Topic Identification (Banking)

Description

This is a Bert-based model, which can be used to classify texts into 77 banking-related classes. This is a Multiclass model, meaning only one label will be returned as an output.

The classes are the following:

• activate_my_card
• age_limit
• apple_pay_or_google_pay
• atm_support
• automatic_top_up
• balance_not_updated_after_bank_transfer
• balance_not_updated_after_cheque_or_cash_deposit
• beneficiary_not_allowed
• cancel_transfer
• card_about_to_expire
• card_acceptance
• card_arrival
• card_delivery_estimate
• card_linking
• card_not_working
• card_payment_fee_charged
• card_payment_not_recognised
• card_payment_wrong_exchange_rate
• card_swallowed
• cash_withdrawal_charge
• cash_withdrawal_not_recognised
• change_pin
• compromised_card
• contactless_not_working
• country_support
• declined_card_payment
• declined_cash_withdrawal
• declined_transfer
• direct_debit_payment_not_recognised
• disposable_card_limits
• edit_personal_details
• exchange_charge
• exchange_rate
• exchange_via_app
• extra_charge_on_statement
• failed_transfer
• fiat_currency_support
• get_disposable_virtual_card
• get_physical_card
• getting_spare_card
• getting_virtual_card
• lost_or_stolen_card
• lost_or_stolen_phone
• order_physical_card
• passcode_forgotten
• pending_card_payment
• pending_cash_withdrawal
• pending_top_up
• pending_transfer
• pin_blocked
• receiving_money
• Refund_not_showing_up
• request_refund
• reverted_card_payment?
• supported_cards_and_currencies
• terminate_account
• top_up_by_bank_transfer_charge
• top_up_by_card_charge
• top_up_by_cash_or_cheque
• top_up_failed
• top_up_limits
• top_up_reverted
• topping_up_by_card
• transaction_charged_twice
• transfer_fee_charged
• transfer_into_account
• transfer_not_received_by_recipient
• transfer_timing
• unable_to_verify_identity
• verify_my_identity
• verify_source_of_funds
• verify_top_up
• virtual_card_not_working
• visa_or_mastercard
• why_verify_identity
• wrong_amount_of_cash_received
• wrong_exchange_rate_for_cash_withdrawal

Predicted Entities

activate_my_card, age_limit, card_acceptance, card_arrival, card_delivery_estimate, card_linking, card_not_working, card_payment_fee_charged, card_payment_not_recognised, card_payment_wrong_exchange_rate, card_swallowed, cash_withdrawal_charge, apple_pay_or_google_pay, cash_withdrawal_not_recognised, change_pin, compromised_card, contactless_not_working, country_support, declined_card_payment, declined_cash_withdrawal, declined_transfer, direct_debit_payment_not_recognised, disposable_card_limits, atm_support, edit_personal_details, exchange_charge, exchange_rate, exchange_via_app, extra_charge_on_statement, failed_transfer, fiat_currency_support, get_disposable_virtual_card, get_physical_card, getting_spare_card, automatic_top_up, getting_virtual_card, lost_or_stolen_card, lost_or_stolen_phone, order_physical_card, passcode_forgotten, pending_card_payment, pending_cash_withdrawal, pending_top_up, pending_transfer, pin_blocked, balance_not_updated_after_bank_transfer, receiving_money, Refund_not_showing_up, request_refund, reverted_card_payment?, supported_cards_and_currencies, terminate_account, top_up_by_bank_transfer_charge, top_up_by_card_charge, top_up_by_cash_or_cheque, top_up_failed, balance_not_updated_after_cheque_or_cash_deposit, top_up_limits, top_up_reverted, topping_up_by_card, transaction_charged_twice, transfer_fee_charged, transfer_into_account, transfer_not_received_by_recipient, transfer_timing, unable_to_verify_identity, verify_my_identity, beneficiary_not_allowed, verify_source_of_funds, verify_top_up, virtual_card_not_working, visa_or_mastercard, why_verify_identity, wrong_amount_of_cash_received, wrong_exchange_rate_for_cash_withdrawal, cancel_transfer, card_about_to_expire

Live Demo Open in Colab Download Copy S3 URI

How to use

document_assembler = nlp.DocumentAssembler() \
    .setInputCol('text') \
    .setOutputCol('document')

tokenizer = nlp.Tokenizer() \
    .setInputCols(['document']) \
    .setOutputCol('token')

sequenceClassifier = finance.BertForSequenceClassification \
      .pretrained('finclf_bert_banking77', 'en', 'finance/models') \
      .setInputCols(['token', 'document']) \
      .setOutputCol('class') \
      .setCaseSensitive(True) \
      .setMaxSentenceLength(512)

pipeline = nlp.Pipeline(
  stages=[
    document_assembler, 
    tokenizer, 
    sequenceClassifier])

example = spark.createDataFrame([['I am still waiting on my card?']]).toDF("text")
result = pipeline.fit(example).transform(example)

Results

['atm_support']

Model Information

References

Banking77 dataset (https://paperswithcode.com/dataset/banking77-oos) and in-house data augmentation

Benchmarking

label               Score              
Loss                0.3031957447528839 
Accuracy            0.9363636363636364 
Macro_F1            0.9364655956915154 
Micro_F1            0.9363636363636364 
Weighted_F1         0.9364655956915157 
Macro_Precision     0.9396792003322154 
Micro_Precision     0.9363636363636364 
Weighted_Precision  0.9396792003322155 
Macro_Recall        0.9363636363636365 
Micro_Recall        0.9363636363636364 
Weighted_Recall     0.9363636363636364