Actor Identification in Discourse: A Challenge for LLMs?
This addresses the problem of actor identification in discourse networks for researchers analyzing societal debates, but it is incremental as it builds on existing methods to improve output control in LLMs.
The study tackled the challenge of identifying political actors in discourse, where pronouns often obscure the canonical names, by comparing a traditional NLP pipeline with an LLM. The result showed that the LLM performed worse, but a hybrid model combining the LLM with a classifier for output normalization substantially outperformed both initial models, achieving improved accuracy.
The identification of political actors who put forward claims in public debate is a crucial step in the construction of discourse networks, which are helpful to analyze societal debates. Actor identification is, however, rather challenging: Often, the locally mentioned speaker of a claim is only a pronoun ("He proposed that [claim]"), so recovering the canonical actor name requires discourse understanding. We compare a traditional pipeline of dedicated NLP components (similar to those applied to the related task of coreference) with a LLM, which appears a good match for this generation task. Evaluating on a corpus of German actors in newspaper reports, we find surprisingly that the LLM performs worse. Further analysis reveals that the LLM is very good at identifying the right reference, but struggles to generate the correct canonical form. This points to an underlying issue in LLMs with controlling generated output. Indeed, a hybrid model combining the LLM with a classifier to normalize its output substantially outperforms both initial models.