By Michael Hoefer

This post will serve as a description of my self-self design study, intended to craft a useful interface to better understand my relationships, inasmuch as they are mediated through digitally traceable messages. This post is structured around the nine main points of an assignment for the information visualization class.

1. Learning - Target Problem

The primary problem I'm seeking to address is that of digital disorganization — we continually create a stream of messages that are stored on the platforms we use to communicate. iMessage, Facebook messenger, Google Voice, WhatsApp, and others provide the medium for digital connection, and these platforms utilize this data for commercial purposes, without providing a synthetic or informatics benefit to the user generating this data.

The goal of this design study is to create interfaces that are useful for making sense of the digital messages one sends and receives. The goal is that by mining social data, we can discover insights that help individuals better understand and manage their relationships, and potentially uncover false mental models disconnected from reality. Interpersonal perception is known to be a mediating factor in the success and failure of important life relationships. By providing an objective analysis of interpersonal data streams, our natural, automatic perceptions of relationships can be augmented by objective data, and provide a new lens for reflecting on and understanding these relationships.

Specifically, for this project, I will be focusing on visualizing social reciprocity, as this is one potential area for modification in social behavior. Am I neglecting (or being neglected in) certain relationships, as evidenced by digital messages?

2. Winnowing - Description of the Data

The data for this study will come from data dumps of my Google Voice data since 2019 and Facebook Messenger data since 2011. Specifically, the data will be pulled from a SQLite database that contains structured messaging data, including the sender, recipient, timestamp, platform, and message content. This data represents the bulk of my digital messages, especially since I switched from Sprint to Google Voice, and all my text message data is now available with Google Takeout data exports. Combined with Facebook messenger, the corpus is fairly well representative of my digital social life. In a previous assignment, I ran all 68,000 messages through a sentiment analyzer, so I will also include the AI generated sentiment of each message in the data to be explored.

If we assume that the digital traces of my relationships are even a partial proxy for the real relationship that generates the messages, then this data will be useful for helping me to understand my relationships. Of course the digital trace is not a perfect representation, as it fails to capture the in-person interactions I have. However, as I have spent the past few years moving around quite a bit, many of my relationships have become long distance, and take place primarily through text message exchanges. In addition, the digital text frequency is likely correlated with the in-person hang-out frequency, so visualizing the messaging data may prove fruitful regardless.

The data will need to be segmented by contact, such that the time-series stream of messages can be analyzed and visualized on a per-contact basis. I will need to develop metrics for social reciprocity that are derived from these data streams.

3. Discover - Identifying Tasks

Why is a task pursued? (goal)

How is a task conducted? (means)

What does a task seek to learn about the data? (characteristics)

Where does the task operate? (target data)

When is the task performed? (workflow)