Seedlings_ - Qianxun Chen
Seedlings_ is an experimental web interface to explore the artistic value of algorithmic correlations of words in data driven text analysis. Words are planted as seeds and grow into different ‘plants’ with the help of Datamuse API, a data-driven word-finding engine. It is at once an ambient piece in which words and concepts are dislocated and recontextualized constantly, and a playground for the user to create linguistic immigrants and textual nomads.
In Seedlings_, a word can be transplanted into a new context, following pre-coded generative rules that are bundled under the names of plants (ginkgo, dandelion, pine, bamboo, ivy…). These generative rules consist of a series of word-finding queries to the Datamuse API such as: words with a similar meaning, adjectives that are used to describe a noun, words that start and end with specific letters. They are then grouped in modules to represent the visual structure of the corresponding plant and can be constrained with a theme word. A new plant can be grafted on top of the previous plant by switching to a new starting point from the latest generative result. Other than words in monospace font, lines of dashes are the only other visual element in the piece, expressing the minimalist aesthetics in these potentially infinite two-dimensional linguistic beings.
In distributional semantics, words that are used and occur in the same contexts tend to have similar meanings. Based on this hypothesis, words are processed by n-grams, represented and manipulated as vectors in contemporary Machine Learning. With the help of algorithms, we can now identify kinships between words (through similarity or frequent consecutive use) in milliseconds. Seedlings_ reconfigures existing technologies and services in Natural Language Processing as the virtual soil to generate alternative linguistic plants: it seeks new poetic combination of words by encouraging unusual flow of words and concepts.