Story Similarity Data

About the Data

This data set describes human perceptions of story similarity.

58 stories were generated by a narrative planner (an algorithm for telling multi-agent stories). The stories are given in formal syntax in solutions.txt. The first line is story #1, the second line is story #2, etc.

Simple translations of those stories into English are given in english.txt.

64 subjects were shown a reference story and two other stories, labeled A and B. Subjects were asked to choose whether story A or B was more similar to the reference story. Subjects were shown the English translations of the stories.

The results can be found in data.csv. The first three columns give the numbers of the reference story, story A, and story B respectively. The third column is the ID number of the participant making the decision. The fourth column is A if they thought story A was more similar or B if they thought story B was more similar.


The data can be downloaded here.


This data was collected by Rachelyn Farrell at the Narrative Intelligence Lab in the Department of Computer Science at the University of Kentucky in April, 2022, using the crowd-sourcing website Prolific. This data was collected under IRB protocol #69302.

Results from this study were originally published in the following paper. Authors wanting to reference this data set are encouraged to cite this paper.

Rachelyn Farrell, Mira Fisher, Stephen G. Ware. Salience vectors for measuring distance between stories. In Proceedings of the 18th AAAI international conference on Artificial Intelligence and Interactive Digital Entertainment, pp. 95-104, 2022.

Story Domain

Below is a description of the story world from which these stories were generated. Subjects were shown this description before reading any stories.

Tom needs a special potion from the merchant who works at the market. Tom's cottage is connected to the market via a crossroads which also connects to the merchant's house. There is a bandit lurking in the crossroads, who is a known criminal. There is a guard at the market who wants to punish criminals, but the guard does not initially know where the bandit is. At night, the merchant takes all her possessions to her house and goes to sleep.

What characters want:

  • Tom wants to have the potion and be back at his cottage.
  • The merchant wants to sell her items for coins.
  • The guard wants to punish criminals.
  • The bandit wants to have valuable items (coins and potions).

What characters can do:

  • Walk between connected locations
  • Buy something from the merchant (for one coin)
  • Attack someone (while armed with a sword) *
  • Rob someone who is unarmed (while armed) *
  • Rob someone who is asleep *
  • Loot something from a dead person *
  • Arrest a criminal (only the guard can do this)
  • Wait for night (only Tom can do this)

* (Attacking and stealing are crimes.)

What characters have:

  • Tom has one coin.
  • The merchant has the potion and a sword.
  • The guard has a sword.
  • The bandit has a sword and one coin.

Stories end when Tom either dies, gets arrested, or returns to his cottage with the potion.