Article Quality Scoring Using Browsing Behavior Indices
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current article quality scoring methods are flawed as they often rely on click-through rate (CTR) or article length, leading to inaccurate assessments where attractive titles with irrelevant content are scored high and short, high-quality articles are scored low.
Innovation Solution
An article quality scoring method that utilizes browsing behavior information such as reading time index, reading length index, and reading experience index, combined with corresponding coefficients, to calculate a browsing behavior score, which is then averaged across multiple users to determine an article's quality score.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If article quality is scored according to click-through rate (CTR), then articles with attractive titles receive high scores, but articles with irrelevant content are also scored high leading to inaccurate quality assessment
Solution Approach 1:
The patent implements feedback mechanisms by collecting user browsing behavior data (reading time, reading length, positive/negative feedback) and using this feedback to calculate quality scores. This creates a closed-loop system where actual user engagement feedback continuously improves the accuracy of quality assessment, resolving the contradiction between measurement precision and reliability by grounding scores in actual user experience rather than superficial metrics like CTR
Solution Approach 2:
The patent replaces the mechanical/simplified scoring system (based on CTR or article length) with a more sophisticated information processing system that analyzes multiple dimensions of user browsing behavior. This substitution transforms the quality assessment from a crude metric-based system to a comprehensive behavioral analysis system, simultaneously improving both measurement precision and reliability
2Measurement precision
If article quality is scored according to article length, then longer articles receive higher scores, but short high-quality articles are scored low
Solution Approach 1:
The patent changes the scoring parameters from static attributes (article length) to dynamic behavioral parameters (reading time, reading length ratio, feedback indications). This parameter transformation allows the scoring system to adapt to articles of any length by measuring actual user engagement rather than relying on length as a proxy for quality, thereby improving both accuracy and adaptability simultaneously
Solution Approach 2:
The patent creates a universal scoring methodology that works across diverse article types and lengths by using normalized behavioral metrics. The reading length ratio (actual reading length divided by total length) and reading time per unit length provide length-independent measures of engagement that can fairly assess both short and long articles, achieving universality while maintaining precision
3Measurement precision
If multiple browsing behavior metrics are collected and processed, then quality assessment precision improves, but system complexity increases
Solution Approach 1:
The patent segments the quality assessment system into distinct functional modules: data collection module (gathering browsing behavior information), processing module (calculating reading time index, reading length index, feedback index), and scoring module (computing final quality scores). This segmentation allows each module to handle specific tasks independently, improving precision through specialized processing while managing complexity through modular architecture
Solution Approach 2:
The patent introduces intermediary computational layers that transform raw browsing behavior data into standardized indices (reading time index, reading length index, feedback index) before final score calculation. These intermediaries simplify the complexity by creating standardized intermediate representations that are easier to process and combine, thereby maintaining precision while reducing system complexity
Data Source
AI summary
The present application discloses an article quality scoring method and device, a client, a server, and a programmable device. The method includes: obtaining browsing behavior information of a user when the user browses a target article; obtaining a browsing behavior score of the user for the target article according to the browsing behavior information and a corresponding browsing behavior coefficient; and obtaining an article quality score of the target article according to obtained browsing behavior scores of multiple users for the target article.