A review of neural approaches to the Question Answering task
Will Needham, City, University of London, Northampton Square, London, EC1V 0HB
The Question Answering task, whereby a system receives a plain language question from a user and returns a concise answer from a corpus of documents, has received considerable attention from academia and the commercial world since mid-way through the 20th century. This paper offers a concise overview of this literature, focussing on recent advancements of the state-of-the-art achieved by neural network-based
approaches. The rate of change of these advancements is considerable and has left a sparse landscape of analysis and research still to be conducted. My main contribution in this paper is to shine a light on these gaps in the literature, offering inspiration for future research in this domain.
Information Retrieval, Question Answering, Neural Networks, Word embeddings, Pre-trained language models
Deep Attention-Based Review Level Sentiment Analysis for Arabic Reviews
Nada M. Almani1 and Lilian h.Tang2, 1Department of Information Technology, King Abdul Aziz University, Jeddah, Saudi Arabia and 2Departement of Computing, University of Surrey, Surrey, United Kingdom
Sentiment analysis is a branch of machine learning that concerns about classifying the polarity for a given text. Recently, it gained a lot of interest because of the availability of huge amount of opinionated data that needs to be analyzed and interpreted. Recently, using deep learning Artificial Neural Network (ANN) architecture has showed significant improvements with high tendency to reveal the underlining semantic meaning in the input text. However, the output of these models could not be explained and the efficiency could not be analyzed because ANN models are considered as a black box and the success of these models comes at cost of interpretability. The main motivation of this work is to develop Arabic sentiment analysis system that understands review semantics, without using any linguistic resources. Different scenarios and architectures were examined to test the ability of the proposed model to extract salient words out of the input. The results proved the ability of the proposed model to understand a given review by highlighting the most informative words to the class label. The model detected Arabic linguistic features, such as negation and intensification, efficiently. In addition, proposed models are supporting visualization option to get intuitive explanation of the output. The effect of applying transfer learning technique on the problem of Arabic sentiment analysis is experimented as well.
Natural language Processing, Sentiment Analysis, Neural Network, Deep Learning, Attention Mechanism
COBAALT: A Model for Determining the Degree of Literariness of English Literature Classics
Tess Crosbie, University of Bedfordshire, United Kingdom
Using tools developed mainly for authorship authentication, the study introduces a model named CoBAALT (computer-based aesthetic analysis of literary texts) which enables computers to detect aesthetic qualities in literature. An investigation is carried out into current and historic literary criticism to determine how texts can be classified as ˇ°good literatureˇ±. Taking Classics as a genre, 100 mainly fiction texts are taken from the Gutenberg Project and ranked. Factor analysis and mean averages determine the metrics that determine the literary quality, and these are qualified by the CoBAALT model. In testing the model, it is found to conform to human definitions of literariness and is remarkably adept at differentiating between fiction and non-fiction. To confirm its efficacy, CoBAALT assesses texts by Jane Austen and D. H. Lawrence and calculates the degree to which they conform to peer-reviewed literary criticism.
Aesthetic Qualities, Computational Stylistics, Factor Analysis, Literary Merit, Stylistic Analysis.