@make{slides}
@define{bigger=transparent,size +6}
@define{smaller=transparent,size -6}
@pagefooting{right="@-{@-{@-{RDB Santorini 9/27/94  [@value{page}]}}}"}

@form{drawing=<@begin{center,leftmargin 0,rightmargin 0,
                        below 2 lines}
                @comment[@graphic MUST start in first column]
@graphic{postscript=@parm(file),
                        magnify=@parm(magnify,default 1.0),
                        boundingbox=@parm(boundingbox,default file)}
                @string{drawingtagname=[@parm{tag,default Notag}]}
                @case{drawingtagname,
                        Notag [@comment{No tag given}],
                        Else [@tag{@parm(tag,default [])}]}
                @end{center}>}
@form{scrdump=<@begin{center,leftmargin 0,rightmargin 0, below 2 lines}
@graphic{postscript=@parm(file),magnify=@parm(magnify,default 1.0),
			boundingbox=@parm(boundingbox,default file)}
		@end{center}>}

@comment{-----------------------------------------------------------}

@blankspace(2)
@heading{The MikroKARAT
Distributed Knowledge
Acquisition Environment}

@begin{center}
Dr. Ralf D. Brown

Carnegie Mellon University
Center for Machine Translation

ralf+@@cs.cmu.edu
@end{center}

@comment{-----------------------------------------------------------}

@newpage()
@heading{Mikrokosmos}

Mikrokosmos is a knowledge-based machine translation project
based on multiple interacting microtheories, such as lexical-semantic
dependency, aspect, discourse relations, and style.

It is a multi-site project, with researchers at New Mexico State
University, Carnegie Mellon University in Pennsylvania, and the
Department of Defense in Maryland, as well as outside consultants.
Thus the need for tools supporting geographically-dispersed users.


@comment{-----------------------------------------------------------}

@newpage()
@heading{MikroKARAT}

@begin{format,leftmargin +1.8in}
@bigger(Mikro)kosmos
@bigger(K)nowledge
@bigger(A)quisition
and
@bigger(R)eview
@bigger(A)ssistance
@bigger(T)ools
@end{format}

MikroKARAT is a multi-user, client-server-based knowledge
acquisition environment written in C++ using the Motif
user interface under X Windows.

A first release (still incomplete) was rolled out on
September 22, 1994.


@comment{-----------------------------------------------------------}

@newpage()
@heading{Client-Server Architecture}

@blankspace(0.75in)
@drawing{file="ukarat-arch.pic",magnify 1.3,tag cliserv-arch}


@comment{-----------------------------------------------------------}

@newpage()
@heading{Development Team}

@begin{center}
Constantin "Costya" Arcadievich
Ralf Brown
Oren Glickman
Dean Grannes
Anandeep Pannu
Gerald Reno
Henry Robertson
@end{center}

@comment{-----------------------------------------------------------}

@newpage()
@heading{Goals}

Develop a fast and easy-to-use acquisition environment
which can be used by multiple geographically dispersed
knowledge analysts to simultaneously update multiple
large knowledge bases of various kinds.

Users should be prevented from accidentally obliterating
each other's updates, yet be given the tools with which
to cooperate as if they were in the same room rather than
thousands of miles apart.

Updates should be checked immediately to prevent invalid
data from being entered.


@comment{-----------------------------------------------------------}

@newpage()
@heading{ONTOS}

ONTOS is an ontology acquisition environment developed
in 1988, written in Lisp with an X Windows interface.

Drawbacks:
@begin{enumerate}
Slooooww

Few Consistency Checks

Buggy Display Code
@end{enumerate}


@comment{-----------------------------------------------------------}

@newpage()
@heading{MikroKARAT Client Architecture}

@blankspace(0.75in)
@drawing{file="ukarat-overview.pic",magnify 1.25, tag client-arch}

@comment{-----------------------------------------------------------}

@newpage()
@heading{The MikroKARAT Client User Interface}

@scrdump{file="fullscreen.pic",magnify 0.84}


@comment{-----------------------------------------------------------}

@newpage()
@heading{Knowledge Representation: Frames}

Frames are an associative memory:

@begin{format}
named frames contain
@hsp{1in}named slots which contain
@hsp{2in}named facets which contain
@hsp{3in}actual filler values
@end{format}

Values are retrieved by giving the name of the frame,
slot, and facet containing the value(s) of interest.



@comment{-----------------------------------------------------------}

@newpage()
@heading{FramepaC}

FramepaC is a fast frame manipulation package written in
C++ and patterned after the Lisp-based FrameKit version
2.0.  It trades off some generality present in FrameKit for
raw speed.

Features:
@begin{itemize}
Lisp-style polymorphic type system

inheritance (multiple methods available)

daemons (compile-time option)

persistent frames
@end{itemize}


@comment{-----------------------------------------------------------}

@newpage()
@heading{Frame Format}

FrameKit (old):

@begin{format,size -2,indent +0.25in}
(make-frame-old frame1
   (slot1 (facet1 fillerA fillerB ...)
          (facet2 fillerC fillerD ...))
   (slot2 (facet3 fillerE fillerF ...))
)  
@end{format}


FrameKit (new):

@begin{format,size -2,indent +0.25in}
(make-frame frame1
   (slot1 (facet1 (view1 fillerA fillerB ...))
          (facet2 (view2 fillerC fillerD ...)))
   (slot2 (facet3 (view3 fillerE fillerF ...)))
)
@end{format}


FramepaC:

@begin{format,size -2,indent +0.25in}
[frame1
    [slot1
        [facet1 fillerA fillerB ...]
	[facet2 fillerC flllerD ...]]
    [slot2
        [facet3 fillerE flllerF ...]]]
@end{format}


@comment{-----------------------------------------------------------}

@newpage()
@heading{Frames and Knowledge Bases}

In our system, a knowledge base consists of a set
of linked frames.

Ontologies are a hierarchy of concepts connected by
IS-A and PART-OF relations.

Text Meaning Representations (TMRs) are linked by
relations such as AGENT, THEME, ASPECT, and ATTITUDE.

Lexicons are have sense frames linked by N, V, ADJ, etc. to
superentry frames.


@comment{-----------------------------------------------------------}

@newpage()
@heading{Text Meaning Representation}

@begin{verbatim,size -3}
;; "... the four concerns said ..."
%say_1
    agent           set_1
    theme           %create_1
    temporal        time_1
    aspect          aspect_1
    attitude        attitude_1
    polarity        positive
 
set_1
    member-type     *company
    cardinality     4
    elements        %company_1 %company_2
		    %company_3 %company_4
 
%company_1
    name    "Beijing International Trust and
		 Investment Corp."
    alias   "Bitic"
 
%company_2
    name    "Nederlandsche Middenstandsbank
		 N.V."
 
%company_3
    name    "K.K. Yeung Management
		 Consultants, Ltd."

%company_4
    name    "Scriven Trading, Ltd."
 
time_1
    absolute-time   {past}
 
aspect_1
    phase           end
    iteration       single
    duration        momentary
@end{verbatim}

@comment{-----------------------------------------------------------}

@newpage()
@heading{Text Meaning Representation}

@begin{verbatim,size -4.3}
[heads
    [head-list [value say_1 own_1 own_2 own_3
		own_4 advise_1 engage_1]]]
[say_1
    [instance-of [value say]]
    [agent [value set_1]]
    [theme [value create_1]]
    [temporal [value time_1]]
    [aspect [value aspect_1]]
    [attitude [value attitude_1]]
    [polarity [value positive]]]
[set_1
    [instance-of [value set]]
    [member-type [value *company]]
    [cardinality [value = 4]]
    [elements [value company_1 company_2 
		company_3 company_4]]]
[company_1
    [instance-of [value company]]
    [name [value "Beijing International Trust
		 and Investment Corp."]]
    [alias [value "Bitic"]]]
[company_2
    [instance-of [value company]]
    [name [value "Nederlandsche
		 Middenstandsbank N.V."]]]
[company_3
    [instance-of [value company]]
    [name [value "K.K. Yeung Management
		 Consultants, Ltd."]]]
[company_4
    [instance-of [value company]]
    [name [value "Scriven Trading, Ltd."]]]
[time_1
    [instance-of [value time]]]
    [absolute-time [value {past}]]]
[aspect_1
    [instance-of [value aspect]]
    [phase [value end]]
    [iteration [value single]]
    [duration [value momentary]]]
@end{verbatim}

@comment{-----------------------------------------------------------}

@newpage()
@heading{Permanent Storage of Frames}

FramepaC provides transparent demand-loading of frames
from either a disk file or a server.

Changes necessary to use persistent frames:
@begin{enumerate}
initialization call: initialize_VFrames_disk or
   initialize_VFrames_server

cleanup call: shutdown_VFrames

optional update calls: store_VFrame or
	synchronize_VFrames
@end{enumerate}

Multiple databases of frames can be active simultaneously,
each in a separate name space.


@comment{-----------------------------------------------------------}

@newpage()
@heading{Frame Editor}

A highly-configurable structured editor for frames is
the main knowledge-entering tool.

The editor configuration can specify:
@begin{itemize}
required slots or facets

permissible slots or facets

types of fillers

default values

aliases for slot names
@end{itemize}

The editor thus supports both free-form frames such as
ontological concepts and very rigid frames such as
lexical entries.


@comment{-----------------------------------------------------------}

@newpage()
@heading{Editor Window}

@scrdump{file="edit-ont.pic",magnify 0.9}


@comment{-----------------------------------------------------------}

@newpage()
@heading{Graphical Browser}

The browser provides an overview of a knowledge base,
and the ability to
@begin{itemize}
launch the editor on a frame

create new frames

manipulate links between frames
@end{itemize}


@comment{-----------------------------------------------------------}

@newpage()
@heading{Browser: Ontology}

@scrdump{file="br-ont-h.pic"}


@comment{-----------------------------------------------------------}

@newpage()
@heading{Browser: Ontology}

@scrdump{file="br-ont-v.pic",magnify 0.8}


@comment{-----------------------------------------------------------}

@newpage()
@heading{Browser: TMR}

@scrdump{file="br-tmr.pic"}


@comment{-----------------------------------------------------------}

@newpage()
@heading{Configuration File: Ontology Browser}

@begin{verbatim}
Browser-Title:	"Frame Browser: %F"
;Geometry:
Orientation:	HORIZONTAL
Start-Node:	*ALL
Subtree-Only:	No
Hierarchy:	IS-A, PART-OF
Lazy-Fetch:	5
Homogeneous-Frames: No
;Disable-Commands:
;Editor-Config-File:
;Editor-Config-Number:
@end{verbatim}

@comment{-----------------------------------------------------------}

@newpage()
@heading{Configuration File: Ontology Editor}

@begin{verbatim,size -2}
;; Frame Editor configuration file for ontology
;;	by Ralf Brown and Sergei Nirenburg
;;	last edit: 7/5/94

Editor-Title:	"Ontology: %F"
;Geometry:
;Font-Name:
;Font-Size:
Permitted-Slots: *
Required-Slots:	 IS-A, DEFINITION, TIME-STAMP
Default-Slots:	 DEFINITION, TIME-STAMP
Hidden-Slots:	TIME-STAMP
;Restrictions:
;Disable-Commands:


Slot:	DEFINITION
Expand:	Yes
Valid-Facets:	VALUE
Required-Facets: VALUE
Filler-Type:	VALUE = STRING
Inherit:	None

Slot:	TIME-STAMP
Valid-Facets:	VALUE
Required-Facets:	VALUE
Default-Filler:	VALUE = "%U on %D at %T"
Add-Fillers:	VALUE = "%U on %D at %T"

Slot:	*
Valid-Facets:	*
Required-Facets: SEM
@end{verbatim}

@comment{-----------------------------------------------------------}

@newpage()
@heading{Configuration File: Lexicon Editor}

@begin{verbatim,size -2}
Editor-Title:	"Lexicon Entry: %F"
Permitted-Slots: ANNO_DEFN, ANNO_EX, APPL_CURR,
	APPL_LOCAL, APPL_FIELD, APPL_LANG,
	ORTH_VAR, ORTH_ABBR, PHONOLOGY,
	MORPH_IRREG, MORPH_STEMV, MORPH_PAR,
	SYN_CLASS, SYN_IDIOF, SYNS_CLASS,
	SYNS_FSPAT, SEM_LEXMAP, SEM_MPROC,
	LR_PAR, LR_SYN, PRAG_FORM, PRAG_SIMP,
	PRAG_COLR, PRAG_FORC, PRAG_DIRE,
	PRAG_RESP, PRAG_ANLY, PRAG_GEN,
	ANNO_COMMENT, TIME-STAMP
Required-Slots:	 ANNO_DEFN, ANNO_EX
Default-Slots:	 ANNO_DEFN, ANNO_EX
Hidden-Slots:	TIME-STAMP
Restrictions:	((PRESENT SYNS_CLASS.SEM) &
		    (ABSENT SYNS_FSPAT.SEM)) |
		((PRESENT SYNS_FSPAT.SEM) &
		    (ABSENT SYNS_CLASS.SEM))

Slot:	PRAG_FORM
Alias:			"Formality"
Valid-Facets:		SEM
Required-Facets:	SEM
Filler-Type:		SEM = RANGE(-5,5,0)
Default-Filler:		VALUE = 0
Max-Fillers:		VALUE = 2
  [five similar slot definitions omitted]

Slot:	ANNO_DEFN
Alias:			"Definition"
Valid-Facets:		VALUE
Required-Facets:	VALUE
Filler-Type:		VALUE = STRING
  [two similar slot definitions omitted]

Slot:	TIME-STAMP
Alias:		 "Modification History"
Valid-Facets:  	 VALUE
Required-Facets: VALUE
Filler-Type:	 VALUE = STRING
Default-Filler:	VALUE = "%U created on %D at %T"
Add-Fillers:	VALUE = "%U updated on %D at %T"

Slot:	APPL_CURR
Alias:			"Currency"
Valid-Facets:		*
Required-Facets:	SEM
  [three similar slot definitions omitted]

Slot:	*
Valid-Facets:	*
Required-Facets: SEM
@end{verbatim}

@comment{-----------------------------------------------------------}

@newpage()
@heading{Configuration File: TMR Browser}

@begin{verbatim,size -2}
Browser-Title:	"TMR Browser: %F"
;Geometry:
Orientation:	HORIZONTAL
Start-Node:	HEADS
Subtree-Only:	No
Hierarchy:	HEAD-LIST, AGENT, THEME,
	ATTITUDE, LOCATION, ELEMENTS, SOURCE,
	PURPOSE, ASPECT
Lazy-Fetch:	3
Homogeneous-Frames: No
;Disable-Commands:
Editor-Config-File: "edit-tmr.cfg"
;Editor-Config-Number:
@end{verbatim}

@comment{-----------------------------------------------------------}

@newpage()
@heading{Glosser}

A very shallow, heuristically-driven English generator
for TMRs.

@scrdump{file="glosser.pic"}
@scrdump{file="glosser2.pic",magnify 0.6}

@comment{-----------------------------------------------------------}

@newpage()
@heading{Linguistic Utilities}

MikroKARAT provides access to all the usual utilities:
@begin{itemize}
dictionary lookup

corpus lookup

KWIC searches
@end{itemize}

It also provides a special frame-search capability.  The names of
all frames in a knowledge base which match the specified criteria are
displayed in a pick list, from which the user may select one or more
to edit.


@comment{-----------------------------------------------------------}

@newpage()
@heading{KWIC Searches}

@scrdump{file="kwic.pic",magnify 0.9}

@comment{-----------------------------------------------------------}

@newpage()
@heading{Frame Searches}

@scrdump{file="frsearch.pic",magnify 0.8}

@comment{-----------------------------------------------------------}

@newpage()
@heading{Scanning Found Frames}
@scrdump{file="searchresult.pic"}

@comment{-----------------------------------------------------------}

@newpage()
@heading{Chat Utility}

To support cooperative updates to a knowledge base,
MikroKARAT provides interactive communication between
users.

(not yet implemented, but one-line messages can be
sent to other users)

@scrdump{file="send-message.pic"}


@comment{-----------------------------------------------------------}

@newpage()
@heading{Work Remaining}

@begin{enumerate}
complete structured editor

connect constraint checks

complete and connect Chat utility

connect external linguistic utilities
@end{enumerate}


@comment{-----------------------------------------------------------}

@newpage()
@heading{Planned Enhancements}

@begin{enumerate}
Concept Cluster Manager

To-Do lists

Annotations
@end{enumerate}

@comment{-----------------------------------------------------------}
@comment{extra slides}
@comment{-----------------------------------------------------------}

@newpage()
@heading{Concept Cluster Manager}

What good are concept clusters?

Concept clusters permit an easy overview of the state of knowledge
acquisition for a particular concept.  One can easily determine
whether lexical entries for all supported languages have been entered,
whether there are sufficient examples provided by TMRs, etc.
Incomplete clusters can automatically be flagged for further KA, or
even cause the KA tool to initiate the acquisition automatically.

The Cluster Manager icon:
@drawing{file="cluster.pic"}


@comment{-----------------------------------------------------------}

@newpage()
@heading{Annotations and To-Do Lists}

With multiple users, there must be a method for indicating to others
what has been done, why it has been done a certain way, and what
remains to be done.

By creating a "shadow" frame with annotations for each frame in the
knowledge base proper, these comments and to-do lists can be stored
where they are readily accessible without adding unnecessary data to
be loaded by programs using the knowledge base as a source of
information.


@comment{-----------------------------------------------------------}

@newpage()
@heading{Configuration File: Lexicon Browser}

@begin{verbatim,size -2}
Browser-Title:	"Lexicon Browser: %F"
;Geometry:
Orientation:	HORIZONTAL
Start-Node:	SUPERENTRY_INDEX
Subtree-Only:	No
Hierarchy:	ENTRIES, N, V, ADJ, ADV, PP
Lazy-Fetch:	3
Homogeneous-Frames: No
;Disable-Commands:
;Editor-Config-File:
;Editor-Config-Number:
@end{verbatim}

@comment{-----------------------------------------------------------}

@newpage()
@heading{MikroKARAT Server Display}

@scrdump{file="server.pic",magnify 0.85}