Decision Support System

Published on December 2016 | Categories: Documents | Downloads: 33 | Comments: 0 | Views: 203
of 6
Download PDF   Embed   Report

Decision Support System

Comments

Content

A Knowledge-Driven Educational Decision Support System
Vo Thi Ngoc Chau
Faculty of Computer Science and Engineering Ho Chi Minh City University of Technology Ho Chi Minh City, Vietnam [email protected]

Nguyen Hua Phung
Faculty of Computer Science and Engineering Ho Chi Minh City University of Technology Ho Chi Minh City, Vietnam [email protected]

Abstract—Education always plays an important role in building up every country around the world. Hence, educational decision making support is significant to students, educators, and educational organizations. The support will be more valuable if a lot of relevant data and knowledge mined from data are available for educational managers in their decision making process. Nevertheless, educational decision support system development is non-trivial and different from organization to organization due to the peculiar features of each educational organization. Besides, an academic credit system is nowadays very widely-used in many educational organizations. Due to the flexibility of a credit system of education, applying data mining techniques to educational data for knowledge discovery is challenging. In this paper, we propose a knowledge-driven educational decision support system for education with a semester credit system by taking advantage of educational data mining. During system development, we have figured out that knowledge discovery from educational data in a semester credit system is full of many problems remaining unsolved with a semester credit system. Tackling those problems, our resulting system can provide educational managers with actionable knowledge discovered from educational data. Such knowledge-driven decision support is helpful for educational managers to make more appropriate and reasonable decisions about student’s study and further give support to students for their graduation. Above all, a waste of effort, time, and money can be avoided accordingly for both students and educators. Keywords-educational decision support system; knowledgedriven decision support system; data-driven decision support system; educational data mining; actionable knowledge

the environment where the system is deployed and run. This implies that an EDSS is specific for each educational organization. According to the taxonomy of decision support systems (DSSs) reviewed in [12], there are five types of DSSs: communication-driven, data-driven, document-driven, knowledge-driven, and model-driven. In our work, we focused on knowledge-driven DSSs capable of providing specialized expertise and information (knowledge) for undertaking specific decision making problems by means of various data mining techniques. This is because we believe that decision making is a knowledge-intensive endeavor as stated in [10]. Therefore, a DSS should provide users with problem-relevant knowledge as much as possible. Up to now, educational decision making support has been taken into consideration in several works [6, 8, 17, 18, 19, 20]. References [17, 18] delineated the benefits of an EDSS along with the difficulties that might have faced in such system development. References [8, 19] mainly supported decisions about administrative activities and resource planning policies in their educational organizations. References [6, 20] supported academic decisions about learning activities in a traditional learning environment and an e-learning environment, respectively. Unfortunately, knowledge-driven educational decision support systems were little examined for education of regular undergraduate students with a semester credit system in the existing works. This shortage needs to be considered as a credit system of education is nowadays employed in a very large number of educational organizations all over the world. In this paper, we first figure out the challenging issues emerging with the flexibility of a semester credit system that have not yet been mentioned in the related works. Then, we propose a knowledge-driven educational decision support system by taking advantage of educational data mining for education of regular undergraduate students with a semester credit system. The proposed system is expected to provide the educational managers with problem-relevant data and knowledge prepared by means of data management and visualization, classification, clustering, and association analysis. Using the proposed system, the educational managers can have data and knowledge relevant to the problem they are considering as to student’s study. Together with their own experiences and knowledge, these data and knowledge will be

I.

INTRODUCTION

Regarding human resource development, education always plays an important role in building up every country all over the world along the time. Educational decisions made by educational managers are important and have a strong impact on not only individual students and educators but also our society. If an educational decision were inappropriate, a lot of student and educator’s effort, time, and money would be wasted and bad outcomes would be produced and exist together with our lives for a long time. Thus, it is worth investigating a decision support system (DSS) in the educational application domain which is called educational decision support system (EDSS). Besides, a DSS is a practical application that is tightly associated with the characteristics of

978-1-4673-0309-5/12/$31.00 ©2012 IEEE

useful evidences that support them in producing more appropriate and reasonable decisions in educational management. Once the decisions are effective, a lot of effort, time, and money can be saved for both students and educators. II. RELATED WORKS

trends, etc. and further discover the reasons for decisions about policies and selected processes. In addition to the systems in [8, 19], the system proposed in [20] was developed to support lecturers in distance education, not aiming at administrative support. Reference [20] took advantage of data warehousing technology and open source data mining library to discover who students were, how students worked, and how students used virtual courses. Among the existing EDSSs, the system from our proposed work for knowledge-driven educational decision making support is somewhat similar to the system in [20]. Nevertheless, the system in [20] was developed for distance learning and data mining techniques were applied to individual courses, e.g. “Introduction to multimedia methods” course. Further, [20] did not consider the flexibility of a credit system. B. Related Works on Educational Data Mining Data about studying and lecturing activities in educational organizations have been accumulated day by day over the time line. If this volume of educational data is explored and knowledge discovered from such data is utilized, studying and lecturing activities must be improved and be of much benefit to students, lecturers, educational organizations, and society. So far, [2, 4] have shown an overview on this active educational data mining area. Although of an applied field of the data mining research arena, educational data mining encountered many challenges stemming from the peculiar features of the educational application domain. Many research works on educational data mining have recently been proposed in [3, 5, 7, 11, 14, 16]. These works used educational data gathered from many different education levels: [7] at the pre-university level, [3, 5, 11, 16] at the undergraduate level, and [14] at the graduate level. Moreover, they employed a diversity of data analysis and mining techniques, for example: [3, 7, 14] used statistical analysis and [5, 11, 14, 16] classification. Most of them used a data set gathered through a web-based learning system for a not very long period of time. It is also realized that most of them just examined the knowledge that students obtained from a single course or a few courses, not from all courses of an entire educational programme. Their approach did not comprehensively pay attention to the knowledge that students obtained successfully and the knowledge that students obtained unsuccessfully and needed to study more. In addition, when educational data have been collected for a long time, an educational programme might have some changes according to the educational policies and the peculiar features in an educational organization. However, there was no mention of this problem in the existing works. The contents and requirements of each course were fixed for both data sets and data analysis/mining. The entire educational programme was also supposed to be fixed. III. KNOWLEDGE-DRIVEN EDUCATIONAL DECISION MAKING SUPPORT WITH AN ACADEMIC CREDIT SYSTEM For knowledge-driven educational decision making support with an academic credit system, we first address the difficulties in educational knowledge discovery stemming from the

In this section, we present an overall view of several related works on educational decision support systems and educational data mining in comparison with our work. A. Related Works on Educational Decision Support Systems Among the existing works, [6, 8, 17, 18, 19, 20] are several examples about EDSS development. None of them was a real knowledge-driven EDSS. As one of the first EDSSs, the work introduced in [17, 18] presented the advantages and disadvantages of an educational decision support system as well as of the development of such a system. References [17, 18] also summarized a few types of educational decisions including programmed to nonprogrammed decisions. Moreover, several types of educational information are also determined. References [17, 18] confirmed that the results from these systems were helpful in education. Nevertheless, when their work was proposed, many difficulties existed and had an impact on the development of an EDSS. Those difficulties were data sharing among education levels and among educational associations, information technologies, training problems, an improvement of education management, and the problems of supporting unstructured/nonprogrammed decisions. Anyway, [17, 18] help us take into account the potential problems that might be faced before developing such an EDSS. Recently, [6] has presented an EDSS to provide some student’s result analysis utilities with tables and statistical charts. As shown in [6], this system helped determining whether learning objectives had been achieved through data analysis on student’s learning results including entrance records, the ratio between the number of required semesters and the number of real semesters that students had taken, and the grade point averages. We noted that [6] just provided the simple utilities for presenting the results from statistical analysis as well as considered an overall result of student’s study. Therefore, if a student did not achieve a learning objective, it would be hard to identify what knowledge the student had not gained in particular. Different from [6], [19] developed an EDSS to support strategic planning about educational resources, about distribution and usage of those resources at universities. Educational resources mentioned in [19] consisted of services about courses, supervision, etc. that students used in their study. The result of this system included reports used for decisions making support containing input data and temporary outputs simulating strategic plans about educational resources. In fact, [19] was related to decisions making support for administration rather than for learning-teaching activities in an educational organization. Also for administration, [8] presented an EDSS that made use of educational data exploration to help users identify important factors about volume, distribution, and

flexibility of the credit system. These difficulties have not yet mentioned in any existing related works although significantly considered for knowledge-driven EDSS development. After that, we analyze an intelligence phase of a decision making process on a typical problem in education. The result of this section will help us determine what knowledge should be discovered and provided by a knowledge-driven EDSS. As introduced earlier, we expect actionable knowledge that can enable educational managers to take an action. A. Difficulties in Educational Knowledge Discovery from the Flexibility of a Credit System of Education An academic credit system is very widely employed in many educational organizations all over the world. It is wellknown for the flexibility in the learning-teaching process and for the capability of developing personality and creativity of students. In such a system, students can choose their own learning path to accomplish an educational programme (i.e. to pass every course and to obtain the required total number of credits) within a permitted time-frame. Nevertheless, we have found some major difficulties in knowledge discovery from educational data gathered for a long time because of the flexibility of a credit system of education. Emphasizing the heterogeneity of an educational data set, these difficulties are pointed out as follows. Firstly, an educational programme probably changes after a period of time. Changes create many different versions of the educational programme. Each version is normally used for students enrolled during a few years so that the educational programme studied by students enrolled in a year can be different from one by students enrolled in another year. So, mining models/patterns from the educational data set gathered in the past might not be scored correctly to the current data. Secondly, if the educational data set used to build mining models/patterns is related to only part of the educational programme the students need to study, the courses taken by each student might be different from those by others. This is because each student can register for a course up to their own situation. For example, we would like to classify all third year students who have studied for six main semesters based on their performance of all courses studied so far. In the training data set, a student might pass 16 courses and another student 17 courses, etc. This case can also take place in the test data set. Thirdly, it is essential to consider comprehensively both knowledge of the courses students have successfully obtained and knowledge of the courses students will obtain when we have to make a decision related to the study performance of a student. The knowledge of the courses students will obtain is the missing part the students need to fulfill to complete their study. Nevertheless, it is hard to interpret if the missing part is unreachable to the students because of their poor study performance or because they intend to leave this missing part for a while. It is also possible that the missing part is intended to be left by the students. Such interpretations ask us to get understood more about students’ capability. Fourthly, the last difficulty that we would like to present is the volume and nature of an educational data set. Although education started an extremely long time ago, the volume of an

educational data set that we can gather so far is rather small in reality. Also, the nature of an educational data set requires the expertise of educators (educational managers) in an educational organization for data understanding. As of this moment, there is no practice about data preprocessing on educational data sets as well as no benchmark in educational data mining. All experiments in the existing works were done on the educational data sets specific to their educational organization. As a knowledge-driven EDSS is mainly based on the result of the data mining process on educational data, tackling the aforementioned difficulties will greatly help us make an appropriate data preparation for discovering actionable knowledge. Furthermore, it is realized that the difficulties stated above are useful for any data analysis and mining on educational data from a credit system of education which is ubiquitous nowadays. B. Solutions to an Intelligence Phase in the KnowledgeDriven Educational Decision Making Process In this subsection, we introduce a typical problem of students with poor study performance in any educational organization. Supporting decision making on this problem, we analyze the corresponding intelligence phase so that we can define which knowledge needs to be discovered from educational data. We then design the appropriate data mining functionalities of the proposed knowledge-driven EDSS. The problem is described with a semester credit system of education as follows. Every semester there is a list of students to be warned for poor study performance or to be forced to stop their study for many various reasons one of which is also poor study performance. The rationale behind this policy is that the entire study of each student is viewed as a cumulative process of obtaining knowledge from all courses specified in an educational programme. Poor study performance at some point in time in the process implies that the student has not yet got enough knowledge to move forward to their successful graduation and the remaining knowledge the student needs to study more is too much for the student to be able to obtain in the rest of permitted time. Therefore, the educational managers have to make a decision about who could continue the study from the list. For those who could continue their study, the educational managers will give them another chance in one extra semester so that the students can improve their study and then get out of the list next semester. If a student is capable of keeping their study, the chance given to the student is really appreciated for the student to successfully accomplish the entire educational programme for graduation. Otherwise, the chance given to the student will be wasted in terms of student and educator’s effort, time, and money. The intelligence phase in solving this problem is now carried out by the educational managers. It is hard for them to manually manage and analyze the data as well as consume reports about the teaching-learning activities. It will be helpful if problem-relevant data and knowledge are provided to the educational managers as much as possible on demand so that they can detect those students, who will potentially accomplish the entire programme successfully. On the other hand, if not able to follow the educational programme, the students can have more chance to enter another university and then to succeed in other study

direction and life. This is because we consider the difference between two students who do not succeed in their studying an existing educational programme: one student who stops their study early in the first year in order to still have more time and inspiration in their life for study and another student who stops their study late in the fourth year. Therefore, a knowledgedriven educational decision support is valuable if provided for the educational managers. This situation is summed up with determining the students who should be offered a chance to study one semester more to improve their study result and the students who should not among a group of students that are forced to stop their current study due to poor study performance. The students being considered are not in special cases of the educational policies for wounded soldier and revolutionary martyr’s families, ethnic minority, etc. In other words, this problem is related to a decision on study extension of students with poor study performance. According to the decision making process in [12], the intelligence phase of the problem in our work is conducted by answering the following questions in form of what-if. Based on this intelligence phase, we have a background to determine what information and knowledge the proposed system should offer the educational managers as specified in Fig. 1.

As for a solution to the second issue, we employ data clustering to derive the study result of the student using the study result of the student in the past who is the most similar to the student being considered in terms of accumulated knowledge and credits from courses they took. The study result can let us know if the student’s study trend goes up or down. Besides, clustering is also used to provide us with the probability of each status the student might have every semester during the period of time studying. If the probability of a positive status is equal to or greater than 0.5, it can be supposed that the student can improve their study and it is worth giving the student a chance to keep studying. 2) If a student with poor study performance is certainly capable of studying and completing the program successfully, what will the student study next semester to guarantee a positive performance improvement? This question follows the previous question as it is assumed that a student with poor study performance will study next semester. So, the student’s course registration will have a strong impact on his/her success next semester. With that success, he/she can keep studying the semester after next more confidently and hopefully succeed in that semester. Gradually he/she can finish the program successfully. That means the chance given is really significant to the student and therefore, to our educators. In order to supply required knowledge for the second question as to course registration, we apply course association analysis on course registrations of all students who have the most similar characteristics to the student being considered to find out which groups of courses the students never pass when studying them together. From the resulting groups of courses, the course registration of the student being considered should not consist of any group of those courses in its entirety; however, maybe part of some group. In short, we hopefully reach our ultimate goal to provide the educational managers with valuable actionable knowledge about study performance of the students and further support the educational decisions for more effectiveness and timeliness. As a result, students will be enabled to finish the entire program successfully if supposed to continue their study. Otherwise, students will be suggested that their current study be unsuitable for their capability and they should stop their study as soon as possible to switch to another study direction for a certain success in their career. We believe that such decisions will save much time, cost, and effort of both students and educators. Further, the number of successful students will increase and so will the number of successful young people in our society. THE PROPOSED KNOWLEDGE-DRIVEN EDUCATIONAL DECISION SUPPORT SYSTEM In this section, our proposed knowledge-driven educational decision support system is elaborated. For knowledge-driven decision making support, the proposed system makes the most of data mining techniques, Web and database technologies. IV. A. Details of the Proposed System Supporting the aforementioned problem with relevant data and knowledge required by the educational managers, we develop a knowledge-driven educational decision support system. At present, key users of the system are the educational

Figure 1. The intelligence phase of the problem on study extension with required knowledge discovered from an educational database

1) If a student with poor performance continues studying, what will happen? This question includes two issues. The first issue is to predict whether a student can accomplish the program successfully based on the existing knowledge the student has accumulated up to now. The second issue is “what” part in case “if” part is evaluated to be true. It is related to the study result of the student in the extended semester and the influences of this study result on student’s study in the remaining semesters. Our solution to the first issue is building a classifier to classify all students into the seven groups such as “complete”, “incomplete”, “drop-out”, “first warning”, “second warning”, “extended”, and “other”; and then classifying the student being considered into the most suitable group. This implies that we apply data classification task to the first issue.

managers. In the future, the system is going to be extended to support both students and educators with different levels of education. Architecture of the proposed knowledge-driven EDSS is presented in Fig. 2. It is based on a typical three-tier component-based architecture: presentation tier, business logic tier, and storage tier.

Figure 2. Architecture of the proposed knowledge-driven educational decision support system

1) Presentation tier: The presentation tier is a web interface. It is developed with Java technologies. The choice of a web interface makes our system more convenient for the educational managers as they can have access to the system at any time on any machine where a web browser is available. 2) Business logic tier: The business logic tier is composed of four main components: Knowledge-Driven Educational Decision Supporter, Data Mining Engine, Raw Data Preprocessor, and Data Management Engine. KnowledgeDriven Educational Decision Supporter component is of our own also developed with Java technologies while Data Mining Engine component is based on open source Weka machine learning library, Raw Data Preprocessor component is based on open source Pentaho Data Integration suite, and Data Management Engine component is based on open source MySQL database management system. - Knowledge-Driven Educational Decision Supporter: this component is the heart of our proposed system. It is used to provide the educational managers with problem-relevant data and knowledge. As of this moment, the functions of this component are towards the end-users of the system, i.e. the educational managers, containing data retrieval and visualization, actionable knowledge discovery, postprocessing of discovered knowledge. - Data Mining Engine: this component is responsible for turning our raw data into required knowledge. It is an essential component lying underneath Knowledge-Driven Educational Decision Supporter component. For the current version of this system, we make use of classification, clustering, and association analysis algorithms in the Weka library. - Raw Data Preprocessor: this component is regarded as the interface of our system with other information systems to feed our raw educational data to the proposed system. In this current version of the system, raw data are kept in Excel files and the extracting, transforming, and loading process is done on demand via Web using this component. Integrity constraints are also checked on the spot before the raw data enter the educational database of the system. As our

educational data incrementally come, we use the overwriting mechanism for simplicity because the number of users is not very large and concurrency control is now ignored in our system. So, the educational database is efficiently read-only. - Data Management Engine: this component plays a role in managing not only the educational database but also the mined knowledge base so that any access from other components to the storage tier can take place smoothly. The implementation of this component is carried out with relational database technology for its efficiency and maturity. 3) Storage tier: The storage tier includes our educational database and mined knowledge base. The educational database contains all data about our courses, students, and their study results along the time. This educational database plays a role of a so-called management database of a decision support system in [9]. In addition to such a traditional management database, our knowledge-driven EDSS has a mined knowledge base as an advantage toward knowledge-driven decision support. The mined knowledge base is a collection of mining results obtained so far from the educational database. We materialize the mining results for further post-processing and use. Such persistence of the mining results also helps us save time for knowledge discovery. Besides, we maintain a synchronous connection between the educational database and the mined knowledge base so that consistency for both data and knowledge can always be ensured. With these two bases, our system can provide the educational managers with not only problem-relevant data but also knowledge. On the other hand, we also tackle the difficulties stemming from the peculiar features of an academic credit system mentioned in subsection III.A. We aligned all versions of our educational programmes by setting up the equivalent relationships among courses of the different versions. The equivalence of courses enables us to compare the study performance of a student with one of another student. B. Knowledge Discovery for Educational Decision Support In its current version, the proposed system has four data mining functionalities to discover actionable knowledge from educational data imported from Excel files. Corresponding to the intelligence phase in Fig. 1, they are: final status-based student classification using J48 algorithm; study trend prediction with clustering using k-means algorithm; status prediction with clustering in terms of probability using Expectation-Maximization algorithm; and course association analysis using Apriori algorithm. It is worth noting that these functionalities are mainly devoted to decision making support for problems about students with poor study performance. C. Experiments with the Proposed System We carry out experiments with the proposed system using the educational programmes in a credit system of education at Faculty of Computer Science and Engineering, Ho Chi Minh City University of Technology, Vietnam, [1]. Our educational data set was obtained about 1348 regular undergraduate students enrolled from 2005 to 2008 where there were 367 students in 2005, 344 students in 2006, 322 students in 2007,

and 315 students in 2008. Data about study results of students in 2005-2008 include 64584 records. Regarding actionable knowledge discovery, we got a finalstatus-based student classification model with 97.063% precision and study trend prediction with 93.0682% precision while the other two data mining functionalities for status prediction in terms of probabilities and course association analysis also produce acceptable and reasonable results. As not concentrating on educational data mining, we skip the details of our educational data mining part: data preprocessing, choice of algorithms, parameters tuning, and post-processing, etc. D. Discussion about the Proposed System In this subsection, we would like to discuss the following features of our system to highlight its contributions. 1) The system is effective with knowledge-driven decision making support: In our system, knowledge discovered from the educational database is made actionable as analyzed in subsection III.A. We have flexibly combined all knowledge together from different data mining tasks in supporting a real common problem of students with poor study performance. 2) The system is developed as an open source software: As noted early, open source technologies are used in our implementation. This choice came from the following facts. Open source technologies do not require a lot of resources in our case. The system can be used for non-commercial purposes in any educational organizations employing a semester credit system of education. 3) The system is flexible and extensible with a Web-based interface: As mentioned in [15], Web-based decision support systems are preferred in coming years. We also found a Webbased interface suitable for our system as the educational managers can use the system on any machine simply equipped with a web browser. The system is also extensible as its architecture shown in Fig. 2 is component-based. V. CONCLUSION

to give support on not only predefined but also ad-hoc and unstructured problems. The second future work is rather tough but really significant. The third one is to make the system capable of learning to flexibly provide educational managers with more data-driven and knowledge-driven decision support. REFERENCES
[1] [2] Academic Affairs Office, Ho Chi Minh City University of Technology, Vietnam, http://www.aao.hcmut.edu.vn/dhcq.html, August 15th, 2011. R. S. J. D. Baker and K. Yacef, “The State of Educational Data Mining in 2009: a Review and Future Visions,” J. Educational Data Mining, vol. 1, pp. 3-17, 2009. J. E. Beck, “Difficulties in Inferring Student Knowledge from Observations (and Why You Should Care),” Educational Data Mining, 2007. P. R. Cohen and C. R. Beal, “Temporal Data Mining for Educational Applications,” Int. J. Software and Informatics, vol. 3, pp. 29-44, 2009. G. W. Dekker, M. Pechenizkiy, and J. M. Vleeshouwers, “Predicting Students Drop Out: A Case Study,” Educational Data Mining, 2009. D. Z. Deniz and I. Ersan, “An Academic Decision-Support System Based on Academic Performance Evaluation for Student and Program Assessment,” Int. J. Engng Ed., vol. 18, pp. 236-244, 2002. Y. Gong, D. Rai, J. E. Beck, and N. T. Heffernan, “Does Self-Discipline Impact Students’ Knowledge and Learning?,” Educational Data Mining, 2009. R. Hartley and S. M. Y. Almuhaidib, “User Oriented Techniques to Support Interaction and Decision Making with Large Educational Databases,” Computers & Education, vol. 48, pp. 268-284, 2007. T. Hirouchi and T. Kosaka, “An Effective Database Formation for Decision Support Systems,” Information & Management, vol. 7, pp. 183-195, 1984. C. W. Holsapple, “Decisions and Knowledge,” in Handbook on Decision Support Systems – 1 Basic Themes, International Handbooks on Information Systems, F. Burstein, C. W. Holsapple Eds., 2008, pp. 21-53. D. Kanellopoulos and S. Kotsiantis, “Towards an Ontology-based System for Intelligent Prediction of Student Dropouts in Distance Education,” Int. J. Management in Education, vol. 2, pp. 172-194, 2008. C. P. Lim and L. C. Jain, “Advances in Intelligent Decision Making,” in Handbook on Decision Making, ISRL 4, L. C. Jain and C. P. Lim Eds., 2010, pp. 3-28. A. Merceron and K. Yacef, “Interestingness Measures for Association Rules in Educational Data,” Educational Data Mining, 2008. M. Paliwal and U. A. Kumar, “A Study of Academic Performance of Business School Graduates Using Neural Network and Statistical Techniques,” Expert Systems with Applications, vol. 36, pp. 7865-7872, 2009. D. J. Power, “Understanding Data-Driven Decision Support Systems,” Information Systems Management, vol. 25, pp. 149-154, 2008. O. Taylan and B. Karagozoglu, “An Adaptive Neuro-Fuzzy Model for Prediction of Student’s Academic Performance,” Computers & Industrial Engineering, vol. 57, pp. 732-741, 2009. M. Telem, “DSS in Educational Organizations,” Computers Education, vol. 14, pp. 61-69, 1990. M. Telem, “Educational DSS: Potential Services, Benefits, Difficulties and Dangers,” Computers Education, vol. 14, pp. 71-80, 1990. S. Vinnik and M. H. Scholl, “UNICAP*: Efficient Decision Support for Academic Resource and Capacity Management,” IFIP International Federation for Information Processing 2005, TCGOV 2005, LNAI 3416, 2005, pp. 235-246. M. Zorrilla, D. Garcia, and E. Alvarez, “A Decision Support System to Improve e-Learning Environments,” in Proc. of EDBT 2010, 2010.

[3]

[4] [5] [6]

[7]

[8]

[9]

[10]

[11]

[12]

[13] [14]

In this paper, we have presented our knowledge-driven decision making support for making decisions on a typical problem of study extension for students with poor study performance. The support is significantly considered for the educational managers in the educational organizations employing a popular flexible academic credit system. With our support, problem-relevant data and knowledge can be prepared as much as possible in the educational decision making process. In addition, a knowledge-driven educational decision support system is proposed. We believe that development of such a knowledge-driven educational decision support system is important for the ubiquity of a credit system of education. As the system has been developed still in its infancy, many various works need to be done further in the future. Firstly, we will investigate more advanced mining algorithms and carry out more experiments in a larger scale academic environment to validate the effectiveness and efficiency of the proposed system with some functional metrics. Secondly, we would like

[15] [16]

[17] [18] [19]

[20]

Sponsor Documents

Or use your account on DocShare.tips

Hide

Forgot your password?

Or register your new account on DocShare.tips

Hide

Lost your password? Please enter your email address. You will receive a link to create a new password.

Back to log-in

Close