BiCropRec: A Bi-Classifier Approach for Crop Recommendation Based on Inclusion of Semantic Intelligence and Topic Modelling

Better crop yield is the sole aim for the working farmers all over the world. Finding new and effective technologies can help all the farmers yield good quality crops with better profits across all categories of crops. In this paper, a BiCropRec framework for crop recommendation has been proposed which integrates latent semantic indexing for topic modelling. The BiCropRec model is knowledge centric and uses two distinct classifiers in order to increase the heterogeneity of classification. The model integrates upper domain ontologies and a knowledge graph and uses several semantic similarity computation schemes to yield best in class recommendation maintaining a high diversity index. Experimentations have indicated that the proposed BiCropRec framework has yielded much higher performance measures when compared to the baseline models. BiCropRec has furnished an overall F-measure of 97.66% with a very low false discovery rate of 0.03.