pandas-nlp 0.6.0

Pandas extension with NLP functionalities

Pandas NLP

It's an extension for pandas providing some NLP functionalities for strings.

Setup

Requirements

• python >= 3.8

Installation

Execute:

pip install -U pandas-nlp

To install the default spacy English model:

spacy install en_core_web_md

Key features

Language detection

import pandas as pd
import pandas_nlp

pandas_nlp.register()

df = pd.DataFrame({
    "id": [1, 2, 3, 4, 5],
    "text": [
        "I like cats",
        "Me gustan los gatos",
        "M'agraden els gats",
        "J'aime les chats",
        "Ich mag Katzen",
    ],
})
df.text.nlp.language()

Output

0    en
1    es
2    ca
3    fr
4    de
Name: text_language, dtype: object

with confidence:

df.text.nlp.language(confidence=True).apply(pd.Series)

Output

  language  confidence
0       en    0.897090
1       es    0.982045
2       ca    0.999806
3       fr    0.999713
4       de    0.997995

String embedding

import pandas as pd
import pandas_nlp

pandas_nlp.register()

df = pd.DataFrame(
    {"id": [1, 2, 3], "text": ["cat", "dog", "violin"]}
)
df.text.nlp.embedding()

Output

0    [3.7032, 4.1982, -5.0002, -11.322, 0.031702, -...
1    [1.233, 4.2963, -7.9738, -10.121, 1.8207, 1.40...
2    [-1.4708, -0.73871, 0.49911, -2.1762, 0.56754,...
Name: text_embedding, dtype: object

Closest concept

import pandas as pd
import pandas_nlp

pandas_nlp.register()

themed = pd.DataFrame({
    "id": [0, 1, 2, 3],
    "text": [
        "My computer is broken",
        "I went to a piano concert",
        "Chocolate is my favourite",
        "Mozart played the piano"
    ]
})

themed.text.nlp.closest(["music", "informatics", "food"])

Output

0    informatics
1          music
2           food
3          music
Name: text_closest, dtype: object

Sentence extraction

import pandas as pd
import pandas_nlp

pandas_nlp.register()

df = pd.DataFrame(
    {"id": [0, 1], "text": ["Hello, how are you?", "Code. Sleep. Eat"]}
)
df.text.nlp.sentences()

Output

0    [Hello, how are you?]
1     [Code., Sleep., Eat]
Name: text_sentences, dtype: object