Detection of Reading Absorption in User-Generated Book Reviews: Resources Creation and Evaluation

Proceedings of the 12th Conference on Language Resources and Evaluation (LREC 2020), pages 4835–4841 Marseille, 11–16 May 2020 c⃝ European Language Resources Association (ELRA), licensed under CC-BY-NC 4835 Detection of Reading Absorption in User-Generated Book Reviews: Resources Creation and Evaluation Piroska Lendvai1, Sándor Darányi2, Christian Geng3, Moniek Kuijpers1, Oier Lopez de Lacalle4, Jean-Christophe Mensonides5, Simone Rebora1, Uwe Reichel6 1DH Lab, University of Basel, Switzerland 2University of Borås, Sweden 3Carstens Medizinelektronik GmbH, Germany 4University of the Basque Country, Spain 5LGI2P, IMT Mines Ales, France 6Research Institute for Linguistics, Hungary Corresponding author: piroska.r@gmail.com Abstract To detect how and when readers are experiencing engagement with a literary work, we bring together empirical literary studies and language technology via focusing on the affective state of absorption. The goal of our resource development is to enable the detection of different levels of reading absorption in millions of user-generated reviews hosted on social reading platforms. We present a corpus of social book reviews in English that we annotated with reading absorption categories. Based on these data, we performed supervised, sentence level, binary classification of the explicit presence vs. absence of the mental state of absorption. We compared the performances of classical machine learners where features comprised sentence representations obtained from a pretrained embedding model (Universal Sentence Encoder) vs. neural classifiers in which sentence embedding vector representations are adapted or fine-tuned while training for the absorption recognition task. We discuss the challenges in creating the labeled data as well as the possibilities for releasing a benchmark corpus. Keywords: social reading corpus, sentence embeddings, reading absorption detection, affective computing 1. Introduction Our study aims to contribute to text-based affect recogni- tion research by focusing on the experience-related state of absorption during reading literary fiction. Our goal was to process user-generated book reviews from an online social reading platform, and detect passages that textually express reading absorption, e.g. in terms of phrases such as ’I was completely hooked’. The reviews belong to the genre of non- elicited self-narratives that emerge in online social reading communities; cf. (Cordón-García et al., 2013; Rebora et al., 2019). They are subjective, opinionated, unstructured self-reports of variable length that typically include refer- ences to one’s individual reading experience. As opposed to product reviews, the texts often do not merely contain mentions of evaluative sentiment toward (components of) a book, but also express complementary, affective aspects such as reviewers’ cognitive engagement during their indi- vidual reading experience. Reading absorption (aka narrative absorption) has tradition- ally been investigated in the humanities and social sciences (Hakemulder et al., 2017), e.g. by empirical literary and aesthetics studies. Importantly, absorption has been found to be composed of multiple facets, such as transportation to the fictional world ("I feel like I’ve just returned from a long vacation in Martin’s fantasy kingdom"), focused atten- tion ("a wonderful one that really draws you in"), altered sense of time during reading ("It went by in a blink"), emo- tional engagement ("I instantly connected with Annabel"), and others. 1.1. Previous work There have been …

Content truncated to 500 words