運用文字探勘於醫療導向社群網站服務分類及關鍵詞分析

Abstract

現今社會因網路科技之發展帶動社群網站服務興起後，人們開始習慣透過瀏覽社群媒體取得各種感到興趣的資訊，甚至透過社群網站了解生活各種問題所需的解決方法，據研究指出，尋求醫療保健議題之使用人數有逐年成長之趨勢。 本研究應用文字探勘分析醫療導向社群網站服務文章內容，探討使用者之健康資訊需求並提出服務分類及關鍵字。研究中著重於利用社群網站使用者所分享之內容分析其所需之服務項目，並參考意見探勘中之詞性架構擷取所需特徵詞，以EM分群及貝式分類法了解醫療導向社群網站服務類別及關鍵詞。 實驗分群結果將醫療社群服務歸納為四大類，並利用分群結果進行貝式分類法分析，分類後精確率顯著，顯示本研究所提出之醫療導向社群服務分類進行文章內容分類有其效果及精確程度。

Nowadays, with the prosperous emersion of social networking sites through the advancement of Internet technology, people are getting used to acquire their desired information and solutions by browsing social media. According to research report, the number of users who are searching for healthcare related issues is gradually increasing every year. This paper presents a text mining approach on analyzing the articles from healthcare social networking sites to explore users’ needs of healthcare information and then propose service classification and keywords analysis. The study focuses on analyzing their ideal service items based on the shared contents from social networking users. After acquiring the keywords extracted from the word class derived from opinion mining, the study adopts EM clustering and Bayesian classification to comprehend the service classification and keywords of healthcare social networking sites. Four main categories are generalized based on the experiment results. Through conducting Bayesian classification, the accuracy rate is significant, which reveals the positive reliability and validity of this study.