Literature research on new genomic techniques : A NLP (natural language processing) approach

Literature review is a process, relying on text mining, that makes valuable contributions in scientific research, by offering detailed and summarised insights on the ongoing research landscape. By grasping the existing knowledge, it prevents unintentional research duplication, while also highlighting research gaps. In this report a workflow is developed, using KNIME software, to aid and automatise a part of the literature research process required when monitoring technological and scientific advances in a specific research field. The pipeline's primary function is called 'dictionary tagging', enabling the automatic labelling of documents based on predefined dictionaries of relevant terms. In addition, documents are further processed to cluster them into descriptive topics using Latent Dirichlet Allocation (LDA), that identifies patterns in word distribution of each document. This allows for the targeted filtration of a set of documents based on topics and the combination of terms tagged to each of them, facilitating the extraction of information according to specific research questions or objectives. Finally, the workflow is designed to enhance literature research by providing structured data directly to the user, thereby ensuring a more consistent research process, while saving valuable time. This methodology not only streamlines the literature review process but also enhances the accuracy and depth of insights that can be derived from large datasets.