臺灣博碩士論文加值系統

　　現今的網路使用者都習慣使用搜尋引擎找尋相關資訊，而在醫療搜尋方面更是如此，使用者比起去醫院看醫生更喜歡在網路上找尋醫療答案，而這些使用者查訊到的內容若是有包含一些驚悚的疾病字眼，例如腫瘤或猝死等等，便會開始進行自我診斷的動作，且盲目地將那些疾病的症狀和自己的做比對，如稍有相似便懷疑自己也患有此種疾病，此種情形即為網路恐病症。而會有這種情形可歸納於兩種原因，第一個是使用者的搜尋習慣，使用者不應該使用一般的搜尋引擎查找醫療資訊，並輕信網路上的資訊，尤其是在醫療這方面的資訊來源繁雜且標示不清；第二個是一般搜尋引擎的排序方法並不適用於醫療這個主題上，越熱門的疾病排序越前面是不符合現實中的發生率的，且通常越熱門的疾病，通常都是嚴重地甚至致命的疾病，此種情形往往會加深使用者的恐病症。由於我們無法改變使用者的搜尋行為，因此我們針對第二個原因提出了症狀個資驅動模型，希望藉由適當的排序疾病結果來緩和恐病症的情形發生。
　　症狀個資驅動模型包含了五個特徵函數，分別是症狀年齡特徵函數、症狀性別特徵函數、症狀身體部位特徵函數、疾病症狀特徵函數以及疾病熱門度特徵函數，搭配事先人工整理完的年齡辭典、性別辭典以及身體部位辭典這三個辭典，找出與輸入詞症狀相關的這三個內容為何，在使用對數線性模型找出和這個症狀相關的疾病為何，並產生疾病列表讓使用者參考，因此就給予的查詢詞沒有相關的性別、年齡以及身體部位，我們也能推斷出來並給予相關分數做排序。最後我們的方法與真實發生率做比較的nDCG高達80%左右，代表我們的方法雖然尚未完美，但產生的疾病列表已到達一定的水準可讓使用者觀看參考以減緩恐病症的發生。

關鍵字：醫學搜尋、資訊檢索、個資症狀驅動模型、疾病列表

Nowadays, web users used to imply general search engine to find some interested information, especially in medical search field. Web users prefer find medical information on the Internet to reality doctor, however, some of these information may contain some fatal medical terms (e.g. brain tumor, sudden death), they would think they get these serious disease once their symptoms matched these disease, this is cyberchondria. Such search results is one of the major reason that why web users feels anxious after searching, therefore, we promote Symptom Basic Information Driven Model, we use this model to rerank diseases list related to the symptoms, we try to alleviating cyberchondria with disease ranking mechanisms.
Symptom Basic Information Driven Model including four feature functions, Symptom Age feature function, Symptom Gender feature function, Symptom Body part feature function, Disease Symptom Similarity Feature Function and Disease Popularity feature function, with age lexicon, gender lexicon and body part lexicon, we can find the diseases related to the symptoms by log linear model.Finally, our model reaches almost 80% of nDCG, which means thought our model is not perfect enough, we still reach a good performance.

Keywords: Medical Search, Information Retrieval, Disease Ranking

摘要 IV
誌謝 V
章節目錄 VII
表目錄 IX
圖目錄 X
第一章 序論 1
1.1研究動機與問題　1
1.2研究目的 4
1.3研究問題與想法　5
1.4 論文架構 6
第二章 相關研究 7
2.1 網路恐病症現象　7
2.2 網路搜尋醫療資訊行為及動機　8
2.3 醫療搜尋引擎研究 9
2.4 醫學問答相關研究 11
第三章 研究方法 12
3.1 觀察　12
3.1.1 症狀與疾病的網路搜尋結果　12
3.1.2 資料觀察　17
3.2 人體的生理資訊　20
3.2.1 年齡　20
3.2.2 性別　22
3.2.3 身體部位　23
3.3 研究問題與想法　24
3.3.1 研究問題　24
3.3.2 研究想法　25
3.4 系統架構　26
3.4.1 Training Part 　27
3.4.2 Testing Part　29
3.5 症狀個資驅動模型(Symptom Basic Information Driven Model, SBID Model）　30
3.5.1 症狀年齡特徵函數(SA Feature Function)　32
3.5.2 症狀性別特徵函數(SG Feature Function)　33
3.5.3 症狀身體部位特徵函數(SB Feature Function)　34
3.5.4 疾病症狀特徵函數(DSS Feature Function)　34
3.5.5 疾病熱門度特徵函數(DP Feature Function)　35
第四章 實驗與評估　36
4.1 實驗資料與評估準則　36
4.1.1 資料集　36
4.1.2 評估準則　41
4.2 症狀個資驅動模型(SBID Model)評估　43
4.2.1 估計症狀個資驅動模型特徵函數的權重係數　43
4.2.2 症狀個資驅動模型特徵函數效能評估　 44
4.2.3 症狀個資驅動模型排序結果評估與分析　46
第五章 結論　53
參考文獻 　54
附錄 　56

[1]Ryen W. W and Eric H., Cyberchondria: Studies of the Escalation of Medical Concerns in Web Search ,“ACM Trans, 2008
[2]Elizabeth H., “Cyberchondria, Swallowing News, 2002 Jun Vol.14 No.1.
[3]Taylor, H. Cyberchondriacs Update The Harris Poll,2002
http://www.harrisinteractive.com/vault/Harris-Interactive-Poll-Research-Cyberchondriacs-Update-2002-05.pdf
[4]Taylor, H. “Cyberchondriacs on the Rise? Those who go online for healthcare information continues to increase, The Harris Poll, 2010
http://www.harrisinteractive.com/Insights/HarrisVault.aspx?PID=792
[5]Susannah F., Online Health Search, Pew Internet ＆ American Life Project, October 2006.
http://www.pewinternet.org/Reports/2006/Online-Health- Search-2006.aspx
[6]Susannah F., The Social Life of Health Information, Pew Internet ＆ American Life Project, 2009.
http://www.pewinternet.org/Reports/2009/8-The-Social-Life-of-Health-Information.aspx
[7]Tania. L.,Seeking health information on the internet: lifestyle choice or bad attack of cyberchondria? Media, Culture ＆ Society, 28 (4) (2006), pp. 521–539
[8]Gang L., Chunqiang T., Hao Y. MedSearch: A Specialized Search Engine for Medical Information Retrieval CIKM’08, October 26–30, 2008,
[9]Gang L. Design and evaluation of the iMed intelligent medical search engine. ICDE 2009: 1379-1390.
[10]Gang L. Intelligent output interface for intelligent medical search engine. AAAI 2008: 1201-1206.
[11]Gang L. Lessons learned from building the iMed intelligent medical search engine. EMBC 2009: 5138-5142.
[12]Gang L. Navigation interface for recommending home medical products. JMS, to appear.
[13]Gang L, and C. Tang. On iterative intelligent medical search. SIGIR 2008: 3-10.
[14]Gang L, and C. Tang. Challenging issues in iterative intelligent medical search. ICPR 2008: 1-4.
[15]Erik T. and Gosse B. Developing Offline Strategies for Answering Medical Questions
[16]Kate M. and Freda M. Cyberchondriasis: Fact or fiction? A preliminary examination of the relationship between health anxiety and searching for health information on the Internet.
[17]Kleinberg J., Authoritative sources in a hyperlinkedenvironment, Journal of the ACM, 46(5):604–622, 1999.
[18]Brin S. and Page L., The anatomy of a large-scale hypertextual web search engine. In Proc. of 7th International World Wide Web Conference, May 1998.
[19]Elkan C., Log-linear models and conditional random fields, CIKM, 2008
[20]Carla T. and Cristina R., Context Effect on Query Formulation and Subjective
Relevance in Health Searches. IIiX 2010