CASE REPRESENTATION AND ADAPTATION IN SmartLP

A Web-based Lesson Planning System

Aslina Saad

1,2

, P. W. H. Chung

and C. W. Dawson

Department of Computer Science, Loughborough University, Leicestershire, U.K.

Faculty of Information and Communication Technology, Universiti Pendidikan Sultan Idris, Tanjung Malim, Malaysia

Keywords: Knowledge modelling, Case representation, Case adaptation.

Abstract: Lesson plans help teachers to organize content, materials and methods for their teaching. Appropriate lesson

plans are crucial to accommodate student differences in various aspects. Currently there are limited

mechanisms to support decision making in constructing lesson plans based on the constraints teachers have.

Since lesson plans have a standard format, they can potentially be shared. SmartLP, a web-based lesson

planning system, was developed to assist teachers in preparing suitable lesson plans based on various

constraints; students’ profile, curriculum and facilities. In SmartLP, teachers can make modification to the

retrieved plans according to their constraints, as opposed to generating new ones from scratch.

Implementation of such systems insists on a proper case representation as it facilitates case retrieval and

subsequently case adaptation to handle differences in hand. An ontology for the lesson plan domain has

been built in the form of a taxonomy. This is followed by case definition that consists of problem

description and solution. Cases are represented as attributes - value representation in a case base.

Transformation, a kind of case adaptation, is implemented in the system to facilitate teachers in adding,

deleting or editing the contents of the retrieved lesson plans. The adaptation can be derived from one case or

several cases.

1 INTRODUCTION

Lesson plans are written documents produced by

teachers using a standard format and based on the

same curriculum.

Such plans help teachers to

organize content, materials and methods for their

teaching and these items need to be prepared to meet

the diverse constraints and factors each teacher has.

The constraints might be different from one teacher

to another depending on numerous factors such as

experience, students’ ability, facilities available and

many more.

Currently there are few mechanisms to support

decision making as well as determining suitable

elements in a lesson plan based on constraints

teachers have.

These limitations could be improved through the

implementation of a web-based lesson planning

system whereby best practices in preparing lesson

plans can be shared among teachers. The sharing of

experiences might be useful for teachers to create

new plans or to make modification and improvement

to existing plans according to their own constraints

and students’ profile. It is often more efficient to

customise existing lesson plans as opposed to

generating new ones from scratch.

However, to simply use other teachers’ lesson

plans is often not practicable because of the various

factors and constraints that need to be considered.

Therefore, it is advisable to make some adaptations

to the solution given by the system. Solving a

problem in this system involves obtaining a problem

description, measuring the similarity of the current

problem to previous problems stored in a database,

retrieving one or more similar cases and attempting

to reuse the solution of one of the retrieved cases,

possibly after adapting it to account for differences

in problem descriptions. This process is similar to

Case-based reasoning (CBR) which offers a

potential solution to lesson plan construction by

retrieving relevant cases that solved similar

problems. Teachers can reuse the retrieved lesson

plans after customising the lesson plans according to

their constraints. The adaptation process of the

previous solutions in CBR will fit the current

problem context which subsequently brings in new

440

Saad A., Chung P. and Dawson C..

CASE REPRESENTATION AND ADAPTATION IN SMARTLP - A Web-based Lesson Planning System.

DOI: 10.5220/0003190404400445

In Proceedings of the 3rd International Conference on Agents and Artiﬁcial Intelligence (ICAART-2011), pages 440-445

ISBN: 978-989-8425-40-9

 2011 SCITEPRESS (Science and Technology Publications, Lda.)

solution to the problem.

While the implementation techniques may vary,

most CBR systems include the following five steps

in some form or other (Raman, 1995; Watson and

Marir, 1994):

• representation where problem storage is handled;

• retrieval where the closest-matching precedent is

identified;

• adaptation where a solution is generated from the

retrieved problem;

• validation where the accuracy of the solution is

verified; and finally;

• update, where the database is modified or

updated with the information gained from this

problem solving process.

According to Craw et al. (2006) in design tasks, it is

common for the retrieved solution, to be regarded as

an initial solution that should be refined to reflect the

differences between the new and retrieved problems.

This adaptation is done in SmartLP via a

customisation function. Here, users can edit, add or

delete elements in the retrieved lesson plans.

Hence, to develop a comprehensive web-based

lesson planning system (SmartLP) a good

knowledge representation is crucial as it contributes

to case representation and facilitates case retrieval

and subsequently case adaptation.

Although much previous research indicates the

role played by knowledge representation, very little

research has focussed on knowledge representation

in the educational area, particularly in lesson

planning.

This paper will discuss knowledge modelling,

case representation and case adaptation that was

implemented in the SmartLP system. The modelled

knowledge is presented as cases that are stored in a

case base for retrieval. Adaptation to the retrieved

cases can be executed by a customisation function in

the system, followed by a verification process.

2 KNOWLEDGE

REPRESENTATION

Sun et al. (2003) reported that the organization of

elements in knowledge representation must facilitate

the retrieval of useful information from the case

base. Two criteria that can facilitate retrieval are

problem classification and the expected target. This

is aligning with case representation that consists of

problem descriptions and solution. In addition,

Urosevic et al. (2006) point out that one problem in

knowledge representation is how to store and

manipulate knowledge in an information system in a

formal way, so that it may be used by a mechanism

to accomplish a given task.

Ontology is a formal representation of the

knowledge by a set of concepts within a domain and

the relationships between those concepts. Ontologies

are specific, high-level models of knowledge

underlying all objects, concepts, and phenomena in a

domain. Generally, an ontology is a metamodel

describing how to build models. The good thing

about using such metamodelling is that we never

sacrifice the usefulness of any specific model.

Ontologies do the same for knowledge models

(Sormo et al., 2007).

According to Mizoguchi (2004), ontology

provides us with a guideline for modelling the

world. To do this, it consists of carefully chosen top-

level categories which are reliable enough to explain

lower concepts.

An ontology of the lesson plan domain was built

in the form of a taxonomy. In SmartLP, a taxonomy

of the general lesson plan domain was produced

based on a semantic net that was constructed first, to

see how all elements and concepts in a lesson plan

relate to each other. This type of representation was

chosen due to its acceptability as a standard

modelling mechanism. The structure of a semantic

net is shown graphically in terms of nodes and the

arcs connecting them. Nodes are often referred to as

objects and the arcs as links or edges. Two types of

commonly used links are IS-A and A-KIND-OF

(AKO). The semantic net is an example of a shallow

knowledge structure because all the knowledge is

contained in the links and nodes.

Matching ontologies from their relational (or

external) structure is very powerful because it allows

all the relationships between entities to be taken into

account. This must be grounded on other tangible

properties, which is why it is often used in

combination with internal structural methods and

terminological methods (Euzenat and Shvaiko,

2007). In addition they claim that the most

commonly used structure is the taxonomy. It is the

backbone of ontologies and has received a lot of

attention from designers.

Ontologies are now central to many applications

such as scientific knowledge portals, information

management and integration systems, electronic

commerce, and semantic web services. Noy and

McGuinness (2000) in (Abdollahi, 2007) state that

an ontology is needed to:

CASE REPRESENTATION AND ADAPTATION IN SmartLP - A Web-based Lesson Planning System

441

• share common understanding of the structure of

information;

• reuse domain knowledge;

• make domain assumptions explicit;

• analyse domain knowledge

A taxonomy consists of carefully chosen top-level

categories which are reliable enough to explain

lower concepts. A taxonomy for lesson planning

domain has been built and shown in Figure 1.

Figure 1: Lesson Plan taxonomy.

Lesson plans consist of four main nodes which

are curriculum, students, facilities and content. Each

node is then divided into detailed nodes. The

ontology introduced above is mapped to a case of

SmartLP system as can be viewed in Table 1.

3 CASE REPRESENTATION

Representation is the issue of deciding what to store

and how the memory should be organized in order to

retrieve and reuse old plans effectively and

efficiently. Cases can be represented using a variety

of notations. In SmartLP, the combination of

hierarchical and attribute – value representation is

used. According to Liqing and Kumar (2005), case

representation is generally regarded as one of the

most important issues and is crucial to the success of

case-based reasoning systems.

This is supported by Spalzzi (2001) who insisted

that the efficiency and effectiveness of a case-based

planner heavily depends on its plan representation

and memory organization. This is a natural

consequence of the fact that its problem solver is

primarily based on retrieving and adapting previous

plans. Bergmann et al. (2005) suggest that object

oriented case representation has an expressiveness

similar to frame representations, but have a different

origin. They make use of the data modelling

approach of the object-oriented paradigm, including

is-a and part-of relationships as well as the

inheritance principle. Cases are represented as

collections of objects, each of which is described by

a set of attribute-value pairs. The structure of an

object is described by an object class. They suggest

that object-oriented representations are particularly

suitable for complex domains in which cases with

different structures occur. It seems similar to what

has been discussed by Giarratano (1998) as object-

attribute-value triples (OAV) or triplet.

It is convenient to list knowledge in the form of a

table, and thus translate the table into computer code

by rule induction. If inheritance is not required and

only a single object is to be represented, attribute –

value pairs (AV) may suffice. Many types of real

world knowledge cannot be represented by the

simple structure of a semantic net (Giarratano and

Riley, 1998:66).

According to Abdollahi (2007), the first step in

building a CBR model is the “Representation of

Cases” as well as knowledge. This means how to

define and describe the cases in the model in order

to recall and reuse them for reasoning. He

highlighted four main challenges for case

representation as the following:

• Case searching and matching;

• Integrating new cases into the existing memory

(model);

• Qualitatively and quantitatively data types to

store in cases;

• Organizing and indexing cases for effective

retrieval and reuse.

Components of a case in CBR consist of problem

description and solution. Problems in a lesson plan

context are the various constraints that teachers face

in constructing lesson plans. This is shown in Table

ICAART 2011 - 3rd International Conference on Agents and Artificial Intelligence

442

Table 1: A case of SmartLP.

Problem Node Elements

Students Ability, knowledge, motivation, No of

student per class

Facilities Resources, material, venue

Curriculum Year, subject, learning area, topic,

learning objectives, learning outcome,

skills, suggested activities.

Solution Lesson Plans Appropriate teaching aid, skills,

learning outcome, short description,

introduction, explanation, activity,

timing of each activity, enrichment,

assessment, extension, closure.

In SmartLP, case searching can be done via five

types of search; basic search, weighted search,

terms, expansion search and browsing. As justified

by other researchers, matching and ranking are

applied in these types of search in order to select the

most appropriate lesson plans to the constraints users

have.

Reyes and Sison (2002) state that matching and

ranking is a procedure in case retrieval that selects

which cases are appropriate among the cases in the

case library. As the process of searching the library

is done, the search process asks the matching

function to compute the degree of match among

indexes. Based on the result of the matches, the

search function collects a set of cases that partially

match the new situation. The matching cases are

then ranked to identify which best address the

requirements of the new situation.

The hybrid approach, which combines

computational and representational approach, was

used for the case retrieval and matching process in

the system. Hierarchical representation together with

linear representation, based upon measures of

similarity, was used together with a computational

approach, in terms of weighting. In addition, query

expansion and query weighting are used in this

system to give flexibility for users and to produce a

better search result. Query expansion gives

flexibility for users to choose related terms to the

searched keywords by expanding the query using

words or phrases with a similar name. Searched

keywords may have different importance for

different users. Therefore, query weighting facilitate

users to indicate the importance of their searched

keywords. The weights are taken into account in

calculating the similarity of the searched keywords

and attributes in cases in the case base.

SmartLP used attribute-value representations for

its case which is a lesson plan itself. This is shown

in Table 2.

Table 2: Table lesson.

Attributes Types

LessonID Auto increment

ParentID Int

Date Varchar

Form Int

Subject Varchar

Learning area Varchar

Topic Text

Learning outcome Text

Objectives Text

Ability Varchar

No of Students Int

Minutes Int

Skills Mediumtext

Resources Varchar

Value Varchar

Prerequisite Mediumtext

Introduction Text

Step1 Longtext

Step2 Longtext

Step3 Longtext

Step4 Longtext

Step5 Longtext

Assessment Longtext

Extension Longtext

Closure Text

Reflection Longtext

Verified Varchar

Indexing is applied in the case base to allow the

database server to look up rows more quickly, thus

speed up the retrieval. Several attributes which are

used for indexing the cases, are year, subject,

learning area, topic, learning outcomes, skills,

values, time period, no of student, and ability. Each

of these attributes has their similarity value in

comparison to the searched keywords. Some are

using hierarchical similarity measure and some use

linear similarity measure. The structure of the

similarity table is shown in Table 3.

Table 3: Attribute- value representation for similarity

table.

Field Type

Id Int

Query Varchar

Case Varchar

similarity Float

4 CASE ADAPTATION

The adaptation process is crucial in SmartLP as it is

CASE REPRESENTATION AND ADAPTATION IN SmartLP - A Web-based Lesson Planning System

443

the process whereby users can change the elements

of the retrieved lesson plans to tailor to their own

constraints. As discussed before, users might have

different constraints in constructing lesson plans in

students’ profile such as ability, previous knowledge

and many more. This is achieved by a customisation

function in the system.

Hanney et al. (1995) review a large number of

CBR systems to determine when and what sort of

adaptation is currently used. Their initial taxonomies

show that CBR systems using adaptation are

predominantly used when prediction and design are

required.

Kolodner (1993) cited by Craw et al. (2006)

identify three types of adaptation:

• Substitution - replaces values in the retrieved

solution with new values appropriate for the new

problem (e.g. changing a house price);

• Transformation - alters the retrieved solution by

adding, deleting or replacing parts of the retrieved

solution to suit the new problem (e.g. altering steps

in a plan);

• Special methods apply specialised heuristic

knowledge to repair the retrieved solution, or replay

the method used to derive the retrieved solution for

the new problem.

For adaptation, the task is to recognise when an

adaptation should be applied because the new and

retrieved problems are sufficiently different in some

relevant way, and to perform some changes to the

retrieved solution. An adaptation can be considered

as a situation (problem description)/action (solution)

pair. The situation contains the differences between

the new and retrieved problems. In SmartLP,

transformation was used. The retrieved cases can be

modified by users to suit their constraints in hand.

Although the adaptation process is done manually,

the system makes the process easier via the smart

interfaces it offers.

This adaptation can be made based on one case

or several cases. If just one case is selected to be

modified, users just need to view the details and

click a customise button. All fields will become

editable. Users can modify the elements in this

lesson plan and this plan will be saved as new lesson

plans. The author of this customised lesson plan is

identified by user session.

A new lesson plan can be generated from several

customised cases. Here, two or more lesson plans

can be chosen to be compared. Elements from these

different lesson plans can be chose to be included in

the customised plan. The selected lesson plans will

be compared in a table as shown in Figure 2.

Figure 2: Selected lesson plans to be compared.

Here users can select whatever fields they want

to have in their customised lesson plans. The

selected value will be combined into their particular

elements and can be edited by the user. If users

prefer most elements in a particular lesson plan they

can check a select all button at the bottom of that

lesson plan.

Here, all fields are editable and attachment files

can also be added or deleted. Users can modify the

elements in this lesson plan and they will be saved

as a new generated lesson plan. The author of this

customised lesson plan is identified by user session.

Figure 3: The generated new lesson plan.

5 CONCLUSIONS

The implementation of SmartLP system based on

CBR should solve teachers’ problems in deciding

appropriate elements in lesson plan construction by

customising their own lesson plans. This can be

done by retrieving previous lesson plans, reusing

ICAART 2011 - 3rd International Conference on Agents and Artificial Intelligence

444

and revising those lesson plans and subsequently

retaining them in the same system.

The knowledge modelling discussed in this paper

transforms detailed requirements into complete,

detailed case representations that will be used in the

retrieval phase. While the system-based view is

concerned with efficient search techniques to match

query and document representations, the user-based

view must account for the cognitive state of the

searcher and the problem solving context. These two

views were taken into account during the retrieval

process. The flexible adaptation process based on

one or several cases helps teachers to generate their

own new lesson plans. These new lesson plans will

be verified and subsequently retrieved by other

users. By having this dynamic process, the system

will expand dynamically.

REFERENCES

Abdollahi, J., 2007. Analysing Complex Oil Well

Problems through Case-Based Reasoning, Norwegian

University.

Bergmann, R. et al., 2005. Representation in case-based

reasoning. 00:0(1–4). Cambridge University Press.

Craw, S. et al., 2006. Learning Adaptation Knowledge to

Improve Case-Based Reasoning.17 (16, 17)

https://openair.rgu.ac.uk/bitstream/10059/57/1/aij06-

smc.pdf

Hanney, K. et al., 1995. What Kind of Adaptation do CBR

Systems Need?: A review of Current Practice,AAAI

Technical Report FS-95-04.

Liqing, F., Kumar, 2000 A. S., XML-based Representation

in a CBR System for Fixture Design. 12 (1-4), pp.

339-348.

Mizoguchi, R., 2004. Tutorial on Ontological Engineering:

Part 3: Advanced Course of Ontological Engineering.

22(2), pp. 198 - 220.

Raman, P., 1995. Case-Based Reasoning: An Implemented

Methodology. Technical Report: CSE95-04.

Reyes.L. and Sison, R. C. (2002). Case Retrieval in CBR-

Tutor, Proceedings of the International Conference on

Computers in Education (ICCE’02).

Sun, B. E. A., 2003. Scenario Based Knowledge

Representation In Case Based Reasoning Systems.

20(2), pp. 92-99.

Urosevic et al., 2006. Knowledge Representation and

Case-Based Reasoning in a Knowledge Management

System for Ambient Intelligence Products, 2006,

Proceedings of the 24th IASTED international

conference on Artificial intelligence and applications ,

pp329 - 334.

Watson, I., Marir, F., 1994. Case-Based Reasoning a

Review, The Knowledge Engineering Review, Vol.9

No.4.

CASE REPRESENTATION AND ADAPTATION IN SmartLP - A Web-based Lesson Planning System

445