top of page

Intelligent Data Visualisation
Intelligent User Interfaces

The project’s goal was to build a user experience of intelligent visualizations for Administrative Proceedings data (“enforcements”) for a financial services corporation.

The solution is a recommender system of visualizations that is controlled by voice commands and is presented as an interactive dashboard. 

Group 15 (1).jpg

Role

UX Designer

Team of 5

Methods

User research, AI, Conversational systems, Recommender system

Tools

Figma, Mural

Context

​

To abide by the confidentiality agreement, the sponsor, the financial service corporation will be referred to as ABC Inc.

ABC Inc. delivers services to customers in a highly regulated environment; with regulatory agencies at the federal and state levels who establish rules, investigate potential violations and impose fines. Currently, ABC Inc.’s compliance team understands the rules that apply to ABC Inc.’s  business and works to ensure firm's compliance. Currently this work is reactive, manual, and time consuming. In this context ABC Inc. approached Bentley’s team with a focus to create an intelligent visualisation that is designed to help users understand and navigate the complex relationships in the data.

​

​

​

​

Problem Statement

​​​

The sources of the data that we received is from the administrative proceedings that are made public. It includes information on:

  • Regulator imposing the fine,

  • Rule that was broken,

  • Violation that occurred,

  • Fine amount,

  • Firm being fined, and

  • Cause of the fine.

However, this information is all in a text format, making it difficult to parse out the complex relationships between the data.

We were provided with multiple personas by the client. We chose to target our visualisations at executives, such as:

  • The Chief Compliance Officer (CCO) or

  • The Chief Executive Officer (CEO)

 

The project focus was on developing data visualisations for the executive persona that can:

  • Help users understand and navigate the data relationships.

  • Assess high-level trends in the data, and take actionable decisions based on the data insights.

  • Act as tools to help with resource allocation decision support.

​

My Role

This was a group project with 5 members, my role was as a UX designer. I was involved in each stage of the project, namely:

  • Research on:

    • Financial Industry​

    • System architecture

    • Exploring visualisation representations and tools

  • Design process

  • Final solution

The Process

​

Research

​

Step 1

  • Financial Industry:

    • FINRA, SEC proceedings

    • Given data sets

  • Executive's persona

    • Executives want to be aware of trends in regulator actions and new areas of focus

    • Executives are big picture thinkers so our system must distill the takeaways from the data into easily digestible pieces of information.

  • User Research and information gathering by holding meetings with the client and experts working in the industry to get domain knowledge.

Step 2

  • System architecture & Visualisations

    • ​We envisaged a system that extracts data through a program.

    • We assumed that a data store that would regularly take in the XML RSS feeds and information stored in the database with various entities and data attributes upon which our visualizations pull in the data.

    • Ability to interact with the system using a voice assistant.

    • The voice assistant will process commands using natural language processing and bring up a visualization on the dashboard.

sketch 1.jpg
28.jpg
1-2.jpg
3.jpg

Initial sketches: System Architecture

Initial sketches: Visualising the given data- ability to zoom in and out

​

  • Dashboard design process

29.jpg

Final Design

Key features​​

30.jpg

​

Recommender system that trains on last year’s data to develop rules that can be used to analyse the current data.

​

Intelligent (AI) enough to sort the data & display only the information that the executive would be interested in.​

30.jpg

​​

Uses gathered knowledge about:

​​

  • Learns from the feedback given to the system

​​

  • Draws co-relations from the training data

​​

  • The executive

31.jpg

​​

Alert function is key because it lets the executive know when they need to look at the data.

​

​​

​

Provides insights from industry, thought leaders and present what is relevant to the topic.

​

​

Primary interaction is through voice but it also has the functionality to interact visually.

32.jpg

Notepad feature: Ability to note- take while making observations

32.jpg

Ability to collaborate and share

Dashboard

Prototype video​​

Design Features

​​​

​

1. Homepage​

​

Landing dashboard: divided into 3 sections

  • Overview section: It provides the ability to filter and sort by date-week-month-year.

  • Visualisation section: 3 options, shown in horizontal scrollable format, so that executive can have a quick preview before taking a deeper dive into each visualisation.

  • Each visualisation has an alert functionality so that executive can be alerted about any latest change

  • Executives are expert users and aren’t really overwhelmed as a novice view, therefore the information rich dashboard is unlikely to so overload them.

 

AI feedback:

  • Top insights- AI recommendation: categorised by different levels of importance

  • Thought leader insights:

    • From social media

    • From conferences

           ​

Enforcement action news: It pertains to new regulations and major fines. New regulations are complied through:

  • Analysis of compliance data

  • Analysis of conference talks

  • Analysis of media platforms such as twitter

 

Top Fines by Firms:

  • Key features – 2 ways of visualisation

  • Donut chart: Focuses on firm visualization by fine amount

  • Bubble chart- Shows fines by risk of violation

    • Frequency of violation

    • Type of violation

    • Magnitude of violation

 

Categorisation by Risk:

  • Low risk

  • Medium risk

  • High risk

 

Tool bars

  • Key features: Scroll by fine range

  • Ability to animate the visualisation in a time period

    • Sort: By time period, or amount

    • Compare: Visualizations over two time periods

 

Right navigation

  • Visual interaction: Visual as well as verbal interaction with agent

  • Interaction ability:

    • Fines by amount: Ability to zoom into a firm's fine and allow deeper secondary analysis

      • Ability to focus only on one firm e.g. JP Morgan fine (Donut chart)

    • Bar chart (fines by topic) : ability to interact on one topic and see the break up of fine categories, and compare monthly fine with the monthly average

 

Notepad feature: allowing for note-taking and collaboration

  • We proposed voice feature on desktop as executives are busy individuals. They can use this feature to get instant answers and also use this notepad feature to set up meetings quickly with their team.

Challenges and Limitations

​

During the project various challenges  such as:​

  • Understanding the financial sector and the executive's role: In the beginning of the project, the team faced issues with understanding the problem space as we were not aware of the working of this sector. However, we overcame this challenge by conducting informational interviews with a subject matter expert. An added advantage was that one member on the team was himself an executive and further enhanced our understanding of the subject.

  • Privacy concerns: We did consider that the user interface design that uses the voice activation system (between the executive and assistant responses) some of this information could be considered private. A strategy will have to be considered such that this information can stay confidential. In addition, as in a lawsuit this information would be discoverable, a strong data retention and destruction program needs to be in place. From a business standpoint damage can be intended to the organisation's public image if the recorded conversations get out to the public.

 

Some of our assumptions were:​

  • Constant feed of updated data: Our prototype relies on a constant feed of data to maintain real-time accuracy. Since the system uses alerts to keep the executive up to date, the system needs to immediately capture enforcement actions to determine if they meet the threshold for an alert. Real-time updates of the data will need to be performed, whether that’s through an automated system or a manual process.

​

Results and Next Steps

​

Through our final design of the executive dashboard we were able to enhance the overall user experience and by showing only necessary information to the user. By providing the user a dashboard overview, simple interactions and enhanced data visualisation will allow users to take informed decisions and gauge data patterns to see future trends.

​

To continue this project, some next steps would include:​

  •  In the future, we also have an eye on implementing eye-tracking technology that could measure the executive’s gaze and focus points. This feedback would increase the power of our system by allowing the executive to interact with the system even more effortlessly, by just looking at a chart and causing the system to zoom in for a closer examination.

  • This technology would be a form of real-time evaluation that would confirm what the executive finds valuable on the dashboard.

​

bottom of page