Proceedings of the 8th WSEAS International Conference on APPLIED COMPUTER SCIENCE (ACS'08)
AESGT Extension to support Explanation Facilities That
Answers "How", and "Why Asking" Types of Questions
Ahmed Fouad Said Ahmed Rafea
Central Lab for Agricultural Expert Systems Department of Computer Science
Agricultural Research Center The American University in Cairo
Ministry of Agriculture and Land 113 Kasr Al Aini Street,
Reclamation. P.O.Box 2511,
El-Nour St., P.O. Box 438, Dokki Cairo, Egypt
Giza, Egypt
Hesham Hassan Samhaa R. El-Beltagy
Faculty of Computers and Information Faculty of Computers and Information
Cairo University Cairo University
5 Tharwat Street, Orman, 5 Tharwat Street, Orman,
Giza 12613, Egypt Giza 12613, Egypt
Abstract: - Plant protection is one of the most important applications in agricultural domain.
An agricultural Expert system Generic Tool (AESGT) [1] has been developed for rapid
generation of plant protection expert system. AESGT provides generic knowledge modeling
approaches namely: CommonKADS[2] and Generic Tasks[3], but This tool doesn't have an
explanation facility, which is an important component in any expert system (ES), as is used to
clarify the reasoning process to the users and hence improve the understandability and
acceptability of the expert system. This paper describes our experience in augmenting the
AESGT tool with "How" and "Why Asking" explanation components for different kind of
knowledge representation schemes that the AESGT tool is using. The developed explanation
components can deal with different knowledge representation schemes that are commonly
used by the modeling approaches mentioned above (e.g., tables, structured tables, hierarchy,
constraints and rules), and handle the task level explanation. As well, our explanation
components have been successfully implemented and used to augment a real agricultural
expert system with different kind of explanations.
Keywords: Expert System, Explanation, Knowledge Engineering, Plant Protection.
.
ISSN: 1790-5109 436 ISBN: 978-960-474-028-4
, Proceedings of the 8th WSEAS International Conference on APPLIED COMPUTER SCIENCE (ACS'08)
1 Introduction frame as a knowledge representation where
Providing explanation in ESs applications is an additional slots are added to the agent definitions to
important issue in man-machine interface. Human hold text strings for describing the agents' goals.
seems to have a natural instinct for wanting to MPA puts the "canned" text together to generate
understand and make sense of their environment and explanation taking into consideration the roles of the
things in it. As well, explanation is used to clarify various agents specialists, plans, steps, .. etc.
the reasoning process to users to improve the
understandability and with the help of good
explanation user can know what the ES does, how 3 AESGT Architecture with
the ES works, why the ES do these specially when Explanation Facility Added
this application is used as a high level advisors to This section overviews the overall structure of the
professionals who must retain responsibility for the agricultural Expert system Generic Tool (AESGT)
decisions which are made. Therefore, recognition of after adding the explanation components. As shown
the importance of explanation components for in figure (1) describes this overall structure. The tool
reasoning systems has existed in number of fields consists of the following components: Problem
for many years, for example NEOMYCIN [4], Solving Knowledge, Problem Solver, Working
XPLAIN [5], MPA [6], and WEBEXPLAIN [7] Memory (WM), Explanation facility, Ontology and
Most of the work in this area either completely User Interface.
ignored the problem solver method (PSM) used in
developing ES, or focused on a single PSM. These
limitations motivated us to propose an explanation
framework aim to deal with different PSM that are
used in the most popular ES methodologies like
CommonKADS , and Generic Tasks.
This paper is organized as follows. Section 2:
Related Work. Section 3: AESGT Architecture with
Explanation Facility Added. Section: 4 Knowledge
Representation and Reasoning. Section: 5
Explanation Facilities. Section 6: case study. Section
7: conclusion
Fig.1: Overall Structure of the AESGT tool with
2 Related Work Explanation Facility
Explanation facilities have generally been added to
knowledge-based system to explain the inference This tool supports four types of knowledge
that occurs as the system reasons. NEOMYCIN [4] representations: rules, structured tables, frames
follows the trace of rules to answer the "Why" and hierarchy, and agent hierarchy. XML[10] is used as
"How" questions. NEOMYCIN uses templates and
• Problem Solver: Two PMSs based on two
a representation language.
canned text as a method for explanation generation.
XPLAIN [5] is a digitalis therapy advisor which development methodologies are supported by
advises physicians on the level of the digitalis drug this tool. These PSMs are:
administrated to congestive heart failure patients. It 1. Generate and Confirm hypotheses PSM:
organizes its knowledge as semantic network. It has which is on CommonKADs
an explanation generator to answer "Why", and methodology [9].
"How" questions, based on automatic tracing of goal 2. Routine Design PSM: which is based
hierarchy that is generated during the reasoning
• Working Memory: contains the user input, and
on Generic Task methodology [3].
process. A similar approach has been used in
WEBEXPLAIN [7] to provide a "How" question the derived results. The PSM used to derive
facility in a web-based infrastructure using an these results are associated with each piece of
extended UPML framework [8].
• Explanation Facility: responsible of replying
these results.
A mission planning assistant (MPA) [6] is a
Generic Task program capable of explaining its
explanation coming from problem solving
problem-solving steps and its strategy. MPA was
method which is asked by the end user.
developed based on Routine Design. MPA uses
ISSN: 1790-5109 437 ISBN: 978-960-474-028-4
