Edwin Simpson
YOU?
Author Swipe
View article: Active night-time tweeting is associated with meaningfully lower mental wellbeing in a UK birth cohort study
Active night-time tweeting is associated with meaningfully lower mental wellbeing in a UK birth cohort study Open
It has been suggested that use of social media late at night could lead to worse mental health outcomes. We linked Twitter (‘X’) data to self-reported measures of mental health from the Avon Longitudinal Study of Parents and Children. We a…
View article: Out-of-Distribution Detection with Attention Head Masking for Multi-modal Document Classification
Out-of-Distribution Detection with Attention Head Masking for Multi-modal Document Classification Open
Detecting out-of-distribution (OOD) data is critical for ensuring the reliability and safety of deployed machine learning systems by mitigating model overconfidence and misclassification. While existing OOD detection methods primarily focu…
View article: Efficient Methods for Natural Language Processing: A Survey
Efficient Methods for Natural Language Processing: A Survey Open
Recent work in natural language processing (NLP) has yielded appealing results from scaling model parameters and training data; however, using only scale to improve performance means that resource consumption also grows. Such resources inc…
View article: How well can LLMs Grade Essays in Arabic?
How well can LLMs Grade Essays in Arabic? Open
This research assesses the effectiveness of state-of-the-art large language models (LLMs), including ChatGPT, Llama, Aya, Jais, and ACEGPT, in the task of Arabic automated essay scoring (AES) using the AR-AES dataset. It explores various e…
View article: Designing Essay Questions for Effective Automatic Scoring
Designing Essay Questions for Effective Automatic Scoring Open
The domain of automatic essay scoring (AES) has increasingly garnered attention, buoyed by advancements in natural language processing (NLP) and deep learning. Despite the progress, much of the existing research has narrowly focused on dev…
View article: Active night-time Tweeting is associated with meaningfully lower mental wellbeing in a UK Birth Cohort Study (Preprint)
Active night-time Tweeting is associated with meaningfully lower mental wellbeing in a UK Birth Cohort Study (Preprint) Open
BACKGROUND Legislation or interventions aiming to improve social media safety should be informed by evidence. Most previous research has focused on the impact of the frequency with which people use social media, and finds conflicting resu…
View article: Using Similarity to Evaluate Factual Consistency in Summaries
Using Similarity to Evaluate Factual Consistency in Summaries Open
Cutting-edge abstractive summarisers generate fluent summaries, but the factuality of the generated text is not guaranteed. Early summary factuality evaluation metrics are usually based on n-gram overlap and embedding similarity, but are r…
View article: Out-of-Distribution Detection with Attention Head Masking for Multimodal Document Classification
Out-of-Distribution Detection with Attention Head Masking for Multimodal Document Classification Open
Detecting out-of-distribution (OOD) data is crucial in machine learning applications to mitigate the risk of model overconfidence, thereby enhancing the reliability and safety of deployed systems. The majority of existing OOD detection met…
View article: Automated essay scoring in Arabic: a dataset and analysis of a BERT-based system
Automated essay scoring in Arabic: a dataset and analysis of a BERT-based system Open
Automated Essay Scoring (AES) holds significant promise in the field of education, helping educators to mark larger volumes of essays and provide timely feedback. However, Arabic AES research has been limited by the lack of publicly availa…
View article: Medfluencer: A Network Representation of Medical Influencers' Identities and Discourse on Social Media
Medfluencer: A Network Representation of Medical Influencers' Identities and Discourse on Social Media Open
In our study, we first constructed a dataset from the tweets of the top 100 medical influencers with the highest Influencer Score during the COVID-19 pandemic. This dataset was then used to construct a socio-semantic network, mapping both …
View article: The dynamics of emotion expression on Twitter and mental health in a UK longitudinal study
The dynamics of emotion expression on Twitter and mental health in a UK longitudinal study Open
Introduction & BackgroundAn estimated 4.95 billion people used social media in 2023, with the average user active on around seven platforms for over two hours per day. This widespread use leads to abundant digital footprint data around int…
View article: Automated Radiology Report Generation: A Review of Recent Advances
Automated Radiology Report Generation: A Review of Recent Advances Open
Increasing demands on medical imaging departments are taking a toll on the radiologist's ability to deliver timely and accurate reports. Recent technological advances in artificial intelligence have demonstrated great potential for automat…
View article: Towards Abstractive Timeline Summarisation Using Preference-Based Reinforcement Learning
Towards Abstractive Timeline Summarisation Using Preference-Based Reinforcement Learning Open
This paper introduces a novel pipeline for summarising timelines of events reported by multiple news sources. Transformer-based models for abstractive summarisation generate coherent and concise summaries of long documents but can fail to …
View article: Analysis of ‘One in a Million’ primary care consultation conversations using natural language processing
Analysis of ‘One in a Million’ primary care consultation conversations using natural language processing Open
Background Modern patient electronic health records form a core part of primary care; they contain both clinical codes and free text entered by the clinician. Natural language processing (NLP) could be employed to generate these records th…
View article: Efficient Methods for Natural Language Processing: A Survey
Efficient Methods for Natural Language Processing: A Survey Open
Recent work in natural language processing (NLP) has yielded appealing results from scaling model parameters and training data; however, using only scale to improve performance means that resource consumption also grows. Such resources inc…
View article: Towards Abstractive Timeline Summarisation using Preference-based Reinforcement Learning
Towards Abstractive Timeline Summarisation using Preference-based Reinforcement Learning Open
This paper introduces a novel pipeline for summarising timelines of events reported by multiple news sources. Transformer-based models for abstractive summarisation generate coherent and concise summaries of long documents but can fail to …
View article: Efficient Methods for Natural Language Processing: A Survey
Efficient Methods for Natural Language Processing: A Survey Open
Recent work in natural language processing (NLP) has yielded appealing results from scaling model parameters and training data; however, using only scale to improve performance means that resource consumption also grows. Such resources inc…
View article: Clustering Tags in Enterprise and Web Folksonomies
Clustering Tags in Enterprise and Web Folksonomies Open
Tags lack organizational structure limiting their utility for navigation. We present two clustering algorithms that improve this by organizing tags automatically. We apply the algorithms to two very different datasets, visualize the result…
View article: Assisting Decision Making in Scholarly Peer Review: A Preference Learning Perspective
Assisting Decision Making in Scholarly Peer Review: A Preference Learning Perspective Open
Peer review is the primary means of quality control in academia; as an outcome of a peer review process, program and area chairs make acceptance decisions for each paper based on the review reports and scores they received. Quality of scie…
View article: Ranking Scientific Papers Using Preference Learning.
Ranking Scientific Papers Using Preference Learning. Open
Peer review is the main quality control mechanism in academia. Quality of scientific work has many dimensions; coupled with the subjective nature of the reviewing task, this makes final decision making based on the reviews and scores there…
View article: SemEval-2021 Task 12: Learning with Disagreements
SemEval-2021 Task 12: Learning with Disagreements Open
This repository contains the Post-Evaluation data for SemEval-2021 Task 12: Learning with Disagreement, a shared task on learning to classify with datasets containing disagreements. The aim of this shared task is to provide a unified testi…
View article: SemEval-2021 Task 12: Learning with Disagreements
SemEval-2021 Task 12: Learning with Disagreements Open
This repository contains the Post-Evaluation data for SemEval-2021 Task 12: Learning with Disagreement, a shared task on learning to classify with datasets containing disagreements. The aim of this shared task is to provide a unified testi…
View article: Aggregating and Learning from Multiple Annotators
Aggregating and Learning from Multiple Annotators Open
The success of NLP research is founded on high-quality annotated datasets, which are usually obtained from multiple expert annotators or crowd workers. The standard practice to training machine learning models is to first adjudicate the di…
View article: SemEval-2021 Task 12: Learning with Disagreements
SemEval-2021 Task 12: Learning with Disagreements Open
Alexandra Uma, Tommaso Fornaciari, Anca Dumitrache, Tristan Miller, Jon Chamberlain, Barbara Plank, Edwin Simpson, Massimo Poesio. Proceedings of the 15th International Workshop on Semantic Evaluation (SemEval-2021). 2021.
View article: A Proposal: Interactively Learning to Summarise Timelines by Reinforcement Learning
A Proposal: Interactively Learning to Summarise Timelines by Reinforcement Learning Open
Timeline Summarisation (TLS) aims to generate a concise, time-ordered list of events described in sources such as news articles. However, current systems do not provide an adequate way to adapt to new domains nor to focus on the aspects of…
View article: Improving Factual Consistency Between a Response and Persona Facts
Improving Factual Consistency Between a Response and Persona Facts Open
Neural models for response generation produce responses that are semantically plausible but not necessarily factually consistent with facts describing the speaker's persona. These models are trained with fully supervised learning where the…