Optimizing Data Delivery: Insights from User Preferences on Visuals, Tables, and Text

TL;DR

The potential use of LLMs to replicate user preference data is demonstrated which has major implications for future user modeling and personalization research.

Abstract

In this work, we research user preferences to see a chart, table, or text given a question asked by the user. This enables us to understand when it is best to show a chart, table, or text to the user for the specific question. For this, we conduct a user study where users are shown a question and asked what they would prefer to see and used the data to establish that a user's personal traits does influence the data outputs that they prefer. Understanding how user characteristics impact a user's preferences is critical to creating data tools with a better user experience. Additionally, we investigate to what degree an LLM can be used to replicate a user's preference with and without user preference data. Overall, these findings have significant implications pertaining to the development of data tools and the replication of human preferences using LLMs. Furthermore, this work demonstrates the potential use of LLMs to replicate user preference data which has major implications for future user modeling and personalization research.

Figures (29)

• Figure 1: When answering the user survey, MTurkers were shown this figure as an example of what the data outputs could potentially look like. The leftmost example is what the text answer would look like, the middle is the answer in the form of a table, and the right is the answer in the form of a chart. Then they were asked, "Given a data analysis question, is it most useful to show the user text, data table, or chart?"
• Figure 2: For each question, we aggregate all the user preferences, and derive a distribution, which is shown above (sorted by the value for chart, which is why we see a nice curve for the probability of chart). Notably, we see that as the probability that a user prefers a chart increases, the probability a user prefers text or table decreases.
• Figure 3: User preference by Data Visualization Experience: This shows the data preferences of respondents based on their data vis experience, specifically comparing charts, tables, and text outputs.
• Figure 4: User preference by Data Analysis Experience: This shows the data preferences of respondents based on their data analysis experience, specifically comparing charts, tables, and text outputs.
• Figure 5: Comparison of the users' age to their preference in terms of whether they prefer the answer to be shown to them as a chart, table, or text.