[Search for users] [Overall Top Noters] [List of all Conferences] [Download this site]

Conference heron::euro_swas_ai

Title:Europe-Swas-Artificial-Intelligence
Moderator:HERON::BUCHANAN
Created:Fri Jun 03 1988
Last Modified:Thu Aug 04 1994
Last Successful Update:Fri Jun 06 1997
Number of topics:442
Total number of notes:1429

51.0. "Egeria" by LARVAE::TURNER (Mark Turner * DTN 849-3429 * SPYDER::TURNER) Fri Jan 20 1989 17:18

This note is reserved for discussion of Egeria.  Please feel free 
to post questions, experiences, descriptions, and comments as replies 
to this note.  
T.RTitleUserPersonal
Name
DateLines
51.1ExperiencesLARVAE::TURNERMark Turner * DTN 849-3429 * SPYDER::TURNERFri Jan 20 1989 17:2159
[from Bryan Pogson, AICU, Basingstoke U.K.  
 Bryan is at SPYDER::POGSON or POGSON@USH]

Dear all,

I have just recieved the latest version of Egeria, v1.1. This appears to be
a version with fewer bugs than 1.0 but not much change in functionality.  

The new version comes with 2 books carved out of solid marble. As well as the
manual which is written in TechnoGibberish and is best understood after you've
spent 2 years  programming the thing, there is a 435-page tutorial called
"Expert Systems with Egeria" which looks more suitable for bedtime reading.
(Unless you can think of better things to do at bedtime.)

My first impressions of Egeria are:

	- *very* powerful language for probably all types of ES

		- frames, inheritance, object sets created on the fly
		- one:many, one:one, many:many relationships on the fly
		- very good pattern matching through inheritance, relationships
		  or simple variables; set operations
		- very good control: demons, internal procedural language
		  tasks, user-defined functions
		- good modularity within a knowledge base (scoping local to
		  classes of objects also covers local procedures, tasks and
		  display windows
		- ISI claim their implementation of Bayes' is the only correct
		  one... haven't checked this		
		- strongly typed including enumerated variables
		- can do simple list operations but not Lisp-like
		- can do a sort of predicate logic through relationships but
		  it's not Prolog

	- its a language: no syntax-directed editor, or debugger
	  (there is a run-time browser)

	- no incremental compilation, but no need to VMS-link unless using
	  external functions

 	- moderate integration with VMS (it's really a PC product)
		- can call external functions and could (I think) be called
                - output window editor knows nothing about DECWindows, UIS etc
                  but runtime i/o system could be modified by 'C' programming

	- will take a longish time to learn 

	- likely to be a bit buggy (Savoir was for its first few months)

	- support good from ISI

but I'll report some more when I've got more experience of it (probably in a
few months time).

I'll be happy to discuss it with anyone.

Regards,

		Bryan
51.2RIP EgeriaMALLET::POGSONMy other hull's a DartThu Mar 14 1991 13:0517
    The latest on Egeria is not good: it's dead.  ISI was taken over by
    Expertech in 1989. They could not then fully support it since only 2 or
    3 of the original developers stayed with the new company. Very few
    Egerias, if any, were sold.
    
    Expertech have now been taken over by Inference Corp to form Inference
    (Europe) which continue to market Expetech's Xi Plus along with ART and
    ART im. They have dropped Egeria from their price list. It is not
    clear why: I guess it may be because of overlap with ART which could have
    damaged its sales.
    
    Egeria was potentially a good product: to be an excellent product on
    our kit it needs DECwindows development and user environments and
    better tailored access to the OS: the underlying knowledge
    representation and inferencing is good.  Maybe Inference (if they have
    the rights) would sell it to anyone who wanted to develop it further.