Javascript must be enabled for the correct page display

Human Argument Structure Language

Linde, Jelmer van der (2018) Human Argument Structure Language. Master's Thesis / Essay, Artificial Intelligence.


Download (3MB) | Preview
[img] Text
Restricted to Registered users only

Download (96kB)


Computers reason, but humans argue. For computers to understand humans and human knowledge, they need to understand human argument. Likewise, they need to be understandable by humans. There are multiple abstract models of argumentation, and argumentation software exists to help fill in such models and evaluate them. These models are often presented as graphs. Computers are becoming capable of interpreting argumentative text through such models, but the variety of language, subtle ways of expressing things, and the assumption of the presence of common knowledge when formulating text make interpreting a challenging task. Let alone that filling in abstract models (i.e. picking the right place for parts of each argument) can be a challenging practise on its own. This work unites an abstract argument model with language, creating a tool that translates written argumentation into argument diagrams and vice versa. The resulting program, HASL is a parser and grammar for argumentative text with accompanying interface. It allows you to enter a text and see how the claims inside the text support and attack each other. It also allowed you to reverse this process, and generate text. The abstract argumentation model is connected to grammar rules which together create our language that is similar to English, which makes it intuitive to read and write. We construct and implement this grammar in two experiments: HASL/1 defines a grammar that allows for all possible elements in argument structures to occur. HASL/1 also adds support for anaphora and enthymemes (i.e. being able to fill in missing premises). HASL/2 makes use of a more abstract grammar that only uses discourse markers (i.e. keywords) and structure, allowing it to be used when the text cannot be understood completely. This grammar and the argument model are isomorphic, and can be used to formulate argumentative texts. We conclude with an evaluation of these two experiments applied to syllogisms, Toulmin arguments and Tort law. HASL is a step towards understanding of argumentation by computers, allowing them to explain and discuss in the same language humans do.

Item Type: Thesis (Master's Thesis / Essay)
Supervisor name: Verheij, H.B. and Spenader, J.K.
Degree programme: Artificial Intelligence
Thesis type: Master's Thesis / Essay
Language: English
Date Deposited: 10 Sep 2018
Last Modified: 10 Sep 2018 12:42

Actions (login required)

View Item View Item