Summary of Alcoholism Experiments

Input Web Page, B: Wikipedia Alcoholism site [40]

Extracted Keywords, SB : alcoholism, alcohol, drunk, alcoholic, alcoholics, naltrexone, drink, addiction,

dependence, detoxification, diagnosed, screening, drinks, moderation, abstinence, 2006, disorder, drinking,

behavior, questionnaire, cage, treatment, citation , acamprosate, because, pharmacological, anonymous,

extinction, sobriety, dsm

Input Web Page, B , ("control" page): Medical Terms Site [29]

Extracted "Control" Keywords, SB : ABO blood group, Alarm clock headache, Ankle-foot orthosis,

Ascending aorta, Benign lymphoreticulosis, Breast bone, Carotid-artery stenosis, Chondromalacia patellae,

Congenital, Cystic periventricular leukomalacia, Discharge, DX, Enterococcus, Familial Parkinson disease

type 5, Fondation Jean Dausset-CEPH, Giant cell pneumonia, Heart attack, Hormone, parathyroid, Impetigo,

Itching, Laughing gas, M. intercellulare, Membranous nephropathy, MRSA, Nerve palsy, laryngeal, Oligoden-

drocyte, Pap Smear, Phagocytosis, Postoperative, Purpura, Henoch-Schonlein

∗

Sensitive Keywords, KAlc: alcoholism, alcoholic(s), alcohol

∗

Percentage of words in SB yielding a top hit containing word(s) in KAlc: 23.33%

∗

Percentage of word pairs in SB yielding a top hit containing word(s) in KAlc: 47.82%

Percentage of "control" words in SB yielding a top hit containing word(s) in KAlc: 0.00%

∗

∗

Percentage of "control" word pairs in SB yielding a top hit containing word(s) in KAlc: 9.43%

∗

Example word sets from SB returning a top hit containing a word in KAlc:11

Keywords

URL of Top Hit

naltrexone

http://www.nlm.nih.gov/medlineplus/druginfo /medmaster/a685041.html

acamprosate

http://www.nlm.nih.gov/medlineplus/druginfo /medmaster/a604028.html

http://www.aafp.org/afp/20050201/495.html12

dsm, detoxification

dsm, detoxification, dependence

http://www.aafp.org/afp/20050201/495.html

Figure 4:

Summary of experiments to identify keywords enabling alcoholism inferences.

Redacted Word(s)

Example Link

Sensitivity of Word(s)

Having 50 or more siblings is very

http://multimedia.belointeractive.com/attack

50, 52, 54

characteristic of Osama Bin Laden.

/binladen/1004blfamily.html

http://www.time.com/time/magazine/article

Many of Osama's relatives reside in Boston.13

Boston

/0,9171,1000943,00.html?promoid=googlep

magnate

Osama's father was a building magnate.

http://www.outpostoffreedom.com/bin ladin.htm

denounced, denunciation

A number of groups (including Bin Laden's

http://www.cairnet.org/html/911statements.html

family) have denounced his actions.

A number of groups (including Bin Laden's

http://www.usnews.com/usnews/politics/whispers

condemnation

family) have condemned his actions.

/archive/september2001.htm

Figure 5:

Words redacted as a result of Web-based inference detection. Column 1 is the word or words, column 2 is a link using those words

output by the algorithm, and column 3 explains why the word(s) are sensitive.