| # Generate POSMatcher rules: |
| # |
| # format: |
| # FunctionName1 RegexPattern1 |
| # FunctionName2 RegexPattern2 |
| # FunctionName3 RegexPattern3 |
| # .. |
| # |
| # Generates the following methods inside POSMatcher |
| # uint16 POSMatcher::Get<FunctionalName>Id() { ... } |
| # bool POSMatcher::Is<FunctionalName>(uint16 id) { ... } |
| # |
| # Example: |
| # uint16 POSMatcher::GetFirstNameId() { ... } |
| # bool POSMatcher::IsFirstName(uint16 id) { ... } |
| |
| Functional ^(助詞|助動詞|動詞,非自立|名詞,非自立|形容詞,非自立|動詞,接尾|名詞,接尾|形容詞,接尾) |
| |
| Unknown 名詞,サ変接続 |
| |
| FirstName 名詞,固有名詞,人名,名 |
| LastName 名詞,固有名詞,人名,姓 |
| |
| Number 名詞,数,アラビア数字 |
| KanjiNumber 名詞,数,漢数字 |
| |
| # used in CandidateFilter |
| WeakCompoundPrefix ^(接頭詞,名詞接続|接頭詞,丁寧連用形接続|フィラー) |
| |
| # use in Segmenter |
| # if the following particles exist at the beggining of a segment, |
| # concat them with the next content word. |
| # ex. |
| # にかく => に書く, not に|書く |
| AcceptableParticleAtBeginOfSegment ^助詞,*,*,*,*,*,(が|で|と|に|にて|の|へ|より|も|と|から|は|や)$ |
| |
| # used in UserHistoryRewriter |
| JapanesePunctuations 記号,(句点|読点) |
| |
| # used in SymbolRewriter |
| OpenBracket 記号,括弧開 |
| CloseBracket 記号,括弧閉 |
| |
| # used in SingleKanjiRewriter |
| GeneralSymbol 記号,一般, |
| |
| # zipcode |
| Zipcode 特殊,郵便番号 |
| |
| # 短縮よみ |
| IsolatedWord 特殊,短縮よみ |
| |
| # サジェストのみ |
| SuggestOnlyWord 特殊,サジェストのみ |
| |
| # Used in UsageRewriter |
| ContentWordWithConjugation ^(動詞,自立,*,*,五段|動詞,自立,*,*,一段|形容詞,自立) |
| |
| # 末尾語 |
| SuffixWord ^(助詞|助動詞|動詞,非自立|動詞,接尾|形容詞,非自立|形容詞,接尾|動詞,自立,*,*,サ変・スル) |
| |
| # 助数詞 |
| CounterSuffixWord 名詞,接尾,助数詞 |
| |
| # 固有名詞 |
| UniqueNoun ^名詞,固有名詞 |
| |
| # 一般名詞 |
| GeneralNoun ^名詞,一般,*,*,*,*,*$ |
| |
| # 代名詞 |
| Pronoun ^名詞,代名詞, |
| |
| # Content Noun |
| ContentNoun ^名詞,(一般|固有名詞|副詞可能|サ変接続), |
| |
| # Noun prefix |
| NounPrefix ^接頭詞,名詞接続, |
| |
| # EOS symbols |
| EOSSymbol ^(記号,(句点|読点|アルファベット|一般|括弧開|括弧閉))|^(名詞,数,(アラビア数字|区切り文字)) |
| |
| # 副詞 |
| Adverb ^副詞, |
| |
| # 副詞節 suffix |
| AdverbSegmentSuffix ^助詞,*,*,*,*,*,(から|で|と|に|にて|の|へ|を)$ |
| |
| # 並立助詞 |
| ParallelMarker ^助詞,並立助詞 |
| |
| # ます助動詞 |
| MasuSuffix 助動詞,*,*,*,特殊・(マス|タイ) |
| |
| # て助動詞/助詞 |
| TeSuffix (助詞,接続助詞,*,*,*,*,て|助動詞,*,*,*,特殊・タ,|動詞,非自立,*,*,一段,*,てる|動詞,非自立,*,*,五段・ワ行促音便,*,ちゃう) |
| |
| # カ行音便 タ接続 "書い"-て "書い"-た |
| # This rule is used for removing candidate "書います/書いたい" |
| KagyoTaConnectionVerb ^動詞,(非自立|自立),*,*,五段・カ行(促音便|イ音便),連用タ接続 |
| |
| # ワ行促音便 連用形 "買っ" -て "買っ"-た |
| # This rule is used for removing candidate "買いた" |
| # Do not match ウ音便, because "吸うた" is acceptable. |
| WagyoRenyoConnectionVerb ^動詞,(非自立|自立),*,*,五段・ワ行促音便,連用形 |
