Perspectives AAAI Symposium on Agent Mediated Knowledge Management

Perspectives AAAI Symposium on Agent Mediated Knowledge Management March 26, 2003 Sidney Bailin Knowledge Evolution, Inc. (www. kevol. com) Walt Truszkowski NASA/Goddard Space Flight Center

Problem Scenario • Agents must cooperate in some task • They disagree about what must be done

A Possible Diagnosis and Treatment • Ontology mismatch – Agents mean different things by the same terms – Agents view the world differently • Solution: ontology negotiation – An incremental process for establishing shared understanding – Cf. Bailin & Truszkowski, “Ontology negotiation between intelligent information agents, ” Knowledge Engineering Review, Volume 17, Issue 1, March 2002.

How do the agents know there’s an ontology mismatch?

Simple Model of an Observer Interpretation Knowledge Base Message Stream

Making Sense of an Observation Iterative KB Update p, q, r, s, t, u, v, w. . . Update a, b, c, d, e ? Interpretation Update Problem Message

Examples of Observation • Remote Sensing • Auditory – Name that tune – “What does this mean? ” • Verbal – Intelligence analysis

Knowledge Representations • Propositional – Assertions • Ontological – Assertions about types • Event sets – E. g. , Probabalistic models • Attribute sets – E. g. , Neural net interpretations • Lattice – States of knowledge

Propositional Representation Observation Type Propositional Interpretation KB Contents Auditory · That’s Beethoven’s opus 110 · The tempo is too slow · The walls of this house are too thin · My neighbor is a pianist · My neighbor is an audio buff · Knowledge of classical music · Opinions about music performance · Opinions about house construction · Beliefs about a person · A signature typical of iron compounds · Spectral analysis was recognized over this region · The asteroid surface is marked by · Topographic knowledge large craters · The soil contains a high concentration · Geo-chemical knowledge Remote Sensing of olivine Verbal · The Prime Minister speaks in the · Political events capital today · Tanks are en route to the northern · Military maneuvers border · Preparations for an attack seem to be · Predictions and risk assessment underway

Event Set Representation Observation Type Propositional Interpretation Possible Event Space Auditory · That’s Beethoven’s opus 110 · The tempo is too slow · The walls of this house are too thin · My neighbor is a pianist · My neighbor is an audio buff · A signature typical of iron · Musical compositions · Features of a musical performance · Architectural features · People’s skills · People’s hobbies · Spectral signatures of the elements · Topographic features · Chemical features Remote Sensing Verbal compounds was recognized over this region · The asteroid surface is marked by large craters · The soil contains a high concentration of olivine · The Prime Minister speaks in the capital today · Tanks are en route to the northern border · Preparations for an attack seem to be underway · Political events · Military events

Attribute Set Representation Observer Schema: - Type - Source - Capability - Threat. . . Type: Missile Source: Adversary Capability: Evasive Threat level: High Auto-respond: Yes. . . Interpretation Position: 50 Velocity: 600 km/hr 70 Size: 3 m Shape: oblong . . .

Attribute Set Representation Observation Type Observed Attributes Inferred Attributes Auditory · Instrument: piano · Key: A-flat major · Tempo: slow · Volume: loud · Possible composer: Beethoven Remote Sensing · Spectral features · Visual features · Probable substance: iron · Topographic type: cratered Verbal · Activity: speech · Actor: Prime Minister · Location: capital · Goal: rally public · Risk: low · Activity: movement · Actor: tanks · Possible goal: launch attack · Risk: high · Possible composition: opus 110 · Performance quality: sub standard · Sound insulation: inadequate

Ontological. Representation Observation Type Propositional Interpretation Classification Performed Auditory · That’s Beethoven’s opus 110 · My neighbor is a pianist · My neighbor is an audio buff · Identification of a specific entity by its signature · Classification of an entity by inference from its behavior Remote Sensing · A signature typical of iron compounds was recognized over this region · Recognition of the presence of a type of substance by its signature Verbal · Preparations for an attack seem to be underway · Identification of a probable threat on the basis of associated signs

Integrated Model Ontology Categories “Events” in the classification process are the possible categories to which the observed phenomenon can be assigned. Attributes Type: Missile Source: Adversary Capability: Evasive Threat level: High Auto-respond: Yes. . . “Events” in the initial interpretation process are the possible attribute-values that can be inferred about the observed phenomenon. Message Classification Initial Interpretation Uncertainty Method

What is “Perspective” (1 of 2) • Something that distinguishes one observation from another • Must be a difference in input, interpretation process, or knowledge base

What is “Perspective” (2 of 2) • Choice of input signal – Signal type – Wavelength • State of observer, target of observation, and environment – E. g. , time, position, angle of observation – As represented in the KB • Interpretation function – Choice of significant info in input signal – Mapping of selected signal to KB ontology – This is like human’s “perspective on an issue:

When Observations Conflict • Problem with signal? – Phenomenon: noise – Resolution method: calibration • Difference in state? – Phenomenon: state-dependent events – Resolution method: correlate state with events • Difference in interpretation? – Phenomenon: ontology conflict – Resolution method: ontology negotiation

Summary • We’ve developed a model of observation and perspective • Agents can explicitly represent and reason about perspective (theirs and others’) • Agents can use perspective reasoning to determine how to resolve a conflict – In particular, to determine if and when an ontology conflict is occurring