– Case Study –
Conversational AI for Banking
Flexible data annotation solutions for diverse model challenges
Challenge
One of Canada’s largest commercial banks sought a flexible Data Annotation Partner to help with a variety of one-time annotation projects that each utilize different models and data. While the projects vary, they are always banking related, which means high accuracy, security, and compliance are paramount to success.
High Accuracy, Security, & Compliance across projects
Results
SOLUTION
Innodata leveraged its AI-enabled platform and in-house subject matter experts to complete a range of projects to date including:
- Classification of verbatims using a multilabel model
- Paraphrasing existing questions into multiple synonymic sentences to create a training corpus for their conversational agent
- Validation that sentences, in both English and French, are classified into categories
Each project uses double pass annotation with an arbitrator to break ties. This ensures their accuracy goals are always met, so they can trust their model performance.
IMPACT
Innodata helped the bank enhance and accelerate the ground truth data pipeline used to train their conversational models. Training on precise and relevant utterances helped the chatbot to become more articulate, more responsive, and increase customer satisfaction. Innodata ultimately won their trust and business by being flexible and precise. The client stated they also “appreciate Innodata’s confidentiality, high-quality standards, and in-house local language annotators.”
Meet an Expert
Our Team of Data Experts
A team comprised of data experts with extensive experience in developing AI-based data solutions for clients. Book a time that works for you and let us help develop a custom solution for your unique needs.