System and method for determining originality of data content

• US 8,412,718 B1
• Filed: 09/20/2010
• Issued: 04/02/2013
• Est. Priority Date: 09/20/2010
• Status: Expired due to Fees

• Current Assignee: Amazon Technologies, Inc. (Amazon.com, Inc.)
• Original Assignee: Amazon Technologies, Inc. (Amazon.com, Inc.)
• Inventors: Bilger, Jeffrey Matthew
• Primary Examiner(s): NGUYEN, PHONG H
• Application Number:
  US12/886,343
• Time in Patent Office: 925 Days
• Field of Search: 707/748, 707/749, 707/750, 707/758, 705/14.4
• US Class Current: 707/749
• CPC Class Codes: G06Q 30/0631 Item recommendations

Abstract

The present invention provides systems and methods for determining the originality of data content. In one embodiment, the determined originality of a particular item (e.g., a book) as compared to one or more other items can be used as a factor in recommending the item to a user. For example, in one embodiment, upon a user'"'"'s selection of an item (e.g., a book), one or more items that have content most diverse from the selected item are determined and provided to the user. In another embodiment, various versions of an item are compared to each other to determine how content in each version differs from that in another version. In another embodiment, content in a collection of items are compared against content from publicly (freely) available sources (e.g., web pages) to determine the originality of the content in the collection of items.

Claims

1. A computer-implemented method for providing a recommendation for an item based on originality of content of similar items, the computer-implemented method comprising:

as implemented by one or more computing devices configured with specific executable instructions,

for at least one item in a plurality of items, determining a subset of items in the plurality of items that are similar to the at least one item;

for each item of the subset of items determined to be similar to the at least one item,

generating an originality score for a pairing including the item in the subset of similar items and the at least one item, the originality score indicating a degree to which content of the item in the subset of similar items is diverse from content of the at least one item, and the originality score being generated based at least in part on a comparison of the content of the item in the subset of similar items with the content of the at least one item;

storing the generated originality scores in a diversity matrix;

receiving an indication of a selection of an item of interest, the item of interest being one of the items in the plurality of items for which originality scores have been generated with respect to a plurality of items determined to be similar to the item of interest;

for each item in the plurality of items determined to be similar to the item of interest, obtaining, from the diversity matrix, a generated originality score for a pairing including the item in the plurality of items determined to be similar to the item of interest and the item of interest; and

selecting, based on the obtained originality scores, an item within the plurality of items that are similar to the item of interest as a recommended item, the selected item having content that is most diverse, among items in the plurality of items determined to be similar to the item of interest, from content of the item of interest.

2. The computer-implemented method of claim 1, wherein each of the plurality of items is a book and the content of each of the plurality of items comprises textual data.

3. The computer-implemented method of claim 2, wherein the comparison of the content of the item in the subset of similar items with the content of the at least one item is based on at least one of min-wise independent permutations, term-document frequency, and latent semantic analysis.

4. A computer-readable medium having a computer-executable component for determining originality of data content, the computer-executable component comprising:

an original content determination component for;

for at least one item in a plurality of items, determining a subset of items in the plurality of items that are similar to the at least one item;

for each item of the subset of items determined to be similar to the at least one item,

generating an originality score for a pairing including the item in the subset of similar items and the at least one item, the originality score indicating a degree to which content of the item in the subset of similar items is diverse from content of the at least one item, and the originality score being generated based at least in part on a comparison of the content of the item in the subset of similar items with the content of the at least one item;

storing the generated originality scores;

receiving an indication of a selection of an item of interest, the item of interest being an item in the plurality of items for which originality scores have been generated with respect to a subset of the plurality of items determined to be similar to the item of interest;

for each item in the subset of items determined to be similar to the item of interest, obtaining, from among the stored originality scores, an originality score for a pairing including the item in the subset of items determined to be similar to the item of interest and the item of interest; and

selecting an item within the subset of items determined to be similar to the item of interest as a recommended item, the selecting based at least in part on the obtained originality scores.

5. The computer-readable medium of claim 4, wherein an item is selected that has a lowest associated originality score of the items in the subset of items determined to be similar to the item of interest.

6. The computer-readable medium of claim 4, wherein an item is selected with an associated originality score that meets a score threshold.

7. The computer-readable medium of claim 4, wherein the originality scores are stored in a matrix, the originality scores including the originality score for each pairing.

8. The computer-readable medium of claim 4, wherein each of the plurality of items is a book.

9. The computer-readable medium of claim 8, wherein the content of the book comprises textual data.

10. The computer-readable medium of claim 9, wherein the comparison of content of the item in the subset of similar items with content of the at least one item is based on min-wise independent permutations.

11. The computer-readable medium of claim 10, wherein the min-wise independent permutations are based on detecting common single words within the content of the item in the subset of similar items and the content of the at least one item.

12. The computer-readable medium of claim 10, wherein the min-wise independent permutations are based on detecting common groups of words within the content of the item in the subset of similar items and the content of the at least one item.

13. The computer-readable medium of claim 9, wherein the comparison of content of the item in the subset of similar items with content of the at least one item is based on at least one of term-document frequency and latent semantic analysis.

14. The computer-readable medium of claim 4, wherein the original content determination component is configured to determine a subset of items in the plurality of items that are similar to the at least one item by determining whether each item other than the at least one item in the plurality of items shares at least one of a view similarity with the at least one item and a purchase similarity with the at least one item.

15. The computer-readable medium of claim 4, wherein the original content determination component is configured to determine a subset of items in the plurality of items that are similar to the at least one item by determining whether each item other than the at least one item in the plurality of items shares a common topic with the at least one item.

16. The computer-readable medium of claim 15, wherein the original content determination component is configured to determine that an item other than the at least one item in, the plurality of items shares a common topic with the at least one item if the item and the at least one item both have at least one of:

(1) a common browse node, (2) a common tag, and (3) a similar textual classification.

17. A system for providing a recommendation for an item based on originality of content of similar items, the system comprising:

a data store that stores data relating to a plurality of items; and

a computing device, comprising one or more processors, in communication with the data store that is configured to;

for at least one item in the plurality of items, determine a subset of items in the plurality of items that are similar to the at least one item;

for each item of the subset of items determined to be similar to the at least one item,

generate an originality score for a pairing including the item in the subset of similar items and the at least one item, the originality score indicating a degree to which content of the item in the subset of similar items is diverse from content of the at least one item, and the originality score being generated based at least in part on a comparison of the content of the item in the subset of similar items with the content of the at least one item;

store the generated originality scores in a diversity matrix;

receive an indication of a selection of an item of interest, the item of interest being one of the items in the plurality of items for which originality scores have been generated with respect to a plurality of items determined to be similar to the item of interest;

for each item in the plurality of items determined to be similar to the item of interest, obtain, from the diversity matrix, a generated originality score for a pairing including the item in the plurality of items determined to be similar to the item of interest and the item of interest; and

select, based on the obtained originality scores, an item within the plurality of items that are similar to the item of interest as a recommended item, the selected item having content that is most diverse, among items in the plurality of items determined to be similar to the item of interest, from content of the item of interest.

18. The system of claim 17, wherein each of the plurality of items is a book and the content of each of the plurality of items comprises textual data.

19. The system of claim 17, wherein the comparison of the content of the item in the subset of similar items with the content of the at least one item is based on at least one of min-wise independent permutations, term-document frequency, and latent semantic analysis.