Abstract: Technology is described for computationally intensive distributed computing systems and methods. A storage process for representing data objects in memory using a serialization format that provides low replication cost and thread-safe reads may include receiving a data object by a computing hub. A distributed shared memory of the computing hub may write the data object to a memory device associated with the computing hub using a byte array broken into a plurality of sections. The plurality of sections may describe the in-memory content of the data object and include information on how to read the plurality of sections to obtain a current state of the data object. The distributed shared memory may process requests from processing applications to modify data associated with the data object. The distributed shared memory may write mutations to the data object caused by the requests to the memory device using an ordered, log-based format to provide versioned snapshots of state.

Abstract: A facility for producing an item recommendation for a selected user is described. The facility accesses an information resource that, for each of a number of buckets that each correspond to a different collection of personal information, identifies users for a members of the bucket into each of whom the entire collection of personal information applies. The facility selects a bucket among the plurality of buckets of which the selected user is a member. The facility accesses a number of item ratings that were each contributed by a member of the selected bucket other than the selected user. For each item rated among the accessed item ratings, the facility aggregates the ratings of the item. On the basis of items' aggregated ratings, the facility selects one or more rated items for recommendation to the selected user.

Abstract: A facility for producing an item recommendation for a selected user is described. The facility accesses an information resource that, for each of a number of buckets that each correspond to a different collection of personal information, identifies users for a members of the bucket into each of whom the entire collection of personal information applies. The facility selects a bucket among the plurality of buckets of which the selected user is a member. The facility accesses a number of item ratings that were each contributed by a member of the selected bucket other than the selected user. For each item rated among the accessed item ratings, the facility aggregates the ratings of the item. On the basis of items' aggregated ratings, the facility selects one or more rated items for recommendation to the selected user.

Abstract: Technology is described for computationally intensive distributed computing systems and methods. A storage process for representing data objects in memory using a serialization format that provides low replication cost and thread-safe reads may include receiving a data object by a computing hub. A distributed shared memory of the computing hub may write the data object to a memory device associated with the computing hub using a byte array broken into a plurality of sections. The plurality of sections may describe the in-memory content of the data object and include information on how to read the plurality of sections to obtain a current state of the data object. The distributed shared memory may process requests from processing applications to modify data associated with the data object. The distributed shared memory may write mutations to the data object caused by the requests to the memory device using an ordered, log-based format to provide versioned snapshots of state.