iKNOWx Forum Search: a system for case retrieval in online medical forums

Information Retrieval;Search Engine;Medicine

Wang, Curtis ; Zhai ; ChengXiang

关键词: Information Retrieval;    Search Engine;    Medicine;   

Others  :  https://www.ideals.illinois.edu/bitstream/handle/2142/44483/Curtis_Wang.pdf?sequence=1&isAllowed=y

美国|英语

来源: The Illinois Digital Environment for Access to Learning and Scholarship

【 摘 要 】

This thesis describes a novel retrieval model for case retrieval from online medical forums. This model uses semantic query weighting to obtain a more accurate representation of a case query. Semantic query weighting involves identifying descriptive words, such as those describing symptoms or medi- cation, and weighting those terms more heavily during the scoring process while simultaneously lowering the weight of less important words. Our ex- perimental results show that by adding semantic query weighting to Okapi BM25, we are able to achieve, on average, better search performance when compared with the standard BM25 model. For example, precision at 5 was improved by 8.5% while recall at 100 was improved by 5.31%.In addition, we describe in detail the techniques required to build a medical forum search engine using the iKNOWx Forum Search retrieval model, which would allow a user to search medical forums for thread discussions that are similar to an input query case. Such a system would be useful in many ways. It can help inform users so they can decide on a best course of action when sick, potentially saving both time and money on healthcare costs. Also, it can easily integrate threads from multiple medical forums, allowing an easy way for users to aggregate information from various sources.