Enabling Data Recommendation in Scientific Workflow Based on Provenance

The comparing method plays an important role in scientific research. Scientists often make discoveries by studying differences. Particularly in life science research, the sequence alignment is accomplished by searching for similar structures in reference data files. As the scale of scientific data grows, scientists have to spend much time selecting appropriate data files in experiments, in which trust plays a critical role. This paper presents a method to make recommendations for scientists based on trust. We first propose an extended provenance model that captures users´ behavioral information during scientific workflow execution. Such provenance information can be used to compute the user´s trust in data and mutual trust degree between users. Then based on predicted trust value, data files can be recommended to users. We also design and implement a prototype system to enhance the scientific workflow system´s usability by providing scientific data recommendations. Our experiments show that, the recommended data files do a good job in helping scientists to execute workflow successfully.