Analyzing website content for improved R&T collaboration planning
Abstract
A well-known problem in research and technology (R&T) planning is the selection of suited R&T collaboration partners. We investigate the use of textual information from the website content of possible collaboration candidates to identify their suitability. This improves the selection of collaboration partners and it enables a successful processing of R&T-projects. In a case study 'defense R&T', organizations and companies that have proven their suitability as collaboration partner in former R&T projects are selected (positive examples) as well as organizations and companies that have not. Latent semantic indexing with singular value decomposition and logistic regression modeling is used to identify semantic textual patterns from their websites' content. As a result of prediction modeling, some of these textual patterns are successful in predicting new organizations or companies as (un-) suited R&T collaboration partners. These results support the acquisition of new coll aboration partners and thus, they are valuable for the planning of R&T.
