ABSTRACT
One of the most important activities of biological text mining is the retrieval of biomedical entities. Currently, many deep learning methods have been used to identify biomedical entities with good outcomes. However, the most significant challenge with deep learning is the ability to provide enough data for the proper utilization of the deep learning model. Several deep learning approaches for biomedical entity recognition (BioNER) have been employed that produce positive results. Biomedical identity data is a fragmented asset, and every dataset is often an exciting component of a specific subset of individuals. Furthermore, many biomedical entities hold similar names, which creates ambiguity and is one of the most significant hurdles in BioNER.
To target the insufficient data and disambiguation problem, a proposed model needs to be capable enough to handle multiple named entity recognition models combined. A proposed model connected different models that trained different biomedical datasets. However, the individual model first trains the given biomedical dataset, and then it sends to targeted models where the model collects the information from the collaborator models for reducing the false positives. In our experimental results, it has been clearly shown that the proposed model successfully reduces the false positive rate and improper classification problem of biomedical entities containing ambiguous words. In terms of precision and recall, the proposed model outperforms as a contrast with previously proposed other models.
The proposed model shows numerous advantages to integrating the BioNER model. The proposed model successfully overcomes the problem of improper classification of the biomedical entity. Also, it improves the performance by leveraging multiple annotated datasets for various types of entities. The state-of-the-art performance of the suggested model increases the accuracy of text mining applications related to biomedical downstream, or to find out the relation between the biomedical-entity relationships.
