400

ICALT-200430 Aug-1 Sep, 2004, Joensuu, Finland: 

ICALT-2004 30 Aug-1 Sep, 2004, Joensuu, Finland I. Hatzilygeroudis, Ch. Giannoulis, C. Koutsojannis A Web-Based Education System for Predicate Logic University of Patras, Dept of Computer Engin. andamp; Informatics andamp; Research Academic Computer Technology Institute (RACTI)

Outline: 

Outline Motivation Our Proposal System Architecture Hypermedia Application Adaptation Facilities Implementation Issues System Evaluation Conclusions Future Work

Motivation: 

Motivation Most existing logic education systems teach logic as a tool to construct correct formal proofs. Most are restricted to propositional logic. They use isolated open interaction facilities. They do not offer adaptation capabilities. Students have problems with specific aspects of using logic (FOPC) as a KRandamp;R language.

Our Proposal: 

Our Proposal Create an education system for teaching first-order logic (FOPC) as a KRandamp;R language. Use open interaction facilities incorporated into the educational framework of the system. Provide adaptation capabilities.

KR&R with FOPC: 

KRandamp;R with FOPC FOPC Formulas Clausal Form Conversion of NL sentences to FOPC (ad hoc process) Conversion of FOPC to Clausal Form (CF) (systematic process) Inference process E.g. 'Dogs are pets' 'Every dog is a pet' x dog(x) =andgt; pet(x) (NL  FOPC) x dog(x) =andgt; pet(x) (dog(x), pet(x)) (FOPC  CL)

System Architecture: 

System Architecture

Hypermedia Application: 

Hypermedia Application Navigation area Content area Info Bar Tool Bar Next Previous Start Back User Interface

Hypermedia Application: 

Hypermedia Application Topics are organized in a tree More complex topics (higher up) Less complex topics (lower down) Simple topics (leaves)  topic pages Topic pages include concepts (ordered list) Concepts  concept pages Content Structure

Hypermedia Application: 

Hypermedia Application Content Structure FOPC Introduction Syntax Vocabulary Formulas NL to FOPC Conversion Clause Form simple topics complex topics

Adaptation Facilities: 

Adaptation Facilities The student can disregard the order of the concepts in a topic page. The student decides on which prerequisite concepts are necessary in a concept page. Concepts Order

Adaptation Facilities: 

Adaptation Facilities Concepts Order

Adaptation Facilities: 

Adaptation Facilities It concerns the NL to FOPC and the FOPC to CF conversion topics. The student selects the difficulty level and specifies the complexity level of exercises. Explanations for the wrong answers are provided. Self-Evaluation Facility

Adaptation Facilities: 

Adaptation Facilities Self-Evaluation Facility

Adaptation Facilities: 

Adaptation Facilities It concerns the FOPC to CF conversion topic. The student can try any formula. It is achieved by calling LISP code. Open Exercising Facility

Adaptation Facilities: 

Adaptation Facilities Open Exercising Facility FOPC formula Next step description Result of executed step

Implementation Issues: 

Implementation Issues It’s a distributed architecture system (HA and DB execute their functions on the server; LA on the client). HA is implemented in ASP. DB is implemented in MS Access. LA is part of a LISP-based theorem prover and is called as a Java applet.

System Evaluation: 

System Evaluation On an AI class of 30 senior (4th year) computer engineering students The students were asked to use the system in a specific way for NL to FOPC and FOPC to CL conversions Use of a questionnaire of 10 questions of various types (open, closed, combined)

System Evaluation: 

System Evaluation

System Evaluation: 

System Evaluation FOPC-to-CF Feb-2003: (x) ((a(x)  b(x)) =andgt; (y) c(x, y)) Feb-2004: (x) ((a(x)  b(x)) =andgt; (c(x)  (y) d(x, y)))

Conclusions: 

Conclusions AI education can be improved by using web-based education systems. Adaptability plays a positive role in the learning process. Incorporation of open subsystems in a closed web-based education system facilitates learning too.

Future Work: 

Future Work Add 'intelligence' to the system to make it adaptive. How much is 'intelligence' going to improve the system? Where is 'intelligence' necessary?