Modeling semantics of inconsistent qualitative knowledge for quantitative Bayesian network inference

Area of Science:

• Computational Statistics
• Bioinformatics
• Systems Biology

Background:

• Qualitative knowledge is often used in Bayesian inference.
• Inconsistent qualitative knowledge poses a challenge for standard Bayesian methods.
• Integrating diverse knowledge sources is crucial for complex modeling.

Purpose of the Study:

• To develop a novel framework for quantitative Bayesian inference from potentially inconsistent qualitative knowledge.
• To enable robust modeling by reconciling conflicting information sources.
• To apply the framework to a realistic biomolecular interaction problem.

Main Methods:

• A hierarchical Bayesian model is proposed to integrate inconsistent qualitative knowledge.
• Prior belief distributions are calculated based on knowledge features.
• Model classes are defined for each inconsistent knowledge component.
• Quantitative inference is approximated using model averaging with Monte Carlo methods.

Main Results:

• The framework successfully integrates inconsistent qualitative knowledge.
• Benchmarking on the ASIA network demonstrates method validity.
• Application to breast cancer bone metastasis reveals effective biomolecular interaction modeling.
• The method enables consistent modeling and quantitative inference.

Conclusions:

• The proposed framework provides a robust approach for quantitative Bayesian inference with inconsistent qualitative knowledge.
• This method reconciles conflicting information, leading to more reliable models.
• It has potential applications in complex systems biology and bioinformatics problems.