Automatic detection and contextual reformulation of implicit device-related queries. Modern multi-device query understanding — when users query devices implicitly, the system understands and routes appropriately.
Patent Overview
- Inventor
- Susan T. Dumais, Adam Fourney, others
- Assignee
- Microsoft Technology Licensing, LLC
- Filed
- 2016
- Granted
- 2022-07-12
The Challenge
The Challenge
Users often query about device-specific issues without naming the device. 'Battery drain' from a phone user implies that device. 'Won't print' implies the user's connected printer. Implicit-device query detection and contextual reformulation routes queries to appropriate device-specific resources.
- Implicit Device Context Is Common — Per query, users often omit device context they assume implicit.
- Detection Requires Multi-Source Signals — Per (user, query), device-context signals combine.
- Reformulation Adds Device Context — Per detected implicit-device query, reformulation adds device explicitly.
- Privacy Must Be Preserved — Per user, device context handled with privacy.
- Multi-Device Era Strategic Significance — Per modern user, multi-device usage makes implicit-device queries pervasive.
Innovation
How The System Works
The system detects implicit-device queries via multi-source signals, identifies the implied device per (user, query), reformulates queries with explicit device context, and routes reformulated queries through standard retrieval.
- Capture Multi-Source Signals — Per user, current device, recent device interactions, query patterns captured.
- Detect Implicit-Device Queries — Per query, classifier detects implicit-device intent.
- Identify Implied Device — Per (user, query), implied device identified.
- Reformulate Query — Per query, reformulation adds device explicitly.
- Route Through Retrieval — Reformulated query routed through standard retrieval.
- Validate — Per query, validation against user engagement.
- Privacy Preserve — Per user, device signals handled with privacy.
Implicit-Device Detection And Reformulation
The patent's load-bearing idea is that multi-device users often query implicitly about their current device. Detecting and reformulating these queries routes them to appropriate device-specific resources.
Context-Aware Query Reformulation
Per (user, query), implicit context detected and added explicitly. Reformulation bridges user-side context and query-side specificity.
- Multi-Source Signal Capture — Current device, recent interactions, query patterns combine.
- Implicit-Device Detection — Per query, implicit-device classifier runs.
- Contextual Reformulation — Per query, device context added explicitly.
Technical Foundation
Technical Foundation
The patent specifies the signal capturer, implicit-device classifier, device identifier, reformulator, retrieval router, validator, and privacy layer.
- Signal Capturer — Per user, multi-source signals captured.
- Implicit-Device Classifier — Per query, classifies implicit-device intent.
- Device Identifier — Per (user, query), identifies implied device.
- Reformulator — Per query, adds device context.
- Retrieval Router — Per reformulated query, routes through retrieval.
- Privacy Layer — Privacy safeguards on signals.
The Process
The Process
Per query, implicit-device pipeline runs in real time.
- Capture Signals — Multi-source signals captured.
- Receive Query — Query arrives.
- Classify Implicit-Device — Per query, classification runs.
- Identify Device — Per (user, query), device identified.
- Reformulate — Query reformulated.
- Route — Reformulated query routed.
- Validate — Engagement validates.
Quality Control
Quality Control
Wrong device identification damages query. The patent specifies safeguards.
- Classifier Validation — Per query, classification validated.
- Device-Identification Confidence — Per (user, query), confidence threshold.
- Privacy Preservation — Per user, signals handled with privacy.
- Pass-Through Default — Low-confidence cases default to no reformulation.
- Continuous Recalibration — Models refresh.
Real-World Application
Implicit-device query reformulation underpins modern multi-device search across Bing, Cortana, and Microsoft 365. The pattern of context-aware reformulation routes users to appropriate resources without requiring explicit context specification.
- Multi-source Signal Combination — Current device, interactions, query patterns combine.
- Per-(user, query) Granularity — Each user-query pair gets device inference.
- Privacy-preserved Architecture — Privacy safeguards on signals.
Why Device-Specific Content Wins For Implicit-Device Queries
Per reformulated query, device-specific content matches reformulated intent. Pages with clear device-specific framing earn ranking benefit on implicit-device-reformulated queries.
Why Multi-Device Content Strategy Matters
Per user, multi-device usage produces implicit-device queries across many devices. Content strategy that addresses device-specific needs per device compounds across reformulation paths.
<\/section>What This Means for SEO
What This Means for SEO
The system detects queries that implicitly reference a device the user is on, reformulates them with explicit device context, and routes them to device-specific resources. SEO implication: device-specific framing wins these reformulated queries, so name the device, model, and platform your content serves.
- Name The Device Explicitly — Reformulation adds the implied device to the query. Content that clearly states which device, model, or OS it addresses matches the reformulated query; vague troubleshooting content does not. Be specific about the hardware and platform.
- Symptom-To-Device Pages Capture Implicit Queries — Users type symptoms ('battery drain', 'won't print') without the device. Pages that pair the symptom with the specific device context map directly onto reformulated intent. Build symptom-plus-device articles.
- Cover Each Device Separately — Multi-device users generate implicit-device queries across many devices. Distinct content per device compounds across the reformulation paths, whereas one generic page tries to serve all and matches none precisely.
- Model And Version Detail Improves Match — Device-specific framing earns the ranking benefit on reformulated queries. Including model numbers, OS versions, and configuration specifics sharpens the match to the reformulated, device-explicit query.
- Structure Content For Device Routing — The system routes reformulated queries to appropriate device-specific resources. Clear, well-labeled device sections and pages make your content the obvious routing destination for a given device.
- Anticipate The Implied Device Context — Detection relies on multi-source signals about the user's device. Content that anticipates the most common device a query implies (a phone for a mobile-app issue) aligns with the likely reformulation.
- Multi-Device Strategy Is Now Baseline — The patent frames multi-device usage as making implicit-device queries pervasive. A content strategy that systematically covers the device matrix for your topic captures a query class that generic content misses entirely.