The SuNFiSH Project

What Is SuNFiSH?

SuNFiSH is a putative European-funded project to build a Scalable Networked Federated Search Hierarchy (The ``u'' and ``i'' don't stand for anything. It's hard to make good acronyms!)

What's Special About SuNFiSH?

Yes, the world is full of distributed searching systems; indeed there are plenty of European-funded projects that have built systems broadly similar in spirit to SuNFiSH - for example, Aquarelle, ELVIL, UNIverse, Decomate II and Elise II.

SunFiSH differs from these and other projects in several ways:

• Most importantly, it uses a ``plug-compatible'' architecture, in which brokers can transparently substitute for either servers or clients, meaning that arbitrary hierarchies of clients, brokers and servers can be constructed.
• Administratively, this gives autonomy not only to the maintainers of individual archives, but also to higher-level organisations which represent aggregates of datasets (or aggregates of aggregates, or ...)
• The SuNFiSH system explicitly addresses the implementation issues usually encountered by a broker aggregating result sets from disparate sources - for example, the problem of one server being much slower than another, or not responding at all.
• The use of a multilingual thesaurus allows queries to be phrased independently of language; this facility is useful not only in translating between different natural languages, but also between technical and ``everyday'' languages.
• The plug-compatibility and open definition of interfaces makes it feasible for third parties to create front-ends for media other than the web - for example, WAP phones or text-only command-line systems.

What's Happening?

The project is at a very early stage, but here's what's happened so far:

• An initial overview document has been written, focussing on the technical aspects of the system. (WARNING: this document is somewhat out of date.)
• Several potential project partners have been approached. Some are on board, others are considering whether or how to participate, and yet others are still to emerge.
• A speculative costing document has been written, breaking the project down into work packages with very very rough time estimates. This is available to project partners on request - send email to admin@sunfish.eu.com
• An ``proof of concept'' application of the system has been been designed: DinoSearch uses a network of SuNFiSH components to allow searching for technical papers related to dinosaur palaeontology. The result of the process will be, in the ideal case, a downloadable soft-copy of the paper; and otherwise, instructions on how to obtain a hard-copy.
• The Dinosaur mailing list has been contacted with an announcement of the DinoSearch application. Participation has been invited from palaeontologoical organisations in Europe.
• A new mailing list has been set up to discuss the DinoSearch application. Its home page is www.topica.com/lists/dinosearch (the simplest way to join is to send an email to <dinosearch-subscribe@topica.com>) and messages can be sent to dinosearch@topica.com.
• A message has been sent to the DinoSearch list outlining the system architecture in simple terms.
• A list of other systems similar in intent to DinoSearch is being compiled, with the intent that we can check our functionality against the systems on this list.