sgml.java

一个自然语言处理的Java开源工具包。LingPipe目前已有很丰富的功能

        SGML_MAP.put("Euml",'\u00CB'); // LATIN CAPITAL LETTER E WITH DIAERESIS
        SGML_MAP.put("euml",'\u00EB'); // LATIN SMALL LETTER E WITH DIAERESIS
        SGML_MAP.put("excl",'\u0021'); // EXCLAMATION MARK
        SGML_MAP.put("exist",'\u2203'); // THERE EXISTS
        SGML_MAP.put("Fcy",'\u0424'); // CYRILLIC CAPITAL LETTER EF
        SGML_MAP.put("fcy",'\u0444'); // CYRILLIC SMALL LETTER EF
        SGML_MAP.put("female",'\u2640'); // FEMALE SIGN
        SGML_MAP.put("ffilig",'\uFB03'); // LATIN SMALL LIGATURE FFI
        SGML_MAP.put("fflig",'\uFB00'); // LATIN SMALL LIGATURE FF
        SGML_MAP.put("ffllig",'\uFB04'); // LATIN SMALL LIGATURE FFL
        SGML_MAP.put("filig",'\uFB01'); // LATIN SMALL LIGATURE FI
        // SGML_MAP.put("fjlig",'?'); // fj ligature
        SGML_MAP.put("flat",'\u266D'); // MUSIC FLAT SIGN
        SGML_MAP.put("fllig",'\uFB02'); // LATIN SMALL LIGATURE FL
        SGML_MAP.put("fnof",'\u0192'); // LATIN SMALL LETTER F WITH HOOK
        SGML_MAP.put("forall",'\u2200'); // FOR ALL
        SGML_MAP.put("fork",'\u22D4'); // PITCHFORK
        SGML_MAP.put("frac12",'\u00BD'); // VULGAR FRACTION ONE HALF
        SGML_MAP.put("frac13",'\u2153'); // VULGAR FRACTION ONE THIRD
        SGML_MAP.put("frac14",'\u00BC'); // VULGAR FRACTION ONE QUARTER
        SGML_MAP.put("frac15",'\u2155'); // VULGAR FRACTION ONE FIFTH
        SGML_MAP.put("frac16",'\u2159'); // VULGAR FRACTION ONE SIXTH
        SGML_MAP.put("frac18",'\u215B'); // VULGAR FRACTION ONE EIGHTH
        SGML_MAP.put("frac23",'\u2154'); // VULGAR FRACTION TWO THIRDS
        SGML_MAP.put("frac25",'\u2156'); // VULGAR FRACTION TWO FIFTHS
        SGML_MAP.put("frac34",'\u00BE'); // VULGAR FRACTION THREE QUARTERS
        SGML_MAP.put("frac35",'\u2157'); // VULGAR FRACTION THREE FIFTHS
        SGML_MAP.put("frac38",'\u215C'); // VULGAR FRACTION THREE EIGHTHS
        SGML_MAP.put("frac45",'\u2158'); // VULGAR FRACTION FOUR FIFTHS
        SGML_MAP.put("frac56",'\u215A'); // VULGAR FRACTION FIVE SIXTHS
        SGML_MAP.put("frac58",'\u215D'); // VULGAR FRACTION FIVE EIGHTHS
        SGML_MAP.put("frac78",'\u215E'); // VULGAR FRACTION SEVEN EIGHTHS
        SGML_MAP.put("frasl",'\u2044'); // FRACTION SLASH
        SGML_MAP.put("frown",'\u2322'); // FROWN
        SGML_MAP.put("gacute",'\u01F5'); // LATIN SMALL LETTER G WITH ACUTE
        SGML_MAP.put("Gamma",'\u0393'); // GREEK CAPITAL LETTER GAMMA
        SGML_MAP.put("gamma",'\u03B3'); // GREEK SMALL LETTER GAMMA
        SGML_MAP.put("gammad",'\u03DC'); // GREEK LETTER DIGAMMA
        // SGML_MAP.put("gap",'?'); // greater-than, approximately equal to
        SGML_MAP.put("Gbreve",'\u011E'); // LATIN CAPITAL LETTER G WITH BREVE
        SGML_MAP.put("gbreve",'\u011F'); // LATIN SMALL LETTER G WITH BREVE
        SGML_MAP.put("Gcedil",'\u0122'); // LATIN CAPITAL LETTER G WITH CEDILLA
        SGML_MAP.put("gcedil",'\u0123'); // LATIN SMALL LETTER G WITH CEDILLA
        SGML_MAP.put("Gcirc",'\u011C'); // LATIN CAPITAL LETTER G WITH CIRCUMFLEX
        SGML_MAP.put("gcirc",'\u011D'); // LATIN SMALL LETTER G WITH CIRCUMFLEX
        SGML_MAP.put("Gcy",'\u0413'); // CYRILLIC CAPITAL LETTER GHE
        SGML_MAP.put("gcy",'\u0433'); // CYRILLIC SMALL LETTER GHE
        SGML_MAP.put("Gdot",'\u0120'); // LATIN CAPITAL LETTER G WITH DOT ABOVE
        SGML_MAP.put("gdot",'\u0121'); // LATIN SMALL LETTER G WITH DOT ABOVE
        SGML_MAP.put("gE",'\u2267'); // GREATER-THAN OVER EQUAL TO
        SGML_MAP.put("ge",'\u2265'); // GREATER-THAN OR EQUAL TO
        // SGML_MAP.put("gEl",'?'); // greater-than, double equals, less-than
        SGML_MAP.put("gel",'\u22DB'); // GREATER-THAN EQUAL TO OR LESS-THAN
        SGML_MAP.put("ges",'\u2265'); // GREATER-THAN OR EQUAL TO
        SGML_MAP.put("Gg",'\u22D9'); // VERY MUCH GREATER-THAN
        SGML_MAP.put("Ggr",'\u0393'); // GREEK CAPITAL LETTER GAMMA
        SGML_MAP.put("ggr",'\u03B3'); // GREEK SMALL LETTER GAMMA
        SGML_MAP.put("gimel",'\u2137'); // GIMEL SYMBOL
        SGML_MAP.put("GJcy",'\u0403'); // CYRILLIC CAPITAL LETTER GJE
        SGML_MAP.put("gjcy",'\u0453'); // CYRILLIC SMALL LETTER GJE
        SGML_MAP.put("gl",'\u2277'); // GREATER-THAN OR LESS-THAN
        // SGML_MAP.put("gnap",'?'); // greater-than, not approximately equal to
        SGML_MAP.put("gne",'\u2269'); // GREATER-THAN BUT NOT EQUAL TO
        SGML_MAP.put("gnE",'\u2269'); // GREATER-THAN BUT NOT EQUAL TO
        SGML_MAP.put("gnsim",'\u22E7'); // GREATER-THAN BUT NOT EQUIVALENT TO
        SGML_MAP.put("grave",'\u0060'); // GRAVE ACCENT
        SGML_MAP.put("gsdot",'\u22D7'); // GREATER-THAN WITH DOT
        SGML_MAP.put("gsim",'\u2273'); // GREATER-THAN OR EQUIVALENT TO
        SGML_MAP.put("Gt",'\u226B'); // MUCH GREATER-THAN
        SGML_MAP.put("gt",'\u003E'); // GREATER-THAN SIGN
        SGML_MAP.put("gvnE",'\u2269'); // GREATER-THAN BUT NOT EQUAL TO
        SGML_MAP.put("hairsp",'\u200A'); // HAIR SPACE
        SGML_MAP.put("half",'\u00BD'); // VULGAR FRACTION ONE HALF
        SGML_MAP.put("hamilt",'\u210B'); // SCRIPT CAPITAL H
        SGML_MAP.put("HARDcy",'\u042A'); // CYRILLIC CAPITAL LETTER HARD SIGN
        SGML_MAP.put("hardcy",'\u044A'); // CYRILLIC SMALL LETTER HARD SIGN
        SGML_MAP.put("harr",'\u2194'); // LEFT RIGHT ARROW
        SGML_MAP.put("hArr",'\u21D4'); // LEFT RIGHT DOUBLE ARROW
        SGML_MAP.put("harrw",'\u21AD'); // LEFT RIGHT WAVE ARROW
        SGML_MAP.put("Hcirc",'\u0124'); // LATIN CAPITAL LETTER H WITH CIRCUMFLEX
        SGML_MAP.put("hcirc",'\u0125'); // LATIN SMALL LETTER H WITH CIRCUMFLEX
        SGML_MAP.put("hearts",'\u2665'); // BLACK HEART SUIT
        SGML_MAP.put("hellip",'\u2026'); // HORIZONTAL ELLIPSIS
        SGML_MAP.put("horbar",'\u2015'); // HORIZONTAL BAR
        SGML_MAP.put("Hstrok",'\u0126'); // LATIN CAPITAL LETTER H WITH STROKE
        SGML_MAP.put("hstrok",'\u0127'); // LATIN SMALL LETTER H WITH STROKE
        SGML_MAP.put("hybull",'\u2043'); // HYPHEN BULLET
        SGML_MAP.put("hyphen",'\u002D'); // HYPHEN-MINUS
        SGML_MAP.put("Iacgr",'\u038A'); // GREEK CAPITAL LETTER IOTA WITH TONOS
        SGML_MAP.put("iacgr",'\u03AF'); // GREEK SMALL LETTER IOTA WITH TONOS
        SGML_MAP.put("Iacute",'\u00CD'); // LATIN CAPITAL LETTER I WITH ACUTE
        SGML_MAP.put("iacute",'\u00ED'); // LATIN SMALL LETTER I WITH ACUTE
        SGML_MAP.put("Icirc",'\u00CE'); // LATIN CAPITAL LETTER I WITH CIRCUMFLEX
        SGML_MAP.put("icirc",'\u00EE'); // LATIN SMALL LETTER I WITH CIRCUMFLEX
        SGML_MAP.put("Icy",'\u0418'); // CYRILLIC CAPITAL LETTER I
        SGML_MAP.put("icy",'\u0438'); // CYRILLIC SMALL LETTER I
        SGML_MAP.put("idiagr",'\u0390'); // GREEK SMALL LETTER IOTA WITH DIALYTIKA AND TONOS
        SGML_MAP.put("Idigr",'\u03AA'); // GREEK CAPITAL LETTER IOTA WITH DIALYTIKA
        SGML_MAP.put("idigr",'\u03CA'); // GREEK SMALL LETTER IOTA WITH DIALYTIKA
        SGML_MAP.put("Idot",'\u0130'); // LATIN CAPITAL LETTER I WITH DOT ABOVE
        SGML_MAP.put("IEcy",'\u0415'); // CYRILLIC CAPITAL LETTER IE
        SGML_MAP.put("iecy",'\u0435'); // CYRILLIC SMALL LETTER IE
        SGML_MAP.put("iexcl",'\u00A1'); // INVERTED EXCLAMATION MARK
        SGML_MAP.put("iff",'\u21D4'); // LEFT RIGHT DOUBLE ARROW
        SGML_MAP.put("Igr",'\u0399'); // GREEK CAPITAL LETTER IOTA
        SGML_MAP.put("igr",'\u03B9'); // GREEK SMALL LETTER IOTA
        SGML_MAP.put("Igrave",'\u00CC'); // LATIN CAPITAL LETTER I WITH GRAVE
        SGML_MAP.put("igrave",'\u00EC'); // LATIN SMALL LETTER I WITH GRAVE
        SGML_MAP.put("IJlig",'\u0132'); // LATIN CAPITAL LIGATURE IJ
        SGML_MAP.put("ijlig",'\u0133'); // LATIN SMALL LIGATURE IJ
        SGML_MAP.put("Imacr",'\u012A'); // LATIN CAPITAL LETTER I WITH MACRON
        SGML_MAP.put("imacr",'\u012B'); // LATIN SMALL LETTER I WITH MACRON
        SGML_MAP.put("image",'\u2111'); // BLACK-LETTER CAPITAL I
        SGML_MAP.put("incare",'\u2105'); // CARE OF
        SGML_MAP.put("infin",'\u221E'); // INFINITY
        SGML_MAP.put("inodot",'\u0131'); // LATIN SMALL LETTER DOTLESS I
        SGML_MAP.put("inodot",'\u0131'); // LATIN SMALL LETTER DOTLESS I
        SGML_MAP.put("int",'\u222B'); // INTEGRAL
        SGML_MAP.put("intcal",'\u22BA'); // INTERCALATE
        SGML_MAP.put("IOcy",'\u0401'); // CYRILLIC CAPITAL LETTER IO
        SGML_MAP.put("iocy",'\u0451'); // CYRILLIC SMALL LETTER IO
        SGML_MAP.put("Iogon",'\u012E'); // LATIN CAPITAL LETTER I WITH OGONEK
        SGML_MAP.put("iogon",'\u012F'); // LATIN SMALL LETTER I WITH OGONEK
        SGML_MAP.put("Iota",'\u0399'); // GREEK CAPITAL LETTER IOTA
        SGML_MAP.put("iota",'\u03B9'); // GREEK SMALL LETTER IOTA
        SGML_MAP.put("iquest",'\u00BF'); // INVERTED QUESTION MARK
        SGML_MAP.put("isin",'\u2208'); // ELEMENT OF
        SGML_MAP.put("Itilde",'\u0128'); // LATIN CAPITAL LETTER I WITH TILDE
        SGML_MAP.put("itilde",'\u0129'); // LATIN SMALL LETTER I WITH TILDE
        SGML_MAP.put("Iukcy",'\u0406'); // CYRILLIC CAPITAL LETTER BYELORUSSIAN-UKRAINIAN I
        SGML_MAP.put("iukcy",'\u0456'); // CYRILLIC SMALL LETTER BYELORUSSIAN-UKRAINIAN I
        SGML_MAP.put("Iuml",'\u00CF'); // LATIN CAPITAL LETTER I WITH DIAERESIS
        SGML_MAP.put("iuml",'\u00EF'); // LATIN SMALL LETTER I WITH DIAERESIS
        SGML_MAP.put("Jcirc",'\u0134'); // LATIN CAPITAL LETTER J WITH CIRCUMFLEX
        SGML_MAP.put("jcirc",'\u0135'); // LATIN SMALL LETTER J WITH CIRCUMFLEX
        SGML_MAP.put("Jcy",'\u0419'); // CYRILLIC CAPITAL LETTER SHORT I
        SGML_MAP.put("jcy",'\u0439'); // CYRILLIC SMALL LETTER SHORT I
        // SGML_MAP.put("jnodot",'?'); // latin small letter dotless j
        SGML_MAP.put("Jsercy",'\u0408'); // CYRILLIC CAPITAL LETTER JE
        SGML_MAP.put("jsercy",'\u0458'); // CYRILLIC SMALL LETTER JE
        SGML_MAP.put("Jukcy",'\u0404'); // CYRILLIC CAPITAL LETTER UKRAINIAN IE
        SGML_MAP.put("jukcy",'\u0454'); // CYRILLIC SMALL LETTER UKRAINIAN IE
        SGML_MAP.put("Kappa",'\u039A'); // GREEK CAPITAL LETTER KAPPA
        SGML_MAP.put("kappa",'\u03BA'); // GREEK SMALL LETTER KAPPA
        SGML_MAP.put("kappav",'\u03F0'); // GREEK KAPPA SYMBOL
        SGML_MAP.put("Kcedil",'\u0136'); // LATIN CAPITAL LETTER K WITH CEDILLA
        SGML_MAP.put("kcedil",'\u0137'); // LATIN SMALL LETTER K WITH CEDILLA
        SGML_MAP.put("Kcy",'\u041A'); // CYRILLIC CAPITAL LETTER KA
        SGML_MAP.put("kcy",'\u043A'); // CYRILLIC SMALL LETTER KA
        SGML_MAP.put("Kgr",'\u039A'); // GREEK CAPITAL LETTER KAPPA
        SGML_MAP.put("kgr",'\u03BA'); // GREEK SMALL LETTER KAPPA
        SGML_MAP.put("kgreen",'\u0138'); // LATIN SMALL LETTER KRA
        SGML_MAP.put("KHcy",'\u0425'); // CYRILLIC CAPITAL LETTER HA
        SGML_MAP.put("khcy",'\u0445'); // CYRILLIC SMALL LETTER HA
        SGML_MAP.put("KHgr",'\u03A7'); // GREEK CAPITAL LETTER CHI
        SGML_MAP.put("khgr",'\u03C7'); // GREEK SMALL LETTER CHI
        SGML_MAP.put("KJcy",'\u040C'); // CYRILLIC CAPITAL LETTER KJE
        SGML_MAP.put("kjcy",'\u045C'); // CYRILLIC SMALL LETTER KJE
        SGML_MAP.put("lAarr",'\u21DA'); // LEFTWARDS TRIPLE ARROW
        SGML_MAP.put("Lacute",'\u0139'); // LATIN CAPITAL LETTER L WITH ACUTE
        SGML_MAP.put("lacute",'\u013A'); // LATIN SMALL LETTER L WITH ACUTE
        SGML_MAP.put("lagran",'\u2112'); // SCRIPT CAPITAL L
        SGML_MAP.put("Lambda",'\u039B'); // GREEK CAPITAL LETTER LAMDA
        SGML_MAP.put("lambda",'\u03BB'); // GREEK SMALL LETTER LAMDA
        SGML_MAP.put("lang",'\u2329'); // LEFT-POINTING ANGLE BRACKET
        // SGML_MAP.put("lap",'?'); // less-than, approximately equal to
        SGML_MAP.put("laquo",'\u00AB'); // LEFT-POINTING DOUBLE ANGLE QUOTATION MARK
        SGML_MAP.put("Larr",'\u219E'); // LEFTWARDS TWO HEADED ARROW
        SGML_MAP.put("larr",'\u2190'); // LEFTWARDS ARROW
        SGML_MAP.put("lArr",'\u21D0'); // LEFTWARDS DOUBLE ARROW
        SGML_MAP.put("larr2",'\u21C7'); // LEFTWARDS PAIRED ARROWS
        SGML_MAP.put("larrhk",'\u21A9'); // LEFTWARDS ARROW WITH HOOK
        SGML_MAP.put("larrlp",'\u21AB'); // LEFTWARDS ARROW WITH LOOP
        SGML_MAP.put("larrtl",'\u21A2'); // LEFTWARDS ARROW WITH TAIL
        SGML_MAP.put("Lcaron",'\u013D'); // LATIN CAPITAL LETTER L WITH CARON
        SGML_MAP.put("lcaron",'\u013E'); // LATIN SMALL LETTER L WITH CARON
        SGML_MAP.put("Lcedil",'\u013B'); // LATIN CAPITAL LETTER L WITH CEDILLA
        SGML_MAP.put("lcedil",'\u013C'); // LATIN SMALL LETTER L WITH CEDILLA
        SGML_MAP.put("lceil",'\u2308'); // LEFT CEILING
        SGML_MAP.put("lcub",'\u007B'); // LEFT CURLY BRACKET
        SGML_MAP.put("Lcy",'\u041B'); // CYRILLIC CAPITAL LETTER EL
        SGML_MAP.put("lcy",'\u043B'); // CYRILLIC SMALL LETTER EL
        SGML_MAP.put("ldot",'\u22D6'); // LESS-THAN WITH DOT
        SGML_MAP.put("ldquo",'\u201C'); // LEFT DOUBLE QUOTATION MARK
        SGML_MAP.put("ldquor",'\u201E'); // DOUBLE LOW-9 QUOTATION MARK
        SGML_MAP.put("lE",'\u2266'); // LESS-THAN OVER EQUAL TO
        SGML_MAP.put("le",'\u2264'); // LESS-THAN OR EQUAL TO
        // SGML_MAP.put("lEg",'?'); // less-than, double equals, greater-than
        SGML_MAP.put("leg",'\u22DA'); // LESS-THAN EQUAL TO OR GREATER-THAN
        SGML_MAP.put("les",'\u2264'); // LESS-THAN OR EQUAL TO
        SGML_MAP.put("lfloor",'\u230A'); // LEFT FLOOR
        SGML_MAP.put("lg",'\u2276'); // LESS-THAN OR GREATER-THAN
        SGML_MAP.put("Lgr",'\u039B'); // GREEK CAPITAL LETTER LAMDA
        SGML_MAP.put("lgr",'\u03BB'); // GREEK SMALL LETTER LAMDA
        SGML_MAP.put("lhard",'\u21BD'); // LEFTWARDS HARPOON WITH BARB DOWNWARDS
        SGML_MAP.put("lharu",'\u21BC'); // LEFTWARDS HARPOON WITH BARB UPWARDS
        SGML_MAP.put("lhblk",'\u2584'); // LOWER HALF BLOCK
        SGML_MAP.put("LJcy",'\u0409'); // CYRILLIC CAPITAL LETTER LJE
        SGML_MAP.put("ljcy",'\u0459'); // CYRILLIC SMALL LETTER LJE
        SGML_MAP.put("Ll",'\u22D8'); // VERY MUCH LESS-THAN
        SGML_MAP.put("Lmidot",'\u013F'); // LATIN CAPITAL LETTER L WITH MIDDLE DOT
        SGML_MAP.put("lmidot",'\u0140'); // LATIN SMALL LETTER L WITH MIDDLE DOT
        // SGML_MAP.put("lnap",'?'); // less-than, not approximately equal to
        SGML_MAP.put("lnE",'\u2268'); // LESS-THAN BUT NOT EQUAL TO