Recent survey delivers the first systematic map of LLM tool-learning, dissecting why tools supercharge models and how four-stage workflows (plan→select→call→respond) perform, while cataloging benchmarks and open challenges to jump-start newcomers.

SharpSMT first fuses DPLL(T) polytope enumeration with new factorization, variable-elimination and caching pre-processing, cutting runtime on instances from program-analysis benchmarks by orders of magnitude.

The rapid spread of animal diseases and pathogen evolution have long posed significant threats to the healthy development of animal husbandry. Traditional nucleic acid detection methods such as PCR rely on expensive equipment and professional personnel, making them difficult to meet the needs of on-site scenarios such as farms and rural areas.

Methane is one of the key greenhouse gases affecting global climate. Although its atmospheric lifetime is shorter than that of carbon dioxide, reducing methane emissions in the short term is more effective in slowing global warming.

 SFDS introduces the first bilateral attribute-based access control and puncturable-encryption key update that eliminates the trusted KGC, cutting key-generation and decryption costs while blocking unauthorized uploads and downloads.

 Research team unveils ProSyno, a context-free prompt-learning model that taps Wiktionary descriptions and a dynamic matching encoder to spot synonyms across domains without KGs or corpora, setting new benchmarks on four datasets.

Routine newborn screening (NBS) has transformed early disease detection. However, traditional biochemical tests limit the range of conditions that can be identified at birth. Next-generation sequencing is being explored as a complementary screening tool. A review published in Pediatric Investigation examines how next-generation sequencing could expand NBS from single-disease assays to genome-enabled, multi-disease screening approaches.