Methods and systems for automated generation of personalized messages
A system includes a set of crawlers that find and retrieve documents from an information network, an information extraction system, a knowledge graph storing nodes and edges that connect them, wherein each node represents a respective entity of a corresponding entity type of a plurality of entity types, and wherein the knowledge graph further stores event data relating to events detected by the information extraction system, a machine learning system that trains models that are used in connection with at least one of entity extraction, event extraction, recipient identification, and content generation, a lead scoring system that scores the relevance of information to an individual and references information in the knowledge graph, and a content generation system that generates content of a personalized message to a recipient who is an individual for which the lead scoring system has determined a threshold level of relevance.
1. A computer-implemented method comprising:
dynamically updating a knowledge data structure to reflect a new entity identified by a crawler of a crawling system extracting information from an external data source over a communication network by:
identifying, by a machine learning system utilizing a classification model, the new entity from the information extracted by the crawler; and
dynamically updating the knowledge structure for the new entity by creating a new node and a new edge between the new node and an existing node within the knowledge data structure, wherein the new node represents the new entity and the new edge represents a new relationship identified by the machine learning system between the new entity and an existing entity represented by the existing node, wherein the knowledge structure is populated with:
entity data representing business entities and individual entities stored as objects within a customer relationship database (CRM) system and tracked using the knowledge structure;
relationship data of relationships between the business entities and individual entities; and
event data relating to events associated with the business entities and individual entities, wherein the event data is associated by the knowledge structure with the new node and the new edge connecting the new node to the existing node based upon the event data specifying an event that occurred between the new entity represented by the new node and the existing entity represented by the existing node;
determining, by a processing system using attributes of an ideal recipient identified based upon an objective of a message and historical data related to outcomes associated with previously sent messages generated to achieve the objective of the message, a recipient list based on a recipient profile and the knowledge data structure that stores entity data relating to entities and relationships between the entities;
generating and providing, by the processing system, a personalized message personalized to an individual in the recipient list based on the entity data and the relationships, wherein the processing system automatically generates the personalized message by:
automatically inferring a message template from historical data of users;
constructing personalized content based upon an objective of communicating with the individual; and
populating the message template with the personalized content to automatically generate the personalized message, wherein the message template is populated with directed content generated using the event data associated by the knowledge structure with the new node and the new edge.
2. The method of claim 1 , wherein the recipient list identifies individuals that are more likely to result in a successful outcome given the recipient profile and information represented in the knowledge data structure.
3. The method of claim 1 , wherein the knowledge data structure stores a plurality of nodes and a plurality of edges that connect respective nodes from the plurality of nodes, wherein each node represents a respective entity of a respective entity type and each edge corresponds to a respective relationship of a respective relationship type.
4. The method of claim 1 , wherein the knowledge data structure stores event data relating to a plurality of detected events relating to the entities and the relationships represented in the knowledge data structure; and the method further comprising:
extracting, by the processing system, a new event corresponding to the new entity, wherein extracting the new relationship is further based on the new event.
5. The method of claim 4 , wherein the new event is extracted using an event classification model that is trained to identify events indicated in documents.
6. The method of claim 1 , comprising:
utilizing a lead scoring system to filter entities from the knowledge data structure using an ideal recipient profile to create the recipient list, wherein lead scores are assigned to the entities using a lead scoring model.
7. The method of claim 1 , wherein the new entity is extracted using a classification model that is trained to identify entities indicated in digital documents crawled by the crawler.
8. The method of claim 1 , wherein the recipient profile indicates attributes of an ideal recipient of the message to be sent on behalf of a user.
9. The method of claim 8 , wherein the determining the recipient list comprises filtering one or more entities from the entities represented in the knowledge data structure based on the attributes.
10. The method of claim 8 , wherein the determining the recipient list comprises, for each individual of a subset of the individuals represented in the knowledge data structure, determining a lead score of each individual based on the attributes of the recipient profile using a machine-learned scoring model.
11. The method of claim 10 , wherein the lead score of each individual is further based on an event related to an organization of each individual.
12. The method of claim 1 , wherein the generating the personalized message for the individual comprises:
generating directed content based on retrieved entity data retrieved from the knowledge data structure, wherein the directed content comprises a phrase with information corresponding to the retrieved entity data, wherein the personalized message is generated based upon the directed content and the message template.
13. The method of claim 12 , wherein the directed content is generated based on a machine-learned generative model that is trained to generate text given entity data of an entity and a particular objective of the personalized message, and wherein message data indicates the particular objective of the personalized message.
14. The method of claim 1 , wherein the knowledge data structure stores event data relating to a plurality of detected events relating to the entities and the relationships represented in the knowledge data structure, wherein the personalized message is generated using directed content and the message template, and wherein the directed content is derived from the event data.
15. The method of claim 14 , wherein the directed content is generated based on a machine-learned generative model that is trained to generate text given a particular objective of the personalized message and the event data relating to a known event that occurred with respect to an entity and the particular objective of the personalized message, and wherein message data indicates the particular objective of the personalized message.
16. The method of claim 1 , wherein the processing system uses natural language processing to extract the new entity from one or more digital documents obtained from the crawler.
17. A system comprising:
memory comprising instructions; and
a processor configured to execute the instructions to perform operations comprising:
dynamically updating a knowledge data structure to reflect a new entity identified by a crawler of a crawling system extracting information from an external data source over a communication network by:
identifying, by a machine learning system utilizing a classification model, the new entity from the information extracted by the crawler; and
dynamically updating the knowledge structure for the new entity by creating a new node and a new edge between the new node and an existing node within the knowledge data structure, wherein the new node represents the new entity and the new edge represents a new relationship identified by the machine learning system between the new entity and an existing entity represented by the existing node, wherein the knowledge structure is populated with:
entity data representing business entities and individual entities stored as objects within a customer relationship database (CRM) system and tracked using the knowledge structure;
relationship data of relationships between the business entities and individual entities; and
event data relating to events associated with the business entities and individual entities, wherein the event data is associated by the knowledge structure with the new node and the new edge connecting the new node to the existing node based upon the event data specifying an event that occurred between the new entity represented by the new node and the existing entity represented by the existing node;
determining, by a processing system using attributes of an ideal recipient identified based upon an objective of a message and historical data related to outcomes associated with previously sent messages generated to achieve the objective of the message, a recipient list based on a recipient profile and the knowledge data structure that stores entity data relating to entities and relationships between the entities;
generating and providing, by the processing system, a personalized message personalized to an individual in the recipient list based on the entity data and the relationships, wherein the processing system automatically generates the personalized message by:
automatically inferring a message template from historical data of users;
constructing personalized content based upon an objective of communicating with the individual; and
populating the message template with the personalized content to automatically generate the personalized message, wherein the message template is populated with directed content generated using the event data associated by the knowledge structure with the new node and the new edge.
18. The system of claim 17 , wherein the knowledge data structure stores event data relating to the entities, and wherein the event data is used to generate the personalized message.
19. The system of claim 17 , wherein the crawler obtains digital documents include documents obtained from public internet websites.
20. The system of claim 17 , wherein the crawler is seeded with a set of resource identifiers from which to obtain digital documents.
21. A computer-implemented method comprising:
dynamically updating a knowledge data structure to reflect a new entity identified by a crawler of a crawling system extracting information from an external data source over a communication network by:
identifying, by a machine learning system utilizing a classification model, the new entity from the information extracted by the crawler; and
dynamically updating the knowledge structure for the new entity by creating a new node and a new edge between the new node and an existing node within the knowledge data structure, wherein the new node represents the new entity and the new edge represents a new relationship identified by the machine learning system between the new entity and an existing entity represented by the existing node, wherein the knowledge structure is populated with:
entity data representing business entities and individual entities stored as objects within a customer relationship database (CRM) system and tracked using the knowledge structure;
relationship data of relationships between the business entities and individual entities; and
event data relating to events associated with the business entities and individual entities, wherein the event data is associated by the knowledge structure with the new node and the new edge connecting the new node to the existing node based upon the event data specifying an event that occurred between the new entity represented by the new node and the existing entity represented by the existing node;
determining, by a processing system using attributes of an ideal recipient identified based upon an objective of a message and historical data related to outcomes associated with previously sent messages generated to achieve the objective of the message, a recipient list based on a recipient profile and the knowledge data structure that stores entity data relating to entities and relationships between the entities;
generating and providing, by the processing system, a personalized message personalized to an individual in the recipient list based on the entity data and the relationships, wherein the processing system automatically generates the personalized message by:
automatically inferring a message template from historical data of users;
constructing personalized content based upon an objective of communicating with the individual; and
populating the message template with the personalized content to automatically generate the personalized message, wherein the message template is populated with directed content generated using the event data associated by the knowledge structure with the new node and the new edge.
22. The method of claim 21 , wherein the knowledge data structure stores event data relating to a plurality of detected events, and wherein the personalized message is generated based upon the event data.
23. The method of claim 22 , further comprising:
extracting, by the processing system, a new event corresponding to the new entity, wherein extracting the new relationship is further based on the new event.
24. The method of claim 21 , wherein the knowledge data structure stores a plurality of nodes and a plurality of edges that connect respective nodes from the plurality of nodes, wherein each node represents a respective entity of a respective entity type and each edge corresponds to a respective relationship of a respective relationship type.