As we expect visitor traffic to our branches to increase as more offices and shops open up during phase three, Group Channels and Digitalisation have worked together with Group Enterprise AI, Data Management Office to launch a Branch Crowd Status feature on the UOB website. This feature enables our customers to plan ahead and make informed decisions about the best time to visit the branch of their choice.
They can also choose to visit a branch where the waiting times are shorter. Coupled with our SMS queue ticketing system, customers can minimise their time spent waiting in the queue. The Branch Crowd Status feature works by predicting the expected crowd and uses artificial intelligence (AI) and machine learning (ML) by monitoring branch traffic trends and charting out the crowd levels.
The accuracy of the predictive crowd model will be enhanced over time as we continue to gather more data, and leverage on the use of AI and ML.
The COVID-19 pandemic has deeply impacted the lives and livelihoods of everyone. UOB is committed to supporting our customers through this challenging period and ensuring that we provide a safe environment.
The Branch Crowd Status feature is just one of many others initiatives by UOB to provide our customers alternatives when they are fulfilling their banking needs. You may visit our website to understand how UOB is making banking seamless for our customers.
At UOB, we have implemented a suite of Natural Language Understanding (NLU) solutions for various business units.
The approach uses pre-processing techniques to extract structured data from free-form analysts' reports. Then, we apply advanced natural language understanding algorithms to surface insights such as sentiment levels, and identify underlying topics and events.
An example of a business-specific NLU application is the market insights dashboard that analyses multiple years of analysts' reports, enabling executives to better understand sentiments, pre-empt potential queries and surface opportunities relating to market analysis.
In particular, unsupervised machine-learning techniques were used to identify topic clusters and highlight trending events and recurring themes on a quarterly basis.
In addition to topical trends, classification techniques are leveraged to provide businesses with a time series view of positive and negative sentiment. For example, according to analysts’ reports, sentiment levels for the digital cluster have been largely positive in the past 5 years.
The solution automatically analyses a large number of reports, distills and presents deep insights via an interactive, self-service dashboard.