US 12,591,914 B2 · Real-time collateral recommendation

Abstract

Methods, systems and products for real-time collateral recommendation for business development representatives (‘BDR’) is presented, including receiving communications between a BDR and a customer across disparate communications channels, each channel comprising a channel type, channel communications protocol, and channel form. Embodiments include normalizing the disparate communications in dependence upon normalization rules, synchronizing the disparate communications in dependence upon a set of rules appropriate to the end task including an occurrence of the disparate communications in time order, summarizing the normalized communications in dependence upon summarization rules, creating semantic triples in dependence upon the summarized and normalized communications including applying a sales-directed taxonomy and ontology; identifying one or more insights in dependence upon the semantic triples; and selecting, in real-time in dependence upon the identified insights, real time collateral for the BDR related to the communications between the BDR and the customer; and presenting the selected real-time collateral to the BDR.

A method of real-time multimedia content recommendation using a voice server during communication between one or more of speech-enabled devices supporting an intelligence assistance and one or more of user devices, the method comprising:

electrically receiving communications, by the voice server, between a speech-enabled device and a user device-across disparate communications channels, each channel comprising a channel type, channel communications protocol, and channel form, wherein the voice server includes at least one computer processor and a memory which is connected through a high-speed memory bus and a bus adapter to the at least one computer processor, and wherein the electrically receiving communication comprises:
accepting, by the speech-enabled device, speech from a user;
recognizing, by the speech-enabled device, the speech from the user, wherein recognizing the speech includes converting, at least one of real-time speech contributions and recorded speech contributions of the communications, to text by performing a natural language processing and automatic speech recognition (NLP-ASR);
and
expressing, by the speech-enabled device, voice prompts and speech responses to the user;
electrically normalizing, by the voice server, the disparate communications in dependence upon normalization rules;
electrically automatically synchronizing, by the voice server, the at least one of real-time speech contributions and recorded speech contributions of the communications across disparate communications channels in dependence upon a set of rules appropriate to the end task including an occurrence of the disparate channel communication in time order;
electrically summarizing, by the voice server, the normalized communications in dependence upon summarization rules;
electrically creating, by the voice server, semantic triples in dependence upon the summarized and normalized communications, between the one or more of speech-enabled devices supporting the intelligence assistance and the one or more of user devices, wherein electrically creating the semantic triples further comprises:
applying a multimedia content-specific taxonomy and ontology;
receiving the text converted from the at least one of real-time speech contributions and recorded speech contributions of the communications;
and
segmenting the text by identifying portions of the text corresponding to the multimedia content-specific taxonomy and ontology to assign one or more identifications to each identified portion of the text;
and
storing the segmented and identified portions of the text as searchable semantic triples in a semantic graph database, wherein each searchable semantic triple comprises an internal pointer to an originating multimedia content database, a link to a document location, and a hyperlink to a webpage to reference back to the originating multimedia content;
electrically identifying, by querying the semantic graph database using the voice server, one or more insights in dependence upon the searchable semantic triples generated and stored in the semantic graph database based on the normalized and summarized communications;
electrically selecting, by the voice server, in real-time in dependence upon the identified insights, real-time multimedia content related to the communications between the speech-enabled device and the user device, wherein selecting the real-time multimedia content comprises:
pre-processing, using the multi-media content-specific taxonomy and ontology, the multimedia content to generate semantic triples specific to the multimedia content;
generating a repository of the pre-processed multimedia content, accessible by the intelligence assistance, to provide relevant multimedia content to the speech-enabled device;
generating, based on the repository, an enterprise knowledge graph of the semantic graph database by storing multimedia-based semantic triples;
receiving, using a query based on the semantic triples in dependence upon the summarized and normalized communications, a candidate multimedia content from the enterprise knowledge graph of multimedia-based semantics;
and
selecting, based on the received candidate multimedia content, the real-time multimedia content during communications between the speech-enabled device and the user device;
and
electrically presenting, by the voice server, the selected real-time multimedia content to the speech-enabled device, wherein presenting the selected real-time multimedia content comprises:
displaying the selected multimedia content to the one or more devices via an agent dashboard or a user interface;
and
providing a digital copy of the selected multimedia content in transmittable form to the one or more users.

25 more elements

Real-time collateral recommendation

Life of this patent

Real-time collateral recommendation

Life of this patent

Abstract

Claims18