Feuding Families and Former Friends: Unsupervised Learning for Dynamic Fictional Relationships

By Mohit Iyyer, Anupam Guha, Snigdha Chaturvedi, Jordan L. Boyd-Graber, Hal Daumé Iii. 2016

152 citations recorded by the source corpus.

Understanding how a fictional relationship between two characters changes over time (e.g., from best friends to sworn enemies) is a key challenge in digital humanities scholarship. We present a novel unsupervised neural network for this task that incorporates dictionary learning to generate interpretable, accurate relationship trajectories. While previous work on characterizing literary relationships relies on plot summaries annotated with predefined labels, our model jointly learns a set of global relationship descriptors as well as a trajectory over these descriptors for each relationship in a dataset of raw text from novels. We find that our model learns descriptors of events (e.g., marriage or murder) as well as interpersonal states (love, sadness). Our model outperforms topic model baselines on two crowdsourced tasks, and we also find interesting correlations to annotations in an existing dataset.

Subfields: Artificial Intelligence, Literature and Literary Theory, Natural Language Processing

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Literature and Literary Theory

In Natural Language Processing

Citations in the corpus

The NarrativeQA Reading Comprehension Challenge (2017, background): “…fine grained processing by inducing character types (Bamman et al., 2014b; Bamman et al., 2014a), understanding relationships between characters (Iyyer et al., 2016; Chaturvedi et al., 2017), or understanding plans, goals, and narrative structure in terms of abstract narratives (Schank and…”
Entity-Centric Contextual Affective Analysis (2019, methodology): “Prior to the proposed power, agency, and sentiment framework, initial approaches to personcentric analyses used graphical models to identify personas in narratives (Bamman et al., 2013; Card et al., 2016; Iyyer et al., 2016; Chaturvedi et al., 2017), where personas are distributions over nouns, adje”
Unsupervised Learning of Evolving Relationships Between Literary Characters (2017, methodology): “.5 Comparison to RMN We now compare our model with another unsupervised approach , RMN (Iyyer et al. 2016) (described in Sec.2).”
Generating Character Descriptions for Automatic Summarization of Fiction (2019, background): “Iyyer et al. (2016) proposed an unsupervised neural network to model the changes of relationships between two characters over the story.”
An Efficient LSTM Network for Emotion Recognition From Multichannel EEG Signals (2020, methodology): “To achieve these, we adopt the contrastive max-margin loss which has been used in many studies [54], [55].”
An Unsupervised Neural Attention Model for Aspect Extraction (2017, methodology): “We adopted the contrastive max-margin objective function used in previous work (Weston et al., 2011; Socher et al., 2014; Iyyer et al., 2016).”
Story Comprehension for Predicting What Happens Next (2017, background): “, 2014, 2015), inter-character relationships (Iyyer et al., 2016; Chaturvedi et al., 2016, 2017), and social networks of characters (Elson et al.”
Incorporating Structured Commonsense Knowledge in Story Completion (2018, background): “Previous studies on narrative comprehension include character roles identi-ﬁcation (Valls-Vargas, Onta ˜ n ´ on, and Zhu 2015), narratives schema construction (Chambers and Jurafsky 2009), and plot pattern identiﬁcation (Jockers 2013).”
Leveraging Just a Few Keywords for Fine-Grained Aspect Detection Through Weakly Supervised Co-Training (2019, background): “Recently, neural topic models (Iyyer et al., 2016; Srivastava and Sutton, 2071; He et al., 2017) have been shown to produce more coherent topics than earlier models such as Latent Dirichlet Allocation (LDA) (Blei et al., 2003).”
The Importance of Modeling Social Factors of Language: Theory and Practice (2021, background): “In practice, such social relations often can be reasonably inferred from text (Krishnan and Eisenstein, 2015; Iyyer et al., 2016; Rashid and Blanco, 2017; Rashid et al., 2020).”
Applications of Topic Models (2017, background): “Jo and Oh [2011] develop a model that first draws sentiment distributions that constrain the aspects discussed in a document. Downstream models can also capture these divisions as well. Nguyen et al. [2013] predict whether a speaker is Republican or Democrat1 based on the versions of topics they dis”
A dataset and baselines for sequential open-domain question answering (2018, methodology): “Instead, we use a continuous representation of relationships (Iyyer et al., 2016).”
STORIUM: A Dataset and Evaluation Platform for Machine-in-the-Loop Story Generation (2020, methodology): “Our topic model is a simpliﬁed version of the relationship modeling network ( RMN ) proposed by Iyyer et al. (2016).”
Inducing Semantic Micro-Clusters from Deep Multi-View Representations of Novels (2017, methodology): “Given a book b and a character c , S , given a book b and a pair of characters c 1 and c 2 , contains linearly ordered text spans which mention both c 1 and c 2 , but no other character, The rest of the input preparation follows Iyyer et al. (2016) as follows.”