Dynamic keyword suggestion and image-search re-ranking

1. A computer-implemented method comprising:

performing content-based re-ranking (CBR) based at least in part on a target image that is selected from a plurality of image search results, the CBR to identify re-ranked image search results;

extracting keywords associated with the re-ranked image search results;

outputting at least a portion of the keywords as suggested keywords with the re-ranked image search results, the suggested keywords used to query refined image search results; and

performing a refined CBR based on the target image and a received selection of at least one of the suggested keywords, the refined CBR to output the refined image search results.

2. The computer-implemented method as recited in claim 1 , wherein each keyword is assigned a score, and wherein a suggested keyword is selected based on the respective score.

3. The computer-implemented method as recited in claim 2 , wherein each score is calculated based at least in part on a frequency and a CBR weight of each of the keywords.

4. The computer-implemented method as recited in claim 3 , wherein the CBR includes a distance value used to obtain the re-ranked image search results, and wherein the CBR weight is inversely proportional to the distance value of the associated image search result.

5. The computer-implemented method as recited in claim 1 , wherein the keywords are at least one of user-assigned words associated with an image, a portion of a filename, or an attribute of the image.

6. The computer-implemented method as recited in claim 1 , wherein the keywords are selected from a collection of words associated with the re-ranked image search results that omit stop words and an original search term used to obtain the plurality of image search results that include the target image.

7. The computer-implemented method as recited in claim 1 , wherein the outputting the at least a portion of the keywords with the re-ranked image search results includes outputting a representative image with the at least a portion of the keywords.

8. The computer-implemented method as recited in claim 1 , further comprising:

causing a first display that includes the image search results;

causing a second display that includes the re-ranked image search results; and

causing a third display that includes the refined image search results.

9. A system comprising:

one or more processors; and

memory to store computer-readable instructions that, when executed on the one or more processors, cause the one or more processors to perform acts comprising:

receiving a selection of a target image from search results based on a search phrase;

generating re-ranked search results from content-based re-ranking (CBR) of the search results based on the target image;

extracting keywords associated with the re-ranked search results;

outputting at least a portion of the keywords as modified search phrases with the re-ranked search results;

generating refined search results from a refined CBR based on the target object and a received selection of at least one of the modified search phrases; and

outputting the refined search results.

10. The system as recited in claim 9 , wherein the generating refined search results further includes:

selecting candidate objects using the modified search phrase, and

re-ranking the candidate objects based on the target image.

11. The system as recited in claim 9 , wherein each keyword is assigned a score, and wherein the at least a portion of the keywords are selected based on the respective score.

12. The system as recited in claim 11 , wherein the score is calculated based on a frequency and a CBR weight of each of the keywords.

13. The system as recited in claim 12 , wherein the CBR includes using a distance value associated with each re-ranked search result to select the re-ranked search results, and wherein the CBR weight is inversely proportional to the distance value of the associated search result.

14. The system as recited in claim 9 , wherein the outputting the at least a portion of the keywords with the re-ranked image search results includes outputting a representative image with the at least a portion of the keywords.

15. A method, comprising:

generating re-ranked search results from a content-based re-ranking (CBR) of search results based on a target object that is selected from the search results;

creating a list of suggested keywords from keywords associated with the re-ranked search results;

outputting at least a portion of the suggested keywords; and

generating refined search results from a refined CBR based on the target object and a received selection of at least one of the suggested keywords.

16. The method as recited in claim 15 , wherein the keywords include a score representative of at least one of the frequency of occurrence of each keyword or a distance value of an associated object of each keyword.

17. The method as recited in claim 16 , wherein the list of suggested keywords is assembled by selecting keywords based on the score of each keyword.

18. The method as recited in claim 15 , wherein the list of suggested keywords includes a representative image for each suggested keyword.

19. The method as recited in claim 15 , wherein the keywords are user-assigned words associated with the re-ranked search results.

20. The method as recited in claim 15 , wherein the target object is one of an image or a video.