Patent · US Active

Cognitive searches based on deep-learning neural networks

US10803055B2 · kind B2 · utility

6Cited by

0References

29Claims

0Family size

Assignee

ACCENTURE GLOBAL SOLUTIONS LIMITED · IE

Inventors

Jadran Sirotkovic · Dublin, IE
Nicholas McCarthy · Dublin, IE

Key dates

Filing date	Dec 15, 2017
Grant date	Oct 13, 2020
Priority date	—
Expiry date	Aug 1, 2038

Classification

Technology area (CPC G)Physics
CPC primaryG06N3/09
WIPO fieldComputer technology
WIPO sectorElectrical engineering

Abstract

This disclosure relates to a development and application of a deep-learning neural network (DNN) model for identifying relevance of an information item returned by a search engine in response to a search query by a user, with respect to the search query and a profile for the user. The DNN model includes a set of neural networks arranged to learn correlations between queries, search results, and user profiles using dense numerical word or character embeddings and based on training targets derived from a historical search log containing queries, search results, and user-click data. The DNN model help identifying search results that are relevant to users according to their profiles.

Source: USPTO / EPO open patent data. Objective bibliographic and citation counts.