Event categorization and key prospect identification from storylines

A non-transitory machine-readable medium storing instruction which, when executed by one or more computing machines of a server computer system, cause the one or more computing machines to implement operations comprising:

• accessing, by a server computer system via a network, a plurality of storylines from one or more social networking systems, each storyline having a time and a geographic location;
• determining, using one or more computing machines of the server computer, themes in the plurality of storylines;
• identifying, using the one or more computing machines of the server computer, a plurality of events comprising an event, each of the plurality of events having an event time, an event geographic location, and an event theme;
• generating, using the one or more computing machines of the server computer system, an event category score for each of the plurality of storylines based on performing machine learning model analysis of correspondences between the time, the geographic location, and the themes associated with each of the plurality of storylines and the event time, the event geographic location, and the event theme of each of the plurality of events, wherein performing the machine learning model analysis is distributed among in-memory computing resources of the one or more computing machines and comprises:
• generating a plurality of event category scores by a plurality of different machine learning models each trained to score a correspondence between storyline time, storyline geographic location, and storyline theme with event time, event geographic location, and event theme for a corresponding event category, and
• categorizing each storyline into an event, including categorizing one or more storylines into an event category associated with the identified event through machine learning, when event category scoring satisfies a categorization threshold associated with each of the one or more events including the identified event;
• identifying, using the one or more computing machines of the server computer, one or more prospects that are mentioned with a highest frequency in social media posts from which storylines categorized into the identified event are constructed, the one or more identified prospects being relevant to the identified event, and the one or more prospects being one or more persons or one or more organizations;
• and
• providing, as a digital transmission via the network, an output representing the one or more identified prospects.

A server computer system, comprising:

• processing hardware of one or more computing machines;
• a memory storing instruction which, when executed by the processing hardware, cause the one or more computing machines to implement operations comprising:
• accessing, via a network, a plurality of storylines from one or more social networking systems, each storyline having a time and a geographic location;
• determining themes in the plurality of storylines;
• identifying a plurality of events comprising an event, each of the plurality of events having an event time, an event geographic location, and an event theme;
• generating an event category score for each of the plurality of storylines based on performing machine learning model analysis of correspondences between the time, the geographic location, and the themes associated with each of the plurality of storylines and the event time, the event geographic location, and the event theme of each of the plurality of events, wherein performing the machine learning model analysis is distributed among in-memory computing resources of the one or more computing machines and comprises:
• generating a plurality of event category scores by a plurality of different machine learning models each trained to score a correspondence between storyline time, storyline geographic location, and storyline theme with event time, event geographic location, and event theme for a corresponding event category, and
• categorizing each storyline into an event, including categorizing one or more storylines into an event category associated with the identified event through machine learning, when event category scoring satisfies a categorization threshold associated with each of the one or more events including the identified event;
• identifying one or more prospects that are mentioned with a highest frequency in social media posts from which storylines categorized into the identified event are constructed, the one or more identified prospects being relevant to the identified event, and the one or more prospects being one or more persons or one or more organizations;
• and
• providing, as a digital transmission via the network, an output representing the one or more identified prospects.