Context-aware recommender system (CARS) is a promising technique for recommending research resources to users (researchers) by predicting their preferences (resources) under different situations. If the contextual information given to such a system is inappropriate, it will certainly have a negative effect on the nature of recommendation output generated by the system as well as making the system to have high dimensionality complexity. Currently, several CARS recommendation algorithms have been developed but they have failed to bring to bear the means and importance of experimentally validating the contextual information used in different domains of application of CARS. Hence, this paper experimentally validates the contextual variables in the domain of research resources by splitting a research resource (article) into three major sections (introduction, review and methodology). These sections are the contextual variables validated in order to authenticate their viability as context that could be used in recommending research resources based on the specific section of an article a researcher is interested in. The result of our experiment shows that irrespective of the domain of articles, journal articles have higher variability in their citations at introduction, very significant variability between the articles in the review and high variability in the methodology contextual variable respectively than the articles in the proceeding under the three contextual variables. This experiment shows that these three variables could be used as context .It also shows the percentage of splitting that could be used within journals and proceedings for context-aware research resources recommendations.