医疗查询回答系统中的拼写纠正：方法、检索影响及实证评估

Healthcare question-answering (QA) systems encounter a significant challenge due to the high rate of spelling errors in user-submitted queries, which exceeds that found in professional documents. This research presents the first controlled study of spelling correction as a retrieval preprocessing step in healthcare QA, utilizing real consumer queries. An error census was conducted across two public datasets: the TREC 2017 LiveQA Medical track, comprising 104 consumer health questions, and HealthSearchQA, which includes 4,436 health queries from Google autocomplete. Findings reveal that 61.5% of real medical queries contain at least one spelling error, with a token-level error rate of 11.0%. Four correction methods were evaluated, including conservative edit distance and standard edit distance, analyzing their impact on retrieval effectiveness.