No abstract

Non-Factoid Question Answering (QA) is the next generation of textual QA systems, which gives passage level summaries for a natural language query, posted by the user. The main issue lies in the appropriateness of the generated summary. This paper proposes a framework for non-factoid QA system, which has three main components: (i) A deep neural network classifier, which produces sentence vector considering word correlation and context. (ii) Zero shot classifier that uses a multi-channel Convolutional Neural Network (CNN), to extract knowledge from multiple sources in the knowledge accumulator. This output acts as a knowledge enhancer that strengthens the passage level summary. (iii) Summary generator that uses Maximal Marginal Relevance (MMR) algorithm, which computes similarity among the query related answer and the sentences from zero shot classifier. This model is applied on the datasets WikiPassageQA and ANTIQUE. The experimental analysis shows that this model gives comparatively better results for WikiPassageQA dataset.

Usage of online learning platforms increases day by day and henceforth, there emerges the need for automated grading systems to assess the learner’s performance. Evaluating these answers demands for a well-grounded reference answer which aids a strong foundation for better grading. Since reference answers impacts the exactness of grading answers of learners, its correctness remains a great concern. A framework that addresses the reference answer exactness in Automated Short Answer Grading (ASAG) systems was developed. This framework includes material content acquisition, clustering collective content, expert answer as key components which was later fed to a zero-shot classifier for a strong reference answer generation. Then, the computed reference answers along with student answers and questions from Mohler dataset were fed to an ensemble of transformers to produce relevant grades. The aforementioned models’ RMSE and correlation values were compared against the past values of the dataset. Based on the observations made, this model outperforms the previous approaches.