Peaks, Slopes, Canyons, Plateaus: Identifying Technology Trends throughout the Life Cycle

Efimenko I., Vladimir Khoroshevsky, Ena O.

Automated identification of new technology trends (trend monitoring, trend hunting, trend watch) is among the hot topics in technology management. Despite many beneficial results in this field, almost no solutions allow users to escape from getting too general or garbage results which make it impossible to identify trends at the stage of weak signals. Lack of attention is paid to automated labeling and merging (for the ‘same’ trends).

Our approach aimed at overcoming such drawbacks is based on the ‘BlackBox’ principle. The concept of a technology trend (TT) is characterized by a complex nature, low formalization level, blurred boundaries, and high degree of domain dependency leading to the need for expert knowledge. For all that, ‘Big Data’ in IT and ‘Genome Editing’ in Healthcare should have some similar features which actually allow us to name both phenomena ‘a TT’. This leads us to an idea of hunting for domain independent ‘external signs’ (trend indicators) while letting a TT itself stay a black box for an observer.

We employ Gartner’s Hype Cycle in our methodology. We build an elaborate ontology of a TT and a system of indicators of TTs ‘presence’ in documents of various genres. The indicators are interrelated with the ontology through linguistic and extra linguistic markers. Both markers and text genres are mapped onto the phases of a technology life cycle. The ontology-driven information extraction (IE) is carried out. 

Keywords: ontology, weak signals, Knowledge Extraction, trend watch, New technology trend monitoring, hybrid approach, Gartner Hype Cycle, Technology Life Cycle, trend hunting, ontology-driven information extraction, black box principle

In book

Proceedings of the 4th Global TechMining Conference (Leiden, Netherlands)

Leiden: [б.и.], 2014.