Mapping the Landscape of Patterns Across Domains – Vol. 2: Linguistic and Semantic Analysis
This is the second set of results from the Survey Mapping the Landscape of Patterns across Domains initiated by the BCSSS Research Group Systems Science and Pattern Literacy in early 2018.
Our intention here was to extract linguistic and semantic regularities from the survey answers where respondents defined patterns in a few sentences.
We assumed that we would find frequently used general terms and associations between terms. We therefore sought to extract couplings between linguistic and semantic structures which could be generalized and re-used for the identification of patterns and the development of patterns knowledge. Such „knowledge units“ would support the development of a formal language and a diagrammatic presentation of patterns.
We applied manual analysis, tree diagrams and software tools from the open source platform Voyant Tools, such as Cirrus, which displays word frequencies as wordclouds, as well as tailed-Stochastic Neighbor Embedding (t-SNE) and Principal Component Analysis (PCA), which help reduce dimensions of complexity and capture the essential structure of the data.
We drilled down in the depth of the corpus of answers to extract regularities from varied and ambivalent pattern definitions.
t-SNE and PCA tools supported the identification of general manifestations and attributes of patterns, as well as associations and consistent combinations of terms. Some persistent associations revealed mature concepts behind them.
We also observed that the application of simple software tools like Cirrus constrained the interpretation of definitions provided in different contexts. More sophisticated linguistic and semantic structures require more sophisticated tools and a multi-dimensional approach.
We will continue to develop a re-usable toolkit for extraction and detection of patterns. These will include tools for text analysis to complement those presented in this report.
A sample of 140 definitions is not sufficient to make all embracing conclusions about existing concepts and perspectives of seeing patterns. Yet it can generate some insights and generalizations to be validated and fine-tuned with further experience and research. After having worked though multiplicity of definitions, we can see that patterns are a kind of a homonym, used to define multiple things.
Pulling multiplicity of perspectives and dimensions together, patterns seem to possess a sufficient variety to absorb complexity, embrace dualities, bridge the gaps and transcend boundaries. If we want to detect something as complex as a pattern, then we also need something that embraces all the regularities we have identified, and remains dynamic and flexible – something what we can call a „ pattern of knowledge“ about patterns.
We will continue our way following the mystery of patterns. The findings from the survey inspire our curiosity and a future research – we are moving forwards and keeping you informed. Our pattern research is research-in-progress and interpretations we make are dynamic and ready for change.
The details of this analysis can be found below. Our interpretation of the survey data is not final and we welcome proposals and ideas concerning objectives, interpretations and application of tools.
First insights of the survey can be seen in Mapping the Landscape of Patterns across Domains – Vol 1: Overview of Survey Results.
Maria Lenzi and Helene Finidori
Abonnieren Sie unseren Systems Newsletter
- Keynote auf Einladung von CIRCULAR4.0 “Der European Green Deal – Wirtschaftspotenziale für KMUs”
- Keynote auf Einladung der Energie Steiermark „Potenziale der Kreislaufwirtschaft“
- Experten-Interview mit Circular Business Academy (CBA), Slowenien zum Innovationspotenzial der Kreislaufwirtschaft für KMU
- UNIDO: Globale Konsultationen zur Kreislaufwirtschaft
- Mission Innovation Austria 2021: Eine Plattform für alle Treiber und Gestalter von Innovation und dem Energiesystem der Zukunft