=Paper=
{{Paper
|id=Vol-52/paper-17
|storemode=property
|title=Agent-based Structures, Agent Ontology Preservation and Enterprise Modeling
|pdfUrl=https://ceur-ws.org/Vol-52/oas01-nourani.pdf
|volume=Vol-52
}}
==Agent-based Structures, Agent Ontology Preservation and Enterprise Modeling==
https://ceur-ws.org/Vol-52/oas01-nourani.pdf
Agent-based Structures, Agent Ontology Preservation
and Enterprise Modeling
Cyrus. F. Nourani
projectmetaai@cs.com
ABSTRACT We refer to Alg[A] and Alg [F] are what we call ontology
Design techniques with software agents and Abstract algebras. The implementation mapping I defines wrappers
Intelligent Implementations are presented. Agent to resources in a manner preserving the ontology algebra.
morphisms are defined and applied to preservation Ontology algebras are multi-sorted algebras defining
principles. The object level definitions for individual multiagent systems defined by formal agents, e.g. hysteretic
modules can be automatically programmed by source or knowledge level agents and agent morphisms. A formal
abstract syntax tree to target abstract syntax tree definition is provided. The Ontology Preservation
morphisms. AII techniques are applied to define an Principle: The AII is correct only if it preserves the
Ontology Preservation Principle for Heterogeneous KB ontology algebras. It will be abbreviated by AIIOPP.
Design and implementation. Widerhold’s domain knowledge base algebra DKB consists
of matching rules linking domain ontology. There are three
The stages of conceptualization, design and operations defined for DKB. The operations are
implementation are defined by AI agents and Mediators. Intersection- creating subset ontology and keeping sharable
Multiagent implementations are applied to software design entries. Union- creates a joint ontology merging entries.
techniques, which incorporates object level Difference- creates a distinct ontology and removing
nondeterministic knowledge learning and knowledge shared entries. Mapping functions must be shown to
representation developed in [12]. Objects, message passing preserve ontology. Applying AIIOPP we can state specific
actions, and implementing agents are defined by syntactic preservation principles as follows. The DKB Preservation
constructs, with agents appearing as functions. By defining Principle- AII implementations must preserve ontology
specified agent activators events and activity are computed under Intersection, Union, and Difference operations. . The
for the AII agents. The proposed AII techniques provide a algebras Alg[A] and Alg[F] define wrappers for the
basis for an approach to automatic implementations from mediators as functions for interacting with resources. A
intelligent syntax trees. Interpretability is defined by wrapper is a tool to access known resources and translate
mediators implementing objects and agents their objects.
1. Abstract Modeling with Computing Agents 1.1 Agents
The paper on which the enclosed abstract is based presents Starting with hysteretic agents [5], the agent has an
a formal basis to agent ontology modeling and ontology internal state set I, which the agent can distinguish its
preservation is specific designs. The notion of abstract membership. The agent can transit from each internal state
implementation defined by this author in [1,4] are either to another in a single step. Actions by agents are based on I
algebraic or model-theoretic (algebraic logic) definitions. and board observations. There is an external state set S,
We refer to specifications of the form , modulated to a set T of distinguishable subsets from the
abbreviating , as observation viewpoint. An agent cannot distinguish states
presentations. A design realization applies a triple I: in the same partition defined by a congruence relation. A
with agents with specific ontologies, sensory function s :S → T maps each state to the partition it
where I is an agent ontology preserving mapping. belongs. Let A be a set of actions which can be performed
Informally the process of design realization was defined by by agents. A function action can be defined to characterize
the author to be that of encoding the algebraic structure of an agent activity action:T →A.
the conceptualization of a problem with agents onto the There is also a memory update function mem: I x T → I.
algebra that specified a specific system. Thus design To define agent at arbitrary level of activity knowledge
realization is via morphisms on agent-based algebras. A level agents are defined. All excess level detail is
multiagent System design might be defined as an eliminated. In this abstraction an agent’s internal state
"enterprise" by is implemented by agents that consists entirely of a database of sentences and the agent’s
characterize the implementation function actions are viewed as inferences based on its database. The
I: → is to be defining a action function for a knowledge level agent maps a
mapping database and a state partition t into the action to be
I: → performed by an agent in a state with database and
observed state partition t. action: Dx T→ A
The update function database maps a state and a state symbols are implemented by computing agents. A set of
partition t into a new internal database. function symbols in the language, referred to by AF, is the
database: D x T → D set modeled in the computing world by AI Agents with
across and/or over board capability. Thus the language
A knowledge-level agent is an environment is an 8-tuple defined by the signature has designated function symbols
shown below. The set D in the tuple is an arbitrary set of called AF. The AF function symbols define signatures
predicate calculus databases, S is a set of external states, T which have specific message paths defined for carrying
is the set of partitions of S, A is a set of actions, see is a context around an otherwise context free abstract syntax. A
function from S into T, do is a function from A S into S, set of function symbols in the language, referred to by AF,
database is a function from D x T into D, and action is a are agents with nontrivial capability.
function from D x T into A.
Knowledge level agents Definition 1.2 We say that a signature is intelligent iff it
are hysteretic agents. has intelligent function symbols. We say that a language
has intelligent syntax if the syntax is defined on an
intelligent signature.
1.2AgentMorphisms Definition 1.3 A language L is said to be an intelligent
Let HA be a set of sextuples defining hysteretic agents. language iff L is defined from an intelligent syntax.
Define HA morphims by a family of functions defined The above mathematical basis might be applied to the
component-wise on the sextuple above. KQML agent language paradigms. However the author has
Definition 1.1 A HA morphism is a function F : HA → not had the time an occasion to explore the applications.
HA’ defined component-wise by F[i]: I→ I’; F[S]: S → The example of intelligent languages [3] we could present
S’, F[T]: T →T’, F[A]: A →A’; F[s]: S→ T’; F[d]: A’ x S’ are composed from triples as control structures.
→ S’ and F[internal]: I’ x T’→ I’. The A's have operations that also consist of agent message
passing. The functions in AF are the agent functions
Definition 1.1 implies F defines new hysteretic agents
capable of message passing. The O refers to the set of
from HA by a morphism. The definition might become
objects and R the relations defining the effect of A's on
further transparent in view of definitions. Component-wise
objects. Amongst the functions in AF only some interact
definitions for morphism might be viewed as functions on
by message passing. There is a new frontier for theoretical
a multi-sorted signature carrying the sextuple. Similar
development of the algebras and that of the AII
morphisms can be defined for knowledge level agents.
foundations. is a pair of algebras,
1.3 Agents, Languages, and Models (see section 3), connected by message
By an intelligent language we intend a language with passing and AII defines techniques for implementing such
syntactic constructs that allow function symbols and systems. To define AII we define homorphisms on
corresponding objects, such that the function symbols are intelligent signature algebras.
implemented by computing agents in the sense defined by Definition 1.4 An I-homorphism is a homoprphism
this author in (Nourani 1993c, 96a). Sentential logic is the defined on algebras with intelligent signature I.
standard formal language applied when defining basic
models. The language R is a set of sentence symbol closed
To define agent specific designs we apply HA-morphisms
via the following definition.
by finite application of negation and conjunction to
sentence symbols. Once quantifier logical symbols are Definition 1.5 Let A and B be I-algebras with signatures
added to the language, the language of first order logic can containing an agent signature HA. A HA-homomorphism
bedefined. from A to B is an I-homomorphism with defined HA-
A Model G for is a structure with a set A. There are
morphism properties.
structures defined for R such that for each constant symbol 2. Agent Ontology Preservation Theorems
in the language there corresponds a constant in A. For each
Let us apply the definition for HA agents and HA
function symbol in the language there is a function defined
morphisms to state a preservation theorem. Let A and B be
on A; and for each relation symbol in the language there is
I-algebras with the signature I containing HA agents. Let
a relation defined on A. For the algebraic theories we are
Alg[B] be an I-algebra defined from B implementing [1] a
defining for intelligent tree computing in the forthcoming
specified functionality defined by A. An AII is an
sections the language is defined from signatures as in the
implementation for Alg[A] by Alg[B].
logical language is the language of many-sorted equational
logic. Definition 2.6 Let A and B be I-algebras with intelligent
signature I containing agents. An I-ontology is an I-algebra
The signature defines the language by specifying the
with axioms for the agents and functions on the signature.
function symbols' arities. The model is a structure defined
on a many-sorted algebra consisting of S-indexed sets for S
a set of sorts. By an intelligent language we intend a Theorem 2.1 Let A and B be I-algebras with the signature
language with syntactic constructs that allow function I containing HA agents. The AII with HA morphisms
symbols and corresponding objects, such that the function
defined from A to B preserve I-ontology algebras iff FL.
defined by HA-homorphisms. [2]G.Wiederhold: "Interoperation, Mediation and
Theorem 2.2 Let A and B be I-algebras with the signature Ontologies''; Proc.Int.Symp. on Fifth Generation Comp
I containing KL agents. The AII with KL morphisms Systems, ICOT, Tokyo, Japan, Vol.W3, Dec.1994, pages
preserve I-ontology algebras iff defined by KL- 33-48.
homomorphisms. [3] Nourani, C.F.," Intelligent Languages- A Preliminary
Syntactic Theory," May 15, 1995, Mathematical
DKB mappings are specific AII's were the ontology Foundations of Computer Science;1998, 23rd International
algebra operations are the same at source and target. the Symposium, MFCS'98, Brno, Czech Republic, August
DKB mappings are proved AIIOPP consistent. 1998, Jozef Gruska, and Jiri Zlatuskanbsp;(Eds.): Lecture
3. Meditors and Ontologies Notes in Computer Science;1450, Springer, 1998, ISBN 3-
540-64827-5, 846 pages.
A mediator is a software module that exploits encoded [4] Nourani, C.F.” AII and Heterogenous Software Design,
knowledge about certain sets or subsets of data to create May 10, 1995, MAAMAW'97, Eighth European Workshop
information for a higher layer of applications. and the on MODELLING AUTONOMOUS AGENTS IN A
definition goes on to state `It should be small and simple, MULTI-AGENT WORLD May 1997, University of
so that it can be maintained by one expert or, at most, a Karlskrona/Ronneby, Dept of Computer Science and
small and coherent group of experts' Mediator instantiation Business Administration Ronneby, SWEDEN. Specific
is to populate a domain-independent service or tool with track on WWW announced papers only.
domain-specific knowledge. We define Mediator [5] Genesereth, M.R. and Nilsson, N.J., Logical
Specifications with agent ontologies consisting of a tuple Foundations of Artificial Intelligence,"Morgan-
engine agent-based computing system := Kaufmann,1987.
, consisting of Design_Agents [6] Nourani, C.F.," Slalom Tree Computing," 1994, AI
:= and CoAgents;= . The design is Communications, December 1996, IOS Press, Amsterdam.
depicted by the following figure. RNA are normal actions [7]ADJ- Goguen, J.A., J.W. Thatcher, and E.G. Wagner
and RFA the faults, exceptions, and remedial functions. “An Initial Algebra Approach To Specification,
Correctness and Implementation of Abstract Data Types, in
Current Trends in Programming Methodology, Vol IV,
1978, R. Yeh, editor, Prentice-Hall, Englewood-Cliffs, NJ,
pages80-149.
AffilationsAcademia USA last appointment UCSB
ScientificURL
http://members.fortunecity.com/crisfn/metaai.html.doc
Fax 415-430-2167 x1342
Telephone 310-754-6000 x3036
.
Figure 1-The pairs are modules composed to
define .
The modules are defined from multiple objects. Actions
could be in form of operations or message communication
from one object to another. A set of computing agents
forms Design_Agents and a dual set forms CoAgents.
CoAgents are agents running parallel checking faults and
unplanned events presenting alternatives. The algebras
Alg[A] and Alg[F] define wrappers for the mediators as
functions for interacting with resources. A wrapper is a
tool to access known resources and translate their objects.
The Design_Agents corresponds to an algebra Alg[A] of
Normal Activities and CoAgents to an algebra Alg[F] for
unplanned events, computing faults recovery.
References
[1] Abstract Implementation Techniques for A.I. By
Computing Agents,: A Conceptual Overview, Technical
Report, March 3, 1993, Proc. SERF-93, Orlando, Florida,
November 1993. Published by the University of West
Florida Software Engineering Research Forum, Melbourne,