Analyzing impact of parental occupation on child's learning performance: a semantics-driven probabilistic approach

Scientific research on the effect of parent's socioeconomic status on child's learning performance is a popular topic from the last century. However, majority of these researches are based on traditional statistical models and involve subjective analysis of the survey data which often becomes impractical when the data volume is huge. Consequently, there remain ample scopes for developing improved formal methods that can replace the subjective analysis with appropriate embedding of data semantics, and thereby, can generate similar or even new insights into this domain. This paper proposes a semantically enhanced probabilistic model based on a variant of semantic Bayesian network to analyze the impact of the parents' profession on their child's learning performance. The novelty of the work lies in incorporating occupational semantics in the child-performance analysis model and developing a probabilistic framework to study the impact. Experimentation using real-world datasets from a number of state-level schools in India reveals interesting facts, especially regarding regional influence of parents' occupation on the child performance. The results of comparative study demonstrate that the proposed model with embedded data semantics has better implication power than the techniques with no incorporated semantics.