Document replication based on distributional semantics
Embodiments of the present invention are directed toward systems, methods, and computer storage media for using a neural network language model to identify semantic relationships between file storage specifications for replication requests. By treating file storage specifications (or at least a portion thereof) as “words” in the language model, replication vectors can be determined based on the file storage specifications. Instead of determining the relationship of the file storage specifications based on ordering within a document, the relationship can be based on proximity of the replication requests in a replication session. When a replication request is received from a user, the replication vectors can be used to determine a semantic similarity between the received replication request and one or more additional replication requests.
1. At least one non-transitory computer-readable storage media having instructions stored thereon, which, when executed by at least one processor of a computing device, cause the computing device to:
receive a document replication request that includes a file storage specification;
determine a replication vector based on a language model and at least a portion of the received file storage specification, wherein the language model is trained based on other document replication requests;
determine a scalar product based on the replication vector and a plurality of stored replication vectors;
generate a plurality of additional document replication requests based on the scalar product;
replicate, to another computing device, at least one document corresponding to the document replication request and at least one replication request selected from the generated plurality of additional document replication requests.
2. The method of claim 1 , wherein the received file storage specification corresponds to a storage location.
3. The media of claim 1 , wherein the language model is generated based further on an order of the received plurality of other document replication requests.
4. The media of claim 1 , wherein the language model is trained based further on a time proximity between the other document replication requests.
5. The media of claim 1 , wherein the replication vector corresponds to at least a portion of the file storage specification.
6. A computer-implemented method, comprising:
receiving, by a computing device, a document replication request that includes a file storage specification;
determining, by the computing device, a replication vector based on a language model and at least a portion of the received file storage specification, wherein the language model is trained based on other document replication requests;
determining, by the computing device, a scalar product based on the replication vector and a plurality of stored replication vectors;
generating, by the computing device, a plurality of additional document replication requests based on the scalar product;
replicating, by the computing device, at least one document corresponding to the document replication request and at least one document replication request selected from the generated plurality of additional document replication requests to another computing device.
7. The method of claim 6 , wherein the replication vector corresponds to at least a portion of the file storage specification.
8. The method of claim 6 , wherein the file storage specification corresponds to a storage location.
9. A system comprising:
a replication request generation means for generating a plurality of additional document replication requests based on
a document replication request including at least a portion of a file storage specification,
a replication vector determined based at least in part on a language model and the portion of the file storage specification, wherein the language model is trained based on other document replication requests, and
a scalar product determined based on the replication vector and a plurality of stored replication vectors; and
a replication means for replicating, from a server to a client, at least one document corresponding to the document replication request and a at least one document replication request selected from the generated plurality of additional document replication requests.
10. The system of claim 9 , wherein the file storage specification corresponds to a storage location.