Conversational assistant using extracted guidance knowledge

A system comprising:

• one or more hardware processors;
• and
• a memory storing instructions that, when executed by the one or more hardware processors, causes the one or more hardware processors to perform operations comprising:
• receiving, via an interface of a machine agent, a request for information that compares a pair of products, the request comprising a natural language question inputted by a user during an interactive natural language conversation;
• in response to receiving the request, accessing, by the machine agent from a product knowledge database, textual statements about one or more features corresponding to the pair of products, the textual statements obtained by a machine-based offline knowledge extraction process that extracts the textual statements from product information of one or more of the pair of products and at least one product review of one or more of the pair of products;
• based on the accessed textual statements and an overall empirical utility of each of the accessed textual statements estimated using past machine agent conversations with multiple users over a period of time, determining, by a hardware processor of the machine agent, one or more statements comparing the one or more features between the pair of products to convey to the user, the overall empirical utility of each of the accessed textual statements quantifying a likelihood of communicating a corresponding accessed textual statement of the accessed textual statements resulting in product purchase;
• and
• causing, by the machine agent, presentation of the one or more statements comparing the one or more features between the pair of products as a natural language response to the user.

A non-transitory machine-readable medium storing instructions that, when executed by one or more processors of a machine, cause the machine to perform operations comprising:

• initiating, by a machine agent, an interactive natural language conversation between a user and the machine agent to provide the user with guidance on one or more products;
• receiving, via an interface of the machine agent, a request for information that compares a pair of products, the request comprising a natural language question inputted by the user during the interactive natural language conversation;
• in response to receiving the request, accessing, by the machine agent from a product knowledge database, textual statements about one or more features corresponding to the pair of products, the textual statements obtained by a machine-based offline knowledge extraction process that extracts the textual statements from product information of one or more of the pair of products and at least one product review of one or more of the pair of products;
• based on the accessed textual statements and an overall empirical utility of each of the accessed textual statements estimated using past machine agent conversations with multiple users over a period of time, determining, by a hardware processor of the machine agent, one or more statements comparing the one or more features between the pair of products to convey to the user, the overall empirical utility of each of the accessed textual statements quantifying a likelihood of communicating a corresponding accessed textual statement of the accessed textual statements resulting in product purchase;
• and
• causing, by the machine agent, presentation of the one or more statements comparing the one or more features between the pair of products as a natural language response to the user.