David DoddsDavid Dodds

david_dodds_2001@yahoo.com

David DoddsDavid Dodds

TOPICS:

Metaphor, Metaphor Processing, Metadata (DAML, RDF), SVG,

Ontologies

David DoddsDavid Dodds

Title of this paper:

Metaphor Processing in Knowledge Technology and

Artificial Intelligence

KT2002

Metaphor Example ProgramMetaphor Example Program

Associated with this set of slides is a program, anim01.svg, which is an SVG program which performs an animation. (Use an SVG viewer or browser plugin to see it.)

The purpose of this animation is to show how tacit and metaphorical information can be captured and represented in a computer in a functional / operational way.

Metaphor ProcessingMetaphor Processing

Is about computing using patterns that are NOT exact match.

Business processing is only exact match.Metaphor processing is not the same as

fuzzy logic or fuzzy sets.Metaphors are used in science and

technology to extend knowledge, discover new ideas and concepts.

The Albatross MetaphorThe Albatross Metaphor

Functional observation of birds in flight have long been used to glean notions that further the design and development of machines that fly.

The U2 aircraft was modeled on the Albatross’ ability to glide efficiently, due to its long narrow wings.

Albatross Metaphor continuedAlbatross Metaphor continued

The size and shape of the wings of the Albatross were determined to be the key factor in its efficient flight, and these aspects were copied in the plane engineering.

Other aspects of the Albatross were not transferred, like the feathers because they were not relevant to the goal (of aircraft design from bird).

SVG Animation MetaphorSVG Animation Metaphor

The SVG animation file which accompanies this set of slides is an instance of deBono diagram combined with Lakoff spatial metaphor.

This particular deBono diagram depicts goal oriented group behaviour in pursuit of achieving a goal but is blocked by an obstruction.

Lakoff spatial metaphors use orientation in three dimensional Euclidean space as representation.

deBono Diagrams 1deBono Diagrams 1 deBono diagrams, shown in Atlas of

Management Thinking, depict particles, which have trails, progressing from a starting point moving along a path to some end point. The particles represent agents or actors (people) and the path represents spatial progress toward some location. The agents move toward a labeled goal area, with the passage of time.

deBono Diagrams 2deBono Diagrams 2

There are “physical constraints” in the picture which prevent the paths from going just anywhere. These constraints represent contexts, such as “focus”.

The diagrams have an implied “gravity” (“naïve physics” model) and visual objects (like walls) are not permeable. There is a temporal factor as well as spatial one.

deBono Diagram 3deBono Diagram 3

Visually the deBono diagram shows that the spatial progress of two “agents” is halted by an obstruction, which moves into place as they approach near the goal, and the HML terms referenced depict that there is a concommitant socio-cultural aspect to such deliberate deflection or blockage.

The Metaphor Explicitly 1The Metaphor Explicitly 1

this animation, then, visually presents through time the following metaphor. “Actors” proceed from initial starting points directly towards a goal area. The path represents effort or a journey and in this example is directly oriented to arriving at the goal area after some time has passed in the transit.

The Metaphor Explicitly 2The Metaphor Explicitly 2Just as they are approaching near the goal

area a physical barrier slides in front of them, preventing their further approach to the goal area. There is a tunnel object in the picture which prevents the path from going to either side to circumvent the obstruction. This is a visual depiction of constrained activity via the actors. It is a depicted context.The non- permeable barrier (obstruction) object visually represents a constraint.

The Metaphor Explicitly 3The Metaphor Explicitly 3Another actor, the obstructor, causes the

occurance of the obstruction and it enters the context of the animation near the end of the animation duration time. The animation stops with the actors blocked from achieving the collective goal.

Lakoff spatial metaphors have an “origin” (locus) or “starting place” (“zero”) based on the body of the person using the metaphor.

With substantial ease adult humans assign the origin of this space to being at the eyes of the self, or generally the outerbody surface (“my outside”) and as needed the innerbody volume bounded by that outerbody surface (“my insides”).

Humans are able to effectively “move” the origin from the body of self to some other location. This other place then gives the “zero perspective”.

What is the Metaphor here? 1What is the Metaphor here? 1The metaphor expressed via this computer

representation is that people act through time to achieve a goal. Metaphorically they take a journey along a path and ever more closely near the goal (fixed goal in this case). Constraints in the world affect their path and control their efforts. Achieving a goal takes time, it is not instantaneous and not without constraints.

What is the Metaphor here? 2.What is the Metaphor here? 2.An actor can introduce deflections or

blockages of progress. There is concomitant social and other affect which precedes actions and which is produced as the result of actions and events. These are modeled by use of HML.

deBono diagram paths and nonpermeable barriers are visual metaphors for activity and constraints in the world.

Lakoff Spatial Metaphor 1Lakoff Spatial Metaphor 1Lakoff spatial aspects in this visual

animation of metaphor are that there is directionality imposed on the visual space. The top is “UP”, and the bottom of the visual space is “DOWN”. While this seems commonsensical computers do not have any such built in notions and must be programmed with that information.

Lakoff Spatial Metaphor 2Lakoff Spatial Metaphor 2Superimposed on top of deBono’s naïve

physics model of agent paths is Lakoff’s metaphor “SUCCESS IS UP”.

The closer the agent gets to the UP location (in this case the top of the picture) the greater his SUCCESS is.

This metaphor is often accompanied by the “GOOD IS UP” one as well, resulting in “I felt really up having achieved my goals”.

Lakoff Spatial Metaphor 3Lakoff Spatial Metaphor 3Because the two meaning contexts (deBono

diagram and Lakoff spatial) are superimposed the animated diagram can depict, in a way meaningful to the computer; agent activity, constraints on that activity and can model or associate relevant HML terms with the events and the actors participating.

Lakoff Spatial Metaphor 4Lakoff Spatial Metaphor 4Example: for most of the period of the

animation the actors are increasingly nearing the goal area, they are becoming increasingly successful. When the obstruction to further progress occurs they no longer have increasing success. There is a measure of success but incomplete. There is thwarting of goal achievement. Frustration could be inferred to result from this.

Lakoff Spatial Metaphor 5Lakoff Spatial Metaphor 5HML is able to provide representation for

this and a production system which uses HML terms can be used to infer (“predict” / detect) likely social outcomes from events in the actors physical world (such as deliberate thwarting).

HMLHML

The Human Markup Language (HML) structured vocabulary provides a standardized reference for the representation of socio-cultural information conveyed and implied in the deBono (Lakoff) diagram.

Contexts and inferencing provide means for computer determine appropriate terms use.

Constructing MetaphorsConstructing Metaphors Because the metaphors are depicted visually and

use XML SVG and other XML technologies, it is possible for the computer to do the following:

The locations, colors, sizes etc of all the graphics elements constituting an animation can be read by a program and used in inferencing.

The SVG code can be created by XSLT to produce animations anew. The animations could be designed to reflect a set of situations depicted by a graph structure like context graphs and or those indicated by HML terms, which may have been participants in a chain of inferences.

OpenCYC and SUMOOpenCYC and SUMO

DAML is used in OpenCYC & SUMO to express taxonomic interrelationships, amongst the general physical, cultural and social knowledge coded there. Terms like #$PurposefulAction and #$performedBy are related to other CYC concepts represented, in such a way that a reasoner can perceive “connections” not directly stated in input.