Towards AI-Complete Question Answering: A Set of Prerequisite Toy Tasks

TL;DR

This paper introduces a set of toy tasks designed to evaluate reasoning and comprehension skills in AI question answering systems, aiming to benchmark progress towards AI-complete language understanding.

Contribution

It proposes a new benchmark suite of prerequisite tasks for AI question answering and extends the Memory Networks model to evaluate its capabilities on these tasks.

Findings

Memory Networks can solve some tasks but not all

The tasks reveal specific reasoning skill gaps in current systems

Benchmark helps identify areas for improvement in AI comprehension models

Abstract

One long-term goal of machine learning research is to produce methods that are applicable to reasoning and natural language, in particular building an intelligent dialogue agent. To measure progress towards that goal, we argue for the usefulness of a set of proxy tasks that evaluate reading comprehension via question answering. Our tasks measure understanding in several ways: whether a system is able to answer questions via chaining facts, simple induction, deduction and many more. The tasks are designed to be prerequisites for any system that aims to be capable of conversing with a human. We believe many existing learning systems can currently not solve them, and hence our aim is to classify these tasks into skill sets, so that researchers can identify (and then rectify) the failings of their systems. We also extend and improve the recently introduced Memory Networks model, and show it is able to solve some, but not all, of the tasks.