Review Scoring

Review Scoring related modeling class

class pororo.tasks.review_scoring.PororoReviewFactory(task: str, lang: str, model: Optional[str])

Bases: pororo.tasks.utils.base.PororoFactoryBase

Regression based Review scoring using Review Corpus

English (roberta.base.en.review)

• dataset: Multilingual Amazon Reviews Corpus (Phillip Keung et al, 2019)

• metric: Pearson (86.85), Spearman (86.60)

Japanese (jaberta.base.ja.review)

• dataset: Multilingual Amazon Reviews Corpus (Phillip Keung et al, 2019)

• metric: Pearson (85.07), Spearman (85.05)

Chinese (zhberta.base.zh.review)

• dataset: Multilingual Amazon Reviews Corpus (Phillip Keung et al, 2019)

• metric: Pearson (80.12), Spearman (80.01)

Korean (brainbert.base.ko.review_rating)

• dataset: Internal data

• metric: Pearson (78.03), Spearman (77.93)

Examples

>>> review = Pororo(task="review", lang="en")
>>> review("Just what I needed! Perfect for western theme party.")
4.79
>>> review("Received wrong size.")
2.65
>>> review = Pororo(task="review", lang="ja")
>>> review("充電あまりしません! 星5だったのに騙されました!")
0.86
>>> review("迅速な対応ありがとうございます。 今後ともよろしくお願いします。")
4.7
>>> review = Pororo(task="review", lang="zh")
>>> review("买的两百多的，不是真货，和真的对比了小一圈！特别不好跟30多元的没区别，退货了！不建议买！")
1.47
>>> review("锅外型好可爱，家人喜欢，很适合3口之家使用")
4.88
>>> review = Pororo(task="review", lang="ko")
>>> review("그냥저냥 다른데랑 똑같숩니다")
2.96
>>> review("좋습니다 만족해요 배송만 좀 더 빨랐으면..")
3.92

static get_available_langs()

static get_available_models()

load(device: str)

Load user-selected task-specific model

Parameters

device (str) – device information

Returns

User-selected task-specific model

Return type

object

class pororo.tasks.review_scoring.PororoBertReviewScore(model, config)

Bases: pororo.tasks.utils.base.PororoSimpleBase

predict(sent: str, **kwargs) → float

Conduct review rating scaled from 1.0 to 5.0

Parameters

sent – (str) sentence to be rated

Returns

rating score scaled from 1.0 to 5.0

Return type

float