Yakha i-Radar Yansuku zonke Yezitayela ze-AI nge-new-term-radar

Okwenziwa yi-new-term-radar

new-term-radar iyithuluzi lokuqapha nsuku zonke elithola amagama aqala ukuduma ku-AI. Lakhelwe ukuthola amamodeli, imikhiqizo, amathuluzi, amawebhusayithi, nama-repository amasha ngaphambi kokuba abonakale ezinkundleni ezijwayelekile zokucwaninga amagama angukhiye.

Le phrojekthi ilusizo kakhulu kubantu abakha amakhasi amathuluzi e-AI, ama-directory, izincwadi zezindaba, izihloko, noma imikhiqizo. Esikhundleni sokulinda inani lokusesha elisunguliwe, iqapha izimpawu zomphakathi zokuqala ezishiywa yigama elisha.

1. Ilabhorethri, umthuthukisi, noma i-akhawunti enethonya ikhuluma ngegama ku-X, noma imodeli ingena ohlwini oluthrendayo lwe-Hugging Face.
2. I-repository ye-GitHub esanda kwakhiwa iqala ukuqongelela ama-star ngokushesha ngendlela engajwayelekile.
3. Ama-directory namawebhusayithi aqhudelanayo ashicilela amakhasi amasha, noma i-domain esanda kubhaliswa iqala ukubalwa ezingeni.
4. Iwebhusayithi yakho ithola ama-impression okuqala ku-Search Console alelo gama.
5. Ngokuhamba kwesikhathi i-Google Trends ibonisa ukwanda okulinganisekayo.

Amathuluzi amaningi amagama angukhiye asebenza eduze nesigaba sokugcina. I-new-term-radar ifunda izigaba ezine zokuqala nsuku zonke, bese isebenzisa i-Google Trends kamuva ukuhlola ukuthi imikhondo yokuqala yaba amathrendi angempela okusesha yini.

Izici nezimpawu ezibalulekile

Ukuqapha ama-akhawunti e-X

I-radar ifunda ama-akhawunti amisiwe ohlwini lwama-akhawunti akho bese ihlonza amagama angazange asetshenziswe yiwo ezinsukwini ezingu-60 ezedlule. Lesi siginali esikhokhelwayo sidinga ukhiye we-AIsa API. Ukusizakala kwaso kuncike kakhulu ezingeni lamalabhorethri, abathuthukisi, nabahlaziyi okhetha ukubalandela.

Amamodeli athrendayo e-Hugging Face

Isiginali ye-Hugging Face ithola amamodeli asanda kuvela kwabangu-50 abaphezulu abasethrendeni. Amakhophi ancishisiwe ahlanganiswa ngaphansi kwegama lemodeli yawo eyisisekelo. Lesi siginali simahhala futhi asidingi mininingwane yokungena yangaphandle.

Ama-repository e-GitHub aqhamukayo

Umtshina we-GitHub ubheka ama-repository aneminyaka engaphansi kwezinsuku ezingu-14 anesilinganiso okungenani sama-star angu-200 ngosuku noma athole okungenani ama-star angu-400 kusukela kusithombe-simo sangaphambilini. Ungakhiqiza futhi izinguqulo zokusesha ezisekelwe kumnikazi nasegameni le-repository, njenge-niko strata.

Lesi siginali simahhala. I-token ye-GitHub ayiphoqelekile, nakuba ukuyisebenzisa kukhulisa umkhawulo wezinga le-GitHub API.

Izinguquko kuma-sitemap

Umfundi wama-sitemap uthola amagama abiwe ama-URL amasha amaningana esizeni lapho lawo magama ayengavamile kuma-URL aso amadala. Ubala ama-URL ahlukene uma kuqhathaniswa nesisekelo sesayithi uqobo, okunciphisa amathuba okuthi i-URL eyodwa ende idale ukwanda okungamanga.

Ngaphambi kokwengeza i-sitemap, funda i-robots.txt yesayithi bese ulungisa ikhefu okungenani elilingana nesikhathi salo sokukhasa.

Amasayithi aqala ukubalwa ezingeni

Isiginali yokuzikhethela yamasayithi amasha isesha ama-domain abhaliswe futhi aqala ukukhonjwa ezinsukwini ezingu-90 ezedlule, kodwa asevele ebala ezingeni lamagama angukhiye okungenani angu-20. Idinga i-AIsa futhi kuhloswe ukuthi isebenze masonto onke.

I-README iphawula ukuthi, kusukela ngo-Okthoba 12, 2026, inkinga yokuqinisekisa ye-AIsa esezingeni eliphezulu ibangela ukuthi lesi siginali sibike iphutha. Umbiko osele uyaqhubeka nokusebenza. Sebenzisa i---skip newsites kuze kulungiswe inkinga esezingeni eliphezulu.

Google Search Console

Uma ixhunywe ku-property yakho, i-radar ithola imibuzo eqale ukukhiqiza ama-impression esayithi lakho okokuqala. Lokhu kuhlanganiswa kuyazikhethela, kumahhala, futhi kusebenzisa i-token yokufunda kuphela egcinwe emshinini wakho.

Ukuhlukanisa nokubuyekeza amagama

Imodeli encane yolimi ikala ukuthi ikhandidethi ngalinye yigama lento eyodwa ethile noma igama elivamile nje. Amagama amukelwayo avela ngokugqamile, kanti amagama avamile afakwa ohlwini oluhlukile. Ngemva kwezinsuku ezingu-14, inqubo yokubuyekeza ihlola izinguqulo zokusesha ku-Google Trends bese iqopha izinga lokuphumelela nenani lezinsuku isiginali ebihamba phambili ngazo.

Indlela ithebula lamagama angukhiye elisebenza ngayo

Igama lomkhiqizo omusha lingakhula ngokushesha ngaphandle kokuchaza umsebenzi ohlala njalo abantu abafuna ukuwenza. Isibonelo, igama lomkhiqizo womtshina lingaba elesikhashana, kanti umusho ojwayelekile ofana ne-ai image detector umele isidingo esiqhubekayo. Ithebula lamagama angukhiye elingakhethwa lixhumanisa igama ngalinye elisha nemisebenzi kanye nosesho olujwayelekile olungemuva kwalo.

1. Ukuqoqa ubufakazi: I-radar ifunda okuthunyelwe ku-X okuthole ama-like amaningi nokukhuluma ngegama, kanye ne-README kuma-repository.
2. Ukukhiqiza imishwana: Imodeli iphakamisa imishwana ejwayelekile yokusesha yemisebenzi eboniswe ebufakazini futhi ifake izingcaphuno ezisekelayo.
3. Ukuhlunga imishwana: Ucingo lwesibili lwemodeli lususa imishwana umkhiqizo oqanjwe ngegama ongayiphenduli ngokusondelene ngokwanele.
4. Ukulinganisa umthamo: Imiqulu yosesho yanyanga zonke ye-Google Ads ilandwa ngeqoqo elilodwa, bese kususwa imishwana engaphansi kwe-min_volume.
5. Ukukala amathrendi: I-Google Trends isetshenziswa kanye ukuhlola ukuma kwethemu ngayinye, iphinde isetshenziswe ukuqhathanisa ubukhulu bayo negama eliyisibonelo, eliba yi-gpts ngokuzenzakalelayo.

Izinqumo ezivelayo yilezi: prioritize, watch, noma skip. Ithemu ibekwa ku-prioritize uma inyukile ezinsukwini ezingu-30 ezedlule, isaqhubeka nokukhuphuka, futhi ifinyelela umkhawulo omisiwe wosesho lwansuku zonke. Amathemu amancane, asaqala, noma afunwa ngokuqhubekayo abekwa ku-watch. Amathemu anciphayo noma amancane angaguquki abekwa ku-skip.

Igama ngalinye lihlaziywa lapho livela okokuqala futhi futhi ngemva kwesikhathi sokuhlola kabusha esimisiwe, esiba yizinsuku ezintathu ngokuzenzakalelayo. Lokhu kubalulekile ngoba okuthunyelwe kosuku lokwethulwa kuvame ukuchaza isimemezelo kunokuba kubonise indlela abantu abasebenzisa ngayo umkhiqizo ngempela.

Ukubhekwa kokuqala kwephrojekthi kwathola ukuthi amagama emikhiqizo ayenyuka ngaphambi kwamagama ajwayelekile ahlobene nawo. Phatha ithebula lamagama angukhiye ngokuyinhloko njengemephu yesidingo esivele sikhona esisetshenziswa umkhiqizo omusha, hhayi njengobufakazi bokuthi wonke umusho ojwayelekile uqala ukwanda.

Izidingo

• I-Python 3.10 noma entsha.
• I-Git yokwenza i-clone yendawo yokugcina ikhodi.
• Ukhiye we-API we-AIsa uma ufuna amasignali akhokhelwayo, ukuhlukaniswa kwamagama, ukubuyekezwa kwe-Google Trends, noma ithebula lamagama angukhiye.
• Izifakazelo ozikhethela zona ze-Google, Resend, ne-GitHub ze-Search Console, ukuthunyelwa kwe-imeyili, nemikhawulo ephezulu ye-API.

I-radar eyisisekelo isebenzisa kuphela umtapo ojwayelekile we-Python. Amaphakheji angeziwe adingeka kuphela ekuhlanganisweni ozikhethela kona nasemisebenzini yokuthuthukisa.

Faka futhi ulungiselele i-radar

1. Yenza i-clone yendawo yokugcina ikhodi

git clone https://github.com/zhuyansen/new-term-radar
cd new-term-radar

2. Dala ukucushwa kwakho

cp config.example.json config.json

Hlela i-config.json ukuze ucacise ama-akhawunti e-X, ama-sitemap, izihloko, amagama aziwayo, amagama okumiswa, nemikhawulo efanele isayithi noma incwadi yakho.

3. Faka ukhiye we-AIsa ozikhethela wona

export AISA_API_KEY=your_key_here

Izinsizakalo ezikhokhelwayo zidluliswa nge-AIsa, okuhlanganisa okuthunyelwe ku-X, idatha yezizinda namathrendi esekelwe ku-DataForSEO, kanye nemodeli esetshenziselwa ukuhlola amagama aphakanyisiwe. Ngaphandle kwalo khiye, amasignali amahhala e-Hugging Face, GitHub, sitemap, ne-Search Console asengasebenza. Nokho, umbiko ubala amagama ngaphandle kokuwahlela ngoba umhloli wamagama awutholakali.

4. Qalisa umbiko wansuku zonke

python -m radar daily --out report.md

Ukuqalisa kokuqala kusungula izifinyezo zamagama asetshenziswa ku-X, izibalo zezinkanyezi zamareposit have, nama-URL ama-sitemap. Njengoba kungekho sifinyezo sangaphambilini sokuqhathanisa naso, umbiko wokuqala uvamise ukuba nezigaba ze-GitHub kuphela, ezingadingi idatha yayizolo. Kusukela ekuqalisweni kwesibili, i-radar ingabika izinguquko kuwo wonke amasignali anikwe amandla.

Lungiselela izilungiselelo ezibalulekile

Ukucushwa kwesibonelo kubhala zonke izinkambu ezisekelwayo. Qala ngezilungiselelo ezinquma lokho i-radar ekubukayo nokuthi iwahlunga ngamandla angakanani amagama aphakanyisiwe.

• lang: Setha ulimi lombiko libe yi-en noma zh.
• x_accounts: Faka izibambo zamalabhorethri nababhlogi. Amalabhorethri avame ukumemezela amagama kuqala, kanti ababhlogi abakhethiwe bakhombisa indlela amagama asakazeka ngayo emkhakheni wakho.
• sitemaps: Chaza izindawo zama-sitemap, amaphethini ama-shard ozikhethela wona, izikhathi zokulinda izicelo, nokususa iziqalo ze-locale.
• topics: Nikeza igama nesisho esivamile ukuze udale izigaba zombiko eziqondene nezihloko.
• known_terms: Faka amagama osuwahlanganisa noma ongafuni ukuwasebenzisa ekubuyekezweni. Imikhondo ehambisanayo iyamakwa futhi ikhishwe ekubuyekezweni okulandelayo.
• stop_words: Faka amagama avamile avela emkhakheni wakho. Uhlu olwakhelwe ngaphakathi lulungiselelwe amamodeli e-AI namathuluzi ama-agent.
• judge: Khetha imodeli yengxoxo ye-AIsa bese usetha amaphuzu amancane egama, azenzakalela ku-0.5.
• breakout: Lawula ubudala bendawo yokugcina ikhodi, inani elincane lezinkanyezi, isivinini sezinkanyezi sansuku zonke, noshintsho kusukela izolo.
• review: Setha ukuthi kulindwe izinsuku ezingaki ngaphambi kokubuyekezwa nokuthi mingaki imikhondo ezocutshungulwa ngosuku.
• usecases: Lungiselela ithebula lamagama angukhiye, imodeli, i-benchmark, umkhawulo wevolumu, umkhawulo wokusesha kwansuku zonke, nesikhawu sokuhlola kabusha.
• gsc: Setha impahla yakho ye-Search Console nefayela lethokheni, noma shiya le nto ingenalutho ukuze ukhubaze ukuhlanganiswa.

Ukuvala ngokuphelele ithebula lamagama angukhiye, setha inani layo lansuku zonke libe nguziro:

{
  "usecases": {
    "per_day": 0
  }
}

Ungaphinda ulikhubaze ngomyalo owodwa usebenzisa --skip usecases.

Izindlela eziyisisekelo zokusebenzisa

Qalisa konke kanye ngosuku

Uma ufuna ukuqalisa okuhleliwe okukodwa kuphela, sebenzisa umyalo wansuku zonke ngaphandle kokweqa noma yimaphi amasignali:

python -m radar daily --out report.md

Umyalo ubhala umbiko we-Markdown futhi udale ne-report.json eceleni kwawo. Ifayela le-JSON liqukethe idatha ehlelekile esetshenziselwa ukwakha umbiko futhi liwusizo ekushicileleni okwenziwe ngokwezifiso noma ekucutshungulweni okulandelayo.

Hlola imithombo esheshayo izikhathi eziningana ngosuku

Indawo yokugcina ikhodi ihlukanisa ukuqapha imithombo esheshayo nenqubo yansuku zonke esindayo:

python -m radar sources --out sources.md
python -m radar daily --skip x,hf --out report.md

Umyalo we-sources uhlola kuphela i-X ne-Hugging Face futhi ubhala umbiko kuphela uma uthola igama elisha. Wuqalise njalo emahoreni ambalwa. Umyalo wansuku zonke ungabe usweqa lokho kuhlola okubukhoma okubili, uhlanganise amagama agcinwe yi-sources emahoreni angu-24 adlule.

Yeqa amamojula akhethiwe

Inketho ethi --skip yamukela x, hf, breakout, sitemaps, newsites, gsc, review, judge, kanye ne-usecases.

python -m radar daily --skip x,newsites,usecases --out report.md

Qalisa ukuhlolwa kwamasayithi amasha kwamasonto onke ngokushesha

python -m radar daily --new-sites --out report.md

Hambisa isimo esigcinwa njalo

I-radar incike ezithombeni ezigciniwe nase-database yayo ukuze inqume ukuthi yini eshintshile. Sebenzisa --state uma ufuna lawa mafayela abe ku-directory ezinikele:

python -m radar daily --state ./radar-state --out report.md

Qonda umbiko okhiqiziwe

Umbiko ubeka amagama okungenzeka abe wusizo kuqala futhi ubonise amasignali awasekelayo. Umkhondo we-GitHub ungafaka indawo yokugcina ikhodi, izinkanyezi zamanje, isilinganiso sezinkanyezi ngosuku, amaphuzu egama, isimo segama elaziwayo, izinhlobo zokusesha, kanye nesibonelo esifushane. Amagama ahlolwe njengamagama avamile avela ngokuhlukile eduze nasekupheleni.

Isigaba sokubuyekeza sibuyekeza amagama abikiwe ngedethi yangaphambili emisiwe. Sibika ukuthi inguqulo yokusesha inyukile yini nokuthi i-radar iyithole izinsuku ezingaki ngaphambi kokunyuka okulinganisekayo kwe-Google Trends.

Inguqulo ibhekwa njengeyinyukile lapho inani layo eliphakeme ngemva kombiko lidlula isisekelo sayo sangaphambilini ngokuphindwe kabili nangamaphuzu ayishumi. Le ndlela esekelwe ezingxenyeni isiza ngamagama asevele enencazelo engahlobene. Igama elibanzi lingase libe nesidingo esikhulu somlando kakhulu ukuze lifaneleke, kanti umbuzo onembile womdali nomkhiqizo ungaveza umkhuba omusha.

Engeza i-Google Search Console

Faka uncike ekugunyazweni ongawukhetha bese usebenzisa iskripthi sokugunyaza esifakiwe:

pip install google-auth-oauthlib
python scripts/gsc_auth.py gsc_client.json

Iskripthi sivula isiphequluli kanye bese sibhala gsc_token.json. Ifayela lokufaka kumele libe yiklayenti le-OAuth lohlobo lwe-Desktop app, elidalwe kuphrojekthi ye-Google Cloud lapho i-Search Console API inikwe amandla khona.

Bese ulungiselela i-property:

{
  "gsc": {
    "site": "sc-domain:your-site.com"
  }
}

I-token ifundwa kuphela, ihlala emshinini wakho, futhi isivele ivikelwe yi-.gitignore yendawo yokugcina ikhodi.

Thumela imibiko nge-imeyili

Inketho ye---mail ithumela umbiko nge-Resend. Setha okuguquguqukayo kwemvelo okudingekayo:

export RESEND_API_KEY=your_resend_key
export EMAIL_FROM=radar@your-verified-domain.com
export RADAR_EMAIL_TO=you@example.com,editor@example.com

Faka iphakheji ye-Markdown uma ufuna inguqulo ye-imeyili ye-HTML:

pip install markdown
python -m radar daily --out report.md --mail

Yenze ngokuzenzakalela nge-GitHub Actions

I-workflow efakiwe ethi .github/workflows/radar.yml isebenzisa i-sources njalo emahoreni ayisishiyagalombili kanye ne-daily kanye ngosuku. Igcina isimo ku-cache ye-Actions futhi ibhala imibiko esifingqweni sokuqalisa kwe-workflow. I-workflow ihlala ivaliwe kuze kube uyayivumela ngokusobala.

1. Yenza i-fork yendawo yokugcina ikhodi. I-fork eyimfihlo yamukelekile.
2. Susa i-config.json ku-.gitignore bese wenza i-commit yokumisa. Ifayela akufanele libe nezimfihlo.
3. Engeza imfihlo yendawo yokugcina ikhodi ethi AISA_API_KEY.
4. Uma kudingeka, engeza i-RESEND_API_KEY, EMAIL_FROM, RADAR_EMAIL_TO, kanye ne-GSC_TOKEN_JSON.
5. Dala i-variable yendawo yokugcina ikhodi ethi RADAR_ENABLED bese uyibeka ku-true.

Ukugcina i-cache yesimo kubalulekile ngoba ukuqhathanisa kuncike kuma-snapshot omlando. Ngaphandle kwesimo esigciniwe, ukuqaliswa okuphindaphindiwe kuziphatha kakhulu njengokuqalisa kokuqala.

Amathiphu okulungisa okuthuthukile

Khetha ama-akhawunti e-X ngenhloso

Isitholi se-X asikwazi ukuthola igama elingashiwo yinoma yimaphi ama-akhawunti amisiwe. Faka izinhlangano ezimemezela imikhiqizo kanye nabaphawuli abathembekile okusabalala ngabo amagama abalulekile. Ukufunda ama-akhawunti ambalwa noma ukunciphisa imvamisa yokuqalisa nakho kwehlisa ukusetshenziswa kwe-API ekhokhelwayo ngokulinganayo.

Hlela amagama okumisa ngaphandle kwamathuluzi e-AI

Ukuhlanza okwakhelwe ngaphakathi nokuziphatha kwamagama okumisa kulungiselelwe amamodeli e-AI nama-agent. Uma usebenzisa i-radar kwenye inkambu, nweba i-stop_words ngesilulumagama esivamile saleyo ndawo ukuze amagama ajwayelekile angawugcwalisi umbiko.

Hlela imingcele yokugqashuka ngokwendawo yakho

Okuzenzakalelayo kuhloswe ekukhuleni kwe-GitHub okushesha ngokungajwayelekile. I-ecosystem encane ingadinga amanani aphansi e-min_stars, velocity_min, noma delta_min. I-ecosystem esebenza kakhulu ingadinga imingcele ephezulu ukuze umbiko uhlale ugxile.

Sebenzisa amagama aziwayo ukunciphisa umsebenzi ophindaphindiwe

Engeza ku-known_terms amagama osuvele unamakhasi awo, kanye namagama ongahlosile ukuwamboza. I-candidate equkethe eyodwa yalezi zinto, noma equkethwe enye yazo, imakwa bese ikhishwa ekubuyekezweni okuzayo.

Ungawathembi amaphuzu amasha kakhulu e-Trends

Idatha ye-Google Trends yezinsuku ezimbili zakamuva ayikaqini futhi ingadala ukweqa okungamanga. Ukuqaliswa kususa lezo zinsuku ezimbili futhi kudinga ukuthi ukunyuka kuqhubeke okungenani izinsuku ezimbili. Gwema ukushintsha lokhu kuziphatha ngaphandle uma unenye indlela yokuphatha idatha engaphelele.

Qinisekisa isidingo ngaphambi kokushicilela

I-radar iwuhlelo lokuxwayisa kusenesikhathi, ayithathi indawo ephelele yocwaningo lwamagama angukhiye. Amavolumu e-Google Ads angase aphuze cishe izinyanga ezimbili, kanti i-Trends ayithembekile kumagama anokusesha okumbalwa nje kwamakhulu ngenyanga. Lapho kukhona izilinganiso kuthebula lamagama angukhiye elingakhethwa, zibheke njengezikhombisi-ndlela. Hlola imiphumela yamanje yokusesha, ukuhambisana, izingozi zobunikazi, kanye nesidingo ngaphambi kokwakha ikhasi.

Linganisa ikhwalithi yesignali yakho

Iphrojekthi ayisho ukuthi inamazinga empumelelo aqinisekisiwe kuwo wonke umthombo. Uhlelo lwayo lokubuyekeza ludinga okungenani izinsuku eziyi-14 ukuze umkhondo ngamunye uvuthwe. Sebenzisa isigaba sokubuyekeza esiqoqiwe ukuhlola ukuthi yiziphi izimpawu ezisebenzela uhlu lwama-akhawunti, isihloko, kanye nemakethe yakho, kunokucabanga ukuthi wonke ama-candidate azoba umkhuba.

Qalisa iqoqo lokuhlola

Ekuthuthukiseni, faka okuncike ekuhloleni naku-Markdown, bese uqhuba ukuhlola:

pip install pytest markdown
python -m pytest -q tests

Ukuhlola kuhlanganisa ukukhishwa kwamagama, ukukala, ukugcinwa, kanye nokwenziwa kwemibiko ngaphandle kokwenza izicelo zenethiwekhi.

Isiphetho

i-new-term-radar iguqula umsebenzi wokuqala womphakathi ube ukuhamba komsebenzi okuphindaphindekayo kokuthola nsuku zonke. Izimpawu zayo zamahhala zingaqapha i-Hugging Face, i-GitHub, ama-sitemap, kanye nedatha yakho ye-Search Console, kanti i-AIsa inika amandla ukuqapha i-X, ukuhlukanisa amagama, ukubuyekeza izitayela, ucwaningo lwezizinda, kanye nethebula lamagama angukhiye.

Qala ngohlu lwemithombo olukhethwe ngokucophelela, sebenzisa ithuluzi isikhathi eside ngokwanele ukuze wakhe izifinyezo, bese ubuyekeza imiphumela yalo ngemva kwezinsuku ezingu-14. Umphumela obaluleke kakhulu awulona nje uhlu lwamagama asemfashinini, kodwa ubufakazi obubonisa ukuthi yimiphi imithombo ehlala ikwazi ukuhlonza amagama abalulekile afanele ngaphambi kokuba amathuluzi ajwayelekile okucinga amagama angukhiye akwenze.