In this lecture, we will study how serialisability (see Lecture 2A) can be used to rank arguments.

We recommend the interested reader to read the full paper in the reading list.

A serialisability-based ranking semantics (Blümel and Thimm, 2022)

Let us consider the argumentation framework displayed below.

Capture d’écran 2024-09-10 à 09.25.12.png

We have a “feeling” that $c$ is better than $b$ or $a$.

But how do we formalize this intuition using serialisability?

We have seen in the previous lecture that we can reconstruct admissible sets by iteratively choosing initial sets.

<aside> ⚠️ DEFINITION

A serialisation sequence for $\mathcal{F} = (A,R)$ is a sequence $(S_1, \dots, S_n)$ with $S_1 \in IS(\mathcal{F})$ and for each $2 \leq i \leq n$, we have $S_i \in IS(\mathcal{F}^{S_1 \cup \dots \cup S_{i-1}})$.

To each admissible set of $\mathcal{F}$, there is at least one corresponding serialisation sequence.

</aside>

In our example with three arguments above, we can see that $\{a\}$ can only be selected after $\{c\}$.