name2nat: a Python package for nationality prediction from a name

name2nat is a Python package that predicts the nationality of any name written in Roman letters. For example, it returns the correct output Korean for my name `Kyubyong Park'. Needless to say, it is not possible to guess somebody's nationality 100% right from their name. After all, nationality can change, you know. However, it is also true that there is a tendency between names and nationality. So it turns out statistical classifiers for this task works to some extent. Details are explained below.

Disclaimer

I am aware that this topic may be viewed from a political perspective. That is absolutely AGAINST my motivation.

NaNa Dataset

Construction

I constructed a new dataset for this project because I failed to find any available dataset that is big and comprehensive enough.

• STEP 1. Downloaded and extracted the 20200601 English wiki dump (enwiki-20200601-pages-articles.xml).
• STEP 2. Iterated all pages and collected the title and the nationality. I regarded the title as a person if the Category section at the bottom of each page included ... births (green rectangle), and identified their nationality from the most frequent nationality word in the section (red rectangles).

* STEP 3. Randomly split the data into train/dev/test in the ratio of 8:1:1 within each nationality group.

Stats

Nationality	# Samples	Train	Dev	Test
Total	1,112,902	890,248	111,286	111,368
Afghan	973	778	97	98
Albanian	2,742	2,193	274	275
Algerian	1,991	1,592	199	200
American	302,215	241,772	30,221	30,222
Andorran	236	188	24	24
Angolan	630	504	63	63
Argentine	11,158	8,926	1,116	1,116
Armenian	2,001	1,600	200	201
Aruban	117	93	12	12
Australian	50,670	40,536	5,067	5,067
Austrian	11,490	9,192	1,149	1,149
Azerbaijani	1,664	1,331	166	167
Bahamian	292	233	29	30
Bahraini	297	237	30	30
Bangladeshi	2,045	1,636	204	205
Barbadian	466	372	47	47
Basque	1,202	961	120	121
Belarusian	2,923	2,338	292	293
Belgian	9,884	7,907	988	989
Belizean	186	148	19	19
Beninese	249	199	25	25
Bermudian	338	270	34	34
Bhutanese	180	144	18	18
Bolivian	822	657	82	83
Bosniak	102	81	10	11
Botswana	315	252	31	32
Brazilian	14,043	11,234	1,404	1,405
Breton	148	118	15	15
British	57,403	45,922	5,740	5,741
Bruneian	144	115	14	15
Bulgarian	4,908	3,926	491	491
Burkinabé	362	289	36	37
Burmese	1,180	944	118	118
Burundian	175	140	17	18
Cambodian	451	360	45	46
Cameroonian	1,286	1,028	129	129
Canadian	42,691	34,152	4,269	4,270
Catalan	2,147	1,717	215	215
Chadian	174	139	17	18
Chilean	3,548	2,838	355	355
Chinese	11,868	9,494	1,187	1,187
Colombian	3,276	2,620	328	328
Comorian	68	54	7	7
Congolese	44	35	4	5
Cuban	2,423	1,938	242	243
Cypriot	1,271	1,016	127	128
Czech	9,056	7,244	906	906
Dane	41	32	4	5
Djiboutian	68	54	7	7
Dominican	1,976	1,580	198	198
Dutch	18,645	14,916	1,864	1,865
Ecuadorian	1,093	874	109	110
Egyptian	3,471	2,776	347	348
Emirati	777	621	78	78
English	96,449	77,159	9,645	9,645
Equatoguinean	242	193	24	25
Eritrean	167	133	17	17
Estonian	2,536	2,028	254	254
Ethiopian	917	733	92	92
Faroese	355	284	35	36
Filipino	4,910	3,928	491	491
Finn	85	68	8	9
French	51,052	40,841	5,105	5,106
Gabonese	226	180	23	23
Gambian	276	220	28	28
Georgian	328	262	33	33
German	52,986	42,388	5,299	5,299
Ghanaian	2,546	2,036	255	255
Gibraltarian	123	98	12	13
Greek	7,469	5,975	747	747
Grenadian	174	139	17	18
Guatemalan	704	563	70	71
Guinean	731	584	73	74
Guyanese	448	358	45	45
Haitian	702	561	70	71
Honduran	626	500	63	63
Hungarian	9,026	7,220	903	903
I-Kiribati	51	40	5	6
Indian	28,365	22,692	2,836	2,837
Indonesian	3,525	2,820	352	353
Iranian	6,263	5,010	626	627
Iraqi	1,566	1,252	157	157
Irish	14,806	11,844	1,481	1,481
Israeli	6,437	5,149	644	644
Italian	36,671	29,336	3,667	3,668
Jamaican	1,778	1,422	178	178
Japanese	26,520	21,216	2,652	2,652
Jordanian	613	490	61	62
Kazakh	31	24	3	4
Kenyan	2,012	1,609	201	202
Korean	9,871	7,896	987	988
Kuwaiti	496	396	50	50
Kyrgyz	20	16	2	2
Lao	33	26	3	4
Latvian	2,117	1,693	212	212
Lebanese	1,558	1,246	156	156
Liberian	368	294	37	37
Libyan	339	271	34	34
Lithuanian	2,474	1,979	247	248
Macedonian	1,374	1,099	137	138
Malagasy	290	232	29	29
Malawian	274	219	27	28
Malaysian	3,228	2,582	323	323
Maldivian	191	152	19	20
Malian	482	385	48	49
Maltese	829	663	83	83
Manx	188	150	19	19
Marshallese	40	32	4	4
Mauritanian	120	96	12	12
Mauritian	329	263	33	33
Mexican	10,810	8,648	1,081	1,081
Moldovan	1,250	1,000	125	125
Mongolian	631	504	63	64
Montenegrin	1,194	955	119	120
Moroccan	1,822	1,457	182	183
Mozambican	263	210	26	27
Namibian	736	588	74	74
Nauruan	40	32	4	4
Nepalese	967	773	97	97
Nicaraguan	357	285	36	36
Nigerian	5,075	4,060	507	508
Nigerien	179	143	18	18
Norwegian	16,891	13,512	1,689	1,690
Omani	247	197	25	25
Pakistani	4,703	3,762	470	471
Palauan	44	35	4	5
Palestinian	660	528	66	66
Panamanian	593	474	59	60
Paraguayan	1,266	1,012	127	127
Peruvian	1,902	1,521	190	191
Portuguese	5,918	4,734	592	592
Qatari	685	548	68	69
Romanian	8,189	6,551	819	819
Russian	26,593	21,274	2,659	2,660
Rwandan	337	269	34	34
Salvadoran	634	507	63	64
Sammarinese	248	198	25	25
Samoan	746	596	75	75
Saudi	1,871	1,496	187	188
Senegalese	1,029	823	103	103
Serb	56	44	6	6
Singaporean	1,646	1,316	165	165
Slovak	3,584	2,867	358	359
Slovene	111	88	11	12
Somali	145	116	14	15
Sotho	62	49	6	7
Sudanese	436	348	44	44
Surinamese	250	200	25	25
Swazi	143	114	14	15
Syriac	98	78	10	10
Syrian	1,309	1,047	131	131
Taiwanese	2,433	1,946	243	244
Tajik	77	61	8	8
Tamil	1,749	1,399	175	175
Tanzanian	784	627	78	79
Thai	3,434	2,747	343	344
Tibetan	332	265	33	34
Togolese	264	211	26	27
Tongan	570	456	57	57
Tunisian	1,340	1,072	134	134
Turk	99	79	10	10
Tuvaluan	83	66	8	9
Ugandan	1,316	1,052	132	132
Ukrainian	7,748	6,198	775	775
Uruguayan	2,834	2,267	283	284
Uzbek	78	62	8	8
Vanuatuan	146	116	15	15
Venezuelan	2,422	1,937	242	243
Vietnamese	1,572	1,257	157	158
Vincentian	10	8	1	1
Welsh	6,588	5,270	659	659
Yemeni	403	322	40	41
Zambian	638	510	64	64

Downloadable Link

• You can download the dataset here.

name2nat

Installation

pip install name2nat

Usage

>>> from name2nat import Name2nat

>>> my_nanat = Name2nat()

>>> names = ["Donald Trump", # American
         "Moon Jae-in", # Korean
         "Shinzo Abe", # Japanese
         "Xi Jinping", # Chinese
         "Joko Widodo", # Indonesian
         "Angela Merkel", # German
         "Emmanuel Macron", # French
         "Kyubyong Park", # Korean
         "Yamamoto Yu", # Japanese
         "Jing Xu"] # Chinese
>>> result = my_nanat(names, top_n=3)
>>> print(result)
# (name, [(nationality, prob), ...])
# Note that prob of 1.0 indicates the name exists
# in Wikipedia.
[
('Donald Trump', [('American', 1.0)])
('Moon Jae-in', [('Korean', 1.0)])
('Shinzo Abe', [('Japanese', 1.0)])
('Xi Jinping', [('Chinese', 1.0)])
('Joko Widodo', [('Indonesian', 1.0)])
('Angela Merkel', [('German', 1.0)])
('Emmanuel Macron', [('French', 1.0)])
('Kyubyong Park', [('Korean', 0.9985014200210571), ('American', 0.000289416522718966), ('Bhutanese', 0.00025851925602182746)])
('Yamamoto Yu', [('Japanese', 0.7050493359565735), ('Taiwanese', 0.12779785692691803), ('Chinese', 0.04263153299689293)])
('Jing Xu', [('Chinese', 0.8626819252967834), ('Taiwanese', 0.09901007264852524), ('American', 0.022995812818408012)])
]

Training

I use a powerful NLP library Flair to train a text classifier model. A bidirectional GRU layer is employed.

python train.py

Evaluation

python predict.py;
python eval.py --gt nana/test.tgt --pred test.pred

Results

K	Precision@K
1	61310/111368=55.1
2	77480/111368=69.6
3	86703/111368=77.9
4	92491/111368=83.0
5	96697/111368=86.8

Applications

Let's predict the nationalities of the first authors of the recent machine learning conferences.

• Check conferences.py and conferences/lrec2020.md
• Contributions (PRs) are welcome!

References

If you use this code for research, please cite:

@misc{park2018name2nat,
  author = {Park, Kyubyong},
  title = {name2nat: a Python package for nationality prediction from a name},
  year = {2020},
  publisher = {GitHub},
  journal = {GitHub repository},
  howpublished = {\url{https://github.com/Kyubyong/name2nat}}
}