Inference Tree Use to Design Arguments in Expository Reports

Inference Tree Use to Design Arguments in Expository Reports

Jens Mende (University of the Witwatersrand, South Africa)
Copyright: © 2009 |Pages: 10
DOI: 10.4018/978-1-59904-845-1.ch056
OnDemand PDF Download:
$30.00
List Price: $37.50

Abstract

When they write essays, many students merely attempt ‘to fill pages with material gathered from sources’ (Erion, 2000). Consequently, they produce inane arguments of the form: Adams said this, Brown said that, Cohen said the other, etc. Conclusion: much has been written about this topic. This is unacceptable both in academic ICT courses and subsequently in the ICT profession. In academe, a written argument should ‘make a leap from the raw materials of the library to an informed opinion’ (Fasel, 1963). In the profession, a written argument should similarly make a leap from a present state of affairs to a desired future state. So in both situations, writers should be able to devise a report that contains an argument from available facts towards an intelligent conclusion. This kind of report is called an ‘expository report’ (Trimble, 1975), or an ‘argumentative report’ (Dykeman, 1974).
Chapter Preview
Top

Introduction

When they write essays, many students merely attempt ‘to fill pages with material gathered from sources’ (Erion, 2000). Consequently, they produce inane arguments of the form:

Adams said this, Brown said that, Cohen said the other, etc.

Conclusion: much has been written about this topic.

This is unacceptable both in academic ICT courses and subsequently in the ICT profession. In academe, a written argument should ‘make a leap from the raw materials of the library to an informed opinion’ (Fasel, 1963). In the profession, a written argument should similarly make a leap from a present state of affairs to a desired future state. So in both situations, writers should be able to devise a report that contains an argument from available facts towards an intelligent conclusion. This kind of report is called an ‘expository report’ (Trimble, 1975), or an ‘argumentative report’ (Dykeman, 1974).

In order to write such reports successfully, ICT writers can get a great deal of useful advice from textbooks of Writing for ICT (e.g., Warner, 1996; Zobel, 1997), as well as textbooks of technical writing (e.g., Andrews & Blickle, 1982; Pauley & Riordan, 1993), business writing (e.g., Ruch & Crawford, 1988), nonfiction (e.g., Fryxell, 1996; Zinsser, 1990), and even prose style (e.g., Strunk & White, 1979; Trimble, 1975). There they will find a variety of methods, such as structuring sentences and paragraphs, introducing a report and ending it, outlining a report and editing it, and so forth, which are useful for writing reports of any kind: narrative, descriptive, imperative, or expository/argumentative. However, they will find little or no advice on devising the argument in an expository/argumentative type of report.

Yet, ICT writers need not despair. Specific argumentation aids are actually available right under their noses in ICT textbooks of artificial intelligence (e.g., Giarratano & Riley, 1989; Turban, 1992). There they will find a tool called the inference tree, and two associated techniques called forward chaining (FC) and backward chaining (BC). Although these three aids were originally intended for the purpose of devising expert systems and related computer applications, writers can easily adapt them for the purpose of devising expository/argumentative reports. The inference tree can be used to outline the argumentation product; the chaining techniques can be used to facilitate the argumentation process.

Top

Background

In such reports, the argument is situated in the paragraphs beyond the introduction. Each of those paragraphs consists of a single core idea supported by several peripheral ideas (Andrews & Blickle, 1982). For example, the following paragraph has a single core idea (in italics), which is supported by four peripheral ideas.

This system has no validation. We examined the system specification, looking for all programs that capture data from human sources. Then, we examined the if-then commands in each program, but found that none of the if’s detect data errors, and none of the then’s produce error messages.

The argument involves inferences between core ideas. Each inference inputs the core ideas of one or more previous paragraphs, and outputs the core idea of a subsequent paragraph. Example:

This system has no validation

If it has no validation, then it captures bad data

So it captures bad data.

The argument usually contains many inferences (Fisher, 1988; Hamblin, 1970; Mende, 2005a, 2005b; Parsons, 1996). For example, Table 1 outlines the argument in a simple report about validation (omitting all peripherals, and indicating the inferences with the keywords so and therefore).

Key Terms in this Chapter

Inference Tree: A diagrammatic outline of an argument. Boxes represent core ideas of paragraphs, and arrows represent inferential connections between core ideas. The boxes are arranged in three columns: premises, intermediates, and conclusion.

Backward Chaining: A technique of generating inferences from a hypothesis of the conclusion, through intermediates, to premises. You start with the concluding hypothesis, match it with related premises to get a new hypothesis, then match it with further premises to get another hypothesis, and so on, until the last hypothesis is confirmed. Then, the concluding hypothesis is also confirmed.

Argument: A series of inferences from premises through intermediates to a conclusion. Premises are core ideas that are not inferred from other core ideas; intermediates are inferred from other cores, and other cores are inferred from them; the conclusion is inferred from other cores, but no other cores are inferred from it.

Reasoning Effectiveness Error: Raises doubts in readers’ minds when they try to understand an argument. A typical example is illusory relevance, where inference inputs appear to be relevant to an output, but are really not. Another example is irrelevance, where an inference insinuates a new idea into the output, which idea is not present in the inputs.

Forward Chaining: A technique of generating inferences from premises through intermediates to a conclusion. You start by matching related premises to get intermediates; then match premises with related intermediates to get further intermediates, and so on, until a conclusion emerges.

Modular Inference Tree: A high-level concluding tree together with several low-level supporting trees. The low-level trees have little or no connection to one another and can be drawn and checked independently, each on a separate page.

Inference: An elementary reasoning step that inputs previous core ideas of an argument and outputs a subsequent core idea.

Reasoning Efficiency Error: Wastes readers time when they try to understand an argument. A common example is inconclusive argument, which leads either to no conclusion at all or to an inane afterthought such as “much has been written about this topic.” Another example is overloaded inference, which could be divided into smaller, simpler inferences.

Complete Chapter List

Search this Book:
Reset