This paper presents a scientific literature information extraction architecture using text mining techniques to assess the human health risk of electromagnetic fields (EMFs) generated by wireless sensor devices in Internet of Things (IoT). The proposed architecture uses three text mining techniques to extract three types of information—purpose statement, research category, and source of EMF exposure—from the scientific literature to help researchers assess the human health risk of EMFs. For the purpose statement, a representative sentence expressing the authors’ intentions and purposes was extracted from the abstract text of the articles through processes of candidate sentence selection, topic lexicon creation, and weighting. For the research category, the articles were classified into three study types—epidemiological, animal experimental, and cell experimental—using a weighting process based on the predefined feature lexicon of each category. Finally, all words representing frequency bands included in the abstract text of the articles were extracted to identify the source of EMF exposure. The aforementioned text mining techniques were used to extract the information from 100 scientific articles and the performance of this architecture was proved through expert verification. The experimental results show that the proposed architecture can extract the desired information to assess the human health risk of EMFs from the scientific literature with high accuracy.