Table 13 Top-20 relevant features and their coefficients computed by the logistic regression classifier for the sensitive class
From: Analysis and classification of privacy-sensitive content in social media posts
Dataset | Feture name (coefficient value) |
|---|---|
SENS2 | Law (0.1075), family (0.0968), OutcomeState (0.0725), health (0.0697), i (0.0617), informal (0.0586), Restriction (0.0537), affect (0.0486), shehe (0.0479), home (0.0463), prep (0.0450), focusfuture (0.0431), ipron (0.0421), Intimacy (0.0408), NormsRequisites (0.0356), ppron (0.0289), work (0.0265), conj (0.0257), friend (0.0228), anx (0.0212) |
SENS3 | Law (0.1928), family (0.1639), affect (0.1133), OutcomeState (0.1006), informal (0.0900), health (0.0865), home (0.0836), Restriction (0.0822), pronoun (0.0822), focusfuture (0.0812), prep (0.0665), i (0.0628), shehe (0.0543), conj (0.0502), money (0.0487), friend (0.0454), reward (0.0417), sad (0.0388), number (0.0303), differ (0.0283) |
OMC | pronoun (0.1831), family (0.0552), OutcomeState (0.0461), i (0.0398), Intimacy (0.0286), negemo (0.0283), bio (0.0263), conj (0.0236), friend (0.0216), sexual (0.0203), feel (0.0189), relativ (0.0188), informal (0.0177), male (0.0169), prep (0.0148), number (0.0145), adj (0.0142), quant (0.0142), posemo (0.0134), female (0.0107) |
WH+TW | sexual (0.1358 ± 0.0312), female (0.1033 ± 0.0103), PrivTtl (0.0978 ± 0.0401), i (0.0833 ± 0.0489), ipron (0.0806 ± 0.1230), male (0.0744 ± 0.0113), cogproc (0.0703 ± 0.0087), ppron (0.0654 ± 0.1355), feel (0.0547 ± 0.0209), social (0.0534 ± 0.0063), conj (0.0483 ± 0.0080), number (0.0446 ± 0.0046), see (0.0427 ± 0.0252), prep (0.0414 ± 0.0051), affect (0.0355 ± 0.0399), article (0.0306 ± 0.0079), body (0.0295 ± 0.0099), health (0.0256 ± 0.0135), quant (0.0242 ± 0.0059), affiliation (0.0242 ± 0.0156) |