Semantic-based system including semantic vectors

• US 7,421,433 B2
• Filed: 10/31/2002
• Issued: 09/02/2008
• Est. Priority Date: 10/31/2002
• Status: Expired due to Fees

• Current Assignee: Hewlett Packard Enterprise Development LP (Hewlett-Packard Enterprise Company)
• Original Assignee: Hewlett-Packard Development Company, L.P. (HP Inc.)
• Inventors: Xu, Zhichen, Mahalingam, Mallik, Tang, Chunqiang
• Primary Examiner(s): Mofiz; Apu
• Assistant Examiner(s): Padmanabhan; Kavita
• Application Number:
  US10/284,093
• Publication Number:
  US 20040088282A1
• Time in Patent Office: 2,133 Days
• Field of Search: 707/3, 707/6, 707/203, 705 26- 27, 709201-203, 711147-153, 711/155, 712 1- 9
• US Class Current: 1/1
• CPC Class Codes: G06F 16/14 Details of searching files based on file metadata
• CPC Class Codes: G06F 16/30 of unstructured textual data document management systems G06F16/93

Abstract

A distributed file system includes a plurality of nodes storing objects; an extractor and a semantic catalogue. The extractor extracts semantic information for the objects. The semantic catalogue includes the semantic information for the objects and may be stored by the plurality of nodes.

Claims

1. A semantic-based system comprising:

at least one client operable to issue a semantic query; and

a file system connected to the at least one client via a network, wherein the file system stores objects and semantic information for the objects that is searchable to execute the semantic query;

wherein the semantic information includes semantic vectors for the objects, each semantic vector identifying predetermined features for an associated object and created from the predetermined features for the associated object, and

wherein the file system is a distributed archival file system operable to store a plurality of base files, each of a plurality of new versions of the base files in the form of a difference between each of the plurality of the new versions and an associated one of the base files, and semantic information for each of the plurality of the new versions and base files;

wherein the semantic query includes a semantic vector stored in the distributed archival file system for one of the objects.

2. The semantic-based system of claim 1, wherein the semantic vectors are searchable to identify objects having similar predetermined features.

3. The semantic-based system of claim 1, wherein the file system further comprises a semantic catalogue including each semantic vector, an associated object name and a location of the associated object in the file system.

4. The semantic-based system of claim 1, wherein the file system comprises at least one extractor for creating a semantic vector for each of the objects of a specific file type.

5. The semantic-based system of claim 4, wherein the file system comprises an extractor registry identifying each extractor in the file system.

6. The semantic-based system of claim 5, wherein the extractor registry is operable to add or remove an extractor from the file system.

7. The semantic-based system of claim 1, wherein the distributed archival file system is overlaid on a peer-to-peer network comprising a plurality of nodes.

8. A distributed file system comprising:

a plurality of nodes storing objects, wherein the objects include at least a first object, and a second object identifying a new version of the first object, the second object is stored as a difference between the new version of the first object and the first object;

at least one extractor extracting semantic information for the objects; and

a semantic catalogue including the semantic information for the objects, the semantic catalogue being stored in the plurality of nodes;

wherein the semantic information includes semantic vectors for the objects, wherein each semantic vector identifies predetermined features for an associated object and is created from the predetermined features for the associated object;

the semantic catalogue further includes a name for each of the objects, a unique identifier different from the name for each of the objects, and a version number for each of the objects that are associated with an associated one of the semantic vectors;

wherein the distributed file system is operable to execute a semantic-based query, the semantic-based query includes a semantic vector in the semantic catalogue stored in the plurality of nodes.

9. The distributed file system of claim 8, wherein the distributed file system is overlaid on a peer-to-peer network comprising the plurality of nodes.

10. The distributed file system of claim 8, wherein the semantic catalogue is distributed among the nodes.

11. A node in a semantic-based distributed file system, the node comprising:

a processor;

a storage device storing objects, wherein the objects includes at least a first object, and a second object identifying a new version of the first object, the second object is stored as a difference between the new version of the first object and the first object;

a semantic catalogue containing semantic information for the objects; and

an extractor, wherein the processor is operable to execute the extractor for extracting and creating the semantic information contained in the semantic catalogue;

wherein the semantic information includes semantic vectors for the objects, wherein each semantic vector identifies predetermined features for an associated object and is created from the predetermined features for the associated object;

wherein the semantic-based distributed file system is operable to execute a semantic-based query, the semantic-based query includes a semantic vector stored in the distributed file system.

12. The node of claim 11, wherein the semantic catalogue comprises at least one entry, the at least one entry including an object name, semantic information for the object, and location of the object.

13. A method for searching a semantic-based file system storing a plurality of objects, the method comprising steps of:

creating a semantic vector associated with each of the plurality of objects stored in the semantic-based file system from predetermined features of the associated object using a latent semantic indexing technique;

storing the created semantic vectors in the file system;

receiving a semantic query, the semantic query includes one of the created semantic vectors;

searching semantic vectors stored in the file system for semantic vectors similar to the one created semantic vector in the semantic query;

generating a result of the search;

receiving a new version of one of the plurality of objects stored in the semantic-based file system;

computing a difference between the new version and the one object;

comparing the difference to a predetermined threshold; and

storing the difference in the semantic-based file system in response to the difference being less than the threshold.

14. The method of claim 13, wherein the semantic-based file system stores a semantic catalogue including an entry for each of the plurality of objects, each entry comprising an object name, a semantic vector associated with the object and the location of the object.

15. The method of claim 14, wherein the step of searching further comprises searching the semantic catalogue for the at least one semantic.

16. The method of claim 15, wherein the step of generating a result further comprises steps of:

identifying at least one object from the catalogue meeting the semantic query;

identifying the location of the at least one object; and

retrieving the at least one object from the location.

17. The method of claim 13, further comprising:

selecting a new object in response to the difference being greater than the predetermined threshold, the new object having similar semantics to the new version;

computing a second difference between the new version and the new object; and

storing the second difference in response to the second difference not being greater than the threshold.

18. The method of claim 17, wherein the selecting the new object comprises:

generating a semantic vector for the new version;

comparing the semantic vector for the new version to semantic vectors for other objects stored in the file system by calculating a Euclidian distance between the semantic vector for the new version and the semantic vectors for the other objects stored in the file system;

identifying one of the semantic vectors for other objects in the plurality of objects stored in the semantic-based filed system that is similar to the semantic vector for the new version; and

selecting the object associated with the identified semantic vector.

19. A semantic-based file system comprising:

means for receiving a semantic query, the semantic query includes a semantic vector stored in the semantic-based file system;

means for searching semantic vectors stored in the semantic-based file system for semantic vectors similar to the semantic vector in the semantic query, wherein each semantic vector is associated with one of the plurality of objects stored in the semantic-based file system and created from predetermined features extracted from the associated object; and

means for generating a result of the search.

20. The semantic-based file system of claim 19 storing a semantic catalogue including an entry for each of the plurality of objects, each entry comprising an object name, a semantic vector associated with the object and the location of the object.

21. The semantic-based file system of claim 20, wherein the means for searching is operable to search the semantic catalogue for the at least one semantic.

22. The semantic-based file system of claim 20, wherein the means for generating a result further comprises:

means for identifying at least one object from the catalogue meeting the query;

means for identifying the location of the at least one object; and

means for retrieving the at least one object from the location.