] with either a leading or a trailing '?'.

Technical stuff

At present the database is organised using the Latin-1 character set. It is intended that this should change to UTF-8 at some stage. The conversion will be transparent to users. It is hoped that native language information will be incorporated into the dataset. For expediency this will be included in the title of the reference and be in [ ... ]. For organisational reasons names of first authors in the linking field have been converted into ASCII using the big hammer approach ... so Förster becomes Forster and Hämäläinen translates to Hamalainen ... etc., however, all names in the references are intended to be in an appropriate Latin script ... CJK names are latinised using standard transliteration tables. For sorting reasons the searchable database will remain latinised.

Contributors

Contributors will be allocated a username and a password on the system which enables privileged access to tables in the accumulation database. Data in this active database can be edited via a web-browser. A full audit trail is maintained. Some data is restricted to special users (this includes creating new species entries and moving species entries into or out of synonymy). Where such changes are required please contact the Administrator. Debate on contentious points may be entered into, but the coordinator's viewpoint will stand (this is not a wiki). Alternative viewpoints will be indicated in the entry by the coordinator.

The active database tables will be scanned and checked regularly and, depending on levels of activity, the updated information will then be melded into the main tables and transferred over to the non-editable passive database. It is then available for public access.

Instructions for contributors:

Contributors need to enter data in a consistent and disciplined way. Probably in a more consistent and a more disciplined fashion than our own initial data loading. Feel free to tidy things as you see them and they irritate you.

Data entry/editing principles:

All data must have an authoritative published source.

1. Before entering data you must establish a reference link to that authority in the reference list. WARNING: a visit to the References table clears the data entry form.


2. The process for finding/entering a reference depends on where you start from. If starting from the 'References' link you can check for references in the Search Form, or you can go directly to the 'Enter a new reference' button. If starting from a data entry field within an entry page you click on the 'Check reference links here' link.
   
   
3. Entering a reference brings up a two line form. Enter the hammered flat (= ASCII coded) 'Surname of first author' and the 'Publication date', this will give a list of all recorded publications for that author and publication year +/- 3 ... If the desired reference is there click the link associated with the selected reference. If the reference isn't there then:
   • after carefully checking the reference really isn't already there (including in nearby years and with different title or pagination!)
   • choose the 'Add a new reference' button
   • enter a 'hammered flat' surname for the first author (= ASCII coded, no compensations - ö does NOT translate to 'oe' but to o ... etc)
   • enter the year of publication and then click on the 'Submit' button If similar entries are detected they will be displayed. If the reference is already there then press 'Cancel', else
   • fill in the form, cutting and pasting appropriate Latin1 text forms to get appropriate characters
   • make sure a latin1 version of the title is present (whether or not in [ ... ]). You may wish to add an English translation (in [ ... ]) if you believe the item will be searched for in English
   • on completing the reference press the 'Submit' button

Once the reference has been located (or entered for a new reference) proceed to use the web-based data form to associate the reference with the data for the species concerned. Pick the reference from the drop-down box then fill in the appropriate fields. Once submitted the data is fixed, but it can be edited through the appropriate 'Edit (whatever) entry' button.

You can edit references by clicking on the author/year link in the left hand column of the References table.

When entering references the evil we most want to avoid are redundancies (repeats) which can end up with separate links to data. Any such fragmenting of the reference/data link will markedly reduce the value and ease of operation of the database and make for an administrative head-ache. Please be careful.

Any redundancies you notice in the Reference listing should be communicated to the Administrator as soon as possible. With any luck nobody has linked to one of them ... which can then be removed. If more than one has data links then it will be a difficult job to fix.

Anyone can enter a reference but removal of references is an Administrator level action ... so please be careful. Once a reference is entered others may link to it, so reference removal can require careful disentangling of threads and transfer of links ...

The 'Locations' button in the header allows for speedy entry of data from checklists. Please choose checklists that cover large areas (e.g. countries), not local checklists.
Enter the checklist bibliographic details and web address if there is one. Then select the biogeographic zone, country and (if necessary) the National subregion (for countries like China, which spans biogeographic zones, this is important). Proceed to the entry form and tick appropriate boxes in the dropdown list of species that is presented. In tests this has proved swift and simple. You can jump between families using the family links on the data entry form.
The checklist data are then available under the 'Locations' button and is also entered into the species entries.
DO NOT enter dubious data (some checklists have 'idiosynchrasies' - we will not propagate these).

Style problems:

Treatment of subgenera and subspecies/infra-subspecific forms:
We didn't intend to handle subgenera, subspecies or infra-subspecific forms ... simply it becomes too difficult in the search phase as the search engine needs to parse complex and messy forms and the potential exists for confusion with seeming synonyms. Unfortunately important information has appeared within both subgenera and subspecies. We have attempted to handle these rankings without breaking the core database structure with, at this stage, mixed success. We would have preferred that subgenera, subspecies and infra-subspecific forms should be handled in the comment (right hand) field. They would have been able to be sought there using an 'includes' search. However, large scale use of subgenera in nineteenth century literature (with most shortly becoming full genera as diversity was recognised) and local use of both subspecific forms and infra-subspecific names means we are having to try. This component should be regarded as dynamic.

Proscribed fields:

To maintain control over species entries the basic species entry capability is only available to the Administrator. Creation of species entries and reduction to/elevation from synonymy are also Administrator class actions (these will involve checking links and melding/separating the entries).

Administrator:

Where an Administrator action is required please contact the Administrator. The response may not be instantaneous as others may need to be consulted before a decision is made. The fact there is a published opinion doesn't necessarily mean the opinion is accepted by the community ...