An interactive map of Human Factors engineering concepts
The book “Designing for People: An Introduction to Human Factors Engineering” contains many Human Factors engineering concepts from the broad areas of:
- Design and evaluation techniques
- Cognitive characteristics of people
- Physical characteristics of people
- Organizational and social characteristics
The interactive visualization below provides a way to explore some of these concepts. Generally, points in this figure that are close to each other are conceptually similar, and hovering over a point shows the lines from the book that are closest to the concept.
This visualization assumes text can be treated as data. Here we consider the contents of “Designing for People: An Introduction to Human Factors Engineering” as data. We processed the text from each chapter to remove punctuation, numbers, and common terms. We then analyzed the resulting words with a word embedding technique. This technique calculates the position of each word in a 100-dimensional space using term frequency and co-occurence weighted by proximity to nearby words in sentences (https://github.com/bmschmidt/wordVectors).
t-SNE collapses the 100 dimensions to 2 dimensions in a way that keeps the close words close to each other, and so maintains local structure. t-SNE is similar to a principle compoents analysis (PCA), but PCA keeps distant words distant and so obscures local structure (For a demonstration applied to time series data). Each color corresponds to a different section of the book and so you might expect that terms from the same section would be grouped together. You might also expect that terms near each other would also have a clear relationship. In many cases this is true, but not always. The computer sometimes finds spurious connections and so a good exercise is to identify those failings. For example, “load” is used in the context of cognitive effort, but also in terms of physical effort. Because the t-SNE transformation collapse a 100-dimensional space to a 2-dimensional space there are invitable distortions and so it is not a perfect map of Human Factors engineering concepts.
