The financial sector generates a lot of valuable data, from individual purchases to large transactions. Given the enormous treasure that all this information holds, expectations of the impact of artificial intelligence (AI) in the banking sector could not be higher. It is anticipated that the industry will save more than $1 trillion by 2030 due to recent developments in AI.
Faced with these amazing opportunities, many banks are beginning to take action. But how can they make the most of artificial intelligence?
Unfortunately, managers who try to stay at the top often find themselves confronted with the same harsh reality: To successfully deploy AI applications, large amounts of data are not enough. The quality of data plays a major role in the quality of results delivered by AI, and this is where most organizations struggle.
Banks spent the last decades gathering data, and in turn, data management has become extremely complex. Data is often disconnected and stored in different formats, creating isolated repositories of information that are not available to an entire organization. This makes bank-wide research ineffective and prevents artificial intelligence applications from discovering insights from data.
Fortunately, a solution already exists. The knowledge graph, a technology that is used by giants such as Amazon, Google and Apple, can connect different databases and make them searchable. Knowledge graphs can also link both structured and unstructured data, allowing AI applications to use information not only inside databases, but also in text documents, for example.
How A Knowledge Graph Works
A knowledge graph is a model of a knowledge domain. It maps all business objects and concepts an enterprise works with, together with their interrelations. Structured as an additional virtual data layer, the knowledge graph lies on top of existing databases and links data together at scale. This is true for structured data such as spreadsheets as well as for unstructured data like text documents.
Since it’s based on knowledge and concepts, the creation of a good knowledge graph must involve subject-matter experts from different areas of an entire organization. This increases the need for collaboration and promotes shared responsibility and transparency in knowledge management. In addition, because the technology doesn’t replace but rather boosts existing IT systems, it is extremely cost-efficient.
Financial institutions trying to develop their knowledge graphs don’t have to start from scratch. The Financial Industry Business Ontology (FIBO) defines sets of business objects that are of interest in financial business applications, as well as how they relate to one another. By using FIBO, organizations can give meaning to any data that describe the business of finance.
Personalized Banking Services With Semantic AI
The combination of knowledge graphs, natural language processing (NLP) and artificial intelligence, often called semantic AI, will be essential for the banking sector’s digital transformation.
One particularly interesting trend is using the technology to improve personalized customer services. This can be done by using a knowledge graph to build recommender systems, which are often used also in online shops to display relevant products to users.
Because knowledge graphs link data in smart ways, they allow recommender systems to make much better recommendations than pure machine learning. Imagine an online store that has a user interested in blue cheese — for example, Roquefort. While most recommenders could suggest other types of blue cheese, like Gorgonzola, to this user, a recommender that is powered by a knowledge graph could take things one step further and suggest a wine that goes well with that specific type of cheese.
The understanding of how different entities relate to each other — in this case, cheese and wine and all their attributes that offer a rich set of additional context information — makes all the difference in the quality of recommendations.
Banks are deploying this technology in self-service portals that show customers a personalized view of information, such as new offerings and services. Along the same lines, they are also using knowledge graphs in online portals to improve customer financial literacy. That is done by building digital assistants that help customers acquire financial knowledge through semantic search over knowledge hubs.
Credit Suisse, for example, helps clients and analysts make informed decisions faster with its semantic AI search engine. The platform is able to retrieve large amounts of information with speed and quality and delivers context-based quality results. It allows clients and analysts to focus only on the information they need and provides access to personalized visual analytics. (Full disclosure: Credit Suisse is a customer.)
Another example is Deutsche Bank, which is striving to implement knowledge graphs to support its AI strategy for several reasons, including automated enrichment via relationship discovery, content contextualization and a better understanding of the meaning of data.
However, the personalization of banking services is just one of the many technology trends in the banking industry. We're also seeing this tech appear in matters that relate to compliance, fraud detection, risk assessment, lease agreements and even loan applications. In all of these use cases, knowledge graphs are essential to achieving optimal results.
Another essential application based on knowledge graphs is "know your customer" (KYC) or customer 360, which also involves the use of linked and holistic views of the customer, enriched with contextual information, to be able to develop accurate communication, make informed decisions or put together an accurate product offer.
How To Start Working With Semantic AI
To get started with semantic AI, banks should begin by defining a concrete use case with a specific goal. By working on a defined project, organizations can understand the technology’s full potential and see other opportunities to apply and eventually deploy it throughout the whole organization.
It is therefore necessary to evaluate the usefulness of knowledge graphs on the basis of individual use cases while building up enough knowledge to be able to embed the methodology in a more comprehensive AI strategy.
When choosing a software to manage your enterprise knowledge graph, you should look for a solution that is based on standards, interoperable with your current architecture, scalable and easy to learn. The biggest bottleneck in the introduction of semantic AI is no longer the technology, but the people who are not willing to trust that it will work for them in the long run.
Want to stay up to date with the latest news, products, and trends in the intelligent industry today? As well as the details of the competitions during the World Intelligence Congress. Subscribe us now and stay informed!Subscribe
世界智能大会 WORLD INTELLIGENCE CONGRESS