Decision-theoretic methods for identifying relevant substructures of a hierarchical file structure to enhance the efficiency of document access, browsing, and storage

• US 7,346,622 B2
• Filed: 03/31/2006
• Issued: 03/18/2008
• Est. Priority Date: 05/04/2001
• Status: Expired due to Fees

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Horvitz, Eric J., Koch, Paul B.
• Primary Examiner(s): Wong; Leslie
• Application Number:
  US11/394,774
• Publication Number:
  US 20060173842A1
• Time in Patent Office: 718 Days
• Field of Search: 707/100, 707/104.1, 709/218
• US Class Current: 1/1
• CPC Class Codes: G06F 16/10 File systems; File servers
• CPC Class Codes: Y10S 707/99945 Object-oriented database structure processing

Abstract

A system and methodology is provided for improving directory operations within a system providing an electronic hierarchical directory of items. The system includes a component which analyzes probabilities and utilities associated with determining potential target directories for storing and accessing data, and a component for building a subset of the potential target directories that are predicted to be the target directory. The probabilities and/or utilities are functions of expected navigation costs associated with traversing from a displayed directory to at least one of the potential target directories. Methods in accordance with the present invention can be coupled with displays of substructures that format the substructures into a coherent hierarchical view.

Claims

1. A computer implemented system for predicting a target file directory, comprising the following computer executable components:

a first component that infers and/or determines expected costs for directory operations associated with navigating potential target directories for storing and accessing data, based on a probabilistic or utility analysis, wherein the target directories are selected from the file directory tree structure residing on one of a local or remote computer based at least on values assigned to directories within the tree structure, the values indicate likelihood of each of the directories being the target directory; and

a second component that displays a subset of the potential target directories determined by selecting target directories with high likelihood values, based at least in part on the expected navigation cost which are inversely proportional to the utility, in order to minimize a cost of traversing directories.

2. The system of claim 1, the utilities being functions of navigation costs associated with traversing from a node associated with a potential target directory under consideration to at least one of the other potential target directories.

3. The system of claim 1, the second component further determines the subset of directories based on expected utilities computed as functions of probabilities of target information being at a node and the navigation costs associated with traversing from the node to at least one of the potential target directories.

4. The system of claim 3, the navigation costs assigned by at least one of user selections and encoded within the system.

5. The system of claim 1, the probabilities being a function of recent and long-term file activity within a directory.

6. The system of claim 5, the long term file activity being determined from a predetermined time horizon.

7. The system of claim 5, the recent file activity being determined from frequency of access to a file.

8. The system of claim 7, further comprising a background monitor to determine file access frequency.

9. The system of claim 1, the second component outputs the subset of directories as tree fragments.

10. The system of claim 1, the subset comprises N potential target directories, N being a predefined integer.

11. The system of claim 1, the subset comprises directories with expected navigation costs below a predetermined level.

12. A method for determining a potential target node for directory operations, comprising:

assigning probabilities and utilities to a plurality of nodes based at least on their likelihoods of being potential target nodes, the utilities being related to navigation costs associated with traversing from a displayed node to at least one of the potential target nodes, the navigation costs are inversely proportional to the utilities being assigned by at least user selections and encoded within the system, wherein the target nodes are selected from the nodes in the file directory tree structure residing on one of a local or remote computer;

determining an expected utility from the probabilities and utilities associated with the plurality of target nodes; and

displaying a candidate list of likely nodes to a user based upon the expected utility.

13. The method of claim 12, further comprising multiplying the assigned probabilities and utilities together to form a product at each of the plurality of target nodes.

14. The method of claim 13, further comprising summing the products from each of the plurality of target nodes together to determine the expected utility for one of the plurality of potential target nodes.

15. The method of claim 12, the probabilities being a function of recent and long-term file activity within a directory, the long-term file activity being determined from a predetermined time horizon and the recent file activity being determined from frequency of access to a file.

16. The method of claim 12, displaying the candidate list in descending order from highest expected utility.

17. The method of claim 12, further comprising removing a potential target node with a maximum expected utility from consideration when evaluating expected utility for the other potential target nodes.

18. A computer implemented system for determining a potential target node for directory operations, comprising:

means for assigning probabilities and utilities to a plurality of nodes based on likelihoods of the plurality of nodes being a target node, wherein the utilities are functions inversely proportion to cost of navigating from one node to another, and the target nodes are selected from the nodes of the file directory tree structure residing on one or more of a local or remote computer;

means for determining an expected utility from the probabilities and utilities associated with the plurality of nodes; and

means for displaying a candidate list of likely nodes to a user based upon the expected utility, the candidate list comprises a subset of the potential target nodes.