Meet ClarifyDelphi: an interactive system that extracts context from social and moral situations using reinforcement learning

Context is the most important element of effective communication. It shapes the meaning of words so that they are correctly heard and interpreted by the listener. Context is important as it informs the speaker and listener how important to consider something, what inferences to make about what is being communicated, and most importantly, it specifies the meaning behind the message. When making moral decisions and using common sense and moral reasoning in circumstances and social acts, context is just as crucial.

The earlier model, called Delphi, models the moral judgments of individuals in a variety of everyday situations, but lacks the necessary knowledge of the surrounding context. To overcome the limitation of Delphi, the team of researchers has proposed CLARIFYDELPHI as a solution to this problem, which is an interactive system that learns to clarify statements in order to extract the relevant context of a situation and improve moral judgments. The authors have mentioned that this model asks questions like ‘Why did you lie to your friend?’ to get the missing context.

According to the authors, the most instructive questions are those that could lead to answers that lead to different moral judgments. In other words, it shows that context is very important in determining moral judgment if different answers to a question lead to varied moral evaluations. To achieve this, the team has created a reinforcement learning framework with a feasibility reward. This framework maximizes the divergence between moral judgments associated with hypothetical answers to a question. The authors have suggested that Proximal Policy Optimization (PPO) can be used to optimize the generation of questions that elicit answers with context.

JOIN the fastest ML subreddit community

Upon evaluation, CLARIFYDELPHI outperforms other referral methods for generating clarification questions by providing more relevant, informative, and disposable questions. The questions that CLARIFYDELPHI generates have meaningful conclusions, demonstrating the efficiency of its method in obtaining crucial contextual data. The authors have also quantified the amount of supervised clarification question training data required for good initial policy and have shown that questions contribute to nullable updates.

The team’s contributions can be summarized as follows:

1. The team has proposed a technique based on reinforcement learning that defines defeatability as a new form of relevance for clarification questions in order to introduce the task of generating clarification questions for social and moral situations.
2. The team has publicly released δ-CLARIFY, which is a dataset of 33k crowdsourced clarification questions.
3. It has also released δ-CLARIFYsilver, which contains generated questions conditional on a nullable inference dataset.
4. Trained models can be accessed, along with their codes.

The adaptability of human moral reasoning involves defining when a moral rule should apply and recognizing legitimate exceptions in light of contextual requirements. CLARIFYDELPHI creates queries that reveal missing context and enable more accurate moral judgments. Compared to the other approaches, ClarifyDelphi generates more questions, leading to either debilitating or empowering answers.

Consequently, CLARIFYDELPHI looks promising and an incredible model for generating informative and relevant questions that are capable of revealing divergent moral judgments.

review the Paper and github link. Don’t forget to join our 22k+ ML SubReddit, discord channel, and electronic newsletter, where we share the latest AI research news, exciting AI projects, and more. If you have any questions about the article above or if we missed anything, feel free to email us at asif@marktechpost.com

Check out 100 AI tools at AI Tools Club

Definitive Guide to Data Labeling in Machine Learning