GOLDEN MEANS: DISCOVERING THE ACADEMIC POTENTIAL OF OUR CHILDREN
PART II
by Miriam J. Masullo, guest columnist
Part 2: An alternative high-tech approach to assessment of academic potential
Data mining holds the promise of discovering the academic potential of students, but only if the attributes for human potential are present in the data examined. The discovery of knowledge about student potential without this data could not take place easily. Data warehouses of new classes of student data must be built.
It is also important to understand where knowledge discovery is headed and how data mining works so that we can start to collect the classes of data that will be needed in order to take full advantage of these techniques in education. The natural expertise that is inherent in parents and educators to interpret these patterns must be tapped and a methodology for gathering this data must be devised. Potential that is difficult to detect by using standardized testing is often intuitively detectable by human contact backed by years of field experience and subjectivity. To discover these attributes by means of a single test, or a single test mechanism is unrealistic, and not surprisingly, parents have always known best "how bright their children are", while tests have and continue to fail to identify for our nation the potential that exists in so many of our children. This has been and still is a tremendous loss to society.
Through the enhancement of alternative assessment methods with data mining and knowledge discovery techniques, we are afforded an opportunity to rethink assumptions about student promise and the selection process, especially as we grapple to meet the challenges of a rapidly changing society. The typical argument from those reluctant to complement standardized testing with alternative methods includes the need for an evaluation mechanism that can be broadly and uniformly applied. I describe below how that may be done.
Data mining techniques that hold some promise for widely and systematically addressing deficiencies in standardized student testing include: decision trees, rule induction, inference engines and neural networks. However, these require very specialized high-tech skills and cannot be easily used by lay-users. Experts and knowledge engineers were trained to deal with such methods from the early stages of their education in artificial intelligence. Such training is typically complex and will be prohibitively expensive for application to educational practice. In order to improve on these approaches in general, techniques for data mining must address the larger issue of knowledge acquisition and discovery by the lay-user.
Data mining, in the practice of alternative assessment, will help lead to knowledge discovery about academic potential as follows. The general principle is construct an enriched pattern around the data through the integration of domain-specific knowledge. Knowledge discovery in alternative assessment will take place when educators are able to understand and consider patterns identified by data mining tools, i.e., the patterns are amenable to interpretation by non technical users. Then, in fact, the ability to recognize and classify intelligence will be based on the identification of cognitive behavior patterns. Additional supportive data such as text, images, video, audio, etc. that can help to build the context necessary to fully interpret these patterns can be incorporated in the data mining approach.
Such an approach lends itself well to harnessing student portfolios and multimedia student projects. A teacher, parent or administrator will be able to easily ask a general or specific question about student performance or potential buried in these portfolios and multimedia folders. These will be interpreted by the data mining tools in such a way that interesting results are highlighted and academic potential is discovered. Using data mining technology we will be able to refine the process of assessment and by using a Web-centric approach, we will be able to institutionalize implementation. Such deployment via the Internet would then produce the systemic change in practices that we desire.
All these things are at hand because all the necessary technologies exist, and because we know what changes are necessary. We must:
If this new approach succeeds, the social impact will be a community of
lay-users (teachers, parents, decision makers) relying more on leading edge
technologies and new assessment methods, and less on learned expectations
and biases. Our educational system will reconcile with the latest technology.
Our communities will share a common Web-centric environment supported by a
pervasive infrastructure that integrates data mining technologies with
alternative assessment practices and with an Internet-based access mechanism.
The new infrastructure for performance assessment will not only expand the
perception of who can achieve in school, but also to expand the numbers of
those who do. And then we can get to the real challenge, namely, to discover
the gold in the future of our children.
---
For more information, see http://www.virtualgold.com