Bi-partite Graph Traversal for Search Query Classification

Solution Overview

Problem

Current search engines lack precision in determining user intent from search queries, leading to inadequate results due to insufficient training data and high computational resource requirements for classification.

Innovation Solution

A search task classification process that uses seed queries and query patterns to construct a bi-partite graph, expanding training data through traversal, and employs a task classification module to classify queries into predefined categories, reducing computational resources and time.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If traditional topic detection or relevance ranking technology is used to determine user intent, then the search engine can process queries using existing methods, but the precision is insufficient to accurately determine the task associated with the query

Engineering Contradiction:

Improveprecision in determining user intentVSAvoidaccuracy in determining task association

Core Design Contradiction:

Measurement precisionVSReliability

Solution Approach 1:

The system performs preliminary classification of search queries into predefined task categories before executing search operations. By pre-defining task categories and training classification models in advance, the system can quickly determine user intent without requiring complex real-time analysis, thereby improving both precision and reliability in task determination

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The patent introduces an intermediate task classification layer between the user query and the search execution. This intermediary component (task classification module) acts as a mediator that translates raw queries into structured task categories, enabling more precise and reliable determination of user intent by bridging the gap between simple keyword matching and complex intent understanding

Inventive Principle:

Principle #24Intermediary (Mediator)

2Measurement precision

If sufficient training data is collected and classified to improve query classification accuracy, then classification precision improves, but computational resources and time required become considerable

Engineering Contradiction:

Improveclassification precisionVSAvoidcomputational resources required

Core Design Contradiction:

Measurement precisionVSUse of energy by moving object

Solution Approach 1:

The system performs preliminary classification of search queries into predefined task categories before executing search operations. By pre-defining task categories and training classification models in advance, the system can quickly determine user intent without requiring complex real-time analysis, thereby improving both precision and reliability in task determination

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The patent changes the parameters of the classification system by defining predefined task categories and using these categories as fixed parameters for classification. This approach transforms the classification problem from an open-ended task to a structured multi-class classification problem with defined parameters, improving precision while reducing computational complexity through the use of predetermined category structures

Inventive Principle:

Principle #35Parameter changes

3Measurement precision

If sufficient training data is collected and classified to improve query classification accuracy, then classification precision improves, but the time required for data collection and classification becomes considerable

Engineering Contradiction:

Improveclassification precisionVSAvoidtime required for data collection and classification

Core Design Contradiction:

Measurement precisionVSLoss of time

Solution Approach 1:

The system performs preliminary classification of search queries into predefined task categories before executing search operations. By pre-defining task categories and training classification models in advance, the system can quickly determine user intent without requiring complex real-time analysis, thereby improving both precision and reliability in task determination

Inventive Principle:

Principle #10Preliminary action

Data Source

PatentUS8838512B2Random walk on query pattern graph for query task classification

Publication Date: 2014.09.16 MICROSOFT TECHNOLOGY LICENSING LLC

AI summary

A classification process may reduce the computational resources and time required to collect and classify training data utilized to enable a user to effectively access online information. According to some implementations, training data is established by defining one or more seed queries and query patterns. A bi-partite graph may be constructed using the seed query and query pattern information. A traversal of the bi-partite graph can be performed to expand the training data to encompass sufficient data to perform classification of the present search task.