GJGard Buen Jenset

Gard Buen Jenset

*This is an algorithmically generated author record based on publicly available information.

Gard Buen Jenset has a total of 21 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    A large-scale, granular topic classification system for scientific documents

    by Gard B. Jenset, Peter J. Bevan, Akarsh Jain

    Knowledge Organisation Systems (KOSs) are crucial for search, retrieval, and analysis of the vast volumes of academic research, but KOSs are challenging to scale in a way that preserves granularity, breadth, and quality. Topics constructed with data-driven algorithms are a key type of KO in scientometrics, and developing high-quality systems for topic construction remains a cornerstone of scientometrics research. We present a topic construction and classification system that advances the state of the art in terms of breadth and granularity, consisting of over 29,000 topics organised into a four-level hierarchy, while achieving a high-quality as measured both quantitatively and via expert judgment. The paper makes three key contributions that address clear gaps in the current state of the art: first, documenting our approach to building a broad and granular topic solution; second, demonstrating that we can train a successful supervised classifier for a large number of topics to assign topics to new documents at scale; third, introducing a new evaluation measure, to measure topic coherence at scale. The paper exemplifies how citation clustering and Natural Language Processing (NLP) can be flexibly wielded together within scientometrics to advance the state of the art.

  • Open Access

    Predicting future grant amounts using topic-level features

    by Gard B Jenset

    This study addresses research funding allocation, in the form of a large-scale empirical investigation into the structural factors, at the topic-level, that predict funding levels in the form of aggregated grant amounts, using a lagged panel-data approach. The topic-centric focus sets this work apart from previous research on which features predict success for individual grant applications. Understanding the topic-level dynamics around research funding provides a crucial complement to the individual-grant perspective, with a potential for informing research strategy and scientific priorities. Employing a data-driven approach based on large-scale data across more than 1,100 topics covering over 130 million publications, the study demonstrates that in addition to topic size, signals of socio-economic impact in the form of links to patents and policy documents, as well as citation patterns and aspects of the researcher community active in the topic, are significant predictors of future funding levels. A model triangulation approach, combining conditional inference trees, linear mixed effects regression, and Random Forest, reveals a clear hierarch of predictor importance across different statistical models. Clear signals of socio-economic impact driving research funding emerge both from this hierarchy of effects, as well as an outlier analysis. Taken together, the results provide compelling evidence for the structural, topic-level traces of socio-economic impact that influence research grants.

  • Open Access

    A close look at children’s and adolescents’ arguments: combining a developmental, educational, and philosophical perspective

    by Chrysi Rapanta, Fabrizio Macagno, Gard Jenset

    It is broadly admitted that social contexts of reasoning may prompt children and adolescents to improve the quality of their reasoning. However, it is not clear how this quality may be assessed when it comes to arguments expressed within oral interactions in diverse settings (whole-class or small-group discussions) by students of different ages and cultural backgrounds. This study aims to offer a methodological contribution to the issue of oral argument assessment of children and adolescents, by looking at a large, annotated corpus of dialogic discussions during 111 lessons taking place in five countries. Our analysis combines a structural (Toulmin Argument Pattern) and functional (Walton’s argumentation schemes) approach to argumentative reasoning. Our findings show significant variations across age groups and social settings. The discussion points out the importance of sociocultural framing of argument reasoning development and the continuation of research in argument assessment methods able to grasp important developmental and cultural variations.


  • Open Access

    Argument structure constructions in competition: The Dat-Nom/Nom-Dat alternation in Icelandic

    by Joren Somers, Gard B. Jenset, Jóhanna Barðdal

    Alternating Dat-Nom/Nom-Dat verbs in Icelandic are notorious for instantiating two diametrically opposed argument structures: the Dat-Nom and the Nom-Dat construction. We conduct a systematic study of the relevant verbs to uncover the factors steering the alternation. This involves a comparison of 15 verbs, five alternating ones, and as a control, five Nom-Dat verbs and five non-alternating Dat-Nom verbs. Our findings show that alternating verbs instantiate the Nom-Dat construction 54% of the time and the Dat-Nom construction 46% of the time on average for four of five verbs when both arguments are full NPs. However, in configurations with a nominative pronoun, the Nom-Dat construction takes precedence over the Dat-Nom construction. Also, for the double-NP configuration, a logistic regression analysis identifies indefiniteness and length as two key predictors, apart from nominative case marking. We demonstrate that the latter systematically correlates with discourse-prominence, which we show, upon closer inspection, correlates with topicality.


  • Open Access

    Investigating patterns of change, stability, and interaction among scientific disciplines using embeddings

    by Barbara McGillivray, Gard B. Jenset, Khalid Salama, Donna Schut

    Multi-disciplinary and inter-disciplinary collaboration can be an appropriate response to tackling the increasingly complex problems faced by today’s society. Scientific disciplines are not rigidly defined entities and their profiles change over time. No previous study has investigated multiple disciplinarity (i.e. the complex interaction between disciplines, whether of a multidisciplinary or an interdisciplinary nature) at scale with quantitative methods, and the change in the profile of disciplines over time. This article explores a dataset of over 21 million articles published in 8400 academic journals between 1990 and 2019 and proposes a new scalable data-driven approach to multiple disciplinarity. This approach can be used to study the relationship between disciplines over time. By creating vector representations (embeddings) of disciplines and measuring the geometric closeness between the embeddings, the analysis shows that the similarity between disciplines has increased over time, but overall the size of their neighbourhood (the number of neighbouring disciplines) has decreased, pointing to disciplines being more similar to each other over time, while at the same time displaying increased specialisation. We interpret this as a pattern of global convergence combined with local specialisation. Our approach is also able to track the development of disciplines’ profiles over time, detecting those that changed the most in the time period considered, and to treat disciplines as compositional units, where relationships can be expressed as analogy equations of the form Discipline1 + Discipline2 ≈ Discipline3. These findings can help researchers, academic institutions and organizations to better understand and react to the dynamics of scientific research, and can support the education sector in designing curricula or in the recruitment of academics and researchers.


  • Open Access

    Enhancing Domain-Specific Supervised Natural Language Intent Classification with a Top-Down Selective Ensemble Model

    by Gard B. Jenset, Barbara McGillivray

    Natural Language Understanding (NLU) systems are essential components in many industry conversational artificial intelligence applications. There are strong incentives to develop a good NLU capability in such systems, both to improve the user experience and in the case of regulated industries for compliance reasons. We report on a series of experiments comparing the effects of optimizing word embeddings versus implementing a multi-classifier ensemble approach and conclude that in our case, only the latter approach leads to significant improvements. The study provides a high-level primer for developing NLU systems in regulated domains, as well as providing a specific baseline accuracy for evaluating NLU systems for financial guidance.


  • Historical corpus annotation

    by Gard B. Jenset, Barbara McGillivray

    Chapter 4 explains the concept and process of annotation for historical corpora, from a theoretical, practical, and technical point of view, and discusses the challenges presented by historical texts. We introduce basic terminology for XML technologies and corpus metadata, and we describe the different levels of linguistic annotation, from spelling normalization to morphological, syntactic, and semantic analysis, and briefly present the state of the art for historical corpora and treebanks. We cover annotation schemes and standards and illustrate the main concepts in corpus annotation with an example from LatinISE, a large annotated Latin corpus.

  • The role of numbers in historical linguistics

    by Gard B. Jenset, Barbara McGillivray

    Chapter 6 discusses the relationship between numbers or frequencies and historical linguistic arguments. Grounded in the principles laid out in Chapter 2, the benefits of using arguments based on quantification are discussed and explicitly connected to the general aims of historical linguistics. The chapter offers an introductory overview of state-of-the art multivariate statistical methods, and discusses why these are preferred over simpler, more commonly used statistical null-hypothesis tests. In particular, the discussion goes into how multivariate techniques are able to handle several potentially explanatory factors simultaneously. An extended case study of the rise of the so-called existential there in Middle English illustrates these points. Using the case of existential there, the chapter shows that simple null-hypothesis tests are insufficient for the complexities of historical linguistic research. Instead, multivariate techniques provide the means to understand the data and assess competing claims from the existing literature.