Character-Centric Narrative Understanding

A narrative is any account of a collection of related events telling a story. It can be found in a variety of forms including novel, journalism, music, film, video, speech, performing arts, painting, and sculpture. This project will develop new AI algorithms for automatic understanding of narratives in novels, with a particular focus on (1) constructing character networks from text; and (2) generating event story maps as a plot summary. The proposed framework will be important for solving downstream tasks such as recommendation, script generation, book review generation and movie scene segmentation.

The project will involve a few tasks including:

• Character identification and event detection. Character identification needs to deal with the challenges of co-referencing and entity linking, while unsupervised event detection approaches need to be developed to recognise main events with an arbitrary set of arguments from text.
• Dynamic character network construction. The relationship between characters may not be static, but evolving over time. Also the relation types cannot be pre-defined. We need to perform dynamic character relation extraction where relation types need to be inferred from data.
• Event story map generation. Traditionally, an event storyline usually organises related events in a chronological order. However, in novels, stories are complex and the narrative could be structured in many different ways including flashbacks, multiple stories happening simultaneously, many smaller stories taking place within a framework of a bigger plot, or even intertwining stories. Creating structured summaries of narratives requires the extraction of an event story map which needs to capture a concise description of stories elaborated in text while at the same time maximising the coverage of salient event information.

How to apply

Candidates must apply via King’s Apply online application system. Details are available at How to apply - King's College London (kcl.ac.uk).

Please indicate Professor Yulan He as the supervisor and and quote the project title “Character-Centric Narrative Understanding” within your application and in all correspondence.

The selection process will involve a pre-selection on documents and, if selected, will be followed by an invitation to an interview. If successful at the interview, an offer will be provided in due time.

Queries

Please direct all queries regarding this project to Professor Yulan He, [Email Address Removed].

(Again for applications - please read the 'How to Apply' and submit via King's Apply)