Generative Adversarial Network for Comment Usefulness Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The abundance of uneven quality comments on the internet makes it difficult for users to efficiently filter out valuable comments from the large number of comments, as conventional methods like crowdsourcing voting yield limited results.
Innovation Solution
A method and apparatus using a generative adversarial network to create an information assessment model by training a predictor model and a discriminator model on comment data, where the predictor model predicts the usefulness of comments and the discriminator model authenticates the labels, iteratively updating parameters based on error and reward functions to improve prediction accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If conventional crowdsourcing voting methods are used to assess comment value, then user participation is maintained, but the efficiency and accuracy of filtering valuable comments deteriorates due to limited voting results
Solution Approach 1:
The patent replaces the mechanical crowdsourcing voting system with an automated machine learning system consisting of a predictor model and discriminator model. The predictor model automatically predicts usefulness labels for comments, while the discriminator model verifies label authenticity, eliminating the need for manual user voting and significantly improving filtering efficiency while maintaining or enhancing assessment accuracy.
Solution Approach 2:
The patent introduces a dual-model intermediary system where the predictor model generates usefulness predictions and the discriminator model acts as a verification layer. This intermediary mechanism between raw comments and final assessment results improves both the efficiency of processing large volumes of comments and the precision of value assessment through cross-validation.
2Quantity of substance
If a large number of comments are processed manually through voting, then more assessment data can be collected, but the time and resources required increase significantly
Solution Approach 1:
The patent implements a self-service assessment system where the predictor and discriminator models automatically process comments without requiring manual user intervention. The system serves itself by using the discriminator model to verify the predictor's outputs, enabling high-volume comment assessment with minimal time investment and resource consumption.
Solution Approach 2:
The patent establishes continuous automated processing of comments through the machine learning models, allowing the system to assess unlimited numbers of comments without interruption. The iterative training process ensures the models continuously improve their assessment capabilities while maintaining high processing speed and low time cost.
3Reliability
If manual verification of comment usefulness is performed, then label authenticity can be ensured, but the complexity and cost of the assessment system increases
Solution Approach 1:
The patent introduces the discriminator model as an intermediary verification layer that automatically checks the authenticity of usefulness labels predicted by the predictor model. This automated intermediary system ensures label reliability without requiring complex manual verification processes, maintaining low system complexity while achieving high trustworthiness in assessments.
Solution Approach 2:
The patent implements a feedback mechanism where the discriminator model's verification results are fed back to the predictor model for iterative improvement. This feedback loop automatically enhances label authenticity over time through training, reducing the need for complex external verification systems while maintaining high reliability in comment usefulness assessment.
Data Source
AI summary
Embodiments of the present disclosure provide a method for generating an information assessment model, a method for determining the usefulness of comment information, apparatus, electronic device, and computer-readable medium. The method may include: acquiring training samples, the training samples including first sample comment information with a usefulness label and second sample comment information without a usefulness label; acquiring a predictor model and a discriminator model respectively constructed based on a generative network and a discrimination network in a generative adversarial network, and pre-training the predictor model using the first sample comment information, the predictor model being used to predict a usefulness label of a piece of comment information, the discriminator model being used to discriminate authenticity of a usefulness label; and training the predictor model and the discriminator model by iteratively performing a plurality of times of training operations, using the trained predictor model as an information assessment model.