The Kellogg School of Management at Northwestern University (also known as Kellogg) is the business school of Northwestern University, located in the Chicago Metropolitan Area, Illinois. Founded in 1908, Kellogg is one of the oldest business schools in the world and has made significant contributions to fields such as marketing, management sciences, and decision sciences. Kellogg's 2-year MBA Program is ranked No. 3 in the U.S. by U.S. News & World Report and Forbes and No. 4 globally by The Economist.Notable Kellogg alumni include Peter G. Peterson, founder of private equity firm The Blackstone Group; Arthur E. Andersen, founder of Accenture; James L. Allen and Edwin G. Booz, founders of Booz Allen Hamilton; Patrick Ryan, founder of Aon Corporation; David Kabiller, founder of hedge fund AQR Capital; and current and former CEOs of Fortune 500 companies such as Allstate, Diageo, DuPont, ExxonMobil, Illinois Tool Works, Kohler, Kraft Foods, Mattel, Motorola, SAP, ServiceNow, Target Corporation, and Union Pacific Corporation.
We propose a new framework for machine-learning-oriented argument analysis tasks. Our proposal involves replacing traditional contextualized word embeddings used in most NLP tasks with logical embeddings, an alternative encoding that directly exploits argumentation structures. In essence, logical embeddings encapsulate the logical semantics of an argument, allowing for a better representation of its meaning. Supporting these embeddings is a mathematical logic-based similarity measure that offers a transparent notion of proximity and is guaranteed to satisfy several desirable theoretical properties that current cosine similarity-based contextualized word embeddings cannot assure. This similarity measure induces a positive semi-definite kernel on the set of arguments, enabling us to uniquely define logical embeddings using the theory of Reproducing Kernel Hilbert Spaces (RKHS). Moreover, we prove that this encoding is optimal, in the sense that no logical information is lost in the process. As with other RKHS applications, logical embeddings can be used in numerous supervised and unsupervised tasks. We provide an implementation of the method and aim to test it against literature benchmarks. Additionally, we demonstrate that logical embeddings outperform most standard embedding methods on a classification task.
Prior studies predominantly examine fundamental performance-driven explanations of loan renegotiations. We contrast with this work by investigating the improvement in secondary loan market trading conditions as a non-fundamental driver of loan renegotiation. Exploiting a regression discontinuity design around the LSTA 100 Index reconstitution, we find that index-included loans are around five times more likely to receive interest-rate-reducing amendments than comparable loans just below the index inclusion threshold. Within-loan-package tests confirm that these renegotiations are not driven by changes in the borrower's fundamental performance. The threat of refinancing likely drives this effect, as the results are more pronounced when such threats are more credible.
The transition to green energy is causing rising demand for critical minerals and expanded mining. While mines must obtain permits from local authorities, they must also achieve a social license to operate in their communities. With this relationship in place, project delays from community 2 opposition are less likely to delay mining projects and communities are likely to be more satisfied with mining projects. Herein, we use public comments about projects in the U.S. and Chile as a data source and propose a modelling framework that combines sentiment analysis to characterize overall public attitudes with topic modelling to identify and rank the mining impacts of greatest public concern. The proposed approach facilitates standardized assessment of public sentiment about mining across multiple cultural contexts and supports more rapid responses to public feedback.
Neonatal sepsis is a leading cause of infant morbidity and mortality in low- and middle-income countries (LMICs), yet despite this, progress in the development and implementation of effective diagnostics has been limited. Current diagnostic methods rely on clinical assessments of nonspecific signs and symptoms, limited laboratory testing and technologies that are inaccessible, unaffordable or unsuitable for resource-constrained settings. There is an urgent need for reliable, affordable and context-appropriate point-of-care diagnostics. This analysis synthesises evidence across current and emerging diagnostic strategies for neonatal sepsis, including clinical algorithms, host-response biomarkers, haematological parameters, continuous monitoring and artificial intelligence-based digital platforms. We assess their performance, operational feasibility in LMICs and alignment with the WHO Target Product Profile (TPP) for neonatal sepsis, including requirements for highly sensitive triage, feasibility in non-hospital settings and suitability for use at the point of care in LMICs. We identified promising existing technologies and emerging approaches that could be used to address the challenges of diagnosing neonatal sepsis in LMICs. These include host-response biomarkers that offer greater accuracy than those currently used, as well as affordable technologies that can improve current predictive algorithms. Taken together, these findings support a move towards integrated, TPP-aligned diagnostic strategies that combine affordable point-of-care testing, routine clinical data and digital decision support to improve early detection, rationalise antibiotic use and reduce neonatal mortality.