This paper discusses ongoing work to build an argumentation inquiry learning system, GAIL. The purpose of GAIL is to support students in constructing scientific arguments in an undergraduate genetics course in order to facilitate deeper learning and improve argumentation skill. Students can construct argument diagrams using a dragand-drop graphical user interface. The system constructs arguments on-the-fly to use as a knowledge source for evaluating the learners' arguments and providing intelligent feedback.
Argumentation plays an important role in science. There
has been significant interest within the field of science
education in argumentation. However, students’
arguments have been shown to be deficient in a number of
ways, e.g., lacking support for claims
To address this problem we are implementing a prototype genetics argumentation inquiry learning system, GAIL. GAIL will support learning to argue about cases in human genetics. This is a field that applies findings from genetics research to biomedical reasoning. GAIL is designed for use in an introductory genetics course for undergraduates that many biology majors find the most challenging course in the biology core curriculum. We hope that use of GAIL will improve argumentation skill, facilitate deeper learning of genetics, and increase interest and engagement in science.
Each GAIL lesson requires learners to construct
arguments for and against certain hypotheses about a
genetics case, e.g., about an infant who may have an
inherited metabolic disorder or someone who inherited a
genetic variant that is associated with increased risk of
colon cancer. A prototype user interface is shown in
Figure 1 (see last page). Information relevant to the lesson
is provided by GAIL on the left-hand side of the screen:
the Problem (to give a certain argument); Hypotheses
(which can be used in the argument, but note that some
are incorrect); Data from medical records about the
patient and the patient’s biological family; and
Connections, a list of facts or principles of genetics. The
center of the screen shows two arguments constructed by a
learner. To construct the arguments, the learner searched
for text components on the left-hand side of the screen,
dragged them into the workspace in the center of the
screen, and connected the components. Arrows point from
support to conclusion. The connection between support
and conclusion – known as the warrant in argumentation
theory
In Figure 1, the problem is to give two arguments for the hypothesis that the patient (referred to as J.B.) has cystic fibrosis, i.e., has two variant alleles of the CFTR gene. The leftmost “chain” of arguments begins with data (at the bottom of the argument diagram) about J.B.’s respiratory problems. The learner used that data to support an intermediate hypothesis that J.B. has thickened mucus in the lungs, which is used to support an intermediate hypothesis that J.B. has abnormal CFTR protein, which is used to support the main hypothesis/conclusion that J.B. has cystic fibrosis. Branching from the right hand side of the diagram, connections (warrants) provide justification for each step of the argument. The second argument for the same hypothesis begins with data about J.B.’s lab test result.
GAIL’s innovation is that the system can generate arguments for evaluating the correctness of the learner’s arguments, rather than requiring the arguments to be constructed by a teacher. Use of the generated arguments enables GAIL to provide intelligent feedback on both the structure and content of the learner’s argument.
GAIL creates an XML-formatted file that contains: (1) strings of natural language text -- the problem, hypotheses, data, and connections -- to be displayed to learners on the left-hand side of the graphical user interface as shown in Figure 1; (2) a specification of an internal causal domain model; and (3) mapping of the natural language strings in (1) to concepts and relations in the domain model. GAIL’s Authoring Tool provides (1) to the user interface, uses (2) to build an internal Knowledge Base, and stores (3) to enable GAIL’s Argument Evaluator to semantically interpret learners’ argument diagrams, to avoid the challenge of interpreting natural language input.
Based on our previous work on modeling genetics
Argumentation schemes are descriptions of acceptable,
but often defeasible, patterns of reasoning
For example, the argumentation scheme for reasoning from effect to cause, shown in Figure 2, can be instantiated from a KB to create an argument that a patient has HNPCC (a mutation in the MLH1 gene, a hereditary
condition predisposing one to colon cancer) based upon the data that genetic testing showed a variant MLH1 allele, and the connection (warrant) that having HNPCC typically leads to that test result. (The exception condition for this argumentation scheme asks whether there is an alternative explanation for the data.) An argument diagram representing this argument is shown in Figure 3; to save space, instead of natural language text from the graphical user interface, the diagram uses letters representing propositions, where A is the conclusion, B is the data, and S+(A,B) is the warrant.
A B
S+(A,B) argument. After each try, the Feedback Generator selects the most general (lowest level) unused message for an error; each time the student makes the same error on a subsequent try, the next more specific (next higher level) message is selected. A positive message is generated when an error is corrected on the next try. Currently, the Feedback Generator displays only the message for the most serious error to the student, but writes all of the detected errors to a logfile for inspection by the teacher.
To illustrate the learner’s interaction with GAIL, suppose the problem was to give an argument for the hypothesis that J.B.’s brother might have malnutrition and poor growth. Internally, GAIL generates a chained argument beginning with the data that J.B.’s brother has been diagnosed as having cystic fibrosis, which supports an intermediate hypothesis that his CFTR protein is abnormal, which supports an intermediate hypothesis that he might have pancreatic abnormality, which supports the main hypothesis that he might have malnutrition and poor growth. (The warrants of GAIL’s argument are not described here to save space.) However, on the first try the learner’s argument contains the main claim that J.B. (rather than J.B.’s brother) has cystic fibrosis, which does not match the problem. Since this type of error has been given the highest severity code, GAIL would tell the student that the main claim of his argument does not match the problem. On the second try, the student fixes the main claim and constructs a new argument. GAIL informs him that the problem noted on the last try has been fixed. However, the student’s argument is missing the intermediate hypothesis that J.B.’s brother might have pancreatic abnormality, so GAIL also informs the student that one or more intermediate hypotheses are missing between J.B.’s brother having abnormal CFTR protein and J.B.’s brother having malnutrition and poor growth. On the third try, the student adds the missing hypothesis but provides an irrelevant warrant. GAIL would inform the student that he has made progress but that the warrant he just added is irrelevant to that subargument.
This paper discusses ongoing work to build an argumentation inquiry learning system, GAIL. The purpose of GAIL is to support students in constructing scientific arguments in an undergraduate genetics course in order to facilitate deeper learning and improve argumentation skill. The system generates arguments onthe-fly to use as a knowledge source for evaluating the learners’ arguments and providing formative and summative intelligent feedback.
All of the components described in this paper have been implemented in Java. Future work includes improvements to the user interface and the Feedback Generator. The Feedback Generator will be made more intelligent to address certain types of errors that we have observed in our formative evaluation studies. For example, a learner “flattened” a chained argument into a one-level structure by conjoining together all of the data and intermediate hypotheses. Note that in this case, the learner has selected the correct content but has just not structured the argument properly into subarguments and shown how one subargument builds upon another subargument. Because the Feedback Generator has access to arguments constructed by GAIL’s Argument Generator, the Feedback Generator will be able to detect this type of error and provide more meaningful feedback than systems that do not have access to content. After these improvements are made, we plan to evaluate GAIL’s effectiveness in an undergraduate genetics course.
Former graduate students Mark Hinshaw, Carl Martensen, Meghana Narasimhan, and Tshering Tobgay contributed to the implementation of GAIL for their MS Projects. Former graduate students Benjamin Wyatt and Chris Cain also contributed to the implementation of GAIL. Wyatt and Martensen received support from a UNCG Regular Faculty grant and Cain received support from the Computer Science Department. Dr. Malcolm Schug of the UNCG Department of Biology has provided helpful feedback on the project.
Figure 1. Screen shot of prototype GAIL user interface