Julie Clover
with the launch of iOS 18.4, Apple introduced a new App Store function, which summarizes several user reviews to provide a resume regarding what people think about the application or game. In a new message on the blog in his blog on the study of machine learning, Apple provides several details on how & zwnj; App Store & Zwnj; Work of reviews work. The user's voice. Apple says that in his resume – “security, justice, truthfulness and usefulness”, at the same time setting out some problems in aggregation & zwnj; App Store & Zwnj; Reviews. Some reviews also include non -core comments or the noise that LLM should filter. Then, the remaining reviews are processed sequentially consistently by LLM modules, which extract key ideas from each review, aggregate topics that repeat, balancing positive and negative arcs, and then generating a length of 100 to 300 characters. During the development of this function, the person’s appraisers were considered to evaluate factors such as usefulness, composition and safety.
