Mnemosyne: P2P Reference manager

The project page on sourceforge
Developer Wiki

What is it?

Architecture

Status

Contacts

What is it?

Disclaimer

Although the following is written in present tense, it describes a vision. Not an existing product. If you want to help make this vision come realize, mail me (replace "-at-" with "@").

..a reference manager

At its basic level, a good tool for managing your references. These could be book or paper references, web locations, real locations, or contacts.

Mnemosyne provides:

A simple form-based interface for adding new references.
A similar interface for search by topic / keyword / title / etc.
Import and export to popular formats, such as BibTex, EndNote and RDF.
An API for providing "reference feeds" by e-mail, media messeging, IM, etc.

..a peer-to-peer knowledge sharing tool

The first exciting thing about Mnemosyne is that it lets you share bits of knowledge, just like you share MP3 files. Use the P2P capabilities to extend your search for references to your peer's knowledge base. In Mnemosyne you can:

Define a list of "buddies"
Define search scope: local / buddies / buddies^n / the world
Define private and sharable references. Default is sharable.
Contact users whose references you find interesting.
Share meta-knowledge, e.g. search strategies, by sending and receiving agents.

..a platform for knowledge engineering

Mnemosyne API allows you to define "knowledge agents", which can added highly sophisticated knowledge management tools to it, such as semantic processing, learning, planning, etc.

For users, this means you can get state-of-the-art power at your hands.

For researchers, this means you can test your ideas in a real-world setting with minimal effort.

..an agent playground

Mnemosyne agent API allows for two types of agents: knowledge processing and user interface. These agents interact with the user as well as with the Mnemosyne reference DB. Using the P2P channel, agents can also interact with each other. Furthermore, since agents are sharable in themselves they are also mobile. In a sense, they are even subject to population dynamics.

Architecture

Components

Core UI	Forms for entering / editing / searching references. Menus for file import / export / save / open / etc. Buddy management window. etc.
Core Engine	Backing store facades Search engine. Agent sandbox etc.
Backing store	Plugable storage: Basic XML file storage, mainly for debugging Simple integrated DB. JDBC interface.
P2P comm layer	Provides services for knowledge sharing: Manage buddies list. Define "trust" and "credibility" levels for buddies. Define relevence matrixes for information from diferent buddies. Send search queries to buddies, collate and rank results. Answer search queries from buddies. Send / handle complex search requests, which involve invoking agents. Transport agents to / from buddies. Mark incoming agents by their source.
Plugin API
Plugin UI space