Combined Language Processing Methods and Mash-Up System for Improving Retrieval in Diabetes Related Patents

摘要

This paper describes a prototype medical patent search system based on combined language processing methods. The fine tuning and customization of the system developed is directed towards diabetes centered drugs. The system incorporates combined methods for improving patents searching and ranking of the retrieved data. The paper presents a detailed scheme of the system along with all processing steps (NLP processing, structural and textual indexing, clustering, patent decomposition, query processing, reordering retrieval results). As a main contribution of the work, a mashed system composed of several services for refinement of searching diabetes related patents is built on top of these methods that will aid individuals; medical personal or pharmaceutical companies refine the complex task of searching diabetes related patents. The evaluation showed that using composed search from document sections resulted in a higher total similarity measure. The Quality of Experience (QoE) evaluation showed a very positive level of user satisfaction.