Abstract screening, a labor-intensive aspect of systematic review, is increasingly challenging due to the rising volume of scientific publications. Recent advances suggest that generative large language models like generative pre-trained transformer (GPT) could aid this process by classifying references into study types such as randomized-controlled trials (RCTs) or animal studies prior to abstract screening. However, it is unknown how these GPT models perform in classifying such scientific study types in the biomedical field. Additionally, their performance has not been directly compared with earlier transformer-based models like bidirectional encoder representations from transformers (BERT). To address this, we developed a human-annotated corpus of 2,645 PubMed titles and abstracts, annotated for 14 study types, including different types of RCTs and animal studies, systematic reviews, study protocols, case reports, as well as in vitro studies. Using this corpus, we compared the performance of GPT-3.5 and GPT-4 in automatically classifying these study types against established BERT models. Our results show that fine-tuned pretrained BERT models consistently outperformed GPT models, achieving F1-scores above 0.8, compared to approximately 0.6 for GPT models. Advanced prompting strategies did not substantially boost GPT performance. In conclusion, these findings highlight that, even though GPT models benefit from advanced capabilities and extensive training data, their performance in niche tasks like scientific multi-class study classification is inferior to smaller fine-tuned models. Nevertheless, the use of automated methods remains promising for reducing the volume of records, making the screening of large reference libraries more feasible. Our corpus is openly available and can be used to harness other natural language processing (NLP) approaches.

The ontological complexity of the twentieth-century Cold War motivates this special issue’s investigation of how social scientists conceptualize institutional novelty and change. We begin by noting the peculiar elision of the Cold War as an explanatory mechanism in mainstream sociology, even while sociologists have theoretical tools for making sense of the phenomenon: war, schema, field, world systems, and empire. All are useful; none are sufficient. We locate the explanatory problem in a tension between notions of structure and event that has organized debate in historical social science for several scholarly generations, and offer a new intellectual tool – medium durée – as a way forward. Medium durée describes phenomena that have sufficient cohesion as ideas and relationships to endure over time, yet remain sufficiently unfixed and ambiguous as to enable multifarious action and sensemaking. Our notion of medium durée is substantially informed by the articles and commentaries assembled for this special issue, which represent three years of dialogue among the authors as well as audiences in serial panels at the 2022 and 2023 annual meetings of the Social Science History Association.

Chinese Stellaria (Stellera chamaejasme L.) is a indicator plant of degraded grasslands. With its robust vitality, once it emerges in grassland ecosystems, it undergoes extensive growth and rapid expansion, leading to grassland degradation as an inevitable consequence. The establishment and invasion of S. chamaejasme disrupts the ecological balance of grasslands in Qilian Mountains. This study was conducted in a grassland on the eastern slope of the Qilian Mountains, employing point pattern analysis to investigate the spatial distribution patterns of S. chamaejasme and the relationships among different age classes. The population was categorized into three growth stages: young, sub-adult, and mature plants. The results revealed that the spatial distribution of this population is primarily dominated by subadult plants, accounting for up to 75.48% of the total, with an overall transition trend from aggregated to random distribution. No significant spatial correlations were observed among different age groups (young, subadult plants, and mature plants), indicating that mature plants do not exert significant inhibitory effects on the growth of young individuals. In high-density areas, the population exhibited a radiating distribution outward from mature plants as the center, with high-density cores predominantly concentrated within 0-5 m. Significant density variations were observed between regions, with the highest total density estimated at approximately 9.57 plants/m2 and the lowest at 2.68 plants/m². The invasion mechanism of S. chamaejasme is closely associated with the spatial independence of its age groups and a distribution pattern dominated by subadult plants. During the initial invasion phase, S. chamaejasme spreads predominantly around mature plants. After securing sufficient growing space (0–1 m), it further competes for territory through shifts in distribution patterns—transitioning from aggregated to random distribution. Additionally, significant differences in distribution density and expansion patterns across regions provide critical theoretical foundations for targeted ecological management strategies.