Note: This page is historical.

Current pages about Yenta are here. Please look at those pages first.

Yenta is still under active development, but this particular page is not. If you're interested in current research papers about Yenta, or obtaining a copy of Yenta, please start here instead.

This page is one of many that were written in late 1994 and early 1995, and are being preserved here for historical purposes. If you're viewing this page, you probably found it via an old link or are interested in the history of how Yenta came to be. These pages have not been actively maintained since 1995, so you'll find all sorts of older descriptions which may not match the current system, citations to old papers and old results, and so forth.

Yenta implementation overview

We are building a toolkit allowing rapid prototyping of potential applications and agent organizations. The basic structure of the toolkit consists of a "core set" of modules which implement a basic set of capabilities that almost any networked, cooperative agent would require. The rest of the toolkit consists of the actual specialized agents that would make up some particular application. Particular modules in the core set are designed to be interchangeably replaced with similar modules; hence, while any given agent could depend on their being some implementation of each module in the core, the particular implementation of any given core module is not fixed.

Where possible, existing implementations will be used, if they exist where we need them. For example, rather then reinventing various cryptographic wheels, the PGP and RSAREF libraries will be used where possible; similarly for clustering mechanisms such as SMART or WordNet.

Amongst the core set modules are:

Basic network-layer communication, ported to many machine architectures.
A simple "installer" to allow relatively painless acquisition of agent-based applications by unsophisticated users. (This simple "installer" would therefore be responsible for doing all of the communication and configuration required to move a possibly large and complicated application from the distribution site.)
Cryptographic technology required for authentication, confidentiality, and key distribution (e.g., similiar to the capabilities provided by PGP in addition to other tools as needed by particular applications (e.g., zero-knowledge-proof subsystems and similar information-hiders). This module in particular requires a design such that export-restricted technology can be removed before international shipment and replaced with local equivalents.
One or more "scanners" that extract user information, to be used in user modelling for matchmaker-type applications. (Such scanners might acquire the user's email or netnews that has actually been read, for example. Such scanners would be responsible for acquiring the "raw bits", but not for a high degree of interpretation, and could be implemented in a large variety ways (such as scanning the user's files, adding hooks to Emacs modes such as RMAIL, and so forth).
One or more "correlators" that attempt to use the results of scanners' work to determine, e.g., matches of user interests with another user. Semantic-net approaches, such as Part-of-speech taggers combined semantic nets such as WordNet, or vector-space approaches, such as SMART, would likely go here.
One or more user interfaces, dependent upon the applications. (Likely interfaces might be standalone X programs, Emacs modes, and so forth. One particularly interesting interface might be to depend on the existence of a WWW browser as the front end to an agent---taking care to use a secure browser if the actual communications path to the agent goes over a network, rather than coming from a local filesystem.
Progress monitoring reporters, to enable evaluation of the spread and utility of such agents, which transmit selected information back to monitoring software elswhere. Such reporters would have to be designed to yield useful monitoring information without breaching their users' privacy, and to give users a chance to verify this behavior.

Lenny Foner

Last modified: Fri Dec 15 04:17:22 1995