Query-focused Summarization (QfS) deals with systems that generate summaries from document(s) based on a query. Motivated by the insight that Reinforcement Learning (RL) provides a generalization to Supervised Learning (SL) for Natural Language Generation, and thereby performs better (empirically) than SL, we use an RL-based approach for this task of QfS. Additionally, we also resolve the conflict of employing RL in Transformers with Teacher Forcing. We develop multiple Policy Gradient networks, trained on various reward signals: ROUGE, BLEU, and Semantic Similarity, which lead to a 10-point improvement over the State-of-the-Art approach on the ROUGE-L metric for a benchmark dataset (ELI5). We also show performance of our approach in zero-shot setting for another benchmark dataset (DebatePedia) – our approach leads to results comparable to baselines, which were specifically trained on DebatePedia. To aid the RL training, we propose a better semantic similarity reward, enabled by a novel Passage Embedding scheme developed using Cluster Hypothesis. Lastly, we contribute a gold-standard test dataset to further research in QfS and Long-form Question Answering (LfQA).

Most of the existing state of the art sentiment classification techniques involve the use of pre-trained embeddings. This paper postulates a generalized representation that collates training on multiple datasets using a Multi-task learning framework. We incorporate publicly available, pre-trained embeddings with Bidirectional LSTM’s to develop the multi-task model. We validate the representations on an independent test Irony dataset that can contain several sentiments within each sample, with an arbitrary distribution. Our experiments show a significant improvement in results as compared to the available baselines for individual datasets on which independent models are trained. Results also suggest superior performance of the representations generated over Irony dataset.