Abstract
Intelligent Tutoring Systems (ITS) are software tools that mimic a teacher’s teaching methods through artificial intelligence techniques. The generalized model of these systems is divided into four main modules: tutoring, student, domain, and interface. Although it has been shown that ITS is very useful in cases where a teacher cannot be present, the development of these systems is expensive and time-consuming, since it requires experts and available programmers. Therefore, this research proposes a framework to develop an authoring tool to build ITS automatically, with a focus on the domain module. We consider that the domain model represents the most important module of the ITS because it contains the knowledge that should be taught and evaluated. Based on this module, the rest of the modules will make decisions.
Similar content being viewed by others
Keywords
1 Introduction
An Intelligent Tutoring System (ITS) is a computer-based teaching-learning system that uses Artificial Intelligence techniques to interact with students to teach them [7]. Currently, ITSs provides effective tracking of the teaching process as it offers personalized tutoring. ITSs offer the following advantages [21]: (1) Constitutes a source of teaching materials. (2) Provides problems for the student to reach a certain level of knowledge. (3) Controls the level of difficulty of the problems so that the students face exercises appropriate to their needs. (4) Helps the student to acquire additional knowledge.
The use of ITS in virtual environments has been shown to improve the teaching-learning process [14]. However, ITS is still not common in classrooms due to time constraints [8], to the intense work [22], to the requirement of experts in both intelligent tutors and in the subject to be developed.
To avoid the above problems, ITS authoring tools have been developed. The aim of this tool is to simplify the development of ITS, to increase the number and diversity of available tutors. It also reduces the complexity of constructing them. Authoring tools also allow rapid ITS prototype design [4, 11]. Accomplishing these objectives will assist ITS developers and users who do not have programming skills construct ITSs [11]. Various components of ITS require different techniques in assisting authoring [6]. As a result, the development of ITS authoring tools has progressed slowly [24].
Literature shows different uses for ITS authoring tools [4], however, little research has focused on the domain module [2]. This module is important because it establishes the bases of knowledge that will be taught and evaluated.
The objective of this research is to establish the framework and foundations in developing an authoring tool to build a domain module that serves as an information base for an ITS. The specific objectives of the research are: To design a generalized model to build the ITS domain module automatically, build an internal representation of the ITS knowledge domain, design an algorithm to determine student knowledge based on representation internal, and build a module of queries on the internal representation to answer and increase knowledge.
This paper is organized as follows: Sect. 2 describes related work. Section 3 shows the topic background. Section 4 displays the methodology. The Sect. 5 show conclusions and references.
2 Related Work
The study of the authoring tools for ITS has been investigated by various authors. Dermeval et al. [4] conducted a systematic literature review (from January 2009 to June 2016) finding 33 articles related to the authoring tools for ITS. Most of these articles focused on the pedagogical model and the domain model. Our research focuses on the domain model.
Dermeval et al. identified features offered by the authoring tools, dividing contributions by ITS module. They used the following features related to the domain module: Definition of problem solutions, authoring by demonstration, automatic domain model generation, definition of hints, reuse of learning content, and human computation.
Our research focuses on working the automatic domain model generation, for which the following investigations were detected:
-
Brawner [2] developed Tools for Rapid Automated Development of Expert Models (TRADEM). TRADEM uses a domain model built as a summarization of provided content mixed into a set of topics, as a part of the Generalized Intelligent Framework for Tutoring (GIFT) Domain Module [20]. The purpose of the TRADEM project has been to rapidly and mostly-automatically create expert models and sequence domain material from initially provided texts. The information is given by the user so that TRADEM builds the knowledge domain. There are benefits in using the TRADEM tool, including aiding in front end analysis of content, automatically summarizing existing documents, providing the foundation of a course. His project is very good, and has the same basis as our proposal. It automatically builds the knowledge domain of the ITS. However, our research aims to build a representation of knowledge based on nodes of concepts with the aim of inferring and evaluating knowledge.
-
Matsuda et al. [6] developed SimStudent. This is a machine-learning agent initially developed to help novice authors to create cognitive tutors without heavy programming. SimStudent helps authors to create an expert model for a cognitive tutor by tutoring SimStudent in solving problems. The expert model represents a how-type of knowledge about the domain. It assumes that the prospective users of the SimStudent authoring system are domain experts who, by definition, know how to solve problems and can identify errors. They consider the expert model as a module that presents problems and provides advice or support in solving them. Matsuda’s work in this article, does not focus on Brawner’s research and our proposal.
-
Mitrovic et al. [8] developed ASPIRE-Author. ASPIRE-Author supports the authoring of the domain model, in which the author is required to provide a high-level description of the domain, as well as examples of problems and their solutions. ASPIRE-Author goal is to reduce the time and effort required for producing ITSs by building an authoring system that can generate the domain model with the assistance of a domain expert and produce a functional system. It describes the processes of the system in seven steps: 1. Specify the domain characteristics 2. Compose the domain ontology 3. Model the problem and solution structures 4. Design the student interface 5. Add problems and solutions 6. Generate syntax constraints 7. Generate semantic constraints 8. Deploy the tutoring system. Step two builds an ontology that represents a hierarchy of concepts, however, it was expected that in later steps this structure would be used to develop the content, this is not the case. In this article, they do not develop concepts of the ontology.
-
Suraweera et al. [22] developed Constraint Authoring System (CAS). CAS was developed to generate the knowledge required for constraint-based tutoring systems, which reduced the effort and the amount of engineering knowledge and programming expertise required. ITS authoring using CAS is a semi-automated process requiring the assistance of a domain expert who initiates the process by modelling the domain as an ontology. The experts then defines the form that solutions will take for problems in this domain using concepts from the ontology. CAS’s constraint-generation algorithms use both the ontology and the solution structure to produce constraints that verify the syntactic validity of solutions. The domain expert is also required to provide sample problems and their solutions, which are used by CAS generators to produce semantic constraints. These verify that a student’s solution is semantically equivalent to a correct solution to the problem. Finally, the author has the opportunity to validate the generated constraints. This work seems to be a continuation of the project of [8], since it is a co-author of Saraweera’s work and basically it is the same proposal.
In another paper outside the research of [4], like the research of Romero et al. [19]. They developed Hedea, a tool for the construction of ITS. This article presents preliminary results obtained by an authoring tool that allows a non-expert teacher in the area of intelligent tutors to develop an ITS based on the definition of a course. This research builds the knowledge representation structure that we want to build, however, Romero et al. builds the network based on the hierarchical breakdown of topics, sub-topics, and concepts. The development of the theory of topics is not addressed in the research.
The work of [2] is the closest to our proposal, since the research proposes to develop the domain model of an ITS, which displays the information or knowledge that a student must have. In addition, our proposal includes turning this knowledge into a network of concepts which allows evaluating the student’s knowledge. This same structure should allow summarize information, and infer knowledge.
3 Background
3.1 Intelligent Tutoring Systems
ITS are computer-based learning systems, which were designed to impart knowledge guiding the student in the learning process through some form of intelligence. These systems exhibit a similar behavior to a human tutor and assists the student with cognitive help. The system adapts to the student’s behavior, identifying the way to solve a problem to offer help when needed [21].
There is a generalized architecture for ITS that considers four basic modules [3]:
-
The tutor module generates learning interactions based on the student’s learning difficulties.
-
The domain module defines the area of knowledge that the ITS teaches.
-
The student module defines the student’s knowledge in the work session.
-
The interface module allows the student to interact with the ITS.
Resources needed to build an ITS come from multiple research fields, including artificial intelligence, the cognitive sciences, education, human-computer interaction, and software engineering. This multidisciplinary foundation makes the process of building an ITS a thoroughly challenging task, given that authors may have very different views of the targeted system [13].
3.2 Authoring Tools for ITS
For several decades developers and researchers have been investigating the possibilities for creating ITS authoring tools with the aim for: (1) reduce the effort and cost of building or customizing ITS, and (2) allow non-programmers, including teachers and domain experts, to participate fully or partly in building or customizing ITS [10].
Authoring tools streamline and accelerate the construction of ITS by providing a framework within which an author can design a learning system. Some authoring systems are general-purpose tools that provide an author with a great deal of leeway. Others embody a set of assumptions about what the authored product will look like and how it will behave. [1]
Murray [9] classifies ITS authoring tools into two main groups:
-
Pedagogy-oriented systems focus on instructional planning and teaching strategies, assuming that the instructional content will be fairly simple. Such systems provide support for curriculum sequencing and planning, authoring tutoring strategies, composing multiple knowledge-types and authoring adaptive hypermedia.
-
performance-oriented systems focus on providing rich learning environments where students learn by solving problems and receiving dynamic feedback. The systems in this category include authoring systems for domain expert systems, simulation-based learning and some special purpose authoring systems focusing on performance.
Woolf et al. [24] proposed a framework for organizing the necessary building blocks found in authoring systems for building ITS. The author identified four layers, each including specific classes of building blocks:
-
Level 1: The knowledge representation level includes tools for easily representing knowledge, the user should adopt the right formalism and select the right language or tool to ease the representation process (Semantic net, constrains, productions rules, frames).
-
Level 2: It is about the type of domain and student models (procedural skills, student effect, student misconceptions).
-
Level 3: Contains tools for implementing teaching knowledge while (content planning, delivery planning, tutoring decision).
-
Level 4: Comprises those for communication knowledge (interface design, pedagogical agent, natural language dialog).
4 Methodology
4.1 Overview of the Project
An ITS is considered a complex system to build, it is necessary to divide the problem for a simpler solution. Thus, the current article focuses on building the domain module. We consider that the domain model represents the most important module of the ITS because it contains the knowledge that should be taught and evaluated. Based on this module, the rest of the modules will take the decision to act as a teacher. Our framework considers the following phases to build a domain module:
-
To design a generalized model to build the ITS domain module automatically: Initially, a general view of the architecture of the proposed system should be proposed, with the aim of defining the scope of the project and specifying the modules that will be developed.
-
To Build an internal representation of the knowledge domain of the ITS: Each topic of the content must form an internal representation, which not only serves to represent the content but also serves to evaluate knowledge, consult and infer information.
-
To design an algorithm that allows evaluating the student’s knowledge based on internal representation: An ITS must be able to evaluate the student’s knowledge, to propose support or reinforcement issues. There are some techniques in the literature that will serve as a base, however, the proposal must solve the deficiencies of the current techniques.
-
To build a query module on internal representation to answer queries and infer knowledge: information on topics must be processed and converted to a structure that allows queries about the information. In addition, the structure must infer knowledge to adequately respond to queries.
-
To develop an application for the knowledge domain of the ITS: Develop a web application that allows generating knowledge domains automatically, that is, to develop topics searching the content in information sources.
-
To implement the internal representation, the evaluation, and query algorithms: Implement algorithms in the web content generation system to build the internal representation, incorporate the knowledge evaluation algorithm, and the query and inference algorithm.
Figure 1 represents the architecture of an ITS, which considers the 4 basic modules: tutoring, student, domain, and interface. However, the scope of this project focuses on the domain module, for this reason, this module is more developed than the rest of modules. Basically, the domain model will be built with Internet documents or e-books, from these, an internal structure that represents the contents will be created to make queries to the information and knowledge inference. It is contemplated that this representation serves to evaluate the student’s knowledge.
4.2 Internal Representation of Knowledge
A computer-based education system teaches what is known as “knowledge domain”. When represented in a way that a computer understands, it is called “Knowledge Representation” (KR). It is important to consider the KR processes and automates information management through computers. In addition, makes inferences that allow decision-making in a human manner in order to improve the tutoring task [16].
Authors such as [17, 18] establish that in the human semantic memory exists a hierarchy of concepts with relations to organize this knowledge. Thus, arises the idea of representing knowledge by means of graphs. There are several techniques to represent this knowledge [16], however, our proposal establishes a combination of Bayesian networks and ontologies. The Bayesian network is a graphical probabilistic model which represents knowledge through nodes and relations [15]. On the other hand, ontology is a representative model of semantic knowledge [5].
We will consider elements which represent knowledge in an educational environment based on [16] as follows:
-
Concepts: A concept is an elemental piece of knowledge. According to the domain expert, it cannot be divided into smaller parts. Therefore, a concept is considered the primary unit of knowledge.
-
Skills: A skill is a cognitive process that interacts with one or more concepts, usually through an application. It has a particular purpose and produces an internal or external result.
-
Relations: Relations are known as links. The relations goal is to know how the concepts are related. Relations can be of three types according to the link direction. These are Unidirectional, bidirectional, and loops.
-
Inference: This is also called reasoning. The inference refers to obtaining deductions or conclusions based on knowledge already established. The main types of inference are abductive, deductive, analogy, and inductive.
-
Type of Knowledge: There are two types of knowledge: (1) Procedural Knowledge focuses on tasks that must be performed to reach a particular objective or goal. (2) Declarative knowledge refers to the representation of objects and events, and about knowledge facts and its relations. It is the knowing whether “something is true or false”. Declarative knowledge is applied in educational institutions. It is easy to represent and structure, so it is the kind of knowledge that is taught by computer-aided systems.
-
Hierarchical Structure: Information for these kind of techniques is organized through a hierarchical structure which forms superclasses, subclasses and shared properties. Based on the hierarchical structure, some elements are considered by different authors as important because they establish a structure which maintains the contents. These elements are: Successors, predecessors, classes, inheritance, among others.
The proposed model considers constructing the nodes based on the textual information of the knowledge domain, forming a structure that can be exploited based on statistical and semantic techniques. Some features considered for the domain model are:
-
The structure of the topics, can be given by an index in a hierarchical form, as an ontology, or a topic based on the breakdown of topics of encyclopedias as is the case of Wikipedia.
-
Important concepts: These concepts are taken based on textual information. Various techniques have been used to consider this concept, such as links, the frequency of the concept, coincidence with title concepts, among others.
-
A concept can have relations with other topics, thereby inheriting the properties of the concept with other concepts or related topics.
-
Relations: Illustrates between concepts can be of two types: (1) Cause-effect: Determines a weight of relation between two concepts. It is useful in determining the knowledge of the student. (2) Semantics: Regarding the meaning relations that will have a label as is_part_of, is_a, kind_of, among others.
The domain model to be developed serves as the basis for the student and tutoring module with aspects such as:
-
Learned topics: A topic learned by the use of examinations and other reinforcement activities.
-
Inference of knowledge: The structure generated can make inference of knowledge to give explanations based on a feedback system.
-
Probability propagation algorithms: The probability propagation algorithms will update the status of concepts, applying forward and backward propagation.
There are different investigations that have addressed this issue, such as [12, 23, 25, 26]. Therefore, the current research has a solid background, which will allow us to take up ideas from other projects.
5 Conclusions
The literature analysis allowed to determine that there is little research to build the domain module automatically, hence, the proposal in this article establishes a framework to build the domain module of an ITS based on text. The proposal considers forming a map of concepts that represents the knowledge that a student must have, using algorithms to evaluate their knowledge, and make the inference of the information to answer questions or feedback automatically.
The project, in general, considers developing a software tool which builds an ITS automatically, by considering the four basic modules: Student, tutoring, domain, and interface. For the scope of this article only the framework of the domain module is established, and the base literature that will be used.
References
Bell, B.: One-size-fits-some: its genres and what they (should) tell us about authoring tools. In: Sottilare, R.A., Graesser, A., Xiangen, H., Brawner, K. (eds.) Design Recommendations for Intelligent Tutoring Systems: Authoring Tools and Expert Modeling Techniques, chap. 3, pp. 31–45 (2015)
Brawner, K.: Rapid dialogue and branching tutors. In: CEUR Workshop Proceedings, vol. 1432, pp. 69–76 (2015)
Carbonell, J.R.: AI in CAI: an artificial intelligence approach to computer assisted instruction. IEEE Trans. Man Mach. Syst. 11, 190–202 (1970)
Dermeval, D., Paiva, R., Ibert, B., Vassileva, J., Borges, D.: Authoring tools for designing intelligent tutoring systems : a systematic review of the literature. Int. J. Artif. Intell. Educ. 28, 336–384 (2017)
Gruber, T.R.: Toward principles for the design of ontologies used for knowledge sharing. Int. J. Hum.-Comput. Stud. 43(5–6), 907–928 (1995). citeulike-article-id:230211
Matsuda, N., Cohen, W.W., Koedinger, K.R.: Teaching the teacher: tutoring simstudent leads to more effective cognitive tutor authoring. Int. J. Artif. Intell. Educ. 25(1), 1–34 (2015). https://doi.org/10.1007/s40593-014-0020-1
Millán, E., Pérez-De-La-Cruz, J.L.: A Bayesian diagnostic algorithm for student modeling and its evaluation. User Model. User-Adapt. Interact. 12, 281–330 (2002). https://doi.org/10.1023/A:1015027822614
Mitrovic, A., Martin, B., Suraweera, P., Zakharov, K., Milik, N., Holland, J., McGuigan, N.: ASPIRE: an authoring system and deployment environment for constraint-based tutors. Int. J. Artif. Intell. Educ. 19(2), 155–188 (2009)
Murray, T.: An overview of intelligent tutoring system authoring tools: updated analysis of the state of the art. In: Authoring Tools for Advanced Technology Learning Environments, pp. 491–544 (2003). https://doi.org/10.1007/978-94-017-0819-7-17
Murray, T.: Theory-based authoring tool design: considering the complexity of tasks and mental models. In: Sottilare, R.A., Graesser, A., Xiangen, H., Brawner, K. (eds.) Design Recommendations for Intelligent Tutoring Systems: Authoring Tools and Expert Modeling Techniques, chap. 2, pp. 9–30 (2015)
Naser, S.S.A.: ITSB : an intelligent tutoring system authoring tool. J. Sci. Eng. Res. 3(5), 63–71 (2016)
Niño Zambrano, M., Jimena Pérez, D., Pezo, D., Cobos Lozada, C., Ramírez González, G.: Procedure for building semantic indexes based on domain-specific ontologies. Entramado 9(1), 262–287 (2013)
Nkambou, R., Bourdeau, J., Psyché, V.: Building intelligent tutoring systems: an overview. In: Nkambou, R., Bourdeau, J., Mizoguchi, R. (eds.) Advances in Intelligent Tutoring Systems. Studies in Computational Intelligence, chap. 18, pp. 361–375. Springer, Heidelberg (2010). https://doi.org/10.1007/978-3-642-14363-218
Noguez, J., Sucar, L.E.: A probabilistic relational student model for virtual laboratories. In: Sixth Mexican International Conference on Computer Science (ENC 2005), pp. 2–9, September 2005. https://doi.org/10.1109/ENC.2005.7
Pearl, J.: Probabilistic Reasoning in Intelligent Systems: Networks of Plausible Inference. Morgan Kaufmann Publishers Inc., San Mateo (1988)
Ramírez-Noriega, A., Juárez-Ramírez, R., Jiménez, S., Martínez-Ramírez, Y.: Knowledge representation in intelligent tutoring system. In: Hassanien, A.E., Shaalan, K., Gaber, T., Azar, A.T., Tolba, M.F. (eds.) Proceedings of the International Conference on Advanced Intelligent Systems and Informatics 2016, pp. 12–21. Springer International Publishing, Egypt (2017). https://doi.org/10.1007/978-3-319-48308-5_2
Rivas Navarro, M.: Procesos cognitivos y aprendizaje significativo. BOCM, Madrid. Servicio de Documentación y Publicaciones (2008)
Rodríguez, R.J.: Herramientas informáticas para la representación del conocimiento software. Subj. Pocesos Cogn. 14(1712), 217–232 (2010)
Romero Inzunza, M.A., Sucar Succar, E., Gómez-gil, P.: Diseño De Hedea: Una Herramienta Para La Construcción De Sistemas Tutores Inteligentes. Memorias del Décimo Encuentro de Investigación. Instituto Nacional de Astrofísica, Óptica y Electrónica, pp. 179–182, Puebla (2009)
Sottilare, R.A., Brawner, K.W., Goldberg, B.S., Holden, H.K.: The Generalized Intelligent Framework for Tutoring (GIFT). US Army Research Laboratory–Human Research & Engineering Directorate (ARL-HRED), Orlando, FL, 1–12 October 2012. https://doi.org/10.13140/2.1.1629.6003
Suárez Granados, J.J., Arencibia Rodríguez del Rey, Y., Pérez Fernández, A.C.: Methodology for developing an intelligent tutoring system based on the web, for engineering students. Universidad y Sociedad 8(4), 108–115 (2016)
Suraweera, P., Mitrovic, A., Martin, B.: Widening the knowledge acquisition bottleneck for constraint-based tutors. Int. J. Artif. Intell. Educ. 20(2), 137–173 (2010). https://doi.org/10.3233/JAI-2010-0005
Toledo-Alvarado, J.I., Guzmán-Arenas, A., Martínez-Luna, G.L.: Automatic building of an ontology from a corpus of text documents using data mining tools. J. Appl. Res. Technol. 10(3), 398–404 (2012)
Woolf, B.P.: Building Intelligent Interactive Tutors: Student-centered Strategies for Revolutionizing e-Learning. Morgan Kaufmann Publishers Inc., San Francisco (2007)
Yarushkina, N., Filippov, A., Moshkin, V., Egorov, Y.: Building a domain ontology in the process of linguistic analysis of text resources. Preprints 2018, pp. 1–10 (2018). https://doi.org/10.20944/preprints201802.0001.v1
Zouaq, A., Nkambou, R.: Building domain ontologies from text for educational purposes. IEEE Trans. Learn. Technol. 1(1), 49–62 (2008). https://doi.org/10.1109/TLT.2008.12
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2019 Springer Nature Switzerland AG
About this paper
Cite this paper
Ramírez-Noriega, A. et al. (2019). Towards the Automatic Construction of an Intelligent Tutoring System: Domain Module. In: Rocha, Á., Adeli, H., Reis, L., Costanzo, S. (eds) New Knowledge in Information Systems and Technologies. WorldCIST'19 2019. Advances in Intelligent Systems and Computing, vol 930. Springer, Cham. https://doi.org/10.1007/978-3-030-16181-1_28
Download citation
DOI: https://doi.org/10.1007/978-3-030-16181-1_28
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-030-16180-4
Online ISBN: 978-3-030-16181-1
eBook Packages: Intelligent Technologies and RoboticsIntelligent Technologies and Robotics (R0)