Online commerce platforms connect suppliers to customers using internet. In online travel industry, the suppliers provide a programmatic interface for the platform to query and book travel services. Typically, multiple competitive platforms connect to the same set of suppliers and any change in inventory level leads to changes in availability and pricing. One of the key problems that online platforms deal with is to ensure the accuracy of inventory availability and pricing, while ensuring low response time to the customers. In this work, we present the design and implementation of DynaCorrect, a dynamically correcting cache for non-cooperative systems. DynaCorrect caches inventory data from remote suppliers to ensure low response time. Since the inventory price and availability may change in the background, DynaCorrect employs random sampling and a feedback loop to correct the cache. Further, in order to ensure high cache hit rate, it refreshes inventory based on popularity, spatial and temporal locality. Finally, it employs deduplication to ensure high resource efficiency for the caching system. We implemented DynaCorrect in production over a period of time and observed significant benefits to achieve the goals of high accuracy, low response time and reduced number of queries to the suppliers.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.