Isaiah Onando Mulang’


pdf bib
Fine-tuning BERT with Focus Words for Explanation Regeneration
Isaiah Onando Mulang’ | Jennifer D’Souza | Sören Auer
Proceedings of the Ninth Joint Conference on Lexical and Computational Semantics

Explanation generation introduced as the world tree corpus (Jansen et al., 2018) is an emerging NLP task involving multi-hop inference for explaining the correct answer in multiple-choice QA. It is a challenging task evidenced by low state-of-the-art performances(below 60% in F-score) demonstrated on the task. Of the state-of-the-art approaches, fine-tuned transformer-based (Vaswani et al., 2017) BERT models have shown great promise toward continued system performance improvements compared with approaches relying on surface-level cues alone that demonstrate performance saturation. In this work, we take a novel direction by addressing a particular linguistic characteristic of the data — we introduce a novel and lightweight focus feature in the transformer-based model and examine task improvements. Our evaluations reveal a significantly positive impact of this lightweight focus feature achieving the highest scores, second only to a significantly computationally intensive system.


pdf bib
Team SVMrank: Leveraging Feature-rich Support Vector Machines for Ranking Explanations to Elementary Science Questions
Jennifer D’Souza | Isaiah Onando Mulang’ | Sören Auer
Proceedings of the Thirteenth Workshop on Graph-Based Methods for Natural Language Processing (TextGraphs-13)

The TextGraphs 2019 Shared Task on Multi-Hop Inference for Explanation Regeneration (MIER-19) tackles explanation generation for answers to elementary science questions. It builds on the AI2 Reasoning Challenge 2018 (ARC-18) which was organized as an advanced question answering task on a dataset of elementary science questions. The ARC-18 questions were shown to be hard to answer with systems focusing on surface-level cues alone, instead requiring far more powerful knowledge and reasoning. To address MIER-19, we adopt a hybrid pipelined architecture comprising a featurerich learning-to-rank (LTR) machine learning model, followed by a rule-based system for reranking the LTR model predictions. Our system was ranked fourth in the official evaluation, scoring close to the second and third ranked teams, achieving 39.4% MAP.