Chris van der Lee

Tilburg University | UVT · Department of Communication and Information Sciences

Medication reconciliation, the process of documenting a patient's medication, is currently a time-consuming and labor-intensive process. To make medication reconciliation more efficient, digital assistants (DAs) offer a promising solution. Especially since human-like digital interfaces tend to be appreciated by more vulnerable populations such as p...

Chatbots have several features that may stimulate self-disclosure, such as accessibility, anonymity, convenience and their perceived non-judgmental nature. The aim of this study is to investigate if people disclose (more) intimate information to a chatbot, compared to a human, and to what extent this enhances their emotional well-being through feel...

This paper is part of the larger ReproHum project, where different teams of researchers aim to reproduce published experiments from the NLP literature. Specifically, ReproHum focuses on the reproducibility of human evaluation studies, where participants indicate the quality of different outputs of Natural Language Generation (NLG) systems. This is...

This study discusses the effect of semi-supervised learning in combination with pretrained language models for data-to-text generation. It is not known whether semi-supervised learning is still helpful when a large-scale language model is also supplemented. This study aims to answer this question by comparing a data-to-text system only supplemented...

We report our efforts in identifying a set of previous human evaluations in NLP that would be suitable for a coordinated study examining what makes human evaluations in NLP more/less reproducible. We present our results and findings, which include that just 13\% of papers had (i) sufficiently low barriers to reproduction, and (ii) enough obtainable...

Background and Importance Medication reconciliation has become standard care to obtain a complete overview of the current medication of a patient. However, it is time-consuming and labour-intensive. Studies have shown promising results for online medication reconciliation preparation done by patients. Nonetheless, there is a need for enhanced patie...

With chatbots and other types of conversational agents becoming more ubiquitous in everyday life, we see a need for tools and frameworks to structure the design and evaluation of this new technology, to optimize its effectiveness and user experience. From existing literature, we have identified eight domains from which the quality of conversational...

In this paper, we describe our reproduction effort of the paper: Towards Best Experiment Design for Evaluating Dialogue System Output by Santhanam and Shaikh (2019) for the 2022 ReproGen shared task. We aim to produce the same results, using different human evaluators, and a different implementation of the automatic metrics used in the original pap...

This study discusses the effect of semi-supervised learning in combination with pretrained language models for data-to-text generation. It is not known whether semi-supervised learning is still helpful when a large-scale language model is also supplemented. This study aims to answer this question by comparing a data-to-text system only supplemented...

While many models attempt to explain the aesthetic experience, most limit themselves to art as their focal point and only a few look into why we arrive at a certain response to a visual aesthetic object. This article attempts to offer an extension to the current models by focusing on the mechanisms that induce emotions in relation to visual aesthet...

This paper introduces a new corpus of paired football match reports, the Multilingual Emotional Football Corpus, (MEmoFC), which has been manually collected from English, German, and Dutch websites of individual football clubs to investigate the way different emotional states (e.g. happiness for winning and disappointment for losing) are realized i...

Currently, there is little agreement as to how Natural Language Generation (NLG) systems should be evaluated, with a particularly high degree of variation in the way that human evaluation is carried out. This paper provides an overview of how (mostly intrinsic) human evaluation is currently conducted and presents a set of best practices, grounded i...

Preregistration refers to the practice of specifying what you are going to do, and what you expect to find in your study, before carrying out the study. This practice is increasingly common in medicine and psychology, but is rarely discussed in NLP. This paper discusses preregistration in more detail, explores how NLP researchers could preregister...

This paper describes the CACAPO dataset, built for training both neural pipeline and end-to-end data-to-text language generation systems. The dataset is multilingual (Dutch and English), and contains almost 10,000 sentences from human-written news texts in the sports, weather, stocks, and incidents domain, together with aligned attribute-value pair...

e-Mental health applications may provide a solution for un-derstaffing issues on the workers' side as well as issues regarding help-seeking (e.g. stigma, high costs) on the patients' side in the mental healthcare domain. Especially the use of conversational AI is seen as a promising solution for these issues. While initial research in this area sho...

With the increasing popularity of visual-oriented social media platforms, the prevalence of visual brand-related User Generated Content (UGC) have increased. Monitoring such content is important as this visual brand-related UGC can have a large influence on a brand's image and hence provides useful opportunities to observe brand performance (e.g.,...

In this paper, we present a novel data-to-text system for cancer patients, providing information on quality of life implications after treatment , which can be embedded in the context of shared decision making. Currently, information on quality of life implications is often not discussed, partly because (until recently) data has been lacking. In ou...

Traditionally, most data-to-text applications have been designed using a modular pipeline architecture, in which non-linguistic input data is converted into natural language through several intermediate transformations. In contrast, recent neural models for data-to-text generation have been proposed as end-to-end approaches, where the non-linguisti...

This study uses two methods to examine whether online daters looking for a long-term relationship behave linguistically different in their profile texts compared to daters seeking casual relationships. To investigate these linguistic differences, 12,310 existing Dutch dating profiles were analyzed using the Linguistic Inquiry and Word Count (LIWC)...