Paper Type
Complete Research Paper
Description
In today´s world, organizations conduct technology assessment (TAS) prior to decision making about investments in existing, emerging, and hot technologies to avoid costly mistakes and survive in the hyper-competitive business environment. Relying on web search engines in looking for relevant information for TAS processes, decision makers face abundant unstructured information that limit their ability to assess technologies within a reasonable time frame. Thus the following qustion arises: how to extract valuable TAS knowledge from a diverse corpus of textual data on the web? To cope with this qustion, this paper presents a web-based model and tool for knowledge mapping. The proposed knowledge maps are constructed on the basis of a novel method of co-word analysis, based on webometric web counts and a temporal trend detection algorithm which employs the vector space model (VSM). The approach is demonstrated and validated for a spectrum of information technologies. Results show that the research model assessments are highly correlated with subjective expert (n=136) assessment (r > 0.91), and with predictive validity valu above 85%. Thus, it seems safe to assume that this work can probably be generalized to other domains. The model contribution is emphasized by the current growing attention to the big-data phenomenon.
TEXT MINING AND TEMPORAL TREND DETECTION ON THE INTERNET FOR TECHNOLOGY ASSESSMENT: MODEL AND TOOL
In today´s world, organizations conduct technology assessment (TAS) prior to decision making about investments in existing, emerging, and hot technologies to avoid costly mistakes and survive in the hyper-competitive business environment. Relying on web search engines in looking for relevant information for TAS processes, decision makers face abundant unstructured information that limit their ability to assess technologies within a reasonable time frame. Thus the following qustion arises: how to extract valuable TAS knowledge from a diverse corpus of textual data on the web? To cope with this qustion, this paper presents a web-based model and tool for knowledge mapping. The proposed knowledge maps are constructed on the basis of a novel method of co-word analysis, based on webometric web counts and a temporal trend detection algorithm which employs the vector space model (VSM). The approach is demonstrated and validated for a spectrum of information technologies. Results show that the research model assessments are highly correlated with subjective expert (n=136) assessment (r > 0.91), and with predictive validity valu above 85%. Thus, it seems safe to assume that this work can probably be generalized to other domains. The model contribution is emphasized by the current growing attention to the big-data phenomenon.