M'ndandanda wazopezekamo[Bisani][Show]
Kuyimba kwa mawu kumathetsedwa m'malo mwazolemba ndi zowonera m'gawo loyankhulana. Malinga ndi kafukufuku wa Facebook, opitilira theka la ogula amakonda kugula kuchokera ku kampani yomwe angalankhule nayo. Kulankhulana kwasanduka njira yatsopano yolankhulirana yovomerezeka ndi anthu.
Zimathandizira mabizinesi kulumikizana ndi makasitomala awo nthawi iliyonse komanso kulikonse. Ma Chatbots akuchulukirachulukira kutchuka pakati pamakampani ndi makasitomala chifukwa chosavuta kugwiritsa ntchito komanso kuchepetsa nthawi yodikirira.
Ma Chatbots, kapena mapulogalamu ochezera okha, amapatsa makasitomala njira yosinthira makonda kuti athe kupeza chithandizo kudzera pamawu. Ma chatbots atsopano a AI amatha kuzindikira funso (funso, lamulo, dongosolo, ndi zina) zopangidwa ndi munthu (kapena bot ina, kuyambika) pamalo enaake ndikuyankha moyenera (yankho, zochita, ndi zina).
Mu positi iyi, tikambirana za ma chatbots, maubwino ake, momwe mungagwiritsire ntchito, komanso momwe mungapangire zanu. kuphunzira kwakukulu chatbot ku Python, mwa zina.
Tiyeni tiyambe.
Ndiye, ma chatbots ndi chiyani?
Chatbot nthawi zambiri imatchedwa imodzi mwamachitidwe apamwamba kwambiri komanso odalirika pamakina amunthu. Othandizira pa digito awa amathandizira makasitomala pakuwongolera kulumikizana pakati pa anthu ndi ntchito.
Nthawi yomweyo, amapatsa mabizinesi zosankha zatsopano kuti akwaniritse njira yolumikizirana ndi kasitomala kuti agwire bwino ntchito, zomwe zingachepetse ndalama zothandizira wamba.
Mwachidule, ndi mapulogalamu a AI omwe amapangidwa kuti azilankhulana ndi anthu m'zinenero zawo zachilengedwe. Ma chatbots awa nthawi zambiri amalumikizana pogwiritsa ntchito njira zomvera kapena zolembera, ndipo amatha kutengera zilankhulo za anthu kuti azilumikizana ndi anthu ngati munthu.
Ma Chatbots amaphunzira kuchokera kuzinthu zomwe amakumana nazo ndi ogwiritsa ntchito, kukhala zenizeni komanso kuchita bwino pakapita nthawi. Amatha kugwira ntchito zosiyanasiyana zamabizinesi, monga kuloleza kugwiritsa ntchito ndalama, kucheza ndi ogula pa intaneti, ndikupanga zotsogola.
Kupanga macheza anu akuya ophunzirira ndi python
Pali mitundu yambiri yama chatbots m'munda wa makina kuphunzira ndi AI. Ma chatbots ena ndi othandizira, pomwe ena amangocheza nawo, pomwe ena amakhala othandizira makasitomala.
Mwinamwake mwawonapo ena mwa omwe amalembedwa ndi mabizinesi kuti ayankhe mafunso. Tipanga chatbot yaing'ono muphunziroli kuti tiyankhe mafunso omwe amafunsidwa pafupipafupi.
1. Kuyika phukusi
Gawo lathu loyamba ndikuyika mapaketi otsatirawa.
2. Deta Yophunzitsa
Tsopano ndi nthawi yoti tidziwe zamtundu wanji womwe tingafunike kuti tipereke ma chatbot athu. Sitifunika kutsitsa ma data onse akulu chifukwa iyi ndi njira yosavuta yochezera.
Tidzangogwiritsa ntchito zomwe tapanga tokha. Kuti muzitsatira bwino phunziroli, mufunika kupanga fayilo ya .JSON yokhala ndi mawonekedwe ofanana ndi omwe akuwonetsedwa pansipa. Fayilo yanga imatchedwa "intents.json."
Fayilo ya JSON imagwiritsidwa ntchito popanga mndandanda wa mauthenga omwe wogwiritsa ntchito amatha kulowetsa ndikuyika mayankho ofunikira. Mtanthauzira mawu uliwonse mufayiloyo uli ndi tagi yomwe imazindikiritsa gulu lomwe uthenga uliwonse uli.
Tigwiritsa ntchito izi pophunzitsa a neural network kugawa chiganizo cha mawu ngati amodzi mwamafayilo athu.
Titha kungoyankha kuchokera m'maguluwa ndikuwapatsa wogwiritsa ntchito. Chatbot idzakhala yabwinoko komanso yovuta ngati mupereka ma tag owonjezera, mayankho, ndi mapatani.
3. Kutsegula kwa data ya JSON
Tiyamba ndikutsitsa mu data yathu ya .json ndikulowetsa ma module ena. Sonkhanitsani fayilo ya your.json mu bukhu lofanana ndi lanu Python script. Deta yathu ya .json tsopano idzasungidwa muzosintha za data.
4. Kuchotsa Deta
Tsopano ndi nthawi yoti tichotse zambiri zomwe tikufuna pafayilo yathu ya JSON. Mapangidwe onse, komanso kalasi / tag yomwe amakhala, amafunikira.
Tidzafunikanso mndandanda wa mawu apadera amitundu yathu (pazifukwa zomwe tidzafotokoza pambuyo pake), kotero tiyeni tipange mindandanda yopanda kanthu kuti tizitsatira izi.
Tsopano tidutsa mu data yathu ya JSON ndikupeza zomwe tikufuna. M'malo mokhala nawo ngati zingwe, tigwiritsa ntchito nltk.word tokenizer kuti tisinthe pateni iliyonse kukhala mndandanda wamawu.
Kenako, mumndandanda wathu wa docs_x, tidzawonjezera pateni iliyonse, pamodzi ndi tag yolumikizidwa, pamndandanda wa docs_y.
5. Kutsika kwa Mawu
Kupeza gwero la mawu kumatchedwa stemming. Mwachitsanzo, tsinde la liwu lakuti “tsinde” lingakhale “icho,” pamene tsinde la liwu lakuti “kudzachitika” likhoza kukhala “kudzachitika.”
Tigwiritsa ntchito njira iyi kuti tichepetse mawu achitsanzo chathu ndikuyesa kudziwa zomwe ziganizo zimatanthawuza. Khodi iyi idzangopanga mndandanda wapadera wa mawu omwe adzagwiritsidwe ntchito mu gawo lotsatira la kukonzekera kwathu deta.
6. Thumba la Mawu
Yakwana nthawi yoti tilankhule za thumba la mawu tsopano popeza tatumiza kunja deta yathu ndikupanga mawu oyambira. Mitundu ya Neural ndi makina ophunzirira makina, monga tonse tikudziwira, amafunikira manambala. Chifukwa chake mndandanda wathu wa zingwe sukudula. Timafunikira njira yoyimira manambala mu ziganizo zathu, pomwe thumba la mawu limalowa.
Chiganizo chilichonse chidzaimiridwa ndi mndandanda wa kutalika kwa chiwerengero cha mawu mu mawu a chitsanzo chathu. Liwu lililonse m'mawu athu lidzayimiridwa ndi malo omwe ali pamndandanda. Ngati malo omwe ali pamndandandawo ndi 1, mawuwa akuwoneka m'mawu athu; ngati ndi 0, mawuwo samawoneka m'chiganizo chathu.
Timachitcha thumba la mawu chifukwa sitidziwa ndondomeko yomwe mawuwo amawonekera m'mawu; zomwe tikudziwa ndikuti zilipo m'mawu athu achitsanzo.
Kuphatikiza pakusintha zomwe talowa, tiyeneranso kupanga zotulutsa zathu kuti neural network imvetse. Tipanga mindandanda yotulutsa yomwe ili kutalika kwa kuchuluka kwa zilembo/ma tag mu dataset yathu, ofanana ndi thumba la mawu. Malo aliwonse pamndandandawo akuyimira chilebulo/chizindikiro chapadera, ndipo malo amodzi mwa malo aliwonsewa akuwonetsa kuti ndi chilebulo/chiti chomwe chikuyimiridwa.
Pomaliza, tidzagwiritsa ntchito NumPy arrays kusunga deta yathu yophunzitsira ndi zotuluka.
7. Chitukuko cha Chitsanzo
Ndife okonzeka kuyamba kupanga ndi kuphunzitsa chitsanzo tsopano popeza takonza kale deta yathu yonse. Tigwiritsa ntchito neural network ya feed-forward yokhala ndi zigawo ziwiri zobisika pazolinga zathu.
Cholinga cha netiweki yathu chikhala kuyang'ana mndandanda wa mawu ndikuwagawa m'kalasi (imodzi mwa ma tag athu kuchokera pa fayilo ya JSON). Tiyamba ndikukhazikitsa kapangidwe kathu kachitsanzo. Kumbukirani kuti mutha kusewera ndi manambala ena kuti mukhale ndi mtundu wabwinoko! Kuphunzira makina nthawi zambiri zimatengera kuyesa ndi zolakwika.
8. Maphunziro a Chitsanzo & Kupulumutsa
Yakwana nthawi yoti tiphunzitse chitsanzo chathu pa data yathu popeza tayiyika! Tikwaniritsa izi pophatikiza deta yathu kuchitsanzo. Chiwerengero cha ma epoch omwe timapereka ndi kuchuluka kwa nthawi yomwe chitsanzocho chidzawonetsedwe ndi deta yomweyi panthawi ya maphunziro.
Tikhoza kusunga chitsanzo ku fayilo ya fayilo tikamaliza kuiphunzitsa. tflearn ndi script yomwe ingagwiritsidwe ntchito m'malemba ena.
9. Kugwiritsa ntchito chatbot
Tsopano mutha kuyamba kucheza ndi bot yanu.
Ubwino wa Chatbot
- Monga bots akuyembekezeka kugwira ntchito masiku 365 pachaka, maola 24 patsiku, osalipira, kuonjezera kupezeka ndi liwiro lakuchita.
- Ma bots awa ndi zida zabwino kwambiri zothanirana ndi ma Vs atatu akuluakulu a data: voliyumu, kuthamanga, ndi mitundu.
- Ma Chatbots ndi mapulogalamu omwe angagwiritsidwe ntchito kuphunzira ndikumvetsetsa ogula akampani.
- Zili ndi mphamvu zapamwamba zomwe zimakhala ndi mtengo wotsika mtengo wokonza pambuyo pokhala ndi ubwino wapamwamba.
- Mapulogalamu a Chatbot amapanga deta yomwe ingasungidwe ndikugwiritsidwa ntchito pofufuza ndi kulosera.
Ntchito
- Kuthetsa mafunso a kasitomala
- Kuyankha mafunso ofunsidwa kawirikawiri
- Kupereka makasitomala kuti athandizire gulu
- Kusonkhanitsa ndemanga za makasitomala
- Kulimbikitsa zotsatsa zatsopano
- Gulani ndi malonda okambirana
- IT Helpdesk
- Kusungitsa malo ogona
- Kutumiza kwa ndalama
Kutsiliza
Ma Chatbots, monga matekinoloje ena a AI, adzagwiritsidwa ntchito kukulitsa luso la anthu ndikumasula anthu kuti akhale opanga komanso oganiza bwino powalola kuti azikhala ndi nthawi yochulukirapo pazinthu zanzeru m'malo mwanzeru.
Mabizinesi, ogwira ntchito, ndi ogula atha kupindula ndi mawonekedwe owongolera a chatbot monga malingaliro ofulumira ndi zolosera, komanso mwayi wopezeka mosavuta pamisonkhano yamakanema atanthauzo apamwamba kuchokera pakukambirana, posachedwapa, AI ikaphatikizidwa ndi chitukuko cha 5G luso.
Izi ndi zina zikufufuzidwabe, koma monga kulumikizidwa kwa intaneti, AI, NLP, ndi kupita patsogolo kwa kuphunzira pamakina, zidzachulukirachulukira.
Chwoo
Moni,
Zikomo chifukwa cha pulogalamuyi.
Ndili ndi funso.
"chikwama_cha_mawu" sichikufotokozedwa. Sindikumvetsetsa cholakwika ichi.
Kodi mungandiuzeko kuti ndingathetse bwanji vutoli??
Zikomo chifukwa cha pulogalamuyi!! Khalani ndi tsiku labwino
Jay
Chonde onjezani ntchito musanagwiritse ntchito gawo lachatbot:
/////////////////////////////////////// //////////////////////////
def bag_of_words(mawu, mawu):
bag = [0 kwa _ mu range(len(mawu))]
mawu = nltk.word_tokenize(s)
s_words = [stemmer.stem(word.lower()) kwa mawu mu mawu_mawu]
kwa se in s_words:
kwa ine, w mu kuwerengera(mawu):
ngati w == se:
chikwama[i] = 1
kubwerera numpy.array(chikwama)
// Idzathetsa vuto lanu. //
//////////////////////////////////////// //////////////////////////
Ndikugawana nanu code yonse, kuti mumvetse bwino.
/////////////////////////////////////// /////////
tumizani nltk
kuchokera ku nltk.stem.lancaster import LancasterStemmer
tsinde = LancasterStemmer ()
import numpy
import tflearn
import tensorflow
import mwachisawawa
tumizani json
tumizani zonona
ndi open("intents.json") ngati fayilo:
data = json.load(fayilo)
yesani:
ndi open(“data.pickle”, “rb”) monga f:
mawu, zilembo, maphunziro, zotuluka = pickle.load(f)
kupatula:
mawu = []
zilembo = []
docs_x = []
docs_y = []
pa cholinga cha data["zolinga"]:
kwa chitsanzo mu cholinga[“patterns”]:
wrds = nltk.word_tokenize(chitsanzo)
mawu.extend(mawu)
docs_x.append(mawu)
docs_y.append(cholinga[“tag”])
ngati cholinga["tag"] sichili m'malebulo:
labels.append(cholinga[“tag”])
mawu = [stemmer.stem(w.lower()) kwa w m'mawu ngati w != “?”]
mawu = osanjidwa(mndandanda(kuyika(mawu)))
zilembo = zosankhidwa (zolemba)
maphunziro = []
zotuluka = []
out_empty = [0 kwa _ mumtundu (len(zolemba))]
kwa x, doc mu enumerate(docs_x):
chikwama = []
wrds = [stemmer.stem(w.lower()) kwa w in doc]
kwa w m'mawu:
ngati w m'mawu:
bag.append(1)
china:
bag.append(0)
output_row = out_empty[:]
output_row[labels.index(docs_y[x])] = 1
training.append(chikwama)
output.append(output_row)
maphunziro = numpy.array(maphunziro)
zotuluka = numpy.array(zotuluka)
ndi open(“data.pickle”, “wb”) ngati f:
pickle.dump((mawu, zilembo, maphunziro, zotuluka), f)
tensorflow.reset_default_graph()
net = tflearn.input_data(shape=[Palibe, len(maphunziro[0])])
net = tflearn.fully_connected(net, 8)
net = tflearn.fully_connected(net, 8)
net = tflearn.fully_connected(ukonde, len(zotulutsa[0]), kuyambitsa=”softmax”)
ukonde = tflearn.regression(net)
model = tflearn.DNN(net)
yesani:
model.load(“model.tflearn”)
kupatula:
model.fit(kuphunzitsa, kutulutsa, n_epoch=1500, batch_size=8, show_metric=Zowona)
model.save(“model.tflearn”)
def bag_of_words(mawu, mawu):
bag = [0 kwa _ mu range(len(mawu))]
mawu = nltk.word_tokenize(s)
s_words = [stemmer.stem(word.lower()) kwa mawu mu mawu_mawu]
kwa se in s_words:
kwa ine, w mu kuwerengera(mawu):
ngati w == se:
chikwama[i] = 1
kubwerera numpy.array(chikwama)
def chat ():
print("Yambani kuyankhula ndi bot (lembani kusiya kuti muyime)!")
pomwe Zowona:
inp = kulowa ("Inu:")
ngati inp.lower() == "siyani":
yopuma
zotsatira = model.predict([chikwama_cha_mawu(inp, mawu)])
results_index = numpy.argmax(zotsatira)
tag = zolemba[results_index]
za tg mu data["zolinga"]:
ngati tg['tag'] == tag:
mayankho = tg['mayankho']
kusindikiza(mwachisawawa.chosankha(mayankho))
kucheza ()
//////////////////////////////////////// //////////////
Zikomo,
Wodala kukod!
Lu
Moni,
Kodi mungandipatseko lingaliro la momwe ndingachitire ngati mukufuna kupanga chatbot mu python, koma chidziwitsocho chimachokera ku kafukufuku wa Excel. Zikomo!