System and method for improving the relevance of search results using data container access patterns

摘要

A system and method for improving the relevance of search results using data container access patterns. An indexing process tracks data access patterns and updates an access data structure. When executing a search operation, a search process first identifies a set of data containers containing the search terms. The search process then utilizes the access data structure to rank the identified data containers based on the collected data access pattern information.

主权项

1. A method for improving relevance of search results, the method comprising:
configuring a switch to connect a first port to a client, connect a second port to a storage system, and connect a third port to a search appliance; processing, in parallel by the storage system and the search appliance, data access requests directed to a plurality of data containers of the storage system, wherein the data access requests are forwarded to the search appliance by the switch; collecting, by the search appliance that examines the data access requests, data container access pattern information for each of the plurality of data containers based on the data access requests, the data container access pattern information including at least a total number of different users that access the data container, and a total number of times the data container has been accessed; indexing the data container access pattern information for the plurality of data containers in a data structure on the search appliance; receiving, from the client, a search request having one or more search terms, the search request locating, by a search process, a plurality of accessed data containers each having corresponding collected data container access pattern information indexed in the data structure on the search appliance, and each located accessed data container containing at least one search term of the search request; ranking the located accessed data containers, containing the at least one search term of the search request received from the client, utilizing the corresponding collected data container access pattern information, wherein a rank is assigned to each of the located accessed data containers by forming a polynomial expression equal to a first weighting factor multiplied by the total number of different users that access the located accessed data container plus a second weighting factor multiplied by the total number times the located accessed data container has been accessed; and sending the ranked located accessed data containers to the client.