Methods and apparatus for intelligent crawling on the world wide web

• US 8,060,816 B1
• Filed: 10/31/2000
• Issued: 11/15/2011
• Est. Priority Date: 10/31/2000
• Status: Expired due to Fees

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Aggarwal, Charu C., Yu, Philip Shi-Lung
• Primary Examiner(s): Paula; Cesar B
• Assistant Examiner(s): Hillery; Nathan
• Application Number:
  US09/703,174
• Time in Patent Office: 4,032 Days
• Field of Search: 715/501.1, 715/513, 715/530, 707/6, 707/3, 707/4
• US Class Current: 715/205
• CPC Class Codes: G06F 16/951 Indexing; Web crawling techniques

Abstract

Methods and apparatus for performing intelligent crawling are provided. Particularly, the intelligent crawling techniques of the invention provide a crawler mechanism which is capable of learning as it crawls in order to focus the search for documents on the information network being explored, e.g., world wide web. This crawler mechanism stores information about the crawled documents as it retrieves the documents, and then uses the information to further focus its search appropriately. The inventive techniques result in the crawling of a small percentage of the documents on the world wide web.

Claims

1. A computer-based method of performing document retrieval in accordance with an information network, the method comprising the steps of:

initially retrieving one or more documents from the information network that satisfy a user-defined predicate, wherein the initial document retrieval operation is performed without assuming a specific model of a linkage structure such that the initial document retrieval operation retrieves the one or more documents without assuming that a relationship exists between a feature of a first one of the one or more documents and a feature of at least another one of the one or more documents that links to the first one;

collecting at least a set of aggregate statistical information and a set of predicate-specific statistical information about the one or more retrieved documents as the one or more retrieved documents are analyzed; and

using the collected statistical information to automatically determine further document retrieval operations to be performed in accordance with the information network, wherein the statistical information using step further comprises learning a linkage structure from at least a portion of the collected statistical information with each successive document retrieval operation such that the learned linkage structure is available for use in performing subsequent document retrieval operations requested by a user.

2. The method of claim 1, wherein the user-defined predicate specifies content associated with a document.

3. The method of claim 1, wherein the statistical information collection step uses content of the one or more retrieved documents.

4. The method of claim 1, wherein the statistical information collection step considers whether the user-defined predicate has been satisfied by the one or more retrieved documents.

5. The method of claim 1, wherein the collected statistical information is used to direct further document retrieval operations toward documents which are more likely to satisfy the predicate than would otherwise occur with respect to document retrieval operations that are not directed using the collected statistical information.

6. The method of claim 1, wherein the collected statistical information is used to direct further document retrieval operations toward documents which are similar to the one or more retrieved documents that also satisfy the predicate.

7. The method of claim 1, wherein the collected statistical information is used to direct further document retrieval operations toward documents which are linked to by other documents which also satisfy the predicate.

8. The method of claim 1, wherein the information network is the world wide web and a document is a web page.

9. The method of claim 8, wherein the statistical information collection step uses one or more uniform resource locator tokens in the one or more retrieved web pages.

10. Apparatus for performing document retrieval in accordance with an information network, the apparatus comprising:

at least one processor operative to;

(i) initially retrieve one or more documents from the information network that satisfy a user-defined predicate, wherein the initial document retrieval operation is performed without assuming a specific model of a linkage structure such that the initial document retrieval operation retrieves the one or more documents without assuming that a relationship exists between a feature of a first one of the one or more documents and a feature of at least another one of the one or more documents that links to the first one;

(ii) collect at least a set of aggregate statistical information and a set of predicate-specific statistical information about the one or more retrieved documents as the one or more retrieved documents are analyzed; and

(iii) use the collected statistical information to automatically determine further document retrieval operations to be performed in accordance with the information network, wherein the statistical information using operation further comprises learning a linkage structure from at least a portion of the collected statistical information with each successive document retrieval operation such that the learned linkage structure is available for use in performing subsequent document retrieval operations requested by a user.

11. The apparatus of claim 10, wherein the user-defined predicate specifies content associated with a document.

12. The apparatus of claim 10, wherein the statistical information collection operation uses content of the one or more retrieved documents.

13. The apparatus of claim 10, wherein the statistical information collection operation considers whether the user-defined predicate has been satisfied by the one or more retrieved documents.

14. The apparatus of claim 10, wherein the collected statistical information is used to direct further document retrieval operations toward documents which are more likely to satisfy the predicate than would otherwise occur with respect to document retrieval operations that are not directed using the collected statistical information.

15. The apparatus of claim 10, wherein the collected statistical information is used to direct further document retrieval operations toward documents which are similar to the one or more retrieved documents that also satisfy the predicate.

16. The apparatus of claim 10, wherein the collected statistical information is used to direct further document retrieval operations toward documents which are linked to by other documents which also satisfy the predicate.

17. The apparatus of claim 10, wherein the information network is the world wide web and a document is a web page.

18. The apparatus of claim 17, wherein the statistical information collection operation uses one or more uniform resource locator tokens in the one or more retrieved web pages.

19. An article of manufacture for performing document retrieval in accordance with an information network, comprising a machine readable medium containing one or more programs which when executed implement the steps of:

initially retrieving one or more documents from the information network that satisfy a user-defined predicate, wherein the initial document retrieval operation is performed without assuming a specific model of a linkage structure such that the initial document retrieval operation retrieves the one or more documents without assuming that a relationship exists between a feature of a first one of the one or more documents and a feature of at least another one of the one or more documents that links to the first one;

collecting at least a set of aggregate statistical information and a set of predicate-specific statistical information about the one or more retrieved documents as the one or more retrieved documents are analyzed; and

using the collected statistical information to automatically determine further document retrieval operations to be performed in accordance with the information network, wherein the statistical information using step further comprises learning a linkage structure from at least a portion of the collected statistical information with each successive document retrieval operation such that the learned linkage structure is available for use in performing subsequent document retrieval operations requested by a user.

20. The article of claim 19, wherein the user-defined predicate specifies content associated with a document.

21. The article of claim 19, wherein the statistical information collection step uses content of the one or more retrieved documents.

22. The article of claim 19, wherein the statistical information collection step considers whether the user-defined predicate has been satisfied by the one or more retrieved documents.

23. The article of claim 19, wherein the collected statistical information is used to direct further document retrieval operations toward documents which are more likely to satisfy the predicate than would otherwise occur with respect to document retrieval operations that are not directed using the collected statistical information.

24. The article of claim 19, wherein the collected statistical information is used to direct further document retrieval operations toward documents which are similar to the one or more retrieved documents that also satisfy the predicate.

25. The article of claim 19, wherein the collected statistical information is used to direct further document retrieval operations toward documents which are linked to by other documents which also satisfy the predicate.

26. The article of claim 19, wherein the information network is the world wide web and a document is a web page.

27. The article of claim 26, wherein the statistical information collection step uses one or more uniform resource locator tokens in the one or more retrieved web pages.