Business Intelligence via Natural Language Search
Speaker: Joel Bjornson, Dealogic
In order to maximize the value of Business Intelligence systems, it is crucial to provide intuitive interfaces for exploring the data, suitable for non-expert users. In this talk I will show how we address this problem by building a Natural Language Search interface for generating reports in the Capital Market domain.
Reports consist of various aggregations based on primary market deals, where investment banks help companies raising capital, for instance by issuing loans, shares and bonds. Deal data has been compiled by Dealogic over a long period of time and new data is added on a daily basis. Reports based on this data provide crucial information for banks operating in the Capital Market; For example by allowing banks to analyze their performance within a specific market section or by identifying new business opportunities.
Using a Natural Language Search interface we are able to provide immediate answers to queries such as:
- Which equity deals are announced in real estate this month?
-Which were the top banks in European healthcare last twelve months?
-How much debt is maturing for Coca-Cola next year?
-Total bank revenue in Brazil this year comparing ECM, DCM and Loan
I will give an overview of how we map user queries to dynamically generated reports running off our data warehouse and discuss some of the challenges involved in this process.