BMBF funded project Dynamic text based output indicators as base for a new innovation metric (2021-2024)
The BMBF-funded DynTOBI project (Dynamic Text Data-based Output Indicators as Base of a New Innovation Metric) is the follow-up project to TOBI, in which innovation indicators were generated through the automated and regular evaluation of digital text mass data. The TOBI project was motivated by the lack of timeliness, granularity and high data collection costs of traditional innovation indicators from surveys. The analysis of online texts from news tickers made it possible to counter these disadvantages. Among other things, the project team was able to show that valide diffusion indicators can be generated automatically from news ticker data by using a specially developed topic model (PVTM), whereby in some cases a lead could be determined compared to comparable indicators. The topic model is available as open-source software.
The DynTOBI project, which continues to be realised in a joint project with ZEW Mannheim, is divided into two sub-projects:
In the JLU sub-project, on the one hand, the approaches based on news ticker data from the TOBI project are being further developed with a focus on mapping the dynamics in innovation events and, on the other hand, preparatory steps are being taken for a "regular operation" for the continuous collection of the resulting innovation indicators.
As a complementary methodological approach, the time courses of the innovation fields endogenously determined from news articles generated via topic modelling are summarised in clusters using Functional Data Analysis (FDA). From the clusters, prototypes are obtained to describe the diffusion process and to derive success forecasts for individual innovations. Core characteristics for each cluster will be identified based on the underlying weights and by means of named entity recognition.
As a further step, the Newsticker tool is to be expanded to collect relevant online news in greater breadth in order to avoid bias through the use of a few fixed sources. This also requires methods for automated detection of changes in access to sources or their formats in order to adapt the tools promptly to changing requirements.
The second sub-project at ZEW comprises the further development of the methods for measuring innovation dynamics developed in the TOBI project and the preparation of a "regular operation" for the continuous collection and provision of standardised, web-based innovation indicators.