Search for: All records

Abstract Large language models (LLMs) are capable of successfully performing many language processing tasks zero-shot (without training data). If zero-shot LLMs can also reliably classify and explain social phenomena like persuasiveness and political ideology, then LLMs could augment the computational social science (CSS) pipeline in important ways. This work provides a road map for using LLMs as CSS tools. Towards this end, we contribute a set of prompting best practices and an extensive evaluation pipeline to measure the zero-shot performance of 13 language models on 25 representative English CSS benchmarks. On taxonomic labeling tasks (classification), LLMs fail to outperform the best fine-tuned models but still achieve fair levels of agreement with humans. On free-form coding tasks (generation), LLMs produce explanations that often exceed the quality of crowdworkers’ gold references. We conclude that the performance of today’s LLMs can augment the CSS research pipeline in two ways: (1) serving as zero-shot data annotators on human annotation teams, and (2) bootstrapping challenging creative generation tasks (e.g., explaining the underlying attributes of a text). In summary, LLMs are posed to meaningfully participate in social science analysis in partnership with humans. 

We built an integrated solid-contact ion-selective electrode (SCISE) system with the functionality of self-calibration. A multiplexed SCISE sensor (K+ and NO3− vs. Ag/AgCl) was fabricated on printed-circuit board (PCB) substrates and was subsequently embedded into a microfluidic flow cell for self-calibration and flow-through analysis. A PCB circuit that includes modules for both sensor readout and fluid control was developed. The sensors showed a fast and near-Nernstian response (56.6 for the K+ electrode and −57.4 mV/dec for the NO3− electrode) and maintained their performance for at least three weeks. The sensors also showed a highly reproducible response in an automated two-point calibration, demonstrating the potential for in situ monitoring. Lastly, the sensor system was successfully applied to measure mineral nutrients in plant sap samples. 

We fabricated and evaluated multiplexed ion-selective electrodes (ISEs) by modifying printed circuit board (PCB). The multiplexed sensor consisted of all-solid-state K+ and NO3- ISEs, together with a Ag/AgCl reference. The sensor was further embedded in a microfluidic microchannel for in-line continuous analysis, and was characterized for up to one week of operation. Both ISEs showed a near-Nernstian response (~52 mV/dec) and reasonable stabilities (baseline drift ~2.9 mV/day). The sensor provides a versatile and low-cost tool for monitoring concentrations of different ions in many biomedical, environmental and agricultural applications. 

Abstract Multiplexed solid‐contact ion‐selective electrodes (SCISEs) are fabricated using printed circuit board (PCB) and mesoporous carbon black (MCB) as ion‐to‐electron transducer (solid contact). Four sensor configurations were examined and showed that in addition to MCB, the sensor configuration plays crucial role in the stability of the potential response. The enhanced sensor stability was also linked with suppression of transmembrane flux of water. The sensors exhibited near‐Nernstian sensitivity (58.1 mV/dec for K⁺ISEs and −55.1 mV/dec for NO₃^‐ISEs), low detection limits (1.5–2.2 μM), and good short‐term stability (∼0.1 mV/min). Sensors can be stored dry and used without preconditioning. This work demonstrates a promising approach to combining PCB technology and carbon black for large‐scale production of low cost ISEs for point‐of‐care testing, wearables, orin situfield measurements.