Personalized Students ’ Profile Based On Ontology and Rule-based Reasoning

Nowadays, most of the existing e-learning architecture provides the same content to all learners due to ”one size fits for all” concept. E-learning refers to the utilization of electronic innovations to convey and encourage training anytime and anywhere. There is a need to create a personalized environment that involves collecting a range of information about each learner. Questionnaires are one way of gathering information on learning style, but there are some problems with their usage, such as reluctance to answer questions as well as guesses the answer being time consuming. Ontology-based semantic retrieval is a hotspot of current research, because ontologies play a paramount part in the development of knowledge. In this paper, a novel way to build an adaptive student profile by analysis of learning patterns through a learning management system, according to the Felder-Silverman learning style model (FSLSM) and Myers-Briggs Type Indicator (MBTI) theory is proposed.


Introduction
In an educational environment, learners with diverse learning capacities and foundation information require particular learning ways [1].The main characteristic of e-learning systems is their ability to recognize students' needs, their educational behavior and also, their capabilities.Learning can be characterized as the procedure of obtaining knowledge or aptitudes [2].It includes three key structures of cooperation: • learner-learner.
In an educational environment, learners with diverse learning capacities and foundation information require particular learning ways [1].The main characteristic of e-learning systems is their ability to recognize students' needs, their educational behavior and also, their capabilities.Learning can be characterized as the procedure of obtaining knowledge or aptitudes [2].It includes three key structures of cooperation: • Questionnaire; • Behavior learning pattern.* Corresponding author.Email: leandros.maglaras@dmu.ac.ukTraditionally, learning styles have mainly been assessed using surveys and questionnaires; asking students to self-evaluate their own behaviors.This is suitable in the traditional way of learning, where it is difficult to observe and analyses understudies' preferences over the entire learning process.However, as with every qualitative survey, this type of assessment endures numerous downsides.Firstly, it can be biased as it relies on upon understudy judgment.Secondly, it is performed only at a single point in time, while learning styles, according to several theories, can change over time.Some of these surveys can reach over 40-questions long, such as Vermunt [4] and Felder-Silverman's and hence, students' dispositions are not easy to keep updated.The uncertainty in the majority of the data gathered from a questionnaire.It has straightforward influences in the quality of learning personalization.
Versatile e-learning frameworks that depend on learning styles by and large utilize distinctive learning style models.This raises the issue of what models and hypotheses are suitable and effective.Likewise, there is an absence of amazing observational assessment with respect to their viability [5] and a scarcity of similar work in connection to these frameworks [6].
Adaptive e-learning systems that are based on learning styles generally by utilize different learning style models.In order to select which model and theories are more suitable in build adaptive learning environment .In addition, there is a lack of high quality empirical evaluation regarding their effectiveness [5] and a paucity of comparative work in relation to these systems [6].Most of the existing learning management system focuses on adaptively in general, whereas others focus more specifically on adaptively based on learning style [7].The objective of this paper is to build an adaptive student profile by analyzing the users behaviour through a learning management system and by matching the learners learning style with their personality with the use of ontologies and rule-based techniques (inference engine).An initial concept of our adaptive learning management system was presented in [8].
We have organized the rest of this paper in the following way; Section 2, introduces motivation and contributions.Section 3, discuss the background which include different learning style models as well as semantic web.Section 4 presents current e-learning systems.Section 5 illustrates proposed Adaptation Process Flowchart.Section 6 presents our proposed adaptive student profile model, in addition to ontological representation of the adaptive model.Finally, Section 7 concludes the article.

Motivation and contributions
This article presents an adaptive student profile based on ontology and inference rules (rules-based ) to match their learning style according to FSLSM and MBTI models.Our research differs from these previous works in relation to several aspects: • We provide personalized student profile based on learner's behavior pattern using two different models namely FSLSM and MBTI.
• We support adaptive learning using different types of techniques such as ontologies and inference rules (rule-based).
• The proposed work is not only intended to ensure the learner's ability to learn, but it is also expected to be useful in providing a learning path and guidance based on individual differences (learning style and personality).
• Personalized guidance is achieved by collecting a student's initial capabilities and preferences and by using semantic rules and rule-based reasoning in order to detect learner behavioural changes.That way the system can determine which learning style is more suitable for the user.
This proposed model addresses the limitations of existing adaptive e-learning models, the principal ones being as follows.
• Most of the existing models assume that the teacher and learner meet frequently during the learning process and that the learning style of the learner is obvious to the teacher.
• The existing models need the complete dataset of the learner's behavior.Since in real environments most of the times incomplete or vague information exist, it is necessary to be able make effective conclusions from incomplete data in order to identify an individual's learning style.

Learning style models
Kolb's Experiential Learning Theory [9] identifies four learning styles, namely, diverging, assimilating, converging and accommodating.They can be tested using the metrics watching, thinking, feeling and doing.Under these lens, experience is very important to the learning process.Whilst Kolb's model is suitable for the conventional type of learning, where the learner meets the tutor directly, this model cannot be applied directly in web based e-learning.
According to Honey and Mumford's model [10] , behaviors are very important for identifying learning styles.They contended that learners fall into one of the categories of: Reflectors, Theorists, Pragmatists and Activists.They introduced the concept of adaptiveness in learning based on behaviour and in order to implement this in e-learning, it is necessary to introduce a neural network model for adjusting the cognitive load dynamically.
Gregorc's model [11] is a cognitive model that refers to four learning preferences: concrete sequential, abstract random, abstract sequential and concrete random.This classification helps to form learning groups that can then be provided suitable learning assistance.However, it is essential to have an effective mechanism for forming such groups, which is possible with the application of clustering algorithms.
Flemming's VAK model [12] is a meta learning theory that terms the different learning styles: Visual, Auditory and Kinaesthetic.This consequently places emphasis on the audio visual features with regards to learning.Audio, video and text mining techniques can be employed with this perspective to understand learner behavior.
Dunn and Dunn's model [13] a biological and experimental model which considers environment and emotion with regard to learner preferences, which is validated using the noise level and persistence metrics.Jackson's model [14] is based on the neuro-psychological theory, where the learning preferences are based on the individual learners' sensation, goal, willingness to achieve, emotion and deep learning to achieve.Both these models consider emotion as a parameter for learning.requires user interaction integrated with machine learning algorithms to derive suitable rules.
Drawing on Carl Jung, Myers and Brigg [15] developed a personality theory, which classifies the personality based on judgment and perception, thinking and feeling, sensing and intuition and extroversion and introversion.They contend that since learning can be adopted based on the personality it is necessary to understand the personality of the learners.However, understanding it is very hard and hence, suitable agents must be introduced into this model for monitoring and hence, comprehending the learner's personality.
Felder-Silverman's psychological theory [16] is helpful to understand the learner's mood, which can be active or reflective.Moreover, he can be sensing or intuitive while learning and the learning is either based on Visual or verbal features.Finally, it considers the sequential and global nature of learners.This we consider to be the most important contribution on learning styles for our purposes when compared with the other perspectives.This is owing to the fact that it can be used in e-learning where the psychology of the learners is considered in advance so that flexible courseware can be prepared and provided suiting the learners' behavior.The following table 1 shows a summary of some existing adaptive systems.

Learning Styles and Personality
This study is based on the widely accepted theory that every student has an individual or particular learning style [16].A learner with a particular learning style can confront difficulties while learning, when there is not bolstered by the instructing environment.Numerous authors have proposed distinctive definitions for learning style.Learning style can be characterized as student's preferences in the way of learning and differences in students' learning, and it is considered as one of the factors influencing learner's achievement [22].
A wide importance, meaning is provided by Keefe [23].Learning styles can be characterized as characteristic cognitive, affective, furthermore psychological behaviors that serve as generally stable indicators of how learners perceive, interact with, and respond to the learning environment.
James and Gardner (1995) define learning style as the "complex manner in which, and conditions under which, learners most efficiently and most effectively perceive, process, store, and review what they are endeavoring to learn " .There is a definition of learning style that was presented by Merriam and Caffarella (1991) which is well known in grown-up education, as the "individual's characteristic method for processing information, feeling, and behaving in learning circumstances" [24].
In our research, we concentrate on two models as explained below.
• First, the Felder-Silverman model [16] (FSLSM) is selected because the authors provide the questionnaire and a comprehensive guide on how to use it .In addition, this model has been turned out to be powerful in numerous adaptive learning systems and it has often been used in technology-enhanced learning [25] [26].In addition, this model is adequately accepted in numerous versatile situations [27], [28].inorder to deliver personalised contents adapted to student's learning styles.the FSLSM describes the learning style of a learner in more detail than other models, distinguishing between preferences across four dimensions :active/reflective, sensing/intuitive, visual/verbal and sequential/global.The dimensions sensory/intuitive and visual/verbal refer to the mechanisms for perceiving information.Whilst the active/reflective and sequential/global are concerned with the way of understanding and processing information [29].The associated questionnaire, named the Index of Learning Styles (ILS), consists of 44 questions with two options, A or B, with each being related to just one of the four dimensions.
• The model we have chosen, the Myers-Briggs Type Indicator (MBTI), has been widely used and validated in the education domain [30] and has long been considered an important instrument by a ducational psychologist's [31].The MBTI questionnaire examines personality traits in four distinct domains: extraverted (E)/introverted (I), sensing (S)/ intuitive (N), thinking (T)/feeling (F), and judging (J)/perceiving (P).
As briefly discussed above, not just can the learning style impact learner performance, but also, additionally, identity has very high impact while determine learners preferences in addition to include them in learning processes.There appear to be critical variables for deciding learner personality while we adapt their profile ; Personality is closely related to preferences for learning materials in that a specific configuration that format reflects learner's preferences regarding receiving an information and making decisions.Identity is firmly fixing to inclinations for learning materials in that a particular Most of personalized e-learning systems did not consider these elements while they build student model, because there is no easy way to model adaptive profile that is based on both learning style and learner personality.The only method thus far is AHA! [20], which identifies the learner's style as "activist/reflector", in view of a self-evaluated personality type.
As previously stated There are a relationship between the Felder-Silverman model and MBTI model.the following figure 1 show the correlation between FSLSM and MBTI personality .

The Semantic Web and Ontologies
The Semantic Web [32] is defined as "an extension of the current web in which information is given well-defined meaning."It imagines a machine-justifiable web with an an explicit semantic representation of fundamental web pages, web information, and other web assets Figure 2 demonstrates the semantic web stack • The first layer is a syntax layer, where XML stands for extensible Markup Language (XML), which allows people to structure their documents by defining and adding their own tags • Web Ontology Language (OWL) is a an ontology language for the Semantic Web, which allows subclasses of the taxonomy to inherit properties and relations of their ancestor classes • Finally, the Proof Layer is used to provide "proofs", for instance, , as , to demonstrate that the joined data is acquired from a trusted source.In the work of ontology based automatic annotation of learning content [34], ontology is used to annotate learning objects with metadata.Similarly, Gasevic et al. [35] have also used domain ontology for semantically marking up the content of a learning object.In the work of Ramezani et al. [36], an algorithm in a Web 2.0 platform is recommended that supports end users collaboratively to evolve ontologies by suggesting semantic relations between new and existing concepts.They use the Wikipedia category hierarchy to evaluate the algorithm and the experimental results show that it produces high quality recommendations.
Gutierrez [37] present the concept of ontology and activity to build an approach of learning activity sequencing.He implements an algorithm to realize this and a dynamically updating learner profile.
Ontology reasoning steps.Ontology reasoning development include the following steps as shown in figure 3 : Every subdomain incorporates points on study that are identifier for that sub-domain [38].For example, the field of MIS can be described as a sub-domain of Information systems.As it is shown in Figure 4, concepts of LDO are learning, learner and learning style.Relations between them is "has style" and "is style of ".

Student profile
A personalized student profile is defined as the ability to provide content and services tailored to the individual based on the knowledge about his preferences and behavior [39].The information regarding these is gathered in the student model.User profile is practically the normal representational of student's data that can be gathered in two ways: from the student or by analyzing his behaviour through a learning management system.If the details are gathered directly from the learner, then subsequently the profile made is called explicit or static profile.Whereas if this information is collected by observing the behavior of the learner then the profile created is known as the implicit or dynamic profile.If we build a learner profile, then the data can be effortlessly adjusted for every learner according to his/her preferences.
Describing Learner Data.Learner data are those pertaining to an individual learner, including the learner profile (personal data), completed content (progress made) and performance data.There are a number of standards for representing learner data : • Public and Private Information (PAPI) [40]; • IMS LIP (Learner Information Package) • FOAF (Friend of a Friend) [42] These vary in terms of their main purpose and the way in which a given system can use their embedded information.Some e-Learning systems use metadata from more than one standards to produce a learner profile; for example, the PAPI standard considers both the student's progress and performance .

Advantages of Using Ontological Profile
There are numerous benefits of building ontological based profile • We could use reasoning toward building the ontologies.We can utilize ontology relations, conditions and restrictions as a premise for deducing extra learner characteristics.
• Ontologies control the uncertainties compared into the data and the user profiles that are acquired by the Ontology model give better results compared to other adaptive techniques.
• Ontology provides shared understanding of the area which helps reuse of the outcomes.Furthermore imparting of learner profiles is the most vital point in utilizing ontological based client profile [43].

Adaptive e-learning models
In recent years, several researchers have focused on applying different data mining techniques in order to analyze learner log files, match them with the appropriate learning style and build personalized learner profiles.In this section we present these works.In our System Adaptive model [45] Fulfillment overview demonstrated that the majority of the group "agree" or "strongly agree" " concerning the ease of use while creating intelligent tutoring system [46] Learners in investigation group were uncovered to finish the lesson in less time than the control bunch.They additionally finished a greater number of lessons than those in control group 70 of the learners found the system helpful and 60 of them found the system adaptive versatile accurate and delivery at fulfilled speed applications. [47] It offers backing to the author in creating in making completely comprehensive materials by proposing accentuation on the design the outline and development of the learning objects the learning objects [48] This is the configuration and improvement of an adaptive system in light of ontology that considers information generally present into the e-learning environment, for the most part that learning styles and metadata with a specific end goal to propose which is adaptation rules. [49] It gives an automatic suggestion to active learners without requiring the explicit input in view of Automatic personalization approach [50] Architecture for an adaptive and personalized tutoring system that completely totally depends on Semantic Web models and advancements.An ontology-based approach is displayed.[46] Authoring tool that permits a high school for web-based intelligent tutoring systems.It takes into account the quick formation of a web-classroom applications.
[51] presents a service-based architecture for a adaptive system based on ontology that technologies to add the personalization capability to open present into the e-learning environment like dynamic, learning networks.Several web-services use the learners' profile to adapt their queries, or to their questions, or to produce suggestions for with requests Table 2. Adaptive models related works model, we utilize ontology with an inference engine (rulebased) to represent and build student learning profile and match it with this learning style that suits his/her preferences and personality.Our focus is to further enhance this area of research by not only adapting the process mining tools, but also presenting a way to introduce semantic-based reasoning for adaptation within the learning process.
Fahland and Van der Aalst [44] note that process mining has been proven to be one of the existing technologies that is able to extract useful information from user log files.
Due to the huge amount of "irrelevant information" in a web log, the original log file cannot be directly used in a web usage mining procedure and consequently, preprocessing of the web log file becomes imperative.To this end, the design of a data e-learning web-house as a supporting structure for future personalized e-learning systems has been proposed [52].
Hang jinghua [53] propose a Semantic Web Based Personalized Learning Service for programming courses in e-learning.This model is based on resource base, ontology base, and strategy base techniques.The proposed model although effective is not suitable for all strategies.
Another model that is using ontologies for generating a student activity report from the log files inside a Moodle-based e-Learning system has been proposed in [54].This research combines two concepts that is, using ontologies and giving recommendations inside the e-Learning mechanism based on knowledge-based reasoning.
Authors in [55] propose to created user profile by collecting information through a meta search of his/her blog, personal/organization, web pages, and any other web sites.WordNet and the Lexico-Syntactic pattern for hyponyms were used to extract features from documents.This profile can be further improved by applying an ontology matching approach to enrich the profile with characteristics other similar users.
Authors in [56] have built a user profile by analyzing the web log with the use of WordNet in order to extract data from documents and solve the semantic inadequacy of the VSM model.A fuzzy technique is employed to classify the learners according to their interests and the Felder-Silverman model, We note that this model is narrow, because it only focuses on analyzing assignments submitted by the students.
The work in [57] describes a context-aware platform which provides personalized services to the learners.It uses an ontology-based context model with accompanying rule-based context-aware algorithms.These algorithms capture the behavior of the learner and provide relevant material.However, it only focuses on learning meta-data for personalized context and this method is not suitable for all learning management system.
Similarly, PASER (Planner for the Automatic Synthesis of Educational Resources) is a retrieval engine for automatic and personalized curricula construction, based on appropriate learning object combinations.The personalization is designed to take into account the learner's profile and his preferences.This model involves first the creation of a repository metadata which includes learning object descriptions, learner profiles and domain ontology; second a deductive objectoriented knowledge base deductive which is responsible for querying and reasoning about RDF/XML metadata, called R-DEVICE ; and finally a planning system called HAPEDU that automatically constructs course plans [58].
The ONTODAPS systems [59] is an ontology-driven disability-aware personalized e-learning system, which personalizes learning resources and services for students with or without disabilities.In addition, it provides appropriate levels of learner control by allowing them to personalize learning resources.The work presented in [59] describes a learning environment that personalizes e-learning relating to pedagogy and a personalized educational process.The framework is based on web services, the description of the semantic information of learning units and the relationship between units.
The work presented in [60] describes a model for building personalized e-learning experiences.This model accounts for different cognitive states and learning preferences of learners.In addition, it supports experts in modeling educational domains using ontologies.Using these models, personalization is achieved through several steps 1-educational domains model based on reference ontologies; 2-modeling of learner cognitive state and preferences (Student Model); 3-build the relationship between metadata and and learning objects 4-modeling of E-Learning experiences (E-Learning experience model)".Another adaptive model is described in [61].This work focuses on the student's cognitive state and cognitive process.It provides a diagnosis related to the student's knowledge state, and achievement quality of the learning objectives.The Student Model is based on ontologies to extract which feature is important in order to build knowledge representation.This design incorporates a number of ontologies including profile ontology (personal information), a student state ontology (progress) and a learning objectives ontology.
The student model can be used for accessing the knowledge in digital libraries by creating a student ontology in [62], which consists of two parts, general student information and information about student behavior in the learning domain dynamically.
A user profile modeling method has been designed in [63] by combining the keywords and ontology concepts.This model takes into account short-term interest and long-term interest of the user.The authors of the proposed system verified that their model improves the efficiency of the information retrieval procedure.A user profile ontology is proposed in [64], which incorporates the concepts and properties deployed to model the user profile.Ontologies related to the domain have been used to create this model.The model is available in two different areas, personal information management and adaptive visualization.
The ALOCoM ontology [65] is designed to generalize the content models and to provide an ontology-based platform to integrate the different ones by explicitly defining the structure of their LOs (Learning Objects).The revised ALOCoM ontology [66] is divided into two different parts: ALOCoM content structure ontology, in order to define the learining obejecrts and its role as well as their components.CoAKTinG project [67] has developed an ontology based system for distributed eScience through the application of advanced knowledge technologies.The EUME Onto [68] is an educational ontology system that contains concepts related to learning resources, learning design and learning content.
The LOFinder [69] is an intellectual Learning Object Metadata which enhance knowledge representations, as well as enables intelligent discovery of learning objects.Cakula et al. [54] have developed a personalized elearning model using methods of ontology.Their aim is to discover overlapping points of KM andbuild personlised e-learning using ontology and metadata in effective manner.HJia et al. [70] has designed a performance oriented workplace e-learning system which aims to overcome the gaps between individual needs and organizational interests and improve the user satisfaction.In order to do so key performance indicators are used in order to clarify organizational training requirements and to aid learners to set up rational learning objectives.Moreover their is also used to develop formal and machine comprehensible conceptualization of the performance oriented learning environment.

Proposed Adaptation Process Flowchart
The following adaptation flowchart represents the adaptation process.On the one hand, the instructor is responsible for adding course material in different formats and adding student cases, which is illustrated in the following figure 5

Proposed adaptive learner profile
We propose a learner ontology model which displays the individual data and learning qualities of distinctive learners.Figure 7 depicts the graphical representation of the learner model.

Student interface
The student Interface is the communication component that controls the interaction between the student and the system.It deals with the account of learner's such as (registration and login) after that student fill learning style questionnaire which related to FSLSM model.

Data collection
In this section we present, the data that are collected from AAST's (Arab Academy for Science and Technology and Maritime Transport) faculty of business.Two types of data are collected from the learners: 1.When they log into the AAST student portal for the first time, they need to fill in the questionnaire based on the index of learning styles (ILS) developed by Felder-Silverman model.
2. The learner behaviour data collected from two sources namely MOODLE and Student portal.

• MOODLE (Modular Object-Oriented Dynamic
Learning Environment) includes learner Personal Information as far as the essential individual data, for example, name, date of conception, email, login record etc.
• Student portal holds information about the learner behaviour Such information comprises categories of knowledge, preferences and behaviour such as the number of visits, time spent on exercises etc.

Adaptive core (data processing)
Data repository.In the data processing stage the collected student's information from the student portal and MOODLE will be placed in a suitable repository for further analysis.
Reference ontology.The system should have a reference ontology for student profile modelling.Ontology creation can broken down into two main parts; the first one is static profile and the second one is dynamic profile in order to match the behavior of the user with the suitable learning style according to FSLSM and MTBI model.The data are collected from two sources data repository and learning style model.
Adaptive engine.In this stage, the system compares the outcomes from questionnaire to these from the reference ontology using inference rules (association rules).Subsequently, it starts to recommend adaptive content based on the personalized profile for the student, as shown in Figure 8.
Inference Engine.Inference engine is the crucial component for constructing adaptive learning.It includes comparing recommendation agent and updating agent that provide personalized student profile dynamically.Whenever new information is available then it will send to the inference engine, which works based on rule-based reasoning.Rule based is most often used to build using a series of if then functions for instance: The framework utilizes this data as a part of request to adjust to learner's individual needs.The framework regulator upgrades the learner models amid the learning procedure, so as to stay informed concerning learner's activities and advancement and perhaps manage the learner as needs be.Learner model is in charge of recovering the attributes of a specific learner, rolling out the fundamental improvements and sending it to the adjustment model through collaboration with the storehouse.The framework additionally gets the information about new learners from the User Interface and stores it in the learner model.Learner model is overhauled when it gets new data about the learner from the adaptive engine.The learner model gets continuously upgraded by incorporating learners' interaction with the framework.In points of interest, learners are occupied with adapting adroitly precharacterized subjects, complete activities and take tests, while the framework ought to consistently perceive changes in the learner's information and capacities as they advance and upgrade the learner model in like manner.

Used tool (Protégé)
In this proposed ontology based mechanism the broadly accessible ontology editor Protégé 4.3 [71] is utilized as a development tool.Ontologies and learning bases can be adjusted intuitively inside Protégé, being accessed with a graphical client interface and Java API.Protégé can be extended using pluggable components to include new functionalities and administration.There is an expanding number of plug-ins offering an assortment of extra elements.Protégé implements a rich arrangement of information demonstrating structures and activities that support the creation, perception, and control of ontologies in different representation designs.There are various structures, for example, RDF(s), OWL and XML Schemes in which protégé philosophy can be exported.Besides, this ontology editor is picked because it enables the construction of domain ontologies and customized data entry forms.In addition, it allows for the definition of classes, class hierarchies, variables, variable-value restrictions, and the relationships between classes as well as the properties of these relationships.

Ontology representation
The first phase of the ontology building process is identifying the ontology goal and scope, in order to specify the domain ontology and identify the required resources.Figure 9 illustrates the components of an ontology that pertain to an adaptive student profile, which is divided into main three classes: basic, static and dynamic information.Students' basic information details are collected from AAST's MOODLE for its faculty of business, comprising name, date of birth, email address etc. which is divided into several subclasses.Students' dynamic information details are then collected from AAST's Student portal such information comprises categories of Knowledge, preferences and behaviour like No. of visits, No. of visits and time spent on exercises Amount of time dealt with reading material etc.
An Ontology has two types of classes : • Defined classes : classes that have at least one set of necessary and sufficient conditions.
• Primitive classes : class that only have necessary conditions.For instance super classes(Learner class).

Object properties (relations)
Relationship is an Object in ontology which link between instances as well as between an object and an attribute which is related to.Some of the Relationships and their properties made for the proposed understudy profile are illustrated in Table 3.

Modeling relationship between personality and learning style
Figure 10 shows the ontological model for the relationship between student personality based on MBTI and learning style based on FSLSM .For instance, for a student who has a thinking personality the most suitable learning style is active and intuitive.

Relation between behavior and style
The style of the learner can be acquired by investigating the learner's behavior while using the framework.Learning styles ordinarily allude to how a user tends to utilize faculties in order to learn.It can be spoken to the learning style in generalization model as indicated by the Felder-Silverman learning style classifications.Figure 11 shows the relationship between student behavior and learning style based on FSLSM model.

Conclusion
In this paper we propose a new model for automatically building s learner profile in an e-Learning environment.It is based on real behavior patterns of students during interaction with the AAST student portal, employing ontology creation and an inference engine to identify learning styles automatically according to the FSLSM model.The ontologies give perspectives of the learner style taking into account the behavior of the student.Personalization can be achieved by coordinating the user's profile with the courses offered in the college.The users subsequently will receive suggestions for courses based on the data collected from their behaviour, thereby avoiding inappropriate recommendations being generated.

Figure 1 .
Figure 1.Matching the four MBTI dominant preferences with the FSLSM dimensions

Figure 10 .Figure 11 .
Figure 10.Modeling relationship between personality and learning style

Table 3 .
classes and properties