Mofuta oa ho tlosa

E tlosa mantsoe le mantsoe a bohlokoa ka ho fetisisa ho tsoa ho mongolo ofe kapa ofe o telele (ngoloa, poso ea blog, transcript, ho hlalosa sehlahisoa). Litsela tse tharo tsa ho tlosa: (1) Lihlooho tsa SEO le ho batla ho batla, (2) lihlooho tsa maemo a holimo + lihlooho, (3) litho tse bitsoang - batho / libaka / lik'hamphani, (4) li-tags tse nang le li-tag tse nang le li-tag tse nang le li-tag, (5) litlhaloso tsa index tsa thuto. Liphetho li bonts'itsoe 0-100 ke TF-IDF, bongata bo sa sebetseng, kapa bo amanang le semantic. Eketsa e le CSV, TXT, kapa JSON.

E-ea - sengoloa sa 1,500-word se hlahisoang ka li-token tsa ~ 350 ka moralo oa Qwen 3 30B, ka mokhoa o sireletsehileng ka hare ho 2,500 o sa tsejoeng kapa 10,000 o ngotsoeng ka letsatsi. Ha ho hlokahale ho ngola bakeng sa ho hlahisoa ha hau sa pele.

Li-tools tsena li nka data ea ho batla ka bongata ba mantsoe a bohlokoa ao u a fang ($ 99 + / mo). Sena se etsa sepheo sa ho etsa - se tlisa mantsoe a bohlokoa a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau a hau

Litsela tse tharo tsa ho ngola: (1) TF-IDF (ka ho sa feleng) e fana ka mantsoe a mangata ka ho fetisisa ho ena lengolo-tsoibila le ho sa tloaelehang ka Senyesemane ka kakaretso - e ntle ka ho fetisisa bakeng sa SEO kaha ke mantsoe a fapaneng. (2) Bongata bo bobe bo lekanya liketsahalo tse nepahetseng - tse ntle ha u batla ho bona seo u se buang ka ho fetisisa. (3) Liphetho tsa bohlokoa ba Semantic ka ho hlaka ho litokomane tsa tesis e ka sehloohong leha ho le joalo mantsoe a hlaha feela ka nako e le 'ngoe - a loketseng ho kopanya lihlooho.

N-gram ke sephethephethe se hokahaneng sa mantsoe a N. "ho fetoha ha maemo" ke 2-gram. "mohlala oa tsebo ea li-artificial" ke 3-gram. Toggle li-chips ho tlosa feela 1-grams (mantsoe a le mong joalo ka "photosynthesis"), 2-grams (li-phrases tsa kaholimo joalo ka "carbon footprint"), kapa 3-grams (long-tail joalo ka "carbon footprint calculator"). Li-workflows tse ngata tsa SEO li batla 2+3-grams - mantsoe a le mong ke a tloaelehileng haholo.

Ehlile - lethathamo la ho tlohela ka ho fetesisa le tlosa "the", "le", "ea", "ke", jj. ka lipuo tse 14. U ka eketsa mantsoe a stopword ka "Lintlha tsa ho tlohela" - a sebetsang bakeng sa ho tlosa lebitso la hau la brand, li-codes tsa sehlahisoa, kapa boilerplate legalese e tla ba le litlamorao tse ling.

Google e hlophisa lipotso ka lebaka la mosebelisi: tlhahisoleseling ("ke eng e leng X"), ho tsamaea ("tsekameli ea ofisi ea nike"), khoebo ("liaparo tse ntle ka ho fetisisa tsa ho tsamaea 2026"), lichelete ("reka mamello ea mamello kantle"). Mochini o tšoaea mantsoe a bohlokoa a hau a hlahisoang ka qeto e ka bang teng hore u tsebe hore na sengoloa sa hau se lokela ho ba le sebaka sa litsebi, barekisi kapa ho sheba li-traffic tse khethehileng.

Li-languages tse 99 ka ho sebelisa Qwen 2.5. Li-languages tse phahameng ka ho fetisisa ke tse kang Senyesemane, Sepanishe, Sefora, Seportigali, Sejeremane, Sechina, Sejapane. Li-languages tse tlase li sebetsa empa ho tlosoa ha ligrama tse 3 ho ka ba le bothata — ho khutlela ho ligrama tse 1+2 bakeng sa Searabia, Sehindi, Seturk, jj.

UI e sebetsa ka mongolo o le mong ka nako. Bakeng sa ho tlosa palo e kholo ea / v1 / chat / API ho script ea Python - e fetole ka ho fetisisa CMS ea hau, POST'mele oa sengoloa ka seng, e tlose JSON. Bearer auth, 1,000 li-calls / hora ka ak'haonte tse mahala, liphelelo tse phahameng ho pro. Docs ho / api /.

RAKE le YAKE ke li-algorithms tsa leksiko la khale - tse potlakileng empa li le bobebe ka ho hlaka (li lahleheloa ke "ho qala theknoloji" haeba kopo e re "ho qala theknoloji" empa eseng "teknoloji" feela). KeyBERT e sebelisa li-embeddings tsa BERT - boleng bo betere ba semantic empa bo potlakile'me bo hloka ho kenya. TextRank e thehiloe ho graph. Tsela ea rona e sebelisa LLM ea 7B-parameter e etsang tsohle tse ka holimo tsa zero-shot le ho arolelana ho batla-ho batla - ho potlaka ka ho rekota empa ho se sebetse, ho mahala, le ho se hlokahale ho ML engineering.

Ka mokhoa oa frequency le TF-IDF, ehlile — mantsoe a bohlokoa ohle a ka tsoa ho tekanyo ea mohloli kapa ka ho fetola ho fokolang. Ka mokhoa oa semantic-relevance, mofuta o ka khutlisa mofuta oa ho laola-moelelo ("pulmonology" leha tekanyo ea hau e re "pulmon doctor visit"). Moea oa li-terms tsa thuto o khetha ho laola moelelo.

E-na — POST ho /v1/chat/ le lengolo-tsoibila le ts'oanang le lebokoseng lena la ho theha (bona sehlooho sa template bakeng sa lengolo-tsoibila le nepahetseng). E khutlisa JSON e hlophisitsoeng. E loketse dashboards ea merero ea litaba kapa li-plugins tsa CMS. Lengolo-tsoibila la ho amohela, li-limits tsa khoeli le khoeli. Litokomane ho /api/.