System and method for generating social summaries

One or more computer-readable memory devices or storage devices comprising instructions which, when executed by one or more processing devices, cause the one or more processing devices to perform acts comprising:

• obtaining a plurality of text communications;
• filtering the plurality of text communications to distinguish a subset of text communications that share a common text string from other text communications that lack the common text string;
• identifying a characteristic of the subset of text communications that share the common text string;
• sorting the subset of text communications that share the common text string into at least a first group of first text communications having first values for the characteristic and a second group of second text communications having second values for the characteristic;
• generating a first summary for the first group of first text communications that share the common text string, the first summary representing at least two of the first text communications from the first group;
• and
• generating a second summary for the second group of second text communications that share the common text string, the second summary representing at least two of the second text communications from the second group.

A method performed by at least one processing unit of a computing device, the method comprising:

• filtering a plurality of communications to identify a subset of the communications that each refer to a common hashtag or a common web link;
• computing similarity scores for the identified subset of the communications, the similarity scores reflecting semantic similarities among individual communications of the subset;
• computing diversity scores for the identified subset of the communications, the diversity scores reflecting semantic differences among individual communications of the subset;
• combining the similarity scores and the diversity scores to determine total scores for the identified subset of the communications;
• selecting one or more of the individual communications based on the total scores;
• and
• generating a summary of the common hashtag or common web link referred to by each of the communications of the subset, the summary representing the one or more selected individual communications.