Systems and methods for ranking document clusters

Abstract

Document cluster ranking systems and methods of ranking document clusters are described. In some example embodiments, the method comprises: obtaining, at a document cluster ranking system, a value associated with a first feature for each of a plurality of document clusters; based on the values associated with the first feature, automatically generating, at the document cluster ranking system, a plurality of first feature bins, each first feature bin defining a range of values and a bin identifier; and obtaining a score for one of the document clusters, by: i) identifying the first feature bin having a range of values which includes the obtained value associated with the first feature for that one of the document clusters; and ii) determining a score for that document cluster based on the first feature bin identifier for the identified first feature bin.

What is claimed is: 
     
         1 . A method of automatically generating a plurality of first feature bins for use in ranking document clusters, the document clusters including two or more documents, the method comprising:
 generating a range of values to define each first feature bin based on values associated with a first feature for a plurality of document clusters;   generating a bin identifier for each first feature bin based on the values associated with the respective first feature bin; and   assigning the bin identifier to the respective first feature bin.   
     
     
         2 . The method of  claim 1 , further comprising:
 obtaining a probability distribution of values of the first feature, and wherein the range of values is generated based on the probability distribution for the values of the first feature.   
     
     
         3 . The method of  claim 2 , further comprising performing peak detection on the probability distribution of values of the first feature, and wherein generating the range of values comprises generating the plurality of first feature bins based on the peaks. 
     
     
         4 . The method of  claim 3 , wherein generating the plurality of first feature bins based on the peaks comprises:
 performing k-means clustering at the detected peaks.   
     
     
         5 . The method of  claim 4 , further comprising, prior to performing peak detection on the probability distribution of values of the first feature, smoothing the probability distribution of values of the first feature. 
     
     
         6 . The method of  claim 2 , wherein the probability distribution of values of the first feature is a histogram. 
     
     
         7 . The method of  claim 1 , wherein the first feature represents the number of documents in the document cluster. 
     
     
         8 . The method of  claim 1 , wherein the first feature is a measure of the portion of the documents in the document cluster which are blog posts. 
     
     
         9 . The method of  claim 1 , wherein the first feature is a measure of the number of the documents in the document cluster which are comments. 
     
     
         10 . The method of  claim 1 , wherein the first feature is a measure of the freshness of the documents in the document cluster. 
     
     
         11 . The method of  claim 1 , wherein the first feature is a measure the portion of the plurality of documents which are micro-blog posts. 
     
     
         12 . A system for automatically generating a plurality of first feature bins for use in ranking document clusters, the document clusters including two or more documents, the system comprising:
 a processor:   a memory coupled to the processor, the memory for storing processor executable instructions which, when executed by the processor cause the processor to:   generate a range of values to define each first feature bin based on values associated with a first feature for a plurality of document clusters;   generate a bin identifier for each first feature bin based on the values associated with the respective first feature bin; and   assign the bin identifier to the respective first feature bin.   
     
     
         13 . The system of  claim 12 , wherein the processor is further configured to:
 obtain a probability distribution of values of the first feature, and wherein the range of values is generated based on the probability distribution for the values of the first feature.   
     
     
         14 . The system of  claim 13 , wherein the processor is further configured to perform peak detection on the probability distribution of values of the first feature, and wherein generating the range of values comprises generating the plurality of first feature bins based on the peaks. 
     
     
         15 . The system of  claim 14 , wherein generating the plurality of first feature bins based on the peaks comprises:
 performing k-means clustering at the detected peaks.   
     
     
         16 . The system of  claim 15 , wherein the processor is further configured to, prior to performing peak detection on the probability distribution of values of the first feature, smooth the probability distribution of values of the first feature. 
     
     
         17 . The system of  claim 13 , wherein the probability distribution of values of the first feature is a histogram. 
     
     
         18 . The system of  claim 12 , wherein the first feature represents the number of documents in the document cluster. 
     
     
         19 . The system of  claim 12 , wherein the first feature is a measure of the portion of the documents in the document cluster which are blog posts. 
     
     
         20 . The system of  claim 12 , wherein the first feature is a measure of the number of the documents in the document cluster which are comments.