A Hybrid Approach Based Sentiment Extraction from Medical Context

In the domain of Bio medical Natural Language Processing (Bio-NLP), the information extraction and context sentiment identification are treated as emerging tasks. Several linguistic features like negation , uni-gram, bi-gram, Part-of-Speech (POS) have been used to extract the medical concepts and their sense-based context level information. Thus, in the present attempt, a hybrid approach which is the combination of both linguistic and machine learning approaches has been introduced to extract the contextual sense-based information from a medical corpus. The extraction of sentiment oriented keywords is the crucial part towards identifying the senses of medical contexts. In our previous work, we have developed a medical sense-based lexicon known as WordNet of Medical Event (WME). Several sentiment lexicons like Senti-WordNet, SenticNet etc. were used to represent WME. In contrast, one of our primary motivations here is to build a sentiment extraction model based on medical contexts to leverage the knowledge of WME using a hybrid approach. The developed model is based on two phases, namely pre-processing phase and learning phase. The prepro-cessing phase is responsible for extracting and preparing structural data from the raw contexts whereas the learning phase helps to identify the sentiment patterns and evaluate the sentiment extraction process. The two phased hybrid model provides us 81% accuracy for extracting the sentiment based medical contexts as positive and negative by employing NaïveBayes and Sequential minimal optimization (SMO) supervised classifiers.