init app

app.py

@@ -0,0 +1,61 @@
+import gradio
+import numpy as np
+import pandas as pd
+import re
+import fasttext
+from huggingface_hub import hf_hub_download
+
+model_path = hf_hub_download(repo_id="facebook/fasttext-id-vectors", filename="model.bin")
+model = fasttext.load_model(model_path)
+new_catalog = pd.read_csv('new_catalog.csv')
+new_prodcatalog_vec = np.load('product_catalog_vec.npy')
+
+def preprocess(text):
+    # remove units
+    regex_f = r'(\d+[.]{0,1}\d*\s*(kg|KG|Kg|ml|ML|Ml|mL|mil|ltr|LTR|liter|lt|Lt|LT|L|l|kilograms|%|gram|Gram|grm|GR|Gr|gr|g|G|wp|Wp|WP|METER|sp|Amper|Ahm|A|Btr|btr|ons|kapsul|cm|inc|cc|SL|sl))'
+    text = re.sub(regex_f, "", text)
+
+    # replace extra characters
+    text = text.replace('@', ' ')
+    text = text.replace('.', ' ')
+    text = text.replace('-', ' ')
+    text = text.replace('(', ' ')
+    text = text.replace(')', ' ')
+    text = text.strip()
+
+    # lowercase
+    text = text.lower()
+    return text
+
+def extract_features(text):
+    return model.get_sentence_vector(text)
+
+def cosine_similarity(v1, v2):
+    return np.dot(v1,v2) / (np.linalg.norm(v1)*np.linalg.norm(v2))
+
+def get_most_similar(target, reference, n=5):
+    score = []
+    for r in reference:
+        if np.count_nonzero(r) == 0:  # untuk menghindari score nan
+            score.append(0.)
+        else:
+            score.append(cosine_similarity(target, r))
+    # ranking descending
+    idx = np.array(score).argsort()[::-1][:n]
+    top_score = np.take(score, idx)
+    return idx, top_score
+
+def predict_catalog(text):
+    text = preprocess(text)
+    X = extract_features(text)
+    idx, top_score = get_most_similar(X, new_prodcatalog_vec)
+    catalog = np.take(new_catalog['Product SKU'].values, idx)[0]
+    return catalog
+
+demo = gradio.Interface(
+    fn=predict_catalog,
+    inputs=gradio.Textbox(placeholder="Enter a product name here..."),
+    outputs="label",
+    examples=[["NPK Kujang 30.6.8 @ 25 Kg"]])
+
+demo.launch(share=True)

new_catalog.csv

@@ -0,0 +1,283 @@
+Product SKU,Product SKU Preprocessed
+Urea Petro,urea petro
+Urea PIM,urea pim
+Urea Nitrea,urea nitrea
+Urea Daun Buah,urea daun buah
+Urea Pusri,urea pusri
+Nitralite,nitralite
+ZA Petro,za petro
+ZA Plus Petro,za plus petro
+ZK Petro,zk petro
+Petro-CAS,petro cas
+SP-36 Petro,sp 36 petro
+Rock Phosphate PETRO,rock phosphate petro
+Phosgreen,phosgreen
+SP-26 Petro,sp 26 petro
+Petro BioFertil,petro biofertil
+Phonska Plus 15-15-15+9S+0.2Zn,phonska plus 15 15 15 9s 0.2zn
+NPK Kebomas 12-12-17+2MgO+0.1Zn+0.2B+0.2Fe,npk kebomas 12 12 17 2mgo 0.1zn 0.2b 0.2fe
+NPK Kebomas 12-6-22+3Mg,npk kebomas 12 6 22 3mg
+NPK Kebomas 15-15-15,npk kebomas 15 15 15
+PETROFERT  16-16-8+13S,petrofert  16 16 8 13s
+Petro Niphos 20-20+13S,petro niphos 20 20 13s
+FERTIGRES  16-20+13S ,fertigres  16 20 13s 
+NPK PIM 15-15-15,npk pim 15 15 15
+Nitroku 16-16-16,nitroku 16 16 16
+NITROSKA  15-9-20+1.5MgO+3S ,nitroska  15 9 20 1.5mgo 3s 
+NPK Kujang 15-15-15 ,npk kujang 15 15 15 
+NPK Kujang 30-6-8,npk kujang 30 6 8
+Polivit-PIM 0-0-12+48S+9MgO,polivit pim 0 0 12 48s 9mgo
+Solution N 28-10-10 + TE,solution n 28 10 10   te
+Pelangi Agro 20-10-10,pelangi agro 20 10 10
+Pelangi Jos 16-16-16 Coating Mikroba,pelangi jos 16 16 16 coating mikroba
+Pelangi 16-16-16,pelangi 16 16 16
+Pelangi 15-15-15,pelangi 15 15 15
+NPK Pusri 15-15-15 ,npk pusri 15 15 15 
+Pelangi 20-10-18,pelangi 20 10 18
+Pelangi 12-12-17-2,pelangi 12 12 17 2
+NPK Pelangi 13-6-27-4,npk pelangi 13 6 27 4
+Petro Nitrat 16-16-16,petro nitrat 16 16 16
+Jeranti 18-10-14+2S+TE,jeranti 18 10 14 2s te
+NPK Pusri 16-16-16,npk pusri 16 16 16
+NPK Pusri 13-6-27-4Mg+0.65B,npk pusri 13 6 27 4mg 0.65b
+NPK Petro Ningrat 12-11-20,npk petro ningrat 12 11 20
+Kapur Pertanian Kebomas,kapur pertanian kebomas
+MerokeZA,merokeza
+MerokeROCK,merokerock
+MerokeTSP,meroketsp
+MerokeKKB 0-0-40-0.8B,merokekkb 0 0 40 0.8b
+Suburkali Butir 0-0-30+17S+10MgO,suburkali butir 0 0 30 17s 10mgo
+MerokeSOP KALI 0-0-52+18S,merokesop kali 0 0 52 18s
+MerokeMOP,merokemop
+KARATE PLUS BORONI 15.5-0-0+0.3S+26CaO,karate plus boroni 15.5 0 0 0.3s 26cao
+Mutiara PARTNER 13-13-24,mutiara partner 13 13 24
+Mutiara 16-16-16,mutiara 16 16 16
+Mutiara PROFESSIONAL 9-25-25,mutiara professional 9 25 25
+Mutiara SPRINTER 20-10-10,mutiara sprinter 20 10 10
+MerokeCPN NK 15-15+TE,merokecpn nk 15 15 te
+MerokeMAP 12-61-0,merokemap 12 61 0
+MerokeMKP 0-52-34,merokemkp 0 52 34
+Meroke Flex-G 8-9-39+3MgO,meroke flex g 8 9 39 3mgo
+SS (AMMOPHOS) 16-20-0+12S,ss (ammophos) 16 20 0 12s
+Mutiara GROWER 15-09-20+TE,mutiara grower 15 09 20 te
+Mahkota ZA,mahkota za
+RP Peru,rp peru
+RP Mesir,rp mesir
+Triple Super Phospate (TSP),triple super phospate (tsp)
+Mahkota Kieserite,mahkota kieserite
+Borat,borat
+Mahkota MOP (KCL Canada),mahkota mop (kcl canada)
+Mahkota NP 16-20-12S,mahkota np 16 20 12s
+Mahkota 15-15-6-4,mahkota 15 15 6 4
+Mahkota 12-6-22-3+TE,mahkota 12 6 22 3 te
+Mahkota 12-12-17-2+TE,mahkota 12 12 17 2 te
+Mahkota 13-6-27-4+0.5B,mahkota 13 6 27 4 0.5b
+Mahkota 13-8-27-4+0.65B,mahkota 13 8 27 4 0.65b
+Mahkota 15-15-15,mahkota 15 15 15
+Mahkota 16-16-16,mahkota 16 16 16
+Urecote,urecote
+ZA Pak Tani,za pak tani
+Fertiphos,fertiphos
+Phosgro,phosgro
+Fertiphos Sawit,fertiphos sawit
+Magnesium Sulfate,magnesium sulfate
+Fertikali,fertikali
+Saprodap  16-20+12S,saprodap  16 20 12s
+Pak Tani 15-15-15+TE,pak tani 15 15 15 te
+Pak Tani Holland  15-15-15,pak tani holland  15 15 15
+Pak Tani  16-16-16 Merah,pak tani  16 16 16 merah
+Pak Tani  16-16-16 Biru,pak tani  16 16 16 biru
+Pak Tani Sawit  15-15-6-4,pak tani sawit  15 15 6 4
+Pak Tani Sawit 12-12-17-2,pak tani sawit 12 12 17 2
+Pak Tani Sawit  13-6-27-4+0.65B,pak tani sawit  13 6 27 4 0.65b
+Pak Tani Sawit  13-8-27-4+0.5B,pak tani sawit  13 8 27 4 0.5b
+Pak Tani Sawit  7-6-35+B+Cu+Zn,pak tani sawit  7 6 35 b cu zn
+Magnum 15-10-22-2Mg-3.8S,magnum 15 10 22 2mg 3.8s
+Pak Tani Padi  21-14-7+2MgO+2S+TE,pak tani padi  21 14 7 2mgo 2s te
+Pak Tani Fertila 13-6-27-4Mg+0.65B+TE,pak tani fertila 13 6 27 4mg 0.65b te
+Pak Tani Fertila  18-6-14+5S,pak tani fertila  18 6 14 5s
+Pak Tani Fertilla 18-8-10+3MgO+TE,pak tani fertilla 18 8 10 3mgo te
+Pak Tani Fertila  21-14-7+4MgO+TE,pak tani fertila  21 14 7 4mgo te
+Pak Tani Fertila 20-10-12+TE,pak tani fertila 20 10 12 te
+Pak Tani Fertila  28-6-13+3MgO+TE,pak tani fertila  28 6 13 3mgo te
+Pak Tani Fertila 8-15-19,pak tani fertila 8 15 19
+Pak Tani Singkong  15-15-15,pak tani singkong  15 15 15
+Pak Tani 12-12-17-2,pak tani 12 12 17 2
+Pak Tani Sawit  12-6-22-3+TE,pak tani sawit  12 6 22 3 te
+Pak Tani  9-15-6,pak tani  9 15 6
+Pak Tani Fertila Qrop   12-6-24,pak tani fertila qrop   12 6 24
+MKP PAK TANI 0-52-32,mkp pak tani 0 52 32
+CPN PAK TANI 15-0-15,cpn pak tani 15 0 15
+ULTRADAP 12-60-0,ultradap 12 60 0
+NEO KRISTALON 12-12-36,neo kristalon 12 12 36
+NEO KRISTALON 18-18-18-0.8Mg-2S+0.02Mn+0.036Zn+0.02B,neo kristalon 18 18 18 0.8mg 2s 0.02mn 0.036zn 0.02b
+FERTILA 6-18-28-0.6Mg-5S+0.1B+0.01Zn+0.01Cu,fertila 6 18 28 0.6mg 5s 0.1b 0.01zn 0.01cu
+FERTILA 12-6-24-0.75Mg-1.5CaO-5.6s,fertila 12 6 24 0.75mg 1.5cao 5.6s
+FERTILA 0-16-17,fertila 0 16 17
+PN PAK TANI 13-0-46+0.55Cl,pn pak tani 13 0 46 0.55cl
+YaraVera,yaravera
+Ammonium Sulfate,ammonium sulfate
+YaraVita TRI-PHOLATE 0-0-0-70Mn+50Zn+25Fe+20B+20Cu+1Mo,yaravita tri pholate 0 0 0 70mn 50zn 25fe 20b 20cu 1mo
+YaraLiva TROPICOTE 15.5-0-0+19Ca,yaraliva tropicote 15.5 0 0 19ca
+YaraTera CALCINIT 15-0-0+26CaO,yaratera calcinit 15 0 0 26cao
+YaraTera KRISTA K 13.7-0-46.3,yaratera krista k 13.7 0 46.3
+YaraTera KRISTA MAP 12-61-0,yaratera krista map 12 61 0
+YaraMila PALMAE 13-11-21+2MgO+0.2B,yaramila palmae 13 11 21 2mgo 0.2b
+YaraLiva NITRABOR 15.5-0-0+25.6CaO+0.3B,yaraliva nitrabor 15.5 0 0 25.6cao 0.3b
+YaraMila WINNER 15-09-20+1.3MgO+3.4S+TE,yaramila winner 15 09 20 1.3mgo 3.4s te
+YaraMila UNIK 16-16-16,yaramila unik 16 16 16
+YaraMila FASTER 25-7-7,yaramila faster 25 7 7
+YARAMILA COMPLEX 12-11-18-3Mg-8S+0.02Zn+0.03Mn-2.56Ca+0.22Fe,yaramila complex 12 11 18 3mg 8s 0.02zn 0.03mn 2.56ca 0.22fe
+YaraTera KRISTA MKP 0-52-34,yaratera krista mkp 0 52 34
+Vrea,vrea
+Nitroplus (ZA),nitroplus (za)
+MK FOS 0-52-34,mk fos 0 52 34
+HSP Astiva,hsp astiva
+KSP,ksp
+CAP TAWON 12-12-17-2Mg,cap tawon 12 12 17 2mg
+CAP TAWON 15-15-15,cap tawon 15 15 15
+CAP TAWON 15-15-6-4,cap tawon 15 15 6 4
+CAP TAWON 16-16-16,cap tawon 16 16 16
+CAP TAWON 18-12-6,cap tawon 18 12 6
+CAP TAWON 19-9-19,cap tawon 19 9 19
+HXAS,hxas
+HX-MROPH,hx mroph
+HX-MKP 0-50-35,hx mkp 0 50 35
+HX-DAP 18-46-0,hx dap 18 46 0
+DGW TSP,dgw tsp
+DGW KNO3,dgw kno3
+COCKHEAD 12-12-17-2Mg,cockhead 12 12 17 2mg
+COCKHEAD 12-6-22-3Mg,cockhead 12 6 22 3mg
+COCKHEAD 13-6-27-4Mg+0.65B,cockhead 13 6 27 4mg 0.65b
+COCKHEAD 15-15-15,cockhead 15 15 15
+COCKHEAD 15-15-6-4Mg,cockhead 15 15 6 4mg
+COCKHEAD 16-16-16,cockhead 16 16 16
+COCKHEAD 7-6-34,cockhead 7 6 34
+COMPACTION DGW 15-15-15+TE,compaction dgw 15 15 15 te
+BOOSTER DGW 12-6-22-3+TE,booster dgw 12 6 22 3 te
+GOLD DGW 16-10-18,gold dgw 16 10 18
+COCKHEAD 13-8-27-4+TE,cockhead 13 8 27 4 te
+DGW 16-16-16+TE,dgw 16 16 16 te
+ENTEC 13-10-20,entec 13 10 20
+NITROPHOSKA 13-10-20,nitrophoska 13 10 20
+ENTEC 20-10-10,entec 20 10 10
+NITROPHOSKA PLUS 12-12-17-2,nitrophoska plus 12 12 17 2
+NITROPHOSKA PERFECT 15-5-20,nitrophoska perfect 15 5 20
+CANTIK CALCINIT,cantik calcinit
+NITROPHOSKA 15-15-15,nitrophoska 15 15 15
+Brucite,brucite
+EMCOTE 13-11-11-7Ca-2Mg-6S+TE(0.08Cu+0.5B),emcote 13 11 11 7ca 2mg 6s te(0.08cu 0.5b)
+EMCOTE 13-6-27-2+TE,emcote 13 6 27 2 te
+MESTI-PATENHIJO 15-10-20+TE,mesti patenhijo 15 10 20 te
+MESTIFOSKA 15-15-15+TE,mestifoska 15 15 15 te
+EMCOTE 15-15-6-4+TE,emcote 15 15 6 4 te
+MESTIFOS 15-20-0+12S,mestifos 15 20 0 12s
+MESTI-PATENBIRU 16-16-16+TE,mesti patenbiru 16 16 16 te
+MESTI-BOMBER 19-9-19+TE,mesti bomber 19 9 19 te
+EMCOTE 20-6-14,emcote 20 6 14
+MESTI-KP 0-52-34,mesti kp 0 52 34
+Mesti-ZA,mesti za
+Mesti-TSP,mesti tsp
+Mesti-Rock,mesti rock
+Mestikali,mestikali
+Mestical,mestical
+Mesti-GAS,mesti gas
+Mestac,mestac
+MestiKisrit,mestikisrit
+Borate Evermax,borate evermax
+KCL MAHKOTA,kcl mahkota
+KCL,kcl
+MINYAK KAYU PUTIH,minyak kayu putih
+KERING,kering
+PRIMA KUAT,prima kuat
+BULDOK,buldok
+ABACEL,abacel
+NITREA,nitrea
+SPONTAN,spontan
+STARBAN,starban
+GRAMOXONE,gramoxone
+INDAMIN,indamin
+CHLORMITE,chlormite
+LINDOMIN,lindomin
+AMISTARTOP,amistartop
+BROADPLUS,broadplus
+SUPREMO,supremo
+SIDAFOS,sidafos
+METINDO,metindo
+ROUNDUP,roundup
+REGENT,regent
+MANUVER,manuver
+SIDATAN XR,sidatan xr
+DACONIL,daconil
+TEBALO,tebalo
+FOLICUR,folicur
+MIPCINTA,mipcinta
+DANGKE,dangke
+AGUS,agus
+ROUND UP,round up
+NAGA,naga
+AQUA,aqua
+KNO PUTIH,kno putih
+EMACEL,emacel
+BENTAN,bentan
+ATONIK,atonik
+GRAMAXONE,gramaxone
+TOPBAN,topban
+ANTRACOL,antracol
+MARSHAL,marshal
+PHONSKA PLUS,phonska plus
+STADIUM,stadium
+MUSI,musi
+SUMO,sumo
+AVIDOR,avidor
+ARJUNA,arjuna
+PAMUNGKAS,pamungkas
+DEMOLISH,demolish
+BABLAS,bablas
+BERAS,beras
+METSULINDO PLUS,metsulindo plus
+DITHANE,dithane
+DECIS,decis
+SEE TOP,see top
+FURADAN,furadan
+GANDASIL B,gandasil b
+KANON,kanon
+ENDURE,endure
+PLENUM,plenum
+GAS,gas
+REGEN,regen
+SIDAMETRIN,sidametrin
+KEJORA,kejora
+ALIKA,alika
+ZIFLO,ziflo
+ZA NS,za ns
+DITHANE M 45,dithane m 45
+RAMBO,rambo
+BABLASS,bablass
+BELT,belt
+GANDASIL BUAH,gandasil buah
+NK,nk
+HOPPER,hopper
+ABOLISI,abolisi
+BENTAN BIRU,bentan biru
+SCORE 250 EC,score 250 ec
+PRIMA LIMA,prima lima
+ZA,za
+MATADOR,matador
+CBA 6,cba 6
+DMA 6,dma 6
+GEMPUR,gempur
+TRISULA,trisula
+DMA,dma
+MKP PAK TANI,mkp pak tani
+ARES,ares
+ELANG,elang
+KSP TAWON,ksp tawon
+UREA PRILL,urea prill
+SIDAGREEN,sidagreen
+SP,sp
+SPGRO,spgro
+MEROKE,meroke
+KNO3,kno3
+UREA PRIL,urea pril
+ROGER,roger

product_catalog_vec.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9a5b0f79682f532a218ba18af27dc9f46d17f5bf7cab16fc2a830a0c1de7e387
+size 338528