Our client, DataOceans distributes other companies’ billing statements that contain advertisements for new products. To make the ads more personalized, DataOceans has developed a user interface that allows businesses to define specific criteria for targeting their advertisements. For instance, a business might target married males over the age of 40, for a motorboat advertisement. The system would filter for users satisfying these criteria with SQL if-else statements and display the corresponding advertisements. However, this interface is cumbersome and complex to maintain, and there are scenarios where end users do not meet the criteria for any ad, resulting in no advertisements being displayed.
Consequently, we proposed several solutions to resolve these issues. To make the system easier for advertisers to use, we have built a natural language processing (NLP) pipeline that allows businesses to input their advertising criteria using plain English. Following this input, we use large language models to translate these descriptions into machine readable queries. This approach aims to resolve the usability issue associated with the user interface.
