How Assistant Systems Combine Information About One Thing from Many Places

This patent describes a system that gathers all known information about a single person, place, or thing from various sources and combines it into one complete profile for an assistant system.

What it covers

This method helps an assistant system provide a complete picture of an 'entity' (like a person or place) by combining information from multiple sources. When a 'client system' requests information about a 'first record' (Claim 1), the system identifies a 'globally unique entity identifier' linked to that record (Claim 1). It then finds other 'second records' also linked to this same unique identifier (Claim 1). All these related records are then 'deduped' and 'compiled' into a single 'fused record' (Claim 1), which is then sent back to the client system for presentation. For example, if you ask a voice assistant about a specific author, it could combine details from their book listings, social media profiles, and interview transcripts into one unified summary.

What it doesn't cover

• —Does not cover combining data about an entity if there is no globally unique identifier linking the different records.
• —Does not cover systems that link data but do not generate and present a single 'fused record' in response to a user request.
• —Does not cover methods of combining entity data that do not involve a deduplication step to associate records with a unique identifier.
• —Does not cover systems that simply store multiple records about an entity without actively compiling and presenting a unified view.

The clever bit

The clever part is how it automatically finds all scattered pieces of information about one specific thing across many different data sources, links them using a unique ID, and then intelligently combines them into one clean, complete summary to show a user. This avoids showing fragmented or repetitive information.

Why it matters

This patent is important for making AI assistant systems smarter and more helpful. By combining information about a person, place, or topic from many different places, it helps assistants give you a complete picture instead of just bits and pieces. This technology is key for companies like Meta Platforms Inc. to build advanced virtual assistants that can answer complex questions by drawing on a wide range of data.

Real-world examples

• 1.Meta AI
• 2.Google Assistant
• 3.Amazon Alexa
• 4.Apple Siri
• 5.Knowledge panels in search engines