Information retrieval in cloud

Information retrieval (IR) is the process of finding relevant documents or information from a large collection to satisfy a user's information need, for example a search for files, e-mails or web pages.

IR in standalone / local information systems

• Scope: a limited amount of data stored in one place (a local database, a file system or a departmental application).
• Data: structured and well known; users know where data are kept.
• Access: folders, menus, simple queries or SQL are enough.
• Significance: less important, because the volume is small, and users can find items by browsing or directory structure.

IR in cloud

• Scope: very large volumes of structured and unstructured data (documents, mail, logs, images, social data) distributed across many servers, data centres and users.
• Data: continuously growing, in many formats, shared by many tenants.
• Access: users cannot browse; they depend on search, indexing and ranking to find data.
• Significance:
  1. Finding the right item among billions of objects is only possible with indexing and search.
  2. Cloud storage services (Drive, OneDrive, S3) and SaaS apps need search for usability.
  3. Big data analytics, e-discovery and compliance require retrieving specific data.
  4. Distributed IR, using MapReduce/Hadoop for index building, gives speed and scalability.
  5. It reduces cost and bandwidth by retrieving only the relevant data.
  6. Supports personalised, ranked and secure (permission-aware) retrieval.
  7. Needed for ads, recommendation, and multi-language search.