README.md

metadata

language:
  - en
dataset_info:
  features:
    - name: id
      dtype: int64
    - name: query
      dtype: string
    - name: product_title
      dtype: string
    - name: product_description
      dtype: string
    - name: median_relevance
      dtype: float64
    - name: relevance_variance
      dtype: float64
    - name: split
      dtype: string
  splits:
    - name: train
      num_bytes: 5156813
      num_examples: 10158
    - name: test
      num_bytes: 14636826
      num_examples: 22513
  download_size: 10796818
  dataset_size: 19793639
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*
      - split: test
        path: data/test-*

Crowdflower Search Results Relevance

• Original source: https://www.kaggle.com/c/crowdflower-search-relevance/overview
• More detailed version: https://data.world/crowdflower/ecommerce-search-relevance

Citation

@misc{crowdflower-search-relevance,
    author = {AaronZukoff, Anna Montoya, JustinTenuto, Wendy Kan},
    title = {Crowdflower Search Results Relevance},
    publisher = {Kaggle},
    year = {2015},
    url = {https://kaggle.com/competitions/crowdflower-search-relevance}
}

Code for generating data

# ! unzip train.csv.zip
# ! unzip test.csv.zip 


df_comp = pd.concat([
    pd.read_csv("./train.csv").assign(split="train"),
    pd.read_csv("./test.csv").assign(split="test"),
])

dataset = DatasetDict(
    train=Dataset.from_pandas(df_comp[df_comp["split"] == "train"].reset_index(drop=True)),
    test=Dataset.from_pandas(df_comp[df_comp["split"] == "test"].reset_index(drop=True)),
)

dataset.push_to_hub("napsternxg/kaggle_crowdflower_ecommerce_search_relevance")