Process Documentation Part. 2

Thoughts, reflection, and key findings in my thesis journey

1/16 Review Feedbacks — Initial concept

Coming back from winter break, Peter and I reviewed feedbacks from faculty reviewers at my mid-point presentations. There were two main concerns.

1. Broad design scope
2. Ambiguity of some vocabularies
3. Lack of preliminary concept

For the first concern, we concluded that my scope is fine as long as it is confined within home furnishing shopping. However, I also agree with the second concern. Particularly, “inspiration” and “creativity” are so abstract that they can be interpreted in many different ways. Hence they need to be clearly articulated what exactly they mean by in the context of my project.

In terms of vocabularies, I tried to define what it refers to in my thesis as follow:

• Inspiration: In my project, inspiration refers to a home furnishing product, a setting(display in store and friends’s house), or pictures of those that trigger an imagination of their desired home. It is a moment thinking “Oh! I’d like to have it/decorate like that so that my home will …” Customers come across inspiration not only in stores but also in their daily lives, such as SNS feed and friends’ house.
• Creativity: Creativity in my project is an ability of an individual to create their home environment by bringing inspiration to life, aligning with their core values. Contribution ranges from DIY craft to designing to leading purchasing decisions.

For the last concern, Peter suggested to turn my direction into a model and test it out with people. With brief ideation, we came up with a simple model of a furniture shopping platform which consists of three steps;

1. Identifying personal values
2. recommending relevant products to the identified value, what we called a mood board
3. Simulate a desired setting

Basically, it is a feature of e-commerce providing personalized suggestions based on customer’s personal values instead of preference like Netflix or apple music.

Apple music onboarding for personalized suggestion.

Quick Ideation & Sketch of The Model

The first step is understanding and distill a user’s core value. There would be pictures already tagged with relevant values. A user will be prompted to selects a set of pictures of home that they prefer or think ideal. Or, the other way around, a user select some picture he/she likes and then labels what value they can get out of from the pictures. This step will reveals a most dominant value of the participant.

Second step is pulling up specific products (or a set of product) which resonate with the drawn values from an inventory. For example, if this model is applied to Ikea, it would suggest a set of Ikea product that matches a user’s value on sustainability.

Finally, the user can create their own space by playing with the suggested products. The model might incorporate assistive features such as simulation and offering reference in order to facilitate this step.

This model could be tested out with potential users as an exercise such as workshop.

1/21 Develop Initial Concept

While I’m trying to flesh out the initial concept, I came up with fundamental questions regarding this concept.

1. What are personal values? Are they different from values of institutions or society? Is it there a competed taxonomy of personal value? How to make a comprehensive but compact personal value taxonomy for my project?
2. How can I determine the right degree of abstraction? There are a number of different values in different degrees of abstraction. For example, fun and pleasure from beauty are more concrete compared to hedonism, which the two values can fall under this overarching concept. Also, there are a various way to frame a value. Saving money can be framed as a better-value or thrift. Easy-to-use can be framed as reliability or functionality.

These questions require extra research (mostly secondary research or expert interview). Also, I found some problems of the initial model and pondered about them.

1. Even if I leverage the power of AI to auto-label values to an image, it is highly unlikely to extract intrinsic value by analyzing some picture. There is deep learning technology that allows machine understand objects, activities, and emotions. However, values are such a high level concept latent deep in mind. Even humans would not be able to conclude certain values of which an image reminds to other without interviewing and listening to underlying thoughts. This is because interpretation of a picture vary from an individual to individuals. Even if it is statistically possible, giving the best guess as one’s core value is a risk. It would immediately lose trust.
2. If I let users label what the image remind of, articulating value would be hard for the users because there are multiple ways of compressing a nebulous feeling into a clear word. The degree of abstract would be various. For example, they might say “I like this interior design (of the picture) because it uses lots of raw materials. Feel like being nature and care about it.” It might hard for them to boil down this idea into peace in the nature, symbiosis, sustainability, or other precise words for values. In fact it is just my assumption. Maybe it might possible. However, this approach is more promising than the AI approach.
3. On that, giving assistive hints may be a way for them to articulate their feeling and thoughts. It seems better rather than giving them an seeming answer. Merely providing statistical result of one’s value wound not be helpful as much as helping them figure out their own core value through self-reflection and analysis by themselves.

1/23

Peter agreed on my approach, an assistant tool for self exploration.

1. How to induce latent core values of an individual?

Let’s customers to be a designer.

Moodboard

For overall process,
1)Start writing from presentation. Think about chapter.For user test or workshop,
1) -> If need enact a technology, think about Wizard of OZ. Let people talk -> collect specimens raw data -> make statistical algorithm
2) Go back to interview and see how they got to the value.Regarding value,
1) Nathan shedroff Making Meaning 15 core meanings -> map them onto vlaue pyramid Regarding general question in thesis process,
1) Is this concept a design concept or a programmer-> Material that I’m dealing with here is AI and what i’m doing is computational thinking. -> design a service through AI with computational thinking

A mood board for interior project. Image from https://www.fohlio.com/blog/mood-board-for-interior-design-2017/

1/27 Personal Values

• Three level of value: global / domain specific / decision specific

1/30 Personalization scenarios & sketch

4 models

2/1 Value Mapping Exercises 1

- Start from small exercise.
- Utilize what I have, like value sets & attributes-end protocol

2/13 Value Mapping Exercises 2

2/17 Interior Design Styles & Similar Services

The Most Popular Interior Design Styles in the US | Joybird

https://rocheledecorating.com.au/14-most-popular-interior-design-styles-explained/

https://www.houzz.in/ideabooks/76918143/list/10-interior-design-styles-explained

Havenly: How This Denver-Based Startup Built An Interior Designer Marketplace

Online Interior Design & Decorating Services | Havenly

modsy

2/19 Personalized Recommenders in ML realm

Artwork Personalization at Netflix

Applying Spotify’s Recommendation Experience to New Markets: a Promising Model?

Recommendation System Algorithms

2/19 Personalized Recommenders

The final model & initial sketch

3/16 Test and Evaluation