A Two-stage Sieve Approach for Quote Attribution

Abstract

We present a deterministic sieve-based system for attributing quotations in literary text and a new dataset: QuoteLi3 1 . Quote attribution, determining who said what in a given text, is important for tasks like creating dialogue systems, and in newer areas like computational literary studies, where it creates opportunities to analyze novels at scale rather than only a few at a time. We release QuoteLi3, which contains more than 6,000 annotations linking quotes to speaker mentions and quotes to speaker entities, and introduce a new algorithm for quote attribution. Our twostage algorithm first links quotes to mentions, then mentions to entities. Using two stages encapsulates difficult sub-problems and improves system performance. The modular design allows us to tune either for overall performance or for the high precision appropriate for many use cases. Our system achieves an average F-score of 87.5% across three novels, outperforming previous systems, and can be tuned for precision of 90.4% at a recall of 65.1%.

Unsupervised inference of implicit biomedical events using context triggers

“…In this architecture, each of the sieves, which is designed to capture specific types of information, is applied one by one to a given document in decreasing order of precision, where earlier sieves inform later ones by transferring precise information. It was first proposed for coreference resolution [30] and then applied to other tasks such as temporal relations and quote attribution [31,32]. It has also been employed for various tasks in the biomedical domains, such as entity linking, coreference resolution, relation extraction, and concept normalization [33][34][35][36].…”

“…This means that it is also easy to assess the contribution of each sieve to the overall performance. This is particularly useful when supervised learning does not work well, as discussed in the section on related work [30][31][32][33][34][35][36].…”

Proceedings of the 3rd Joint SIGHUM Workshop on Computational Linguistics for Cultural Heritage, Social Sciences, Humanities and Literature

“…Grishina and Stede (2017) test the projection of coreference annotations, a task related to speaker attribution, using multiple source languages. Muzny et al (2017) improved on previous work on quote and speaker attribution by providing a cleaned-up dataset, the QuoteLi3 corpus, which includes more annotations than the previous datasets. They also present a two-step deterministic sieve model for speaker attribution on the entity level and report a high precision for their approach 1 .…”

“…The concept of dialogism has been a notable focus in recent computational literary scholarship (Brooke et al, 2017;Hammond and Brooke, 2016;Muzny et al, 2017). As theorized by Russian literary critic Mikhail Bakhtin (2013), a dialogic novel is one in which characters present "a plurality of independent and unmerged voices and consciousnesses, a genuine polyphony of fully valid voices".…”

“…For English, automatic approaches based on machine learning He et al, 2013) or rule-based systems (Muzny et al, 2017) have been developed for this task, and a limited amount of annotated resources already exists. For most other languages, however, such resources are not yet available.…”