Unstructured data is coming into its own as financial institutions deploy machine learning to drive business insights out of the data, and use the data to develop more holistic risk profiles. Use cases such as these demonstrate the huge potential of harnessing unstructured data, but there are also associated challenges that must be addressed.
Ahead of next week’s A-Team Group webinar that will dive into the detail of unstructured data, we caught up with one of the speakers, Gurraj Singh Sangha, formerly head of data science, risk and market intelligence at State Street, to get a flavour of some of the key points that will be up for discussion.Setting the scene, Sangha describes how unstructured data has added myriads of data to more traditional structured data such as securities, company, and time series data. He also notes the rise of machine learning and Natural Language Processing (NLP) that can automate previously manual data extraction.
“There are tremendous streams of unstructured data,” Sangha says. “Rather than having individuals read and extract data from news feeds for risk purposes, you can use machine learning to extract important information, such as how asset classes or markets are moving, and identify investment risk.”
He also notes the ability to create investment opportunities in a timely manner and the operational efficiencies of using algorithms (with human oversight) rather than humans to read and extract information from large documents.
Realising the potential of unstructured data, traditional data vendors provide a vast majority of the data, although some, such as data with privacy or regulatory constraints, is not so easy to source, says Sangha. This data may need to be anonymised, cleansed, corrected and validated before it is useful, raising the next challenge of how to integrate and store unstructured data. Join us at next week’s webinar to discuss the challenges and opportunities of unstructured data.