Keyword Performance Assessment via Classification Scoring

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

There is a need for effective techniques in performance assessment and classification, particularly in electronic advertising, where training data often lacks sufficient information to accurately determine the productivity of search keywords, leading to inefficiencies in targeting audiences and optimizing advertising systems.

Innovation Solution

A system that stores information on keyword occurrences, successful occurrences, and attributes, using a classification scheme to assign scores based on successful occurrences, enabling conversion-rate estimation and performance assessment for keywords, which updates bidding databases for cost reduction.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If machine learning is used to generate classification schemes from training data, then classification accuracy may be improved, but the system becomes ineffective when training data contains scarce information on productivity

Engineering Contradiction:

Improveclassification accuracyVSAvoidinformation scarcity in training data

Core Design Contradiction:

Measurement precisionVSLoss of information

Solution Approach 1:

The system performs preliminary classification of items into groups based on available attributes before assessing productivity. By pre-grouping items with similar characteristics, the system can leverage aggregated data from multiple items in each group, thereby compensating for information scarcity in individual items and improving overall classification accuracy.

Inventive Principle:

Principle #10Preliminary action

2Productivity

If classification is performed on items with sparse data, then performance assessment coverage is improved, but measurement reliability deteriorates

Engineering Contradiction:

Improveperformance assessment coverageVSAvoidmeasurement reliability

Core Design Contradiction:

ProductivityVSReliability

Solution Approach 1:

The system merges data from multiple items into classification groups, aggregating information across items that share similar attributes. This combining approach allows the system to assess performance of items with sparse data by leveraging aggregated statistics from their classification group, thereby maintaining measurement reliability while expanding assessment coverage.

Inventive Principle:

Principle #5Merging (Combining)

3Ease of operation

If traditional performance assessment methods are used with limited information, then implementation simplicity is maintained, but assessment accuracy deteriorates

Engineering Contradiction:

Improveimplementation simplicityVSAvoidassessment accuracy

Core Design Contradiction:

Ease of operationVSMeasurement precision

Solution Approach 1:

The system introduces classification groups as intermediary structures between individual items and performance assessment. These intermediate groups aggregate information from multiple items, providing a mediator that bridges the gap between limited individual data and accurate performance assessment, thereby improving accuracy without significantly complicating the overall process.

Inventive Principle:

Principle #24Intermediary (Mediator)

Data Source

PatentUS8856130B2System, a method and a computer program product for performance assessment

Publication Date: 2014.10.07 KENSHOO

AI summary

A system, a computerized method, and a computer program product for classification of items based on their attributes and on a classification scheme that is defined based on information pertaining to each item of a set of items, and which is indicative of: (a) a quantity of occurrences of the item in a sample; (b) a quantity of successful occurrences of the item in the sample; and (c) at least one attribute of the item with regard to at least one variable out of a set of variables.