!pip install transformers

from transformers import pipeline

classifier = pipeline("text-classification")

No model was supplied, defaulted to distilbert-base-uncased-finetuned-sst-2-english

import pandas as pd

text = """This book is a comprehensive, wide-ranging detailed, book that covers a huge range of different topic areas in great detail.
Make no mistake this book is big - 700+ pages of in-depth content, and much, nay, most of it not aimed at the casual beginner.
Across 19 chapters the authors go through topic such as building good training sets, dimension reduction techniques, best practices, ensembles, PyTorch, Scikit-Learn, CNNs, RNNs, transformers, GANs and much much much more.
I, personally, have not read any other books by these authors, but I have bought and read many Packt books previously so I knew potentially what I would be getting."""

outputs = classifier(text)

pd.DataFrame(outputs)

text = """Content is undoubtedly good as we all know but here book received is Black & White printed with cheap quality pages.
It looks like a Xerox.
Do not buy from Epitome Books."""

outputs = classifier(text)

pd.DataFrame(outputs)

!pip install datasets

from datasets import load_dataset

dataset = load_dataset("amazon_reviews_multi", name="ja")

dataset

DatasetDict({
    train: Dataset({
        features: ['review_id', 'product_id', 'reviewer_id', 'stars', 'review_body', 'review_title', 'language', 'product_category'],
        num_rows: 200000
    })
    validation: Dataset({
        features: ['review_id', 'product_id', 'reviewer_id', 'stars', 'review_body', 'review_title', 'language', 'product_category'],
        num_rows: 5000
    })
    test: Dataset({
        features: ['review_id', 'product_id', 'reviewer_id', 'stars', 'review_body', 'review_title', 'language', 'product_category'],
        num_rows: 5000
    })
})

dataset["train"][0]

{'language': 'ja',
 'product_category': 'shoes',
 'product_id': 'product_ja_0270003',
 'review_body': '普段使いとバイクに乗るときのブーツ兼用として購入しました。見た目や履き心地は良いです。 しかし、２ヶ月履いたらゴム底が削れて無くなりました。また、バイクのシフトペダルとの摩擦で表皮が剥がれ、本革でないことが露呈しました。ちなみに防水とも書いていますが、雨の日は内部に水が染みます。 安くて見た目も良く、履きやすかったのですが、耐久性のなさ、本革でも防水でも無かったことが残念です。結局、本革の防水ブーツを買い直しました。',
 'review_id': 'ja_0388536',
 'review_title': '本革でも防水でもない',
 'reviewer_id': 'reviewer_ja_0454098',
 'stars': 1}

dataset = dataset.remove_columns(["review_id", "product_id", "reviewer_id", "stars", "review_title", "language"])
dataset = dataset.rename_column("review_body", "text")
dataset = dataset.rename_column("product_category", "label_name")

dataset["train"][0]

{'label_name': 'shoes',
 'text': '普段使いとバイクに乗るときのブーツ兼用として購入しました。見た目や履き心地は良いです。 しかし、２ヶ月履いたらゴム底が削れて無くなりました。また、バイクのシフトペダルとの摩擦で表皮が剥がれ、本革でないことが露呈しました。ちなみに防水とも書いていますが、雨の日は内部に水が染みます。 安くて見た目も良く、履きやすかったのですが、耐久性のなさ、本革でも防水でも無かったことが残念です。結局、本革の防水ブーツを買い直しました。'}

import pandas as pd

dataset.set_format(type="pandas")

df_train = dataset["train"][:]
df_validation = dataset["validation"][:]
df_test = dataset["test"][:]

df_train.head()

labels = [
  *df_train["label_name"],
  *df_test["label_name"],
  *df_validation["label_name"]
]

labels = sorted(list(set(labels)))

from functools import reduce

label_map = reduce(lambda acc, x: { **acc, x: labels.index(x) }, labels, {})

label_map

{'apparel': 0,
 'automotive': 1,
 'baby_product': 2,
 'beauty': 3,
 'book': 4,
 'camera': 5,
 'digital_ebook_purchase': 6,
 'digital_video_download': 7,
 'drugstore': 8,
 'electronics': 9,
 'furniture': 10,
 'grocery': 11,
 'home': 12,
 'home_improvement': 13,
 'industrial_supplies': 14,
 'jewelry': 15,
 'kitchen': 16,
 'lawn_and_garden': 17,
 'luggage': 18,
 'musical_instruments': 19,
 'office_product': 20,
 'other': 21,
 'pc': 22,
 'pet_products': 23,
 'shoes': 24,
 'sports': 25,
 'toy': 26,
 'video_games': 27,
 'watch': 28,
 'wireless': 29}

label_mapper = lambda x: label_map[x]

df_train["label"] = df_train["label_name"].apply(label_mapper)
df_test["label"] = df_test["label_name"].apply(label_mapper)
df_validation["label"] = df_validation["label_name"].apply(label_mapper)

df_train.head()

import matplotlib.pyplot as plt

xs = df_train["label_name"].value_counts().plot.barh(figsize=(9, 7))
plt.show()