Abstract: Knowledge engineering (KE) is a subarea of artificial intelligence (AI). Recently, KE paradigms receive increasing attention within the fields of smart education and learning. Developing of Smart learning Systems (SLS) is very difficult from the technological perspective and a challenging task. In this paper, three KE paradigms, namely: case-based reasoning, data mining, and intelligent agents are discussed. This article demonstrates how SLS can benefit from the innovative KE paradigms. So, the paper addresses the advantages of such approaches for the industry of SLS. Moreover, we concentrate our discussion on the challenges faced by knowledge engineers and software developers in developing and deploying efficient and robust SLS.Keywords: Knowledge engineering; Smart Learning Systems; Artificial Intelligence; Intelligent Agents; Data Mining; Case-Based Reasoning1. Introduction Researchers have been applied the KE techniques and methodologies to develop a smart tutoring and learning systems 1. Moreover, the convergence of AI, KE and web science is enabling the creation of a new era of web-based smart systems for all educational and learning tasks. Smart learning represents a collection of smart services that employ smart digital media and communication and information technologies for supporting learning, training and educational processes 2-5. For that reason, SLSs are complex to build and complex to maintain. Based on the recent research during the last five years, Knowledge engineers and software developers have started to investigate the usage of KE to develop robust and new generation of SLS 6-8. The main properties and characteristics of these systems are the ability of perception, inference, reasoning, learning, thinking, and knowledge-based.This paper is organized as follows: the second section introduces an overview of the knowledge management and representation techniques for developing the SLS. In the third section, we present three intelligent approaches used by the knowledge engineers to develop SLS, namely data mining (DM), case-based reasoning (CBR), and ontological engineering (OE). Sections four and five discusses the benefits from suchapproaches and challenges. The last section draws conclusion and perspectives.2. Knowledge Management and Representation Techniques for SLSFrom the knowledge engineering (KE) prospective, the main components in developing an efficient and smart learning/educational system for any task involve the “knowledge base” and the “inference engine/reasoning mechanism”. From the KE point of view, there is a set of knowledge representation and management techniques to build the knowledge base, e.g. semantic networks, trees, lists, frames, scripts, ontologies, production rules and cases. Fig. 1. shows the different techniques used for knowledge representation. Concerning the inference engine/reason mechanism, there are many methodologies of reasoning, e.g. Case-based, commonsense, geometric, non-monotonic, model-based reasoning, fuzzy, probabilistic, causal reasoning, automated reasoning, qualitative, temporal reasoning and spatial. In fact, these methodologies receive increasing attention within the community of smart learning/educational systems 1-2, 9-10. In this section, we will present a summary of the different techniques of representing knowledge.Fig. 1. Knowledge Management and Representation Techniques2.1. Knowledge Representation Techniques for Static and Hierarchical KnowledgeKnowledge can be a vague term. Research is still being done on about how knowledge can be modeled so it can be manipulated and processed by a computer. Among the many approaches used by the knowledge engineers are semantic networks, frames, production rules and scripts 1, 11.Knowledge Management and Representation TechniquesHierarchical Knowledge(Static Knowledge)? Lists? Trees? Frames? Semantic Nets? Scripts? Ontology? Production RulesStereotypical Knowledge(Dynamic Knowledge)? Cases (List of features)? Explanation Patterns? Memory Organization PacketsSemantic Networks as a knowledge structureA Semantic Network (SN) shows relationships among various entities. SNs are a kind of network commonly used to structure more general kinds of information. These are groups of nodes and links in which the nodes refer to concepts and the links present the relationships between them. So, SN shows relationships among various entities. They got their name because they were originally employed to represent the meaning of natural language (NL) expressions. The use of some form of network for modeling of concepts and relations is so widespread in AI systems and all kinds of knowledge –based systems.Frames as a knowledge structureThe frame is grounded on Marvin Minsky’s theory about humans think. “When one faces a new situation”, he says, “One chooses from memory a substantial structure called a frame”. Following to Minsky’s view a frame is a data structure for promoting a stereotyped situation, like being in a certain kind of living room or going to a child birthday party.Production Rules as a knowledge structureProduction rule is the simplest and most popular method to present knowledge. “If-then” rules are considered as the most common form of declarative knowledge representation applied in AI applications. Most people are comfortable reading rules, in contrast to knowledge represented in predicate logic. Every rule can be viewed as a standalone piece of knowledge or unit of information in a knowledge base. New knowledge can be easily added, and existing knowledge can be changed simply by creating or modifying individual rules. Therefore, the rule is simple, modular, appropriate size, procedural, and descriptive.Scripts as a knowledge structureScript is another kind of a knowledge representation technique that it is similar to a frame, but instead of depicting an object, the script depicts a sequence of events. Like the frame, the script represents a stereotyped situation. Dissimilar to the frame, it is typically presented in a particular context. To demonstrate a sequence of events, the script applies the use of series of slots that include information about the actions, objects and people that are involved in the events. Scripts accounted for information about stereotypical events, e.g. visiting the dentist, taking a bus and going to a restaurant. In stereotypical events (common situations), a person has a group of expectations of the props, goals, default setting and behaviors of the other people involved. Scripts are analogous to Minsky’s frames 1, 11, which were proposed in the context of visual processing. Scripts are relevant to autobiographical events and are inherently episodic in origin and use, i.e. scripts arise from experience and are applied to interpret new events. Scripts were proposed as a knowledge structure for a conceptual memory. As a psychological theory of memory, scripts suggested that people would remember anevent in terms of its associated script.2.2. Knowledge Representation Techniques for Stereotypical knowledgeMemory Organization Packets (MOP)MOP is a general knowledge representation technique accounts for the diverse and heterogeneous nature of episodic knowledge 1, 11 MOPs can be viewed as met scripts, e.g. instead of a doctor script or a dentist script, there might be a professional-office-visit MOP that can be instantiated and specified for both the dentist episodes and the doctor episodes. This MOP will involve a generic waiting room scene, thus providing the basis for confusion between dentist and doctor episodes.Reminding and Explanation PatternsSchank (1981) 11 proposed a theory of learning based on reminding. The main characteristics of this theory can be abridged in the following points:? Conform-Driven Learning: When the new situations (or experiences) conform the past cases and events, Thus we can classify a new episode in terms of previous cases.? Failure- Driven Learning: WHEN the new situation does not adapt to the prior case, we have a failure. That is, we had an expectation according to a prior event that did not occur in the new situation. THUS we must classify this new experience as different from the precedent episode. We must store this new experience, and we must learn.? Discrepancy-Driven Learning: WHEN we realize a discrepancy between our predictions and some event, THUS we have something to learn and consequently we need to review our knowledge structure.The mechanism to evolve our knowledge requires explanation. Schank (1986) 11 presented an explicit knowledge structure “explanation patterns”, which is used to generate, index, and test explanations in conjunction with an episodic memory.Cases as a knowledge structureThe “case” is a list of features that lead to a specific outcome, e.g. the information on a patient history and the associated diagnosis 12. Fig. 2 shows the perfect structure of the “case” from the knowledge engineering intelligent perspective. From this figure, it can be seen that: depending on the “case” structure, the “case” can be used for a variety of purposes. In Smart Learning Systems, the “case” can include: (a) a multi-media description of the problem, (b) a description of the correct actions to take including optimal and alternative steps, (c) a multi-media explanation of why these steps are correct, and (d) a list of methods to determine whether learner/students correctly executed the steps.Fig. 2. The “case” structureFrom the intelligent informatics point of view, determining the appropriate “case” features is the principal knowledge engineering task in developing the case-based smart software. This task involves defining the domain terminology and collecting representative “cases” of problem solving by the knowledge engineers. Representation of “cases’ can be in any of various forms (frames, predicate, scribes).Ontological Engineering (OE)The term ontology represents a common vocabulary in a specific task. OE refers to the group of activities that concentrate on the ontology development process, the ontology life cycle, and methodologies for building ontologies, as well as the tool suites and languages that support them. Now OE are frequently used by smart computing and information science communities 13. At present, there are applications of ontologies in commercial, industrial, medical, academicals domains and research focuses 14. The use of Ontologies in smart educational and learning systems may be proposed from several points of view: as a chain between heterogeneous educational systems, as a common and approved vocabulary for multi-agent system, ontologies for pedagogical resources sharing or for sharing data and ontologies used to mediate the search of the learning materials on the web-based environments. The brief specification of a system is involved of functional interconnected elements. These elements communicate using an intelligent interface and a common vocabulary 14.3. Intelligent Approaches for Smart Learning Systems (SLS)This section presents three intelligent and robust approaches used by the knowledge engineers to implement the smart learning systems, namely: data mining, case-based reasoning and intelligent agents.ProblemDescription(1)CaseSolution(2)CaseOutcome(3)DerivingSolutions toNew ProblemsEvaluatingProposedSolutionsEvaluatingNewSituations++3.1. Data Mining and Knowledge Discovery (DM and KD) ApproachData mining (DM) focus on the discovery of hidden knowledge, unexpected patterns and new rules from large databases 15. DM and KD is not a consistent field, it is a dwells upon already well established technologies including neural networks, statistics, rough sets, data preprocessing, pattern recognition, data cleaning, clustering, machine learning, fuzzy sets, etc. Fig. 3 shows the major functional phases of the KD process. The preprocessing phase is often referred to as data cleaning. The cleaned data are located in the warehouse. This is followed by DM phase and its results are provided to an output generator (visualization) generating reports, action lists, or monitor reports. Each phase is supported by different methodologies.Fig. 3. Knowledge discovery processKD process involves the following three processes: (i) using the database with any required selection, preprocessing, sub-sampling, and transformations of it, (ii) applying machine learning and computational intelligence techniques to identify hidden patterns from it, and (iii) evaluating the products of data mining to enumerate the subset of the identified patterns deemed knowledge.The data mining components of the KD process is concerned with the intelligent methodologies by which patterns are extracted and enumerated from data. The overall KD process involves the evaluation and the adequate interpretation of the mined patterns to identify which patterns can be considered as new knowledge. Fundamental issues in KD arise from the very nature of databases and the objects (data) they deal with. They are characterized as follows: (a) huge amounts of data, (b) dynamic nature of data, (c) incomplete or imprecise data, (d) noisy data, (e) missing attribute values, and (f) redundant or insignificant data 16-17.Following to many literatures e.g. 15-17, DM is supported by a host that acquires the character of data in various different ways including:(a) Clustering: The main objective is to find natural groupings (clusters) in highly dimensional data. Clustering is an example of unsupervised learning and it is a part of pattern recognition. In this respect, K-means algorithms are usually used.(b) Regression Models: These derived from standard regression analysis and its applied part known as system identification. The underlying idea is to construct a linear or nonlinear function. Machine learning techniques; Support Vector Machines (SVMs), Decision Trees, Rule induction, Neural Networks are preferable techniques to perform this task.(c) Classification: This concerns learning that classifies data into the predefined categories. Regarding this task there is a huge number of classifiers have been developed. Based on our analysis we found that SVMs, Decision Trees, Neural Networks, Rule induction, and Genetic Algorithms are more appropriate techniques to perform classification tasks.(d) Summarization: This is an approach of describing data with a small number of attributes /features. In the simplest scenario, one can think of standard deviations and a mean as two extremely closely descriptors of the data. This task is often applied in an automated report generation and interactive exploratory data analysis through the multivariate visualization approaches.(e) Link analysis: It is concerned about the identification of relationships (dependencies) among database fields. In a particular case, we may be interested in the determination of the correlation between the variables.(f) Sequence Analysis: A kind of analysis that is oriented for problems of modeling sequential data. Pertinent models support the temporal neural networks, time series analysis and time series models.Recently, researchers investigate different DM methods to help administrators and instructors to enhance e-Learning systems 18-20. Some of the major e-Learning problems or subjects to which data mining techniques have been applied are dealing with.3.2 Case-Based Reasoning (CBR) ApproachCBR concerns about the reasoning from experiences or “old cases” in an effort to fix problems, critique solutions, and explain anomalous situations. CBR is an analogical reasoning method presents both a robust methodology for building case-based reasoning systems, and a cognitive model of people. From the psychological point of view 12, CBR refers to reasoning in which a human problem-solver relies on previous cases thathe or she has encountered. Psychologists have observed the following facts: (a) people are good at using analogues to solve new problems, (b) people are not always good at remembering the right solutions (computers are good at remembering).From the computational perspective 12, CBR refers to a number of concepts and techniques (e.g. data structures and intelligent algorithms), that can be used to perform the following operations: (a) record and index cases, (b) search cases to identify the ones that might be useful in solving new cases when they are presented, (c) modify earlier cases to better match new cases, and (d) synthesize new cases when they are needed.Nowadays, CBR paradigm is becoming popular in developing SLS because it automates applications that are based on precedent or that contain incomplete causal models. Research discloses that learners study best when they are presented with examples of problem-solving knowledge and are then required to apply the knowledge to real situations 2.Knowledge engineer’s and AI researchers have started to use the CBR concepts in enhancing human decision making through developing case-based reasoning systems 2, 21-22. Fig. 4 shows the typical functional diagram of a CBR methodology. When a new problem is appeared in the system, the problem is indexed, and therefore, the indexes are used to extract previous cases from memory.These past cases lead to a set of prior solutions. Subsequently, the previous solutions are modified to adapt to the new situation. Then the proposed solution is tried out. If the solution succeeds, then it is saved as a working solution; if it fails, the working solution must be repaired and tested again. In support of CBR processes, the following rules knowledge structures are necessary to perform the following tasks: case indexing, case memory, similarity, modification, and repair. One of the essential goals of CBR systems is to find the most similar, or most relevant cases for new input problems. New generations of CBR based knowledge systems: (a) applies the CBR methodology as an inference technique, (b) uses an extensive past cases as a knowledge structure, and (c) solves new problems by adapting solutions of similar problems. The effectiveness of CBR systems depends on the quality and quantity of cases in a case memory.Fig. 4. Functional diagram of a CBR methodology3.3. Intelligent Agents (IAs) Approach for Smart Learning Systems (SLS)IAs are artificial entities that have several intelligent characteristics and features, such as being autonomous, responding adequately to changes in their environment, persistently pursuing goals, flexible, robust, and social by interacting with other agents 23-24. From the intelligent software industry, the main advantages of IAs are:(a) Agents have been described as objects that exhibit various important properties that are very attractive for the modeling and design of advanced user interfaces encountered in e-Learning systems: teachers, tutors and students.(b) Generic agent types proven to be effective for the suitable functional decomposition of e-learning systems.(c) Dynamic and interoperability features of agents are very appropriate for supporting extensibility and maintainability of e-Learning systems 25-26.Consequently, an intelligent learning system based on a multi-agent approach consists in a set of intelligent agents, which have to communicate and collaborate through messages 27. Software agents can comprehend and interpret the messages due to an accredited ontology or the interoperability of the private ontologies.4. Benefits of Knowledge Engineering Approaches for Smart Learning SystemsBased on our discussion of the three approaches mentioned in the above sections, we can draw the major benefits of DM, CBR, and IA approaches for SLS (see Table 1).Table 1The major benefits of KE Approaches; Data Mining, Case Based Reasoning, and Intelligent agents approaches for SLSKE ApproachBenefits for SLSData Mining? Evaluation of candidate’s learning performance and learning recommendations according to the candidateslearning behavior.? Assessment of learning material and web-based courses, provide feedback to both teachers and students of e-Learning courses.? Detection of new, useful and interesting knowledge according to the candidate’s usage data.? Grouping learners/users in order to give them differentiated guiding based on their skills and other characteristics.? Identifying learners with little motivation and finding remedial actions in order to lower drop-out rates.? Identification of a typical candidate’s learning behavior.Case-Based Reasoning? With more cases available in the case-memory, learner will be able to benefit from the failures of others.? Retrieval cases process will allow learners to better recognize what is important in a new situation.? During a training process, CBR system provides the learner with a model of the way decision-making ought to be done.? For tasks where there is much to remember, CBR systems can augment the memories of even educators.? From the educational perspectives, both educators and learners tend to focus on too few possibilities when reasoning analogically or to focus on the wrong cases.Intelligent agents? Acting autonomously, i.e. by deciding themselves what to do? Being sociable, i.e. by acting with other software agents.? Mimics human interaction types, e.g. coordination, negotiation, and cooperation.5. Discussion and ChallengesFrom the previous aspects of the KE paradigms, one can summarize the main characteristics of SLS in the following: (a) it is a knowledge-based system, (b) based on heuristic, declarative, interactive and symbolic processing, (c) convergent reasoning, i.e. producing a few results from the big data analytics, and (d) gives recommendations.Case-based reasoning methodology organizes knowledge in terms of “cases or examples” of previous problems and their solutions. Such knowledge structure overcomes the knowledge in a lesson-oriented manner and the automatic generation oftests and exercise. Moreover, CBR methodology addresses the problems of rule-based systems, e.g. knowledge acquisition, performance experience, adaptive solutions and maintenance.Information and data mining is a very promising approach towards the data analysis of learner activities and behavior which collected by learning management systems. Smart data mining techniques can promote online learning for the learners. The big challenge in this respect is, while some tools using data mining techniques to help software developers, knowledge engineers and learners are being developed, the research is still in its infancy.Nowadays, intelligent agents paradigms were proposed to reinforce smart learning systems among at leastwise two dimensions: (a) agents as a modeling and design paradigm for advanced human-computer interaction and (b) agents for smart functional decomposition of complex systems. Agents’ technologies are often considered as incarnations of different forms of AI including reasoning, knowledge engineering, machine learning and information mining. Research interests in agent systems are extended to several topics such as modeling, design, and development of advanced software systems that are appealing for a number of computer applications.6. ConclusionsThe development of smart learning systems is a very burdensome; it is a composite process that promotes a set of technological and research challenges that have to be treated in an interdisciplinary manner. Today the fusion of machine learning and computational intelligence techniques with the knowledge engineering paradigms and web science solves the technical problems and difficulties in designing new generation of SLS. Such convergence will produce a new generation of web-based SLS. The web based of such systems can enhance the online learning/education/training processes through the web.The key to the success of these systems is the selection of the appropriate technique that best fits the domain knowledge and the problem to be solved. That choice relies on the experience of the knowledge engineer. On the other hand, in the current environment of global wired and wireless networks, intelligent agents may play the role of a universal carrier of distributed artificial intelligence. Therefore, the integration of software agents approaches and educational technologies is beneficial for designing efficient, robust and smart learning systems. In addition, ensuring the success of such systems to the cloud is an interesting challenge.