Etsy (United States)

Embedding-based neural retrieval is a prevalent approach to addressing the semantic gap problem, which often arises in product searches with tail queries. In contrast, popular queries typically lack context and have a broad intent where additional context from users' historical interaction can be helpful. In this paper, we share our novel approach to address both: semantic gap problem with an end-to-end trained model for personalized semantic retrieval. We propose learning a unified embedding model incorporating graph, transformer and termbased embeddings end-to-end and share our design choices for optimal tradeoff between performance and efficiency. We share our learnings in feature engineering, hard negative sampling strategy, and application of transformer model, including a novel pre-training strategy and other tricks for improving search relevance and deploying such a model at industry scale. Our personalized retrieval model significantly improves the overall search experience, as measured by a 5.58% increase in search purchase rate and a 2.63% increase in site-wide conversion rate, aggregated across multiple A/B tests -on live traffic.

Abstract The principal explanations in the existing economics literature for the formation of concentrated markets are intellectual property-related entry barriers, economies of scale, and network effects. In each of these explanations, a few firms have an inherent advantage, allowing them to maintain their dominance. Our study’s objective is to show that even when all firms are equally situated, an industry can evolve from a competitive to an oligopolistic structure purely as a result of random chance. We create a stylized model where firms are identical at inception, with none having any competitive advantage. In each period, a firm’s profit is random with zero mean. The randomness of profits is hypothesized to stem from demand uncertainty and production cost fluctuations. Simulation results show that, solely as a result of chance, a competitive industry transitions to a market structure where only a handful dominate. The antitrust implications of our paper pertain to the causes of oligopoly formation. Notwithstanding that in some cases oligopolies can arise as a result of anticompetitive behavior of firms, we show that market concentration can also occur as a benign, natural consequence of evolution of an industry characterized by firms with uncertain profits.

ABSTRACTWe propose an unsupervised approach for linking records across arbitrarily many files, while simultaneously detecting duplicate records within files. Our key innovation involves the representation of the pattern of links between records as a bipartite graph, in which records are directly linked to latent true individuals, and only indirectly linked to other records. This flexible representation of the linkage structure naturally allows us to estimate the attributes of the unique observable people in the population, calculate transitive linkage probabilities across records (and represent this visually), and propagate the uncertainty of record linkage into later analyses. Our method makes it particularly easy to integrate record linkage with post-processing procedures such as logistic regression, capture–recapture, etc. Our linkage structure lends itself to an efficient, linear-time, hybrid Markov chain Monte Carlo algorithm, which overcomes many obstacles encountered by previously record linkage approaches, despite the high-dimensional parameter space. We illustrate our method using longitudinal data from the National Long Term Care Survey and with data from the Italian Survey on Household and Wealth, where we assess the accuracy of our method and show it to be better in terms of error rates and empirical scalability than other approaches in the literature. Supplementary materials for this article are available online.