SearcharxivSearch

arXiv subjects

Kevin Mc Daid

Publications and source records attributed to Kevin Mc Daid.

4 recordsLinked to original sources

An Under-Explored Application for Explainable Multimodal Misogyny Detection in code-mixed Hindi-English

Digital platforms have an ever-expanding user base, and act as a hub for communication, business, and connectivity. However, this has also allowed for the spread of hate speech and misogyny. Artificial intelligence models have emerged as an effective solution for countering online hate speech but are under explored for low resource and code-mixed languages and suffer from a lack of interpretability. Explainable Artificial Intelligence (XAI) can enhance transparency in the decisions of deep learning models, which is crucial for a sensitive domain such as hate speech detection. In this paper, we present a multi-modal and explainable web application for detecting misogyny in text and memes in code-mixed Hindi and English. The system leverages state-of-the-art transformer-based models that support multilingual and multimodal settings. For text-based misogyny identification, the system utilizes XLM-RoBERTa (XLM-R) and multilingual Bidirectional Encoder Representations from Transformers (mBERT) on a dataset of approximately 4,193 comments. For multimodal misogyny identification from memes, the system utilizes mBERT + EfficientNet, and mBERT + ResNET trained on a dataset of approximately 4,218 memes. It also provides feature importance scores using explainability techniques including Shapley Additive Values (SHAP) and Local Interpretable Model Agnostic Explanations (LIME). The application aims to serve as a tool for both researchers and content moderators, to promote further research in the field, combat gender based digital violence, and ensure a safe digital space. The system has been evaluated using human evaluators who provided their responses on Chatbot Usability Questionnaire (CUQ) and User Experience Questionnaire (UEQ) to determine overall usability.

cs.AI

NLP-SIR: A Natural Language Approach for Spreadsheet Information Retrieval

Spreadsheets are a ubiquitous software tool, used for a wide variety of tasks such as financial modelling, statistical analysis and inventory management. Extracting meaningful information from such data can be a difficult task, especially for novice users unfamiliar with the advanced data processing features of many spreadsheet applications. We believe that through the use of Natural Language Processing (NLP) techniques this task can be made considerably easier. This paper introduces NLP-SIR, a Natural language interface for spreadsheet information retrieval. The results of a recent evaluation which compared NLP-SIR with existing Information retrieval tools are also outlined. This evaluation has shown that NLP-SIR is a more effective method of spreadsheet information retrieval.

cs.SE

Evaluation of an Intelligent Assistive Technology for Voice Navigation of Spreadsheets

An integral part of spreadsheet auditing is navigation. For sufferers of Repetitive Strain Injury who need to use voice recognition technology this navigation can be highly problematic. To counter this the authors have developed an intelligent voice navigation system, iVoice, which replicates common spreadsheet auditing behaviours through simple voice commands. This paper outlines the iVoice system and summarizes the results of a study to evaluate iVoice when compared to a leading voice recognition technology.

cs.HC

Voice-controlled Debugging of Spreadsheets

Developments in Mobile Computing are putting pressure on the software industry to research new modes of interaction that do not rely on the traditional keyboard and mouse combination. Computer users suffering from Repetitive Strain Injury also seek an alternative to keyboard and mouse devices to reduce suffering in wrist and finger joints. Voice-control is an alternative approach to spreadsheet development and debugging that has been researched and used successfully in other domains. While voice-control technology for spreadsheets is available its effectiveness has not been investigated. This study is the first to compare the performance of a set of expert spreadsheet developers that debugged a spreadsheet using voice-control technology and another set that debugged the same spreadsheet using keyboard and mouse. The study showed that voice, despite its advantages, proved to be slower and less accurate. However, it also revealed ways in which the technology might be improved to redress this imbalance.

cs.HC