OASiS: Objective Argument Summarization in Search
Conceptually, an argument logically combines a claim with a set of reasons. In real-world text, however, arguments may be spread over several sentences, often intertwine multiple claims and reasons along with context information and rhetorical devices, and are inherently subjective. This project aims to study how to computationally obtain an objective summary of the gist of an argumentative text. In particular, we aim to establish foundations of natural language processing methods that (1) analyze the gist of an argument's reasoning, (2) generate a text snippet that summarizes the gist concisely, and (3) neutralize potential subjective bias in the summary as far as possible.
The rationale of the planned project is that argumentation machines, as envisioned by the RATIO priority program, are meant to present the different positions people may have towards controversial issues, such as abortion or social distancing. One prototypical machine is our argument search engine, args.me, which opposes pro and con arguments from the web in response to user queries, in order to support self-determined opinion formation. A key aspect of args.me and comparable machines is to generate argument snippets, which give the user an efficient overview of the usually manifold arguments. Standard snippet generation has turned out to be insufficient for this purpose. We hypothesize that the best argument snippet summarizes the argument's gist objectively.
Building upon existing argument mining techniques, the project centers around text summarization and style transfer. Both tasks are hard in general, but extensive research in the last years has shown potential in focused domains. For arguments, short content summaries have been generated successfully, yet without attention to the argument's reasoning. Style transfer even remains fully unstudied so far in computational argumentation. We aim to fill these gaps through the project's objectives.
In particular, we plan to (1) create a first corpus of thousands of human-written argument summaries of different kinds. On this basis, we develop and evaluate genuine computational methods that (2) generate a summary of an argument's reasoning and (3) neutralize the style of the summaries to reduce subjective bias. In (4) empirical studies, we explore what kinds of summaries are seen as best and why. We expect to obtain new knowledge about what is important in summarizing subjective language such as argumentation. The corpus enables more systematic research on argument generation, and the methods will be useful for various types of argumentation machines and other applications of computational argumentation. Making all outcomes publicly available, the project substantially contributes to the goals of RATIO in particular, and to natural language processing research in general.