Brainstorming
To improve the efficiency and experience of literature reviews for second language users, I conducted brainstorming and listed possible solutions to this problem. Also, analyzing opportunities for each function, I summarized the following solutions:
User Research
We did secondary research, quantitive research, and users interviews, aiming to have a deep insights about the users’ needs and expectations.
Interview Quotes
Interview users about what they meet when doing literature reviews.
Doctor of Architecture
Q1: Describe your process of writing literature review.
Q2: What problems have you encountered in the process of finding and writing documents?
Q3: Have you used AI for writing and what problems do you think they have?
Phoebe,Shanghai
“The problem of literature arrangement”
AI lacks of search channels for second language literature; It is difficult to match keywords with literature content; How to find the desired documents accurately in the second language;There are problems in the search and reading of literature in non-native and second languages.
“The problem of language expression”
AI speaking is sometimes formulaic, repeated back and forth, do not involve details, anyway, the literature review must be detailed, you want to write aspects to cover all, and to explain the value of your review.
Master of Computer Science
Aaron,Shenzhen
“The problem of finding the research gaps and drawing a conclusion”
AI writing is generally easy to identify, because the contents are vague, they are general descriptions, seldom have their own opinions, and the accuracy of summarizing literature content is difficult to guarantee.
Master of Psychology
Rebecca,Exeter
Quantitive Reserch
Respondents have bachelor's degree or above, and the language level is advanced.
Respondents work in science and engineering, humanities and social sciences.
Most majorities of respondents have a second language reading and writing need.
2.5%
2.5%
7.5%
37.5%
50%
57.5%
2.5%
2.5%
37.5%
57.5%
32.5%
25%
17.5%
What are the most difficult stages in your paper writing in a second language?
Most respondents find writing literature reviews in a second language to be the most challenging task. They also struggle with collecting and reading literature, writing statements, and revising papers.
What specific parts of the writing for the paper you find it difficult? (select three)
Most respondents think that the literature review is the most difficult part to write in an academic paper. They need extra writing assistance like translations and grammar correction.
This projects designed an AI-driven assistant to streamline the English literature review process, particularly aiding second language learners. It simplifies complex academic tasks, making literature reviews more accessible and comprehensive through AI-generated content and visual tools.
Overview
My Role
UX Research
UI / UX Design
8 weeks
Aug 2023
Figma,
Adobe Illustrator
ChatGPT
Xinran Cheng
Huilin Zhang
Yufan Xu
Duration
Tools
Team
Problem Statement
Solutions
Second language learners often struggle with literature reviews due to linguistic barriers and the complexity of academic English. These challenges make it difficult for them to fully grasp research achievements and produce thorough literature reviews.
AIterature addresses these challenges by breaking down the literature review process into four stages: searching, reading, summarizing, and expression. For each process:
Provides a comprehensive literature library and search engine, and automatically sort out papers.
Enhances comprehension of complex texts through advanced AI translation.
Identifies research gaps using AI-driven classification and visual relationship diagrams.
Supports summary writing, ensuring conclusions align with research themes.
Insights Summary
I organized four main pain points and provide solutions for each one, forming a complete flow of literature review. So I can narrow down the design concept and propose the design question.
Pain Point
01
Searching
Hard to arrange the literature
02
Reading
Lack of the vocabulary and grammar knowledge
03
Summarizing
Hard to find the research gap
04
Writing
Hard to draw a conlusion fit the theme of the paper
Solutions
Automaticlly sort out papers
An integrated resource library screen the quality and citation rate of literature in different subdivisions.
AI translation and reading assistance
More accurate AI translation
analysis of long and difficult sentences
Visual literature relationship
AI classification of literature, visual literature relationship diagram and statistical analysis.
AI inspirations and write scripts
AI chatting for writing inspirations, such as generating ideations, paper outline, key paragraphs, and scripts.
Competitive Analysis
To understand the applications of generative AI tools, I researched competitive products and listed functions of different products.
I found that current AIGC products do not exist to cover the whole process of literature review. If users want to complete a literature review with the assistance of AI, you often need to use multiple sites in collaboration.
Design Question
How might we make the literature review process effective and enjoyable for second language users?
Persona
User Journey Map
I listed five stages containing the process from searching and reading to summarizing and writing, and used emojis to express users’ feeling during different processes. Finally, aiming at these pain points reflecting users’ needs, I analyzed opportunities for my product.
User flow
Based on the brainstorming results, I outlined this user flow, a comprehensive system for academic research, featuring modules for paper search and review, reading assistance, writing support, and language translation, all integrated within a user-centric website platform.
User Testing and Feedback
Iteration 1: Data Visualization and Analysis
Users were unable to interactive with the data visualization tools.
Added interactive chat box with the AI assistant.
Wireframing
Paper Prototype
To visualize the results of the user flow, I began with paper prototypes. These low-fidelity prototypes allow me to think quickly and iterate designs efficiently, facilitating rapid changes to the feedback of user testing.
Users expected features supporting grammar checks and style suggestions could be realized.
Added grammar checks and style suggestions.
Iteration 2: Writing and Editing Assistance
The aesthetic could be warmed up slightly to make long periods of work more visually engaging.
Change the overall fonts, style and UI color system.
Iteration 3: Overall Design Aesthetic
Final Design
Home page design aims to integrate main functions and navigate users to paper searches, in-depth reading and analysis, and also writing aid pages.
Search page
Home Page
Reading page
Writing page
Footer
Search
Page
Reading Page
Various display modes are available for papers with and without covers, providing a clear categorization that helps users easily identify recently read literature.
Classfied folders
Summarizing Page
The visual literature diagram illustrates the interconnectedness of collected papers and automatically tags them, offering users insightful inspirations for their research. Also, an interactive AI assistant is on hand to spark further creativity and ideas.
In the search results interface, users apply filters based on their own needs, such as the citations, imapct factors, language, date, relevance, subject, etc., to sift through literature and swiftly obtain the results they require.
Results filter
In the search results interface, users apply filters based on their own needs, such as the citations, imapct factors, language, date, relevance, subject, etc., to sift through literature and swiftly obtain the results they require.
In the reference sorting page, users can select arrangement basis, such as the citations, relevance, years, impact factors, etc., analyzing the research trend and focusing on key literature, and comparing similar literatures effectively.
In the reference sorting page, users can select arrangement basis, such as the citations, relevance, years, impact factors, etc., analyzing the research trend and focusing on key literature, and comparing similar literatures effectively.
Sort reference
AI suggests prompt words to users, who can train it to match their writing style.
AI prompt
Personal training
Writing Page
Users can directly edit AI-generated inspiration snippets for personalized modifications.
Clicking on a word or sentence on the page allows for translation, the grammar and writing style modification.
Auto translation
AI-assited writing
Relationship diagram
My Takeaways
Design research is crucial for defining the narrow concept and solving real problems. Actually, academic writing is a general concept. By talking to the users, I can narrow down the scope of what I am going to build first. So literature review is defined as our product scope.
Conduct the usability testing as soon as possible, let the users tell which idea works and which doesn't. When our tean started wireframing, we had two different color systems for the UI design. We put both of them in front of the users amd got feedback for the final design. This is a effective and high-efficient method to help us avoid heasitation.
What I could have done better. AIGC is an emerging tool that directs a two-way street chat between humans and AI. UX designs and UI layout need more human-centerend and personalized design. This means I should create more customized experiences that adapt to individual user needs, preferences, and behaviors, and test prototypes before finalizing a product. For example, more iterations for AI-powered chatbot.
