The importance of analyzing R&D environment changes and forecasting future technologies for supporting policy decision and efficient resource distribution has been increasingly recognized. Many futurists are forecasting future technology based on Delphi study, brainstorming, expert survey, trend analysis, data mining, etc. However, these processes still need to be formalized. In this paper, we introduce the NEST (New & Emerging Signals of Trends) model, which is a systematic collective intelligence model for collecting information from expert network worldwide and detecting weak signals of emerging technologies, developed by KISTI. The most outstanding feature of NEST model is that it is based on both quantitative and qualitative methods. In the stages of quantitative methods, NEST performs clustering, pattern recognition, scientometrics, and cross impact analysis. In the stages of qualitative methods, NEST conducts environmental scanning, brainstorming, and Delphi study. For illustration purpose, a result of experiment for detecting weak signals of emerging technologies is presented.
This paper introduces a collaborative environmental analysis
model, NEST, for forecasting future trends useful to support
groups‟ or nations‟ decision making and R&D strategy
establishment. This model is designed to find weak signals of
future trends. Weak signals are events, accidents, or strange
issues that are thought to be the beginning of future changes [
NEST model utilizes knowledge of a group of experts from various fields over the world and uses interactive feature of web 2.0 to communicate and deduce new refined knowledge from the shared knowledge.
In the next section, the result of literature review on related
research is presented. In Section 3, a detailed description of
KISTI [
„Weak signal‟ is a small sign in present which has potential of
significant changes in the future. Environmental scanning is a
process of collecting and analyzing the environment information
of an organization or nation to support its decision making. Our
NEST model performs environmental scanning using the GTB,
Global Trends Briefing [
Hot topics, or trends, are detected by grouping documents into concepts, based on a co-word or co-citation analysis.
The NEST consists of both quantitative analysis stages and qualitative analysis stages. The Global Monitoring, an environmental scanning, in Step 1 is the first filtering of NEST process. This process is conducted in the manner of qualitative analysis. In Step 2, the second filtering is performed on the collected information, both in qualitative and quantitative manner, by information analysts in KISTI based on its significance. In Step 3, various quantitative data analysis techniques, such as clustering, pattern recognition, regression, anomaly detection, etc. are used, to detect weak signals of trends, patterns and structures in the information. In Step 4, also a quantitative analysis step, experts detect upward trends using the weak signal tracking board, which is based on cross impact analysis model developed by KISTI.
NEST-Clipping is second filtering procedure on the monitored information performed by information analysts based on its significance.
NEST‟s upward trend detection process is an application of autoregression based extrapolation model. A regular monitoring framework “Study-Watch board”, shown in Figure 3, is devised to detect trends and issues.
In this step, information analysts analyze GTB and NEST-Clip to find the candidates of weak signal. An evaluation index for measuring the strength of impact is defined by the information analysts. Table 1 shows several examples of the index. -staSgtaesg)eofInRd&uDstrilaifliezactyiocnle: Planning Science Development (6 sub- Extension of social system: case regulation treaty rule law - Increase of occurrence frequency: very weak weak neutral
138 thousands of environmental scanning data collected by Global Monitoring Network, which has been operated for 10 years and archived in GTB website, is used as the source data. 57 weak signal candidates were selected after NEST-Signal detection process, the step 3 in Figure 1, and prepared for online Delphi study. Table 2 presents the summary of the 57 weak signal candidates.