Clustering multimedia search

Abstract

A method for clustering a set of web search results is disclosed. A first signature is compared based at least in part on an analysis of multimedia content associated with a first web search result with a second signature based at least in part on an analysis of multimedia content associated with a second web search result. The first web search result is clustered with the second web search result based at least in part on the comparison of the first signature with the second signature.

What is claimed is: 
     
         1 . (canceled) 
     
     
         2 . A method, comprising:
 determining a first level of entropy associated with a first signature for a first multimedia content item;   determining a second level of entropy associated with a second signature for a second multimedia content item; and   clustering the first multimedia content item with the second multimedia content item based at least in part on comparing the first level of entropy associated with the first signature and the second level of entropy associated with the first signature with that of another set of content items.   
     
     
         3 . A method as recited in  claim 2 , further comprising generating a first media content signature based at least in part on an analysis of multimedia content for the first multimedia content item. 
     
     
         4 . A method as recited in  claim 3 , further comprising generating a second media content signature based at least in part on an analysis of multimedia content for the second multimedia content item. 
     
     
         5 . A method as recited in  claim 4 , wherein the first multimedia content item is associated with a first web search result and the second multimedia content item is associated with a second web search result. 
     
     
         6 . A method as recite in  claim 5 , further comprising reducing web search results based at least in part on an analysis of textual metadata. 
     
     
         7 . A method as recited in  claim 4 , wherein clustering is based on an expectation that all else being equal if a low entropy set of multimedia content items have the same degree of similarity as the respective signatures of a high entropy set of content items, the high entropy set is more likely to have similar multimedia than the low entropy set. 
     
     
         8 . A method as recited in  claim 4 , wherein clustering is based on an expectation that a low entropy signature is less likely to uniquely represent a particular multimedia content. 
     
     
         9 . A method as recited in  claim 4 , wherein clustering includes labeling two web search results with similar video signatures and different audio signatures as commentary. 
     
     
         10 . A method as recited in  claim 4 , wherein clustering includes labeling two web search results with similar audio signatures and different video signatures as remixes. 
     
     
         11 . A method as recited in  claim 4 , wherein the comparison includes calculating a distance metric between the first media content signature and the second media content signature. 
     
     
         12 . A method as recited in  claim 11 , wherein the distance metric includes one or a weighted combination of: a Cartesian distance; a Manhattan distance; a Euclidean distance; and a byte difference. 
     
     
         13 . A method as recited in  claim 4 , wherein each media content signature includes a hash value based at least in part on one or more of the following: image features, audio features, and on video features. 
     
     
         14 . A method as recited in  claim 4 , wherein each media content signature includes one or more of the following: a recognized face, a recognized logo, and a recognized facial feature. 
     
     
         15 . A method as recited in  claim 4 , further comprising finding video that sound like a web search result. 
     
     
         16 . A method as recited in  claim 2 , wherein multimedia is any non-textual data or metadata. 
     
     
         17 . A method as recited in  claim 2 , wherein multimedia includes images, video and audio. 
     
     
         18 . A method as recited in  claim 2 , wherein clustering includes consolidating web search results if the distance metric is below an identical-threshold. 
     
     
         19 . A method as recited in  claim 2 , wherein clustering includes highlighting web search results if the distance metric is below a similar-threshold but above an identical-threshold. 
     
     
         20 . A system, comprising:
 a data store configured to store signatures of web search results; and   a processor coupled to the data store and configured to:   determine a first level of entropy associated with a first signature for a first multimedia content item;   determine a second level of entropy associated with a second signature for a second multimedia content item; and   cluster the first multimedia content item with the second multimedia content item based at least in part on comparing the first level of entropy associated with the first signature and the second level of entropy associated with the first signature with that of another set of content items.   
     
     
         21 . A computer program product, the computer program product being embodied in a non-transitory computer readable storage medium and comprising computer instructions for:
 determining a first level of entropy associated with a first signature for a first multimedia content item;   determining a second level of entropy associated with a second signature for a second multimedia content item; and   clustering the first multimedia content item with the second multimedia content item based at least in part on comparing the first level of entropy associated with the first signature and the second level of entropy associated with the first signature with that of another set of content items.