Web Information Extraction for Finding Remedy Based on a Patient-Authored Text: A Study on Homeopathy.

Automatic medical diagnosis and remedy finding is an active research area for decades. The increasing trend of finding health remedies through the internet emerged the necessity of research on the analysis of the patient-authored text. Focussed analysis of the patient-authored text can also help in automatic remedy finding. As the web contains a huge amount of medicine and diagnosis-related information, an intelligent system can extract the relevant information to provide a health remedy given a patient-authored text query. In this paper, we attempted to develop such a system. As the patients’ description of suffering plays a key role in homeopathy remedy finding, here we focussed on the homeopathy domain. As per the best of our knowledge, this is the first attempt in this domain. For the development, first, the patient-authored text is processed to identify the disease name and characteristic symptoms. Then a query is formed and a set of relevant web pages is retrieved. The retrieved pages are then processed in multiple levels to extract the medicine names. The appropriateness of the medicines is computed using a hybrid similarity scoring technique. The medicine having the highest similarity is suggested to the user. The system is tested using a set of real questions collected from various relevant websites. The evaluation results demonstrate that the system recommends a relevant remedy in 96.33% of cases.