AUTOMATED QUESTION GENERATOR SYSTEM: A REVIEW

Abstract

Question generation, the task of automatically creating questions that can be answered by a certain span of text within a given passage, is important for question-answering and conversational system in digital assistants. Automatic generation of questions from text plays a key role in two domains-interactive question answering sessions and educational assessment. Recent sequence to sequence neural models have outperformed previous rule-based system. Existing models mainly focus on using one or two sentences as the input. In proposed system the admin can add text or paragraphs of his/her choice. User will operate that system hence user can enter the paragraph in English language with grammatically correct sentence. The sentence is selected and separated and Stanford POS tagger for POS Tagging is applied. After the input is given, keywords from a data-set are matched to the input text so as to find the sentence/context on which the question can be created. In feature extraction the system will identify the questionable term from that sentence and rearrange the words and automatically create question from entered sentence or paragraph using Bloom's Taxonomy. However, it often requires the whole paragraph as context in order to generate high quality questions. Proposed system uses Stanford tagger for tagging the sentences with gated self-attention encoder to address the challenges of processing long text input for question generation. With sentence-level input, this model outperforms previous approaches with either sentence or paragraph input. Furthermore proposed model can also effectively utilize paragraphs as inputs.