Effective item identifiers (IDs) are an important component for recommender systems (RecSys) in practice, and are commonly adopted in many use cases such as retrieval and ranking. IDs can encode collaborative filtering signals within training data, such that RecSys models can extrapolate during the inference and personalize the prediction based on users' behavioral histories. Recently, Semantic IDs (SIDs) have become a trending paradigm for RecSys. In comparison to the conventional atomic ID, an SID is an ordered list of codes, derived from tokenizers such as residual quantization, applied to semantic representations commonly extracted from foundation models or collaborative signals. SIDs have drastically smaller cardinality than the atomic counterpart, and induce semantic clustering in the ID space. At Snapchat, we apply SIDs as auxiliary features for ranking models, and also explore SIDs as additional retrieval sources in different ML applications. In this paper, we discuss practical technical challenges we encountered while applying SIDs, experiments we have conducted, and design choices we have iterated to mitigate these challenges. Backed by promising offline results on both internal data and academic benchmarks as well as online A/B studies, SID variants have been launched in multiple production models with positive metrics impact.

Smartphones have become increasingly prevalent and important in our daily lives. To meet users' expectations about the Quality of Experience (QoE) of mobile applications (apps), it is essential to obtain a comprehensive understanding of app QoE and identify the critical factors that affect it. However, effectively and systematically studying the QoE of popular mobile apps such as Facebook and YouTube still remains a challenging task, largely due to a lack of a controlled and reproducible measurement methodology, and limited insight into the complex multi-layer dynamics of the system and network stacks.

SUMMARY Transactional memory is a promising technique for enforcing disciplined access to shared data in a multiprocessor system. Transactional memory simplifies the implementation of a variety of concurrent data structures. In this paper, we study the benefits of a modest, real‐time aware, hardware implementation of transactional memory that we callmicro‐transactions. In particular, we argue that hardware support for micro‐transactions allows us to efficiently implement certain data structures. Those data structures are difficult to realize with the atomic operations provided by stock hardware and provide real‐time guarantees for those operations. Our main implementation platform is the Java Optimized Processor system, a field‐programmable gate array (FPGA) implementation of the Java virtual machine, optimized for real‐time Java. We report on the performance of data structures implemented with locks, atomic instructions, and micro‐transactions. Our results suggest that transactional memory is an interesting alternative to traditional concurrency control mechanisms. Copyright © 2012 John Wiley & Sons, Ltd.

This paper presents a constructive solution to the classic problem of land division. It is the first solution that enables the allocation of higher-dimensional resources without degenerating them first into a series of one-dimensional resource allocation problems. We base our allocation procedure on the topology of overlaps among the regions of interest of different agents. Our result is an algorithm suitable for computer implementation, unlike earlier ones that were only existential in nature. It uses the notion of degree of partial overlap to create a sufficiency condition for the existence of a solution, and proposes a procedure to find the overlaps in such a case. The proposed solution is fair, strategy-proof, non-existential, and does not explicitly need the resource to be measurable. The agents do not have to reveal their private utility functions. We extend our earlier result for one-dimensional resource allocation to this two-dimensional one and explain the distinctive issues involved.