Query Compression via Graph-Based Intent Scoring

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Users often enter verbose queries that fail to concisely express their intent, leading to irrelevant search results, extended search sessions, and inefficient resource usage for both users and search engines, resulting in lost revenue for advertisers and search engines due to inefficient ad placement.

Innovation Solution

A computer-implemented technique compresses original queries into sub-queries using graph-based analysis to generate feature values and scores, selecting the most relevant candidate queries that effectively express the user's intent, thereby identifying and presenting the most relevant target items, such as ads, to the user.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If the search engine uses the original verbose query for matching, then the query fully expresses user intent, but the search results become less relevant and require extended search sessions

Engineering Contradiction:

Improvequery intent expression accuracyVSAvoidsearch result relevance efficiency

Core Design Contradiction:

Measurement precisionVSProductivity

Solution Approach 1:

The original verbose query is segmented into multiple candidate sub-queries by removing different combinations of tokens. Each candidate query represents a potential core intent expression. The system then evaluates these segmented queries and selects the most effective one for matching against target items, resolving the contradiction between preserving full intent and achieving relevant results efficiently.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

Instead of directly using the original verbose query for matching, the system inverts the approach by generating simplified candidate queries from the original query and then selecting the best candidate. This inversion allows the system to overcome the verbosity problem while maintaining intent accuracy through the selection process guided by scoring models and graph-based analysis.

Inventive Principle:

Principle #13The other way round (Inversion)

2Reliability

If the user conducts an extended search session to find relevant target items, then the user can find useful results, but communication and processing resources are wasted

Engineering Contradiction:

Improvesearch result usefulnessVSAvoidcommunication and processing resources

Core Design Contradiction:

ReliabilityVSLoss of energy

Solution Approach 1:

The system performs preliminary action by pre-processing the original query into multiple candidate sub-queries and evaluating them using graph-based analysis and scoring models before the actual search matching occurs. This preliminary compression and selection process ensures that the most relevant candidate query is used for matching, reducing the need for extended search sessions and minimizing resource waste while maintaining result usefulness.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The query compression component performs self-service by automatically analyzing the original query structure, generating candidate queries, and selecting the optimal one without requiring user intervention or multiple user-initiated search attempts. This automated self-optimization reduces both user time investment and system resource consumption while ensuring reliable search results.

Inventive Principle:

Principle #25Self-service

3Adaptability or versatility

If ads are placed based on verbose queries, then all query tokens are considered for matching, but ad placement efficiency and revenue are reduced

Engineering Contradiction:

Improvekeyword matching coverageVSAvoidad placement efficiency

Core Design Contradiction:

Adaptability or versatilityVSProductivity

Solution Approach 1:

The system applies partial action by considering only the most relevant candidate query for ad placement matching, rather than evaluating all possible token combinations from the original verbose query. The graph-based analysis and scoring models identify the candidate query that provides the optimal balance between keyword coverage and matching efficiency, enabling effective ad placement without the computational overhead of analyzing all query tokens equally.

Inventive Principle:

Principle #16Partial or excessive action

4Device complexity

If the search engine processes the original query without compression, then processing is straightforward, but search session duration increases and user experience deteriorates

Engineering Contradiction:

Improvequery processing simplicityVSAvoidsearch session duration

Core Design Contradiction:

Device complexityVSLoss of time

Solution Approach 1:

The system performs preliminary query compression and candidate selection before the main search processing occurs. By pre-processing the verbose query into optimized candidate queries using graph-based analysis and scoring models, the system simplifies the subsequent matching process while significantly reducing search session duration. This preliminary action resolves the contradiction by making the overall process more efficient without sacrificing processing simplicity.

Inventive Principle:

Principle #10Preliminary action

Data Source

PatentUS10248967B2Compressing an original query while preserving its intent

Publication Date: 2019.04.02 MICROSOFT TECHNOLOGY LICENSING LLC

AI summary

A computer-implemented technique is described herein for shortening an original query into one or more sub-queries. The technique chooses the sub-query(ies) such that they preserve the original intent of the original query. To accomplish this goal, the technique uses graph-based analysis to generate a set of richly descriptive query-context-specific feature values for each sub-query, and then uses those feature values to score the relevance of that sub-query.