Asking the Right Question: Inferring Advice-Seeking Intentions from Personal Narratives

Proceedings of NAACL-HLT 2019, pages 528–541 Minneapolis, Minnesota, June 2 - June 7, 2019. c⃝2019 Association for Computational Linguistics 528 Asking the Right Question: Inferring Advice-Seeking Intentions from Personal Narratives Liye Fu Cornell University liye@cs.cornell.edu Jonathan P. Chang Cornell University jpc362@cornell.edu Cristian Danescu-Niculescu-Mizil Cornell University cristian@cs.cornell.edu Abstract People often share personal narratives in order to seek advice from others. To properly infer the narrator’s intention, one needs to apply a certain degree of common sense and social in- tuition. To test the capabilities of NLP sys- tems to recover such intuition, we introduce the new task of inferring what is the advice- seeking goal behind a personal narrative. We formulate this as a cloze test, where the goal is to identify which of two advice-seeking ques- tions was removed from a given narrative. The main challenge in constructing this task is finding pairs of semantically plausible advice- seeking questions for given narratives. To ad- dress this challenge, we devise a method that exploits commonalities in experiences people share online to automatically extract pairs of questions that are appropriate candidates for the cloze task. This results in a dataset of over 20,000 personal narratives, each matched with a pair of related advice-seeking questions: one actually intended by the narrator, and the other one not. The dataset covers a very broad array of human experiences, from dating, to career options, to stolen iPads. We use human an- notation to determine the degree to which the task relies on common sense and social intu- ition in addition to a semantic understanding of the narrative. By introducing several base- lines for this new task we demonstrate its feasi- bility and identify avenues for better modeling the intention of the narrator. 1 Introduction “Computers are useless. They can only give you answers.” - Pablo Picasso People often share their personal experiences to elicit advice from others. These personal narra- tives provide the necessary context for properly understanding the informational goals of the nar- rators. Endowing automated systems with the ca- pability to infer these advice-seeking intentions Personal narrative: I am generally a person who needs a lot of sleep, but today I was not able to sleep more than 6 hours and I am extremely tired. My eyes hurt and two hours later I have program- ming [lesson] so I have to be alert. I’ve already drunk a cup of coffee and although I rarely drink coffee, it had no effect on me. I am not at home so I have limited possibilities as for food. I don’t want to do anything too unhealthy such as drink- ing 10 cups of coffee, tho I may consider drink- ing another one. Which advice-seeking question is more likely to have been asked by the narrator: Q1: Is it even possible to be addicted to coffee? Q2: How can I energize myself? Figure 1: An abbreviated instance from the ASQ dataset. A personal narrative is matched with two plau- sible advice-seeking questions, only one of which was actually …

Content truncated to 500 words