US 10,592,841 B2Granted

Automatic clustering by topic and prioritizing online feed items

Assigneesalesforce.com, inc.

InventorsAlon Talmor, Tal Menahem Brown

Life of this patent

Estimating the term…

Loading patent…

Abstract

The technology disclosed relates to presenting important business insights to a sales engineer. In particular, the technology disclosed assembles a set of news feed items for companies of interest to a sales engineer and groups them into topics. It also qualifies some of the news feed items to return or not based on mandatory or prohibited words in the news feed items. Further, it determines a plurality of metric values for each of the returned news feed items that are based on one of a source metric, business metric, company reference metric, social buzz metric, and matched account metric. It then orders the news feed items, based on the determined metric values, with respect to one or more of source reputation, business activity-related vocabulary, company-name mention, social buzz, and correlation with accounts preferred by the sales engineer, and presents the ordered news feed items as business insights about the topics.

Claims20

Dependent claims

A method for automatically prioritizing news feed items, the method including:

assembling a set of the news feed items for companies of interest from a plurality of electronic sources for display to a user;
grouping the set into clusters based on topics by:
constructing a graph of the news feed items in the set and representing a connection strength as edges between nodes representing the news feed items, thereby forming connected node pairs;
calculating the connection strength between the nodes of the connected node pairs based on target unified resource locators (URLs) of the news feed items;
and
grouping the news feed items, with the connection strength above a threshold, into the clusters;
preprocessing the clusters to qualify some of the news feed items to return or not based on one or more occurrences of mandatory or prohibited words in respective titles and descriptions of the news feed items, wherein the news feed items that contain a prohibited word are filtered out and not returned;
determining a metric value for at least two of the news feed items that are returned;
determining an order for the at least two news feed items, based at least in part on the metric value;
and
presenting, to the user, the returned news feed items about the topics, wherein the at least two news feed items are presented according to the determined order.

8 more elements

A system for automatically prioritizing news feed items, the system including:

a processor and a computer readable storage medium storing computer instructions configured to cause the processor to:
assemble a set of the news feed items for companies of interest from a plurality of electronic sources for display to a user;
grouping the set into clusters based on topics by:
constructing a graph of the news feed items in the set and representing a connection strength as edges between nodes representing the news feed items, thereby forming connected node pairs;
calculating the connection strength between the nodes of the connected node pairs based on target unified resource locators (URLs) of the news feed items;
and
grouping the news feed items, with the connection strength above a threshold, into the clusters;
preprocess the clusters to qualify some of the news feed items to return or not based on one or more occurrences of mandatory or prohibited words in respective titles and descriptions of the news feed items, wherein the news feed items that contain a prohibited word are filtered out and not returned;
determine a metric value for at least two of the news feed items that are returned;
determine an order for the at least two news feed items, based at least in part on the metric value;
and
present, to the user, the returned news feed items about the topics, wherein the at least two news feed items are presented according to the determined order.

9 more elements

Dependencies are inferred from the claim text.