In the petroleum industry, a huge amount of data is generated every day by many different sources. The petroleum industry relies on the efficient use of this data to build computational models that represent subsurface petroleum reservoirs. A problem that arises when one is building reservoir models is the fact that the existent data is hardly interoperable. Many companies produce data, but the data is not easily integrated. In this work, we present a novel ontology-based conceptual framework to enhance well logs data quality, in order to improve the construction of reservoir models. Better reservoir models results in a better prediction of the petroleum reservoir quality, supporting a better evaluation of the reservoir economic value. We propose the use of the geological concept of reservoir petrofacies to make explicit the attributes that most contribute to the reservoir porosity and permeability. Thus, with the help of an expert, it is possible identify these attributes in well logs, enabling the possibility to extrapolate these known attributes to other logs in which the expert did not evaluate. The result is the possibility to trace this data along the different modeling activities.