Rationalization: A Neural Machine Translation Approach to Generating Natural Language Explanations

By Brent Harrison, Upol Ehsan, Mark O. Riedl. 2017

222 citations recorded by the source corpus.

We introduce \em AI rationalization, an approach for generating explanations of autonomous system behavior as if a human had performed the behavior. We describe a rationalization technique that uses neural machine translation to translate internal state-action representations of an autonomous agent into natural language. We evaluate our technique in the Frogger game environment, training an autonomous game playing agent to rationalize its action choices using natural language. A natural language training corpus is collected from human players thinking out loud as they play the game. We motivate the use of rationalization as an approach to explanation generation and show the results of two experiments evaluating the effectiveness of rationalization. Results of these evaluations show that neural machine translation is able to accurately generate rationalizations that describe agent behavior, and that rationalizations are more satisfying to humans than other alternative methods of explanation.

Subfields: Artificial Intelligence, 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 Natural Language Processing

Citations in the corpus

Program Induction by Rationale Generation: Learning to Solve and Explain Algebraic Word Problems (2017, background): “This work is thus related to models that can explain or rationalize their decisions (Hendricks et al., 2016; Harrison et al., 2017).”
Reframing Human-AI Collaboration for Generating Free-Text Explanations (2021, background): “…paradigm for producing free-text explanations is via direct supervision, i.e., training an autoregressive, generative language model to predict human-authored explanations directly (Kim et al., 2018; Park et al., 2018; Ehsan et al., 2018; Narang et al., 2020; Wiegreffe et al., 2021, i.a. ).”
Few-Shot Self-Rationalization with Natural Language Prompts (2021, background): “On the other hand, a standard approach to generating free-text explanations is to use human-written explanations (Hendricks et al., 2016; Kim et al., 2018; Hancock et al., 2018; Camburu et al., 2018; Ehsan et al., 2018; Liu et al., 2019; Wu and Mooney, 2019; Narang et al., 2020; Tang et al., 2020; M”
Measuring Association Between Labels and Free-Text Rationales (2020, background): “This is the most common approach to free-text rationalization (Hendricks et al., 2016; Kim et al., 2018; Camburu et al., 2018; Ehsan et al., 2018; Liu et al., 2019; Wu and Mooney, 2019; Narang et al., 2020; Do et al., 2020), but little is understood about its faithfulness.”
Explain Yourself: A Natural Language Interface for Scrutable Autonomous Robots (2018, background): “In previous work, explanations have been categorised as either explaining 1) machine learning as in [11] who showed that they can increase trust; 2) explaining plans [2, 13]; 3) verbalising robot [12] or agent rationalisation [3].”
QED: A Framework and Dataset for Explanations in Question Answering (2020, background): “We take as our departure point the following passage from Ehsan et al. (2018) concerning explain-able AI: Explainability is important in situations where human operators work alongside autonomous and semi-autonomous systems because it can help build rapport, conﬁdence, and understanding be-tween…”
Evaluating XAI: A comparison of rule-based and example-based explanations (2021, unknown): “Several XAI methods try to identify examples to generate such explanations, for example those from [43–47].”
Teach Me to Explain: A Review of Datasets for Explainable NLP (2021, background): “The human-computer interaction (HCI) community even advocates to use the term “rationale” to stress that the goal of explainable AI is to produce explanations that sound like humans (Ehsan et al., 2018).”
Human-centered Explainable AI: Towards a Reflective Sociotechnical Approach (2020, methodology): “Neural Model Configuration We use a 2-layered encoder-decoder recurrent neural network (RNN) [4,31] with attention to teach our network to generate relevant natural language explanations for any given action (for details, see [19]).”
Learning to Faithfully Rationalize by Construction (2020, background): “…predictions than if provided rationales, and their time tends to cost signiﬁcantly more than that of equivalent com-4 ) is x,z ) regularization encour-agerationaleconcisenessandcontiguity input tokens to existing feature 2018; Jain and these efforts by rationales, although secondary aim here.”
Transparency in Deep Reinforcement Learning Networks (2018, methodology): “Similarly, Fig.”
Natural Language Rationales with Full-Stack Visual Reasoning: From Pixels to Semantic Frames to Commonsense Graphs (2020, background): “We follow Ehsan et al. (2018) who define AI rationalization as a process of generating rationales of a model’s behavior as if a human had performed the behavior.”
Translation Tutorial : A Shared Lexicon for Research and Practice in Human-Centered Software Systems (2018, unknown): “To a machine learning researcher, an explanation is a description of the operation of a model, either in general or for a particular test vector, which covers the mechanism used to relate inputs to outputs (e.g. Harrison et al. (2017); DoshiVelez and Kim (2017)).”
Cross-Domain Transfer of Generative Explanations Using Text-to-Text Models (2021, background): “Other previous works have also studied generative explanations for non-NLP tasks [12,16,22].”
Explainable Agents and Robots: Results from a Systematic Literature Review (2019, background): “Other workers relied on; proxemics [84], rationalization [27], and signal detection theory (Psychophysics) [68].”