Files

Abstract

As the rapid development of IT technology and the explosion of big data in the past decades,data mining, especially text mining methods have made a great progress and have been applied to various areas. However, existing methods or models are mostly focusing on a single task, which is not sufficient for the need of large scale general applications in nowadays. In the field of clinics, there has been a lot of diagnosis models for some specific diseases such as cancers, but few for general and common diseases, especially for the lower-resource areas. Based on text mining models, I developed a general diagnosis assistant system for African countries, and built a simple app which relies on little computing resources and internet availability. Text mining can also create values in the field of chemistry. As Whorf hypothesis states, the language of chemistry strongly affects how chemists think about chemistry. I investigated the possibility to improve on the standard way of representing about chemical structure using text mining methods. In the mean time, I also explored the chemical drugdrug interaction based on the medical documents with advanced text mining techniques. On the other hand, Text mining can be further extended to other fields such as manufacturing. While the manufacturing sensor data can be transformed into the same format as texts, text mining methods are applied to do a diagnostic prediction for wafers in Seagate factories. In summary, Multiply text mining techniques are explored and applied on clinics, chemistry and manufacturing, respectively. In the following, details on the techniques and the applications will be further described.

Details

Actions

PDF

from
to
Export
Download Full History