Subjectivity Analysis

This tutorial is available as an IPython notebook at Malaya/example/subjectivity.

This module trained on both standard and local (included social media) language structures, so it is save to use for both.

[1]:
%%time
import malaya
CPU times: user 6.21 s, sys: 1.33 s, total: 7.54 s
Wall time: 8.81 s

Explanation

Positive subjectivity: based on or influenced by personal feelings, tastes, or opinions. Can be a positive or negative sentiment.

Negative subjectivity: based on a report or a fact. Can be a positive or negative sentiment.

[2]:
negative_text = 'Kerajaan negeri Kelantan mempersoalkan motif kenyataan Menteri Kewangan Lim Guan Eng yang hanya menyebut Kelantan penerima terbesar bantuan kewangan dari Kerajaan Persekutuan. Sedangkan menurut Timbalan Menteri Besarnya, Datuk Mohd Amar Nik Abdullah, negeri lain yang lebih maju dari Kelantan turut mendapat pembiayaan dan pinjaman.'
positive_text = 'kerajaan sebenarnya sangat bencikan rakyatnya, minyak naik dan segalanya'

string1 = 'Sis, students from overseas were brought back because they are not in their countries which is if something happens to them, its not the other countries’ responsibility. Student dalam malaysia ni dah dlm tggjawab kerajaan. Mana part yg tak faham?'
string2 = 'Harap kerajaan tak bukak serentak. Slowly release week by week. Focus on economy related industries dulu'
string3 = 'Idk if aku salah baca ke apa. Bayaran rm350 utk golongan umur 21 ke bawah shj ? Anyone? If 21 ke atas ok lah. If umur 21 ke bawah?  Are you serious? Siapa yg lebih byk komitmen? Aku hrp aku salah baca. Aku tk jumpa artikel tu'
string4 = 'Jabatan Penjara Malaysia diperuntukkan RM20 juta laksana program pembangunan Insan kepada banduan. Majikan yang menggaji bekas banduan, bekas penagih dadah diberi potongan cukai tambahan sehingga 2025.'

Load multinomial model

def multinomial(**kwargs):
    """
    Load multinomial emotion model.

    Returns
    -------
    result : malaya.model.ml.Bayes class
    """
[3]:
model = malaya.subjectivity.multinomial()

Predict batch of strings

def predict(self, strings: List[str], add_neutral: bool = True):
    """
    classify list of strings.

    Parameters
    ----------
    strings: List[str]
    add_neutral: bool, optional (default=True)
        if True, it will add neutral probability.

    Returns
    -------
    result: List[str]
    """
[4]:
model.predict([positive_text,negative_text])
[4]:
['neutral', 'negative']

Disable neutral probability,

[5]:
model.predict([positive_text,negative_text], add_neutral = False)
[5]:
['positive', 'negative']

Predict batch of strings with probability

def predict_proba(self, strings: List[str], add_neutral: bool = True):
    """
    classify list of strings and return probability.

    Parameters
    ----------
    strings: List[str]
    add_neutral: bool, optional (default=True)
        if True, it will add neutral probability.

    Returns
    -------
    result: List[dict[str, float]]
    """
[6]:
model.predict_proba([positive_text,negative_text], add_neutral = False)
[6]:
[{'negative': 0.420659316666446, 'positive': 0.5793406833335559},
 {'negative': 0.7906212884104161, 'positive': 0.2093787115895868}]

List available Transformer models

[2]:
malaya.subjectivity.available_transformer()
INFO:root:tested on 20% test set.
[2]:
Size (MB) Quantized Size (MB) macro precision macro recall macro f1-score
bert 425.6 111.00 0.92004 0.91748 0.91663
tiny-bert 57.4 15.40 0.91023 0.90228 0.90301
albert 48.6 12.80 0.90544 0.90299 0.90300
tiny-albert 22.4 5.98 0.89457 0.89469 0.89461
xlnet 446.6 118.00 0.91916 0.91753 0.91761
alxlnet 46.8 13.30 0.90862 0.90835 0.90817
fastformer 458.0 116.00 0.80785 0.81973 0.80758
tiny-fastformer 77.3 19.70 0.87147 0.87147 0.87105

Make sure you can check accuracy chart from here first before select a model, https://malaya.readthedocs.io/en/latest/Accuracy.html#subjectivity-analysis

You might want to use Tiny-Albert, a very small size, 22.4MB, but the accuracy is still on the top notch.

Load Transformer model

All model interface will follow sklearn interface started v3.4,

def transformer(model: str = 'bert', quantized: bool = False, **kwargs):
    """
    Load Transformer subjectivity model.

    Parameters
    ----------
    model : str, optional (default='bert')
        Model architecture supported. Allowed values:

        * ``'bert'`` - Google BERT BASE parameters.
        * ``'tiny-bert'`` - Google BERT TINY parameters.
        * ``'albert'`` - Google ALBERT BASE parameters.
        * ``'tiny-albert'`` - Google ALBERT TINY parameters.
        * ``'xlnet'`` - Google XLNET BASE parameters.
        * ``'alxlnet'`` - Malaya ALXLNET BASE parameters.
        * ``'fastformer'`` - FastFormer BASE parameters.
        * ``'tiny-fastformer'`` - FastFormer TINY parameters.

    quantized : bool, optional (default=False)
        if True, will load 8-bit quantized model.
        Quantized model not necessary faster, totally depends on the machine.

    Returns
    -------
    result: model
        List of model classes:

        * if `bert` in model, will return `malaya.model.bert.BinaryBERT`.
        * if `xlnet` in model, will return `malaya.model.xlnet.BinaryXLNET`.
        * if `fastformer` in model, will return `malaya.model.fastformer.BinaryFastFormer`.
    """
[12]:
model = malaya.subjectivity.transformer(model = 'albert')
INFO:tensorflow:loading sentence piece model
INFO:tensorflow:loading sentence piece model

Load Quantized model

To load 8-bit quantized model, simply pass quantized = True, default is False.

We can expect slightly accuracy drop from quantized model, and not necessary faster than normal 32-bit float model, totally depends on machine.

[14]:
quantized_model = malaya.subjectivity.transformer(model = 'albert', quantized = True)
WARNING:root:Load quantized model will cause accuracy drop.
INFO:tensorflow:loading sentence piece model
INFO:tensorflow:loading sentence piece model

Predict batch of strings

def predict(self, strings: List[str], add_neutral: bool = True):
    """
    classify list of strings.

    Parameters
    ----------
    strings: List[str]
    add_neutral: bool, optional (default=True)
        if True, it will add neutral probability.

    Returns
    -------
    result: List[str]
    """
[15]:
model.predict([negative_text, positive_text])
[15]:
['negative', 'negative']
[17]:
quantized_model.predict([negative_text, positive_text])
[17]:
['negative', 'negative']

Predict batch of strings with probability

def predict_proba(self, strings: List[str], add_neutral: bool = True):
    """
    classify list of strings and return probability.

    Parameters
    ----------
    strings: List[str]
    add_neutral: bool, optional (default=True)
        if True, it will add neutral probability.

    Returns
    -------
    result: List[dict[str, float]]
    """
[18]:
model.predict_proba([negative_text, positive_text])
[18]:
[{'negative': 0.9956738, 'positive': 4.326162e-05, 'neutral': 0.0042829514},
 {'negative': 0.9615872, 'positive': 0.00038412912, 'neutral': 0.038028657}]
[16]:
quantized_model.predict_proba([negative_text, positive_text])
[16]:
[{'negative': 0.9954784, 'positive': 4.521673e-05, 'neutral': 0.0044763684},
 {'negative': 0.9612684, 'positive': 0.00038731584, 'neutral': 0.038344264}]

Open subjectivity visualization dashboard

Default when you call predict_words it will open a browser with visualization dashboard, you can disable by visualization=False.

[12]:
model.predict_words(negative_text)
[13]:
from IPython.core.display import Image, display

display(Image('subjective-dashboard.png', width=800))
_images/load-subjectivity_29_0.png

Vectorize

Let say you want to visualize sentence / word level in lower dimension, you can use model.vectorize,

def vectorize(self, strings: List[str], method: str = 'first'):
    """
    vectorize list of strings.

    Parameters
    ----------
    strings: List[str]
    method : str, optional (default='first')
        Vectorization layer supported. Allowed values:

        * ``'last'`` - vector from last sequence.
        * ``'first'`` - vector from first sequence.
        * ``'mean'`` - average vectors from all sequences.
        * ``'word'`` - average vectors based on tokens.

    Returns
    -------
    result: np.array
    """

Sentence level

[8]:
texts = [negative_text, positive_text, string1, string2]
r = quantized_model.vectorize(texts, method = 'first')
[9]:
from sklearn.manifold import TSNE
import matplotlib.pyplot as plt

tsne = TSNE().fit_transform(r)
tsne.shape
[9]:
(4, 2)
[11]:
plt.figure(figsize = (7, 7))
plt.scatter(tsne[:, 0], tsne[:, 1])
labels = texts
for label, x, y in zip(
    labels, tsne[:, 0], tsne[:, 1]
):
    label = (
        '%s, %.3f' % (label[0], label[1])
        if isinstance(label, list)
        else label
    )
    plt.annotate(
        label,
        xy = (x, y),
        xytext = (0, 0),
        textcoords = 'offset points',
    )
_images/load-subjectivity_34_0.png

Word level

[12]:
r = quantized_model.vectorize(texts, method = 'word')
[13]:
x, y = [], []
for row in r:
    x.extend([i[0] for i in row])
    y.extend([i[1] for i in row])
[14]:
tsne = TSNE().fit_transform(y)
tsne.shape
[14]:
(109, 2)
[15]:
plt.figure(figsize = (7, 7))
plt.scatter(tsne[:, 0], tsne[:, 1])
labels = x
for label, x, y in zip(
    labels, tsne[:, 0], tsne[:, 1]
):
    label = (
        '%s, %.3f' % (label[0], label[1])
        if isinstance(label, list)
        else label
    )
    plt.annotate(
        label,
        xy = (x, y),
        xytext = (0, 0),
        textcoords = 'offset points',
    )
_images/load-subjectivity_39_0.png

Pretty good, the model able to know cluster top side as positive subjectivity, bottom side as negative subjectivity.

Stacking models

More information, you can read at https://malaya.readthedocs.io/en/latest/Stack.html

[6]:
multinomial = malaya.subjectivity.multinomial()
alxlnet = malaya.subjectivity.transformer(model = 'alxlnet')
[12]:
malaya.stack.predict_stack([multinomial, model, alxlnet], [positive_text])
[12]:
[{'negative': 0.19735892950073536,
  'positive': 0.003119166818228667,
  'neutral': 0.1160071232668102}]
[13]:
malaya.stack.predict_stack([multinomial, model, alxlnet], [positive_text], add_neutral = False)
[13]:
[{'negative': 0.7424157666636825, 'positive': 0.04498033797670938}]