{"id":72898,"date":"2020-03-05T20:42:08","date_gmt":"2020-03-05T17:42:08","guid":{"rendered":"https:\/\/prohoster.info\/blog\/administrirovanie\/kak-otkryt-kommentarii-i-ne-potonut-v-spame"},"modified":"2020-03-05T20:42:08","modified_gmt":"2020-03-05T17:42:08","slug":"kak-otkryt-kommentarii-i-ne-potonut-v-spame","status":"publish","type":"post","link":"https:\/\/prohoster.info\/sk\/blog\/administrirovanie\/kak-otkryt-kommentarii-i-ne-potonut-v-spame","title":{"rendered":"Ako otvori\u0165 koment\u00e1re a nepotopi\u0165 sa v spame","gt_translate_keys":[{"key":"rendered","format":"text"}]},"content":{"rendered":"

\"Ako
\n
\nKe\u010f je tvojou pr\u00e1cou vytv\u00e1ra\u0165 nie\u010do pekn\u00e9, nemus\u00ed\u0161 o nej a\u017e tak ve\u013ea rozpr\u00e1va\u0165, preto\u017ee v\u00fdsledok je ka\u017ed\u00e9mu pred o\u010dami. No ak zakr\u00fdva\u0161 n\u00e1pisy na plotoch, nikto si tvoju pr\u00e1cu nev\u0161imne, k\u00fdm s\u00fa ploty v dobrom stave alebo k\u00fdm nezma\u017ee\u0161 nie\u010do nevhodn\u00e9.<\/p>\n

Ka\u017ed\u00e1 slu\u017eba, kde sa d\u00e1 zanecha\u0165 koment\u00e1r, recenzia, posla\u0165 spr\u00e1va alebo nahra\u0165 obr\u00e1zky, sk\u00f4r \u010di nesk\u00f4r \u010del\u00ed probl\u00e9mu spamu, podvodov a nevhodn\u00e9ho obsahu. Tomu sa ned\u00e1 vyhn\u00fa\u0165, ale je potrebn\u00e9 s t\u00fdm bojova\u0165. <\/p>\n

Vol\u00e1m sa Michail, pracujem v t\u00edme Antispamu, ktor\u00fd chr\u00e1ni pou\u017e\u00edvate\u013eov slu\u017eieb Yandexu pred podobn\u00fdmi probl\u00e9mami. Na\u0161a pr\u00e1ca zriedkakedy up\u00fata pozornos\u0165 (a to je dobr\u00e9!), tak\u017ee dnes v\u00e1m o nej poviem viac. Zist\u00edte, v ktor\u00fdch pr\u00edpadoch je moder\u00e1cia zbyto\u010dn\u00e1 a pre\u010do presnos\u0165 nie je jedin\u00fdm ukazovate\u013eom jej efektivity. A e\u0161te sa porozpr\u00e1vame o nevhodn\u00fdch v\u00fdrazoch na pr\u00edklade ma\u010diek a psov a o tom, pre\u010do niekedy pom\u00e1ha \u201eprem\u00fd\u0161\u013ea\u0165 ako hrubozrnn\u00fd \u010dlovek\u201c.
\n<\/a><\/noindex>
\nV Yandexe sa objavuje \u010doraz viac slu\u017eieb, kde pou\u017e\u00edvatelia publikuj\u00fa svoj obsah. M\u00f4\u017eete polo\u017ei\u0165 ot\u00e1zku alebo nap\u00edsa\u0165 odpove\u010f v Yandex.Q, prediskutova\u0165 novinky vo va\u0161ej \u0161tvrti v Yandex.Raione, zdie\u013ea\u0165 dopravn\u00e9 inform\u00e1cie v rozhovoroch na Yandex.Map\u00e1ch. Ale ke\u010f sa publikum slu\u017eby rozrast\u00e1, st\u00e1va sa atrakt\u00edvnym pre podvodn\u00edkov a spammerov. T\u00ed prich\u00e1dzaj\u00fa a zap\u013a\u0148aj\u00fa koment\u00e1re: pon\u00fakaj\u00fa \u013eahk\u00e9 z\u00e1robky, inzeruj\u00fa z\u00e1zra\u010dn\u00e9 prostriedky a s\u013eubuj\u00fa soci\u00e1lne d\u00e1vky. V d\u00f4sledku spammerov niektor\u00ed pou\u017e\u00edvatelia prich\u00e1dzaj\u00fa o peniaze, in\u00ed zas o chu\u0165 tr\u00e1vi\u0165 \u010das na neudr\u017eiavanej, spamom zarastenej slu\u017ebe.<\/p>\n

A to nie je jedin\u00fd probl\u00e9m. Sna\u017e\u00edme sa nielen ochr\u00e1ni\u0165 pou\u017e\u00edvate\u013eov pred podvodn\u00edkmi, ale aj vytvori\u0165 pr\u00edjemn\u00fa atmosf\u00e9ru pre komunik\u00e1ciu. Ak sa \u013eudia v koment\u00e1roch stret\u00e1vaj\u00fa s nevhodn\u00fdmi v\u00fdrazmi a ur\u00e1\u017ekami, s ve\u013ekou pravdepodobnos\u0165ou od\u00eddu a u\u017e sa nevr\u00e1tia. Preto je potrebn\u00e9 aj s t\u00fdmto vedie\u0165 bojova\u0165. <\/p>\n

\u010cist\u00fd Web<\/h3>\n

\nAko to u n\u00e1s \u010dasto b\u00fdva, prv\u00e9 n\u00e1vrhy vznikli v H\u013eadan\u00ed, v tej \u010dasti, ktor\u00e1 bojuje so spamom vo v\u00fdsledkoch. E\u0161te pred desiatimi rokmi sa objavila \u00faloha filtrovania dospel\u00e9ho obsahu pre rodinn\u00e9 vyh\u013ead\u00e1vanie a pre vyh\u013ead\u00e1vania, ktor\u00e9 nepredpokladaj\u00fa odpovede z kateg\u00f3rie 18+. Tak sa objavili prv\u00e9, ru\u010dne zostaven\u00e9 slovn\u00edky pornografie a vulgarizmov, ktor\u00e9 zap\u013a\u0148ali analytici. Hlavnou \u00falohou bola klasifik\u00e1cia vyh\u013ead\u00e1van\u00ed na tie, kde je dovolen\u00e9 zobrazova\u0165 dospel\u00fd obsah a kde nie. Pre t\u00fato \u00falohu sa zhroma\u017e\u010fovala anot\u00e1cia, budovali sa heuristiky, tr\u00e9novali modely. Tak vznikli prv\u00e9 n\u00e1vrhy na filtrovanie ne\u017eelan\u00e9ho obsahu.<\/p>\n

Postupom \u010dasu sa v Yandexe za\u010dal objavova\u0165 UGC (u\u017e\u00edvate\u013eom generovan\u00fd obsah) \u2014 spr\u00e1vy, ktor\u00e9 p\u00ed\u0161u samotn\u00ed pou\u017e\u00edvatelia, a Yandex ich len publikuje. Z vy\u0161\u0161ie pop\u00edsan\u00fdch d\u00f4vodov mnoh\u00e9 spr\u00e1vy nebolo mo\u017en\u00e9 publikova\u0165 bez kontroly \u2014 vy\u017eadovala sa moder\u00e1cia. Potom sa rozhodlo vytvori\u0165 slu\u017ebu, ktor\u00e1 by poskytovala ochranu pred spamom a zlo\u010dincami pre v\u0161etky UGC produkty Yandexu a vyu\u017eila n\u00e1vrhy na filtrovanie ne\u017eelan\u00e9ho obsahu v H\u013eadan\u00ed. Slu\u017ebu nazvali \u201e\u010cist\u00fd Web\u201c. <\/p>\n

Nov\u00e9 \u00falohy a pomoc pracovn\u00edkov<\/h3>\n

\nNa za\u010diatku u n\u00e1s fungovala len jednoduch\u00e1 automatiz\u00e1cia: slu\u017eby n\u00e1m posielali texty, a my sme na nich prev\u00e1dzali slovn\u00edky vulgarizmov, slovn\u00edky pornografie a regul\u00e1rne v\u00fdrazy \u2014 analytici v\u0161etko zostavovali manu\u00e1lne. Ale slu\u017eba sa postupom \u010dasu aplikovala vo st\u00e1le v\u00e4\u010d\u0161om po\u010dte produktov Yandexu, a museli sme sa nau\u010di\u0165 pracova\u0165 s nov\u00fdmi probl\u00e9mami. <\/p>\n

\u010casto namiesto recenzie pou\u017e\u00edvatelia publikuj\u00fa bezv\u00fdznamn\u00fa zbierku p\u00edsmen, sna\u017eiac sa z\u00edska\u0165 si svoje \u00faspechy, ob\u010das propaguj\u00fa svoju spolo\u010dnos\u0165 v recenzi\u00e1ch o spolo\u010dnosti konkurenta, a niekedy jednoducho pom\u00fdlia organiz\u00e1cie a v recenzii o zverimexu nap\u00ed\u0161u: \u201eV\u00fdborne pripraven\u00e1 ryba!\u201c. Mo\u017eno, \u017ee raz umel\u00e1 inteligencia nau\u010d\u00ed ide\u00e1lne zachyti\u0165 v\u00fdznam ak\u00e9hoko\u013evek textu, ale teraz automatiz\u00e1cia niekedy zvl\u00e1dne hor\u0161ie ako \u010dlovek. <\/p>\n

Bol jasno, \u017ee bez ru\u010dn\u00e9ho ozna\u010dovania sa tu nezaob\u00eddeme, a pripojili sme k n\u00e1\u0161mu syst\u00e9mu druh\u00fa f\u00e1zu \u2014 odoslanie na ru\u010dn\u00e9 overenie \u010dlovekom. Sem patrili tie publikovan\u00e9 texty, pri ktor\u00fdch klasifik\u00e1tor nezistil \u017eiadne probl\u00e9my. Rozsah tejto \u00falohy si m\u00f4\u017eete \u013eahko predstavi\u0165, preto sme sa nielen spoliehali na hodnotite\u013eov, ale aj vyu\u017eili \u00abm\u00fadros\u0165 davu\u00bb, teda sme po\u017eiadali o pomoc tolokerov. T\u00ed n\u00e1m pom\u00e1haj\u00fa odha\u013eova\u0165 to, \u010do stroj prehliadol, a t\u00fdm ju takto u\u010dia. <\/p>\n

Inteligentn\u00e9 cachovanie a LSH hashovanie<\/h3>\n

\n\u010eal\u0161\u00ed probl\u00e9m, s ktor\u00fdm sme sa stretli pri pr\u00e1ci s koment\u00e1rmi, je spam, presnej\u0161ie jeho objemy a r\u00fdchlos\u0165 \u0161\u00edrenia. Ke\u010f sa publikum Yandex.Raj\u00f3na r\u00fdchlo rozr\u00e1stlo, pri\u0161li tam spammeri. Nau\u010dili sa obch\u00e1dza\u0165 regul\u00e1rne v\u00fdrazy, mierne zmenou textu. Spam, samozrejme, st\u00e1le nach\u00e1dzali a odstra\u0148ovali, ale v rozsahu Yandexu mohlo by\u0165 neprijate\u013en\u00e9 ozn\u00e1menie zverejnen\u00e9 aj na 5 min\u00fat a videlo ho stovky \u013eud\u00ed. <\/p>\n

\"Ako
\n
\nN\u00e1s to, samozrejme, neuspokojovalo, a vytvorili sme inteligentn\u00e9 cachovanie textov zalo\u017een\u00e9 na LSH (lokalitne citliv\u00e9 hashovanie<\/a><\/noindex>). Funguje to takto: normalizovali sme text, odstr\u00e1nili z neho odkazy a rozrezali ho na n-gramy (sekvencie z n p\u00edsmen). Potom sme po\u010d\u00edtali hashe z n-gramov a na ich z\u00e1klade sme vytvorili LSH vektor dokumentu. Zmysel je v tom, \u017ee podobn\u00e9 texty, aj ke\u010f boli trochu zmenen\u00e9, sa transformovali na podobn\u00e9 vektory. <\/p>\n

Toto rie\u0161enie n\u00e1m umo\u017enilo znovu vyu\u017ei\u0165 verdikty klasifik\u00e1torov a hodnotite\u013eov pre podobn\u00e9 texty. Pri spamovej \u00fatoku, akon\u00e1hle prv\u00e1 spr\u00e1va pre\u0161la kontrolou a dostala sa do cache s verdiktom \u00abspam\u00bb, v\u0161etky nov\u00e9 podobn\u00e9 spr\u00e1vy, aj ke\u010f zmenen\u00e9, dostali ten ist\u00fd verdikt a boli automaticky odstr\u00e1nen\u00e9. Nesk\u00f4r sme sa nau\u010dili tr\u00e9nova\u0165 a automaticky pretr\u00e9nova\u0165 klasifik\u00e1tory spamu, ale tento \u00abinteligentn\u00fd cache\u00bb n\u00e1m ostal a dodnes n\u00e1m \u010dasto pom\u00e1ha.<\/p>\n

Klasifik\u00e1tor dobr\u00fdch textov<\/h3>\n

\nNedok\u00e1zali sme si odd\u00fdchnu\u0165 od boja so spamom, ke\u010f sme si uvedomili, \u017ee 95 % obsahu u n\u00e1s sa moderuje ru\u010dne: klasifik\u00e1tory reaguj\u00fa len na poru\u0161enia a v\u00e4\u010d\u0161ina textov s\u00fa dobr\u00e9. Zatia\u013e \u010do za\u0165a\u017eujeme pracovn\u00edkov, ktor\u00ed v 95 pr\u00edpadoch zo 100 daj\u00fa hodnotenie \"V\u0161etko OK\". Museli sme sa pusti\u0165 do nezvy\u010dajnej pr\u00e1ce \u2014 vytv\u00e1ra\u0165 klasifik\u00e1tory pre dobr\u00fd obsah, na\u0161\u0165astie sa medzit\u00fdm nazbieralo dos\u0165 ozna\u010dkovan\u00fdch \u00fadajov.<\/p>\n

Prv\u00fd klasifik\u00e1tor vyzeral takto: lematizujeme text (privedieme slov\u00e1 k z\u00e1kladnej forme), vyhadzujeme v\u0161etky pomocn\u00e9 \u010dasti re\u010di a aplikujeme vopred pripraven\u00fd \u201eslovn\u00edk dobr\u00fdch lem\u201d. Ak s\u00fa v texte v\u0161etky slov\u00e1 \u201edobr\u00e9\u201d, znamen\u00e1 to, \u017ee ani text ako celok neobsahuje poru\u0161enia. Na r\u00f4znych slu\u017eb\u00e1ch tak\u00fdto pr\u00edstup hne\u010f poskytol od 25 do 35 % automatiz\u00e1cie ru\u010dn\u00e9ho ozna\u010dovania. Samozrejme, tento pr\u00edstup nie je dokonal\u00fd: nie je \u0165a\u017ek\u00e9 skombinova\u0165 nieko\u013eko nevinn\u00fdch slov a z\u00edska\u0165 ve\u013emi ur\u00e1\u017eliv\u00e9 vyjadrenie, ale umo\u017enil n\u00e1m r\u00fdchlo dosiahnu\u0165 dobr\u00fa \u00farove\u0148 automatiz\u00e1cie a dal \u010das na vy\u0161kolenie zlo\u017eitej\u0161\u00edch modelov.<\/p>\n

Nasleduj\u00face verzie klasifik\u00e1torov dobr\u00fdch textov u\u017e zah\u0155\u0148ali aj line\u00e1rne modely, rozhodovacie stromy a ich kombin\u00e1cie. Pri ozna\u010dovan\u00ed vulgarizmov a ur\u00e1\u017eok sk\u00fa\u0161ame napr\u00edklad neur\u00f3nov\u00fa sie\u0165 BERT. Tu je d\u00f4le\u017eit\u00e9 zachyti\u0165 v\u00fdznam slova v kontexte a vz\u0165ah medzi slovami z r\u00f4znych viet a BERT s t\u00fdm celkom dobre zvl\u00e1da. (Mimochodom, ned\u00e1vno kolegovia z Novinami rozpr\u00e1vali<\/a><\/noindex>, ako vyu\u017e\u00edvaj\u00fa technol\u00f3giu na netradi\u010dn\u00fa \u00falohu \u2014 h\u013eadanie ch\u00fdb v nadpisoch.) Nakoniec sa n\u00e1m podarilo automatizova\u0165 a\u017e 90 % toku v z\u00e1vislosti od slu\u017eby. <\/p>\n

Presnos\u0165, \u00faplnos\u0165 a r\u00fdchlos\u0165<\/h3>\n

\nAby sme sa mohli rozv\u00edja\u0165, mus\u00edme rozumie\u0165, ak\u00fa hodnotu prin\u00e1\u0161aj\u00fa tie \u010di on\u00e9 automatick\u00e9 klasifik\u00e1tory, zmeny v nich, ako aj to, \u010di sa nezni\u017euje kvalita ru\u010dn\u00fdch kontrol. Na tento \u00fa\u010del pou\u017e\u00edvame ukazovatele presnosti a \u00faplnosti. <\/p>\n

Presnos\u0165 je podiel spr\u00e1vnych verdiktov medzi v\u0161etk\u00fdmi verdiktmi o zlom obsahu. \u010c\u00edm je presnos\u0165 vy\u0161\u0161ia, t\u00fdm je menej falo\u0161n\u00fdch poplachov. Ak sa nezameriame na presnos\u0165, teoreticky m\u00f4\u017eeme odstr\u00e1ni\u0165 cel\u00fd spam a vulg\u00e1rnosti, a spolu s nimi aj polovicu dobr\u00fdch pr\u00edspevkov. Na druhej strane, ak sa budeme spolieha\u0165 len na presnos\u0165, najlep\u0161ou technol\u00f3giou bude t\u00e1, ktor\u00e1 nikoho nezachyt\u00e1va. Preto existuje aj ukazovate\u013e \u00faplnosti: podiel identifikovan\u00e9ho zl\u00e9ho obsahu medzi celkov\u00fdm objemom zl\u00e9ho obsahu. Tieto dve metriky sa navz\u00e1jom vyva\u017euj\u00fa. <\/p>\n

Na meranie vzorkujeme cel\u00fd prich\u00e1dzaj\u00faci tok z ka\u017ed\u00e9ho slu\u017eby a odosielame vzorky obsahu hodnotite\u013eom na odborn\u00e9 pos\u00fadenie a porovnanie s rie\u0161eniami stroja. <\/p>\n

Ale existuje e\u0161te jeden d\u00f4le\u017eit\u00fd ukazovate\u013e. <\/p>\n

Povedal som, \u017ee neprijate\u013en\u00fa spr\u00e1vu m\u00f4\u017ee za 5 min\u00fat vidie\u0165 aj sto \u013eud\u00ed. Preto po\u010d\u00edtame, ko\u013ekokr\u00e1t sme stihli zobrazi\u0165 zl\u00fd obsah \u013eu\u010fom predt\u00fdm, ako sme ho skryli. Je to d\u00f4le\u017eit\u00e9, preto\u017ee nesta\u010d\u00ed pracova\u0165 kvalitne \u2014 treba pracova\u0165 aj r\u00fdchlo. A ke\u010f sme vybudovali ochranu pred vulgarizmami, naplno sme to poc\u00edtili. <\/p>\n

Proti vulgarizmom na pr\u00edklade ma\u010di\u010diek a ps\u00edkov<\/h3>\n

\nMal\u00e9 liter\u00e1rne odbo\u010denie. Niekto m\u00f4\u017ee poveda\u0165, \u017ee vulgarizmy a ur\u00e1\u017eky nie s\u00fa tak\u00e9 nebezpe\u010dn\u00e9 ako \u0161kodliv\u00e9 odkazy, a nie s\u00fa tak\u00e9 otravn\u00e9 ako spam. Ale usilujeme sa udr\u017eiava\u0165 komfortn\u00e9 podmienky pre komunik\u00e1ciu mili\u00f3nov pou\u017e\u00edvate\u013eov, a \u013eudia sa radi nevracaj\u00fa tam, kde s\u00fa ur\u00e1\u017ean\u00ed. Nie nadarmo je z\u00e1kaz vulgarizmov a ur\u00e1\u017eok uveden\u00fd v pravidl\u00e1ch mnoh\u00fdch komun\u00edt, vr\u00e1tane Habr. Ale odbo\u010dili sme. <\/p>\n

Slovn\u00edky vulgarizmov si neved\u00fa rady so v\u0161etk\u00fdm bohatstvom ru\u0161tiny. Napriek tomu, \u017ee existuj\u00fa len \u0161tyri hlavn\u00e9 zak\u00e1zan\u00e9 korene, z nich je mo\u017en\u00e9 vytvori\u0165 nespo\u010detn\u00e9 mno\u017estvo slov, ktor\u00e9 nijak\u00fdmi regul\u00e1rnymi v\u00fdrazmi nezachyt\u00edte. Navy\u0161e, m\u00f4\u017eete p\u00edsa\u0165 \u010das\u0165 slova translitom, nahr\u00e1dza\u0165 p\u00edsmen\u00e1 podobn\u00fdmi kombin\u00e1ciami, pres\u00fava\u0165 p\u00edsmen\u00e1, prid\u00e1va\u0165 hviezdi\u010dky at\u010f. Niekedy bez kontextu nie je mo\u017en\u00e9 ur\u010di\u0165, \u010do mal pou\u017e\u00edvate\u013e na mysli vulg\u00e1rnym slovom. Respektujeme pravidl\u00e1 Habr, a preto to neuk\u00e1\u017eeme na \u017eiv\u00fdch pr\u00edkladoch, ale na ma\u010di\u010dk\u00e1ch a ps\u00edkoch. <\/p>\n

\"Ako
\n
\n\u201eM\u0148au,\u201c povedal koc\u00farik. Ale ch\u00e1peme, \u017ee koc\u00farik povedal in\u00e9 slovo\u2026<\/p>\n

Za\u010dali sme prem\u00fd\u0161\u013ea\u0165 o algoritmoch \u201enepresn\u00e9ho zladenia\u201c na\u0161ich slovn\u00edkov a o inteligentnej\u0161om predspracovan\u00ed: prev\u00e1dzali sme transliter\u00e1ciu, sp\u00e1jali medzery a interpunkciu, h\u013eadali vzory a p\u00edsali na ne samostatn\u00e9 regul\u00e1rne v\u00fdrazy. Tento pr\u00edstup prin\u00e1\u0161al v\u00fdsledky, no \u010dasto zni\u017eoval presnos\u0165, ned\u00e1vaj\u00fac po\u017eadovan\u00fa \u00faplnos\u0165.<\/p>\n

Vtedy sme sa rozhodli \u201emyslie\u0165 ako vulgaristi\u201c. Za\u010dali sme sami prid\u00e1va\u0165 \u0161um do d\u00e1t: pres\u00favali sme p\u00edsmen\u00e1, generovali preklepy, zamie\u0148ali sme p\u00edsmen\u00e1 podobn\u00e9 svoj\u00edm p\u00edsan\u00edm a podobne. Po\u010diato\u010dn\u00e9 ozna\u010dovanie pre toto sme brali aplikovan\u00edm slovn\u00edkov vulgarizmov na ve\u013ek\u00e9 korpusy textov. Ak vezmeme jednu vetu a skomol\u00edme ju nieko\u013ek\u00fdmi sp\u00f4sobmi, vznikne u\u017e mno\u017estvo viet. T\u00fdmto sp\u00f4sobom m\u00f4\u017eeme zv\u00fd\u0161i\u0165 tr\u00e9ningov\u00fa vzorku desiatkykr\u00e1t. Ost\u00e1valo len nau\u010di\u0165 na z\u00edskanom s\u00fabore nejak\u00fd rozumn\u00fd model, ktor\u00fd zoh\u013ead\u0148uje kontext. <\/p>\n

\"Ako
\n
\nO kone\u010dnom rie\u0161en\u00ed je zatia\u013e pred\u010dasn\u00e9 hovori\u0165. St\u00e1le experimentujeme s pr\u00edstupmi k tomuto probl\u00e9mu, ale u\u017e vid\u00edme, \u017ee jednoduch\u00e1 symbolick\u00e1 konvolu\u010dn\u00e1 sie\u0165 z nieko\u013ek\u00fdch vrstiev vidite\u013ene prevy\u0161uje slovn\u00edky a regul\u00e1rne v\u00fdrazy: dok\u00e1\u017eeme zv\u00fd\u0161i\u0165 aj presnos\u0165, aj \u00faplnos\u0165.<\/p>\n

Samozrejme, ch\u00e1peme, \u017ee v\u017edy sa n\u00e1jdu sp\u00f4soby, ako ob\u00eds\u0165 aj t\u00fa najpokro\u010dilej\u0161iu automatiz\u00e1ciu, obzvl\u00e1\u0161\u0165 ke\u010f ide o tak\u00fa z\u00e1bavn\u00fa \u010dinnos\u0165: nap\u00edsa\u0165 tak, aby blb\u00fa ma\u0161inu nepochopila. Tu, ako pri boji proti spamu, nem\u00e1me za cie\u013e vym\u00fdti\u0165 samotn\u00fa mo\u017enos\u0165 nap\u00edsa\u0165 nie\u010do neslu\u0161n\u00e9, na\u0161ou \u00falohou je zabezpe\u010di\u0165, aby hra nest\u00e1la za n\u00e1mahu.<\/p>\n

Otvorenie pr\u00edle\u017eitosti na zdie\u013eanie svojho n\u00e1zoru, komunik\u00e1ciu a komentovanie \u2014 nie je \u0165a\u017ek\u00e9. Podstatne \u0165a\u017e\u0161ie je dosiahnu\u0165 bezpe\u010dn\u00e9, komfortn\u00e9 podmienky a re\u0161pektuj\u00faci pr\u00edstup k \u013eu\u010fom. A bez toho nebude v\u00fdvoj \u017eiadneho spolo\u010denstva.
\n
Zdroj: habr.com<\/a> <\/p>","protected":false,"gt_translate_keys":[{"key":"rendered","format":"html"}]},"excerpt":{"rendered":"

\u041a\u043e\u0433\u0434\u0430 \u0442\u0432\u043e\u044f \u0440\u0430\u0431\u043e\u0442\u0430 \u2014 \u0441\u043e\u0437\u0434\u0430\u0432\u0430\u0442\u044c \u0447\u0442\u043e-\u0442\u043e \u043a\u0440\u0430\u0441\u0438\u0432\u043e\u0435, \u043e \u043d\u0435\u0439 \u043c\u043e\u0436\u043d\u043e \u043e\u0441\u043e\u0431\u0435\u043d\u043d\u043e \u043d\u0435 \u0440\u0430\u0441\u0441\u043a\u0430\u0437\u044b\u0432\u0430\u0442\u044c, \u043f\u043e\u0442\u043e\u043c\u0443 \u0447\u0442\u043e \u0440\u0435\u0437\u0443\u043b\u044c\u0442\u0430\u0442 \u0443 \u0432\u0441\u0435\u0445 \u043f\u0435\u0440\u0435\u0434 \u0433\u043b\u0430\u0437\u0430\u043c\u0438. \u0410 \u0432\u043e\u0442 \u0435\u0441\u043b\u0438 \u0442\u044b \u0441\u0442\u0438\u0440\u0430\u0435\u0448\u044c \u043d\u0430\u0434\u043f\u0438\u0441\u0438 \u0441 \u0437\u0430\u0431\u043e\u0440\u043e\u0432, \u0442\u0432\u043e\u044e \u0440\u0430\u0431\u043e\u0442\u0443 \u043d\u0438\u043a\u0442\u043e \u043d\u0435 \u0437\u0430\u043c\u0435\u0447\u0430\u0435\u0442, \u043f\u043e\u043a\u0430 \u0437\u0430\u0431\u043e\u0440\u044b \u0432\u044b\u0433\u043b\u044f\u0434\u044f\u0442 \u043f\u0440\u0438\u043b\u0438\u0447\u043d\u043e \u0438\u043b\u0438 \u043f\u043e\u043a\u0430 \u0442\u044b \u043d\u0435 \u0441\u043e\u0442\u0440\u0451\u0448\u044c \u0447\u0442\u043e-\u043d\u0438\u0431\u0443\u0434\u044c \u043d\u0435 \u0442\u043e. \u041b\u044e\u0431\u043e\u0439 \u0441\u0435\u0440\u0432\u0438\u0441, \u0433\u0434\u0435 \u043c\u043e\u0436\u043d\u043e \u043e\u0441\u0442\u0430\u0432\u0438\u0442\u044c \u043a\u043e\u043c\u043c\u0435\u043d\u0442\u0430\u0440\u0438\u0439, \u043e\u0442\u0437\u044b\u0432, \u043e\u0442\u043f\u0440\u0430\u0432\u0438\u0442\u044c \u0441\u043e\u043e\u0431\u0449\u0435\u043d\u0438\u0435 \u0438\u043b\u0438 […]<\/p>\n","protected":false,"gt_translate_keys":[{"key":"rendered","format":"html"}]},"author":1,"featured_media":72899,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[688],"tags":[],"class_list":["post-72898","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-administrirovanie"],"aioseo_notices":[],"aioseo_head":"\n\t\t\n\t\n\t\n\t\n\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\n","aioseo_head_json":{"title":"\ud83e\udd47Ako otvori\u0165 koment\u00e1re a nepotopi\u0165 sa v spame | ProHoster","description":"Ke\u010f je tvoja pr\u00e1ca \u2014 vytv\u00e1ra\u0165 nie\u010do.","canonical_url":"https:\/\/prohoster.info\/sk\/blog\/administrirovanie\/kak-otkryt-kommentarii-i-ne-potonut-v-spame","robots":"max-image-preview:large","keywords":"","webmasterTools":{"miscellaneous":""},"schema":null,"og:locale":"sk_SK","og:site_name":"ProHoster | \u041a\u0443\u043f\u0438\u0442\u044c \u043d\u0430\u0434\u0435\u0436\u043d\u044b\u0439 \u0445\u043e\u0441\u0442\u0438\u043d\u0433 \u0434\u043b\u044f \u0441\u0430\u0439\u0442\u043e\u0432 \u0441 \u0437\u0430\u0449\u0438\u0442\u043e\u0439 \u043e\u0442 DDoS, VPS VDS \u0441\u0435\u0440\u0432\u0435\u0440\u044b","og:type":"article","og:title":"\ud83e\udd47\u041a\u0430\u043a \u043e\u0442\u043a\u0440\u044b\u0442\u044c \u043a\u043e\u043c\u043c\u0435\u043d\u0442\u0430\u0440\u0438\u0438 \u0438 \u043d\u0435 \u043f\u043e\u0442\u043e\u043d\u0443\u0442\u044c \u0432 \u0441\u043f\u0430\u043c\u0435 | ProHoster","og:description":"\u041a\u043e\u0433\u0434\u0430 \u0442\u0432\u043e\u044f \u0440\u0430\u0431\u043e\u0442\u0430 \u2014 \u0441\u043e\u0437\u0434\u0430\u0432\u0430\u0442\u044c \u0447\u0442\u043e-\u0442\u043e.","og:url":"https:\/\/prohoster.info\/sk\/blog\/administrirovanie\/kak-otkryt-kommentarii-i-ne-potonut-v-spame","og:image":"https:\/\/prohoster.info\/wp-content\/uploads\/2021\/11\/logo-350.jpg","og:image:secure_url":"https:\/\/prohoster.info\/wp-content\/uploads\/2021\/11\/logo-350.jpg","og:image:width":350,"og:image:height":350,"article:published_time":"2020-03-05T17:42:08+00:00","article:modified_time":"2020-03-05T17:42:08+00:00","article:publisher":"https:\/\/www.facebook.com\/prohoster","article:author":"https:\/\/www.facebook.com\/prohoster"},"aioseo_meta_data":{"post_id":"72898","title":null,"description":null,"keywords":null,"keyphrases":null,"primary_term":null,"canonical_url":null,"og_title":null,"og_description":null,"og_object_type":"default","og_image_type":"default","og_image_url":null,"og_image_width":null,"og_image_height":null,"og_image_custom_url":null,"og_image_custom_fields":null,"og_video":null,"og_custom_url":null,"og_article_section":null,"og_article_tags":null,"twitter_use_og":false,"twitter_card":"default","twitter_image_type":"default","twitter_image_url":null,"twitter_image_custom_url":null,"twitter_image_custom_fields":null,"twitter_title":null,"twitter_description":null,"schema":{"blockGraphs":[],"customGraphs":[],"default":{"data":{"Article":[],"Course":[],"Dataset":[],"FAQPage":[],"Movie":[],"Person":[],"Product":[],"ProductReview":[],"Car":[],"Recipe":[],"Service":[],"SoftwareApplication":[],"WebPage":[]},"graphName":"","isEnabled":true},"graphs":[]},"schema_type":null,"schema_type_options":null,"pillar_content":false,"robots_default":true,"robots_noindex":false,"robots_noarchive":false,"robots_nosnippet":false,"robots_nofollow":false,"robots_noimageindex":false,"robots_noodp":false,"robots_notranslate":false,"robots_max_snippet":null,"robots_max_videopreview":null,"robots_max_imagepreview":"large","priority":null,"frequency":null,"local_seo":null,"seo_analyzer_scan_date":null,"breadcrumb_settings":null,"limit_modified_date":false,"reviewed_by":null,"ai":null,"created":"2021-02-28 18:43:27","updated":"2022-09-29 22:58:18","focus_keyword":null,"additional_keywords":null,"truseo_locale":null},"gt_translate_keys":[{"key":"link","format":"url"}],"_links":{"self":[{"href":"https:\/\/prohoster.info\/sk\/wp-json\/wp\/v2\/posts\/72898","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/prohoster.info\/sk\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/prohoster.info\/sk\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/prohoster.info\/sk\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/prohoster.info\/sk\/wp-json\/wp\/v2\/comments?post=72898"}],"version-history":[{"count":0,"href":"https:\/\/prohoster.info\/sk\/wp-json\/wp\/v2\/posts\/72898\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/prohoster.info\/sk\/wp-json\/wp\/v2\/media\/72899"}],"wp:attachment":[{"href":"https:\/\/prohoster.info\/sk\/wp-json\/wp\/v2\/media?parent=72898"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/prohoster.info\/sk\/wp-json\/wp\/v2\/categories?post=72898"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/prohoster.info\/sk\/wp-json\/wp\/v2\/tags?post=72898"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}