AI Text analytics dashboard
AI Text analytics dashboard
AI Text analytics dashboard
B2B System, UX Design, Articulate Design Goal & Decision
Building editing features on text analytics products to improve AI accuracy
Building editing features on text analytics products to improve AI accuracy
Building editing features on text analytics products to improve AI accuracy
Role
UX Design Intern
Role
UX Designer Intern
Role
UX Design Intern
Team
Team
Senior UX Designer
Product manager
Engineers
Senior UX Designer
Product manager
Engineers
Team
Senior UX Designer
Product manager
Engineers
Tool
Tool
Figma
Confluence
Figma
Confluence
Tool
Figma
Confluence
Duration
Duration
June - August 2023
June - August 2023
Duration
June - August 2023
Contribution
Contribution
Solve complex problems
I approached the complex challenge of designing editing features for text analytics by methodically analyzing the underlying problems, understanding user needs, and aligning with project goals.
Articulate design decision
I iterated on multiple prototype designs, making well-informed design decisions, and presented them to stakeholders, including the UX designers, PM, and development team.
Articulate design decision
I iterated on multiple prototype designs, making well-informed design decisions, and presented them to stakeholders, including the UX designers, PM, and development team.
Contribution
Solve complex problems
I approached the complex challenge of designing editing features for text analytics by methodically analyzing the underlying problems, understanding user needs, and aligning with project goals.
Articulate design decision
I iterated on multiple prototype designs, making well-informed design decisions, and presented them to stakeholders, including the UX designers, PM, and development team.
Challenge
Define Editing Feature
Before designing the editing feature, I realized the need to clearly define the scenarios that would require editing. I focused on identifying which specific functionalities, processes, and logical flows were necessary for effective editing, ensuring that our design would meet user needs.
Defining Editing Functionality
Before designing the editing feature, I realized the need to clearly define the scenarios that would require editing.
I focused on identifying which specific functionalities, processes, and logical flows were necessary for effective editing, ensuring that our design would meet user needs.
Challenge
Defining Editing Functionality
Before designing the editing feature, I realized the need to clearly define the scenarios that would require editing.
I focused on identifying which specific functionalities, processes, and logical flows were necessary for effective editing, ensuring that our design would meet user needs.
Project Background
Project Background
Project Background
What is Text Analytics?
What is Text Analytics?
What is Text Analytics?
Text analytics will show customers the topic feedback that clients have received from the surveys and perform an overall sentiment analysis of individual reviews.
Text analytics will show customers the topic feedback that clients have received from the surveys and perform an overall sentiment analysis of individual reviews.
Text analytics will show customers the topic feedback that clients have received from the surveys and perform an overall sentiment analysis of individual reviews.
Text analytics will show customers the topic feedback that clients have received from the surveys and perform an overall sentiment analysis of individual reviews.
Understand Feature value - why should we have this feature?
Understand Feature value - why should we have this feature?
Understand Feature value - why should we have this feature?
To recognize the feature value and users' needs, I initiated one-on-one meetings with the product manager and designers to recognize value and team goal.
Scheduling 1on1 meeting with product manager and designers to understand the value of building this feature
To recognize the feature value and users' needs, I initiated one-on-one meetings with the product manager and designers to recognize value and team goal.
Scheduling 1on1 meeting with product manager and designers to understand the value of building this feature
To recognize the feature value and users' needs, I initiated one-on-one meetings with the product manager and designers to recognize value and team goal.
Product need and value
Product need and value
Product need and value
On Market Side
There is a high demand from clients, and many competitors in the market also offer this feature.
Scheduling 1on1 meeting with product manager and designers to understand the value of building this feature
To recognize the feature value and users' needs, I initiated one-on-one meetings with the product manager and designers to recognize value and team goal.
Scheduling 1on1 meeting with product manager and designers to understand the value of building this feature
To recognize the feature value and users' needs, I initiated one-on-one meetings with the product manager and designers to recognize value and team goal.
On Demand Side (users' needs)
Data analysts can use the AI text analysis feature to view qualitative analysis results.
Customer success team can demo this feature to clients, enhancing the product’s selling points and market competitiveness.
Scheduling 1on1 meeting with product manager and designers to understand the value of building this feature
To recognize the feature value and users' needs, I initiated one-on-one meetings with the product manager and designers to recognize value and team goal.
Scheduling 1on1 meeting with product manager and designers to understand the value of building this feature
To recognize the feature value and users' needs, I initiated one-on-one meetings with the product manager and designers to recognize value and team goal.
Current problems
Current problems
Current problems
Identifying current problem: Incomplete Editing Workflow in AI Text Analysis Page
Identifying current problem: Incomplete Editing Workflow in AI Text Analysis Page
Identifying current problem: Incomplete Editing Workflow in AI Text Analysis Page
When I joined the team, the product already had an initial version of the AI text analysis interface. However, the interface was incomplete and required significant improvements to better support user needs.
When I joined the team, the product already had an initial version of the AI text analysis interface. However, the interface was incomplete and required significant improvements to better support user needs.
AI doesn't always get all results correctly. Humans need to modify errors to increase accuracy.
For example, it will mistakenly detect that a "long wait time" is positive, which we know isn't true, so we need to modify it as negative.
Also, there are examples where the AI needs to identify and classify topics accurately. For example, when customers say, "The staff was nice and friendly," we should add a "customer service" topic to catch better and understand these comments.
When I joined the team, the product already had an initial version of the AI text analysis interface. However, the interface was incomplete and required significant improvements to better support user needs.
Understand users and their pain points
Understand users and their pain points
Understand users and their pain points
Our users, Data analysts, frequently face issues with AI text analysis, including incorrect sentiments and missed topic labels. There is a strong need for them to correct these errors to enhance AI accuracy, but there is currently no interface available for visually editing these results.
Our users, Data analysts, frequently face issues with AI text analysis, including incorrect sentiments and missed topic labels. There is a strong need for them to correct these errors to enhance AI accuracy, but there is currently no interface available for visually editing these results.
AI doesn't always get all results correctly. Humans need to modify errors to increase accuracy.
For example, it will mistakenly detect that a "long wait time" is positive, which we know isn't true, so we need to modify it as negative.
Also, there are examples where the AI needs to identify and classify topics accurately. For example, when customers say, "The staff was nice and friendly," we should add a "customer service" topic to catch better and understand these comments.
Our users, Data analysts, frequently face issues with AI text analysis, including incorrect sentiments and missed topic labels. There is a strong need for them to correct these errors to enhance AI accuracy, but there is currently no interface available for visually editing these results.
Problems in AI technique
Problems in AI technique
Problems in AI technique
Two main issues
AI determines incorrect sentiment
AI fails to label topics
Two main issues
AI determines incorrect sentiment
AI fails to label topics
Two main issues
AI determines incorrect sentiment
AI fails to label topics
Recognizing complexity & acknowledgements
Recognizing complexity & acknowledgements
Recognizing complexity & acknowledgements
After understanding the users, current problem, and the necessity of the functionality, I began to consider how I could assist users in correcting AI recognition errors to improve accuracy.
After understanding the users, current problem, and the necessity of the functionality, I began to consider how I could assist users in correcting AI recognition errors to improve accuracy.
After understanding the users, current problem, and the necessity of the functionality, I began to consider how I could assist users in correcting AI recognition errors to improve accuracy.
After understanding the users, current problem, and the necessity of the functionality, I began to consider how I could assist users in correcting AI recognition errors to improve accuracy.
How Might We
How Might We
How Might We
How might we assist users in correcting AI recognition errors?
How might we assist users in correcting AI recognition errors?
How might we assist users in correcting AI recognition errors?
Project goal
Project goal
Project goal
"Design an editing feature that enables users to efficiently modify analysis results."
"Design an editing feature that enables users to efficiently modify analysis results."
"Design an editing feature that enables users to efficiently modify analysis results."
Prototype Design
Prototype Design
Prototype Design
Analyzing elements covered in text analytic page
Analyzing elements covered in text analytic page
Analyzing elements covered in text analytic page
Before designing, I need to clarify what elements will be included in the text analytics interface and what components will be displayed on this screen.
Before designing, I need to clarify what elements will be included in the text analytics interface and what components will be displayed on this screen.
Before designing, I need to clarify what elements will be included in the text analytics interface and what components will be displayed on this screen.
Before designing, I need to clarify what elements will be included in the text analytics interface and what components will be displayed on this screen.
For each individual customer feedback, there will be 3 sections,
1. Overall sentiment
2. Original feedback with full text
3. Topic sentiment
For each individual customer feedback, there will be 3 sections,
1. Overall sentiment
2. Original feedback with full text
3. Topic sentiment
For each individual customer feedback, there will be 3 sections,
1. Overall sentiment
2. Original feedback with full text
3. Topic sentiment
For each individual customer feedback, there will be 3 sections,
1. Overall sentiment
2. Original feedback with full text
3. Topic sentiment
Defining editing scenarios and features
Defining editing scenarios and features
Defining editing scenarios and features
Before designing the editing feature, I need to define the scenarios where editing will be necessary and specify the functionalities that the editing feature should include.
Before designing the editing feature, I need to define the scenarios where editing will be necessary and specify the functionalities that the editing feature should include.
Before designing the editing feature, I need to define the scenarios where editing will be necessary and specify the functionalities that the editing feature should include.
Before designing the editing feature, I need to define the scenarios where editing will be necessary and specify the functionalities that the editing feature should include.
Brainstorming editing scenarios with product team
Brainstorming editing scenarios with product team
Brainstorming editing scenarios with product team
After identifying users' needs, I focused on clarifying and exploring various editing use cases. To facilitate this, I invited product team members, including the product manager and UX designers, to brainstorm potential scenarios where users might require editing features.
After identifying users' needs, I focused on clarifying and exploring various editing use cases. To facilitate this, I invited product team members, including the product manager and UX designers, to brainstorm potential scenarios where users might require editing features.
After identifying users' needs, I focused on clarifying and exploring various editing use cases. To facilitate this, I invited product team members, including the product manager and UX designers, to brainstorm potential scenarios where users might require editing features.
After identifying users' needs, I focused on clarifying and exploring various editing use cases. To facilitate this, I invited product team members, including the product manager and UX designers, to brainstorm potential scenarios where users might require editing features.
Essential stage in defining editing feature flows
Essential stage in defining editing feature flows
Essential stage in defining editing feature flows
This stage was crucial as it helped differentiate between the editing flows for individual feedback and topic categories, clarifying the design logic and significantly saving time and effort in the design process.
This stage was crucial as it helped differentiate between the editing flows for individual feedback and topic categories, clarifying the design logic and significantly saving time and effort in the design process.
This stage was crucial as it helped differentiate between the editing flows for individual feedback and topic categories, clarifying the design logic and significantly saving time and effort in the design process.
This stage was crucial as it helped differentiate between the editing flows for individual feedback and topic categories, clarifying the design logic and significantly saving time and effort in the design process.
Challenges - clarify complex feature logic
Challenges -
clarify complex feature logic
Challenges - clarify complex feature logic
Articulating and Visualizing design logic
Articulating and Visualizing design logic
Articulating and Visualizing design logic
I developed a topic sentiment matching logic that visualizes the matching mechanism. While multiple categories can map to a single evaluation, each topic is assigned only one sentiment.
I developed a topic sentiment matching logic that visualizes the matching mechanism. While multiple categories can map to a single evaluation, each topic is assigned only one sentiment.
I developed a topic sentiment matching logic that visualizes the matching mechanism. While multiple categories can map to a single evaluation, each topic is assigned only one sentiment.
I developed a topic sentiment matching logic that visualizes the matching mechanism. While multiple categories can map to a single evaluation, each topic is assigned only one sentiment.
Design decision
Design decision
Design decision
Referencing to design system, choosing multi-select dropdown menu as our design solution
Referencing to design system, choosing multi-select dropdown menu as our design solution
Referencing to design system, choosing multi-select dropdown menu as our design solution
Considering the relationships of many-to-one and one-to-one logic, I decided to use a multi-select menu design for the editing individual feedback feature.
Considering the relationships of many-to-one and one-to-one logic, I decided to use a multi-select menu design for the editing individual feedback feature.
Considering the relationships of many-to-one and one-to-one logic, I decided to use a multi-select menu design for the editing individual feedback feature.
Final Prototype
Final Prototype
Final Prototype
Learn & Takeaway
Learn & Takeaway
Learn & Takeaway
Explain our own design decision
Explain our own design decision
Explain our own design decision
I critically evaluated each design decision by creating various prototypes learning to articulate and defend my ideas effectively. This invaluable experience enhanced my design process awareness and communication skills.
I critically evaluated each design decision by creating various prototypes learning to articulate and defend my ideas effectively. This invaluable experience enhanced my design process awareness and communication skills.
I critically evaluated each design decision by creating various prototypes learning to articulate and defend my ideas effectively. This invaluable experience enhanced my design process awareness and communication skills.
Consider the real-world constraints
Consider the real-world constraints
Consider the real-world constraints
Our illustrations aligned with our design system and the engineering team's coding environment to ensure consistency and streamline development, navigating constraints beyond academic design practices.
Our illustrations aligned with our design system and the engineering team's coding environment to ensure consistency and streamline development, navigating constraints beyond academic design practices.
