Potential Innovations (New Ideas/Trends) Detection in Information Network

This paper considers the task of semanticinnovation (new idea or trend) detection possibility incorporate information field applying the analysis of dynamicchanges in enterprise information flows. Here we consider suchhypothesis that the sign of innovation appearance is thedramatic increase of new terminology employment (that hasoccurred rarely or has not been used at all), or the appearanceof previously not occurred combinations of old “terms” (words)and “N-grams” (sequences of words). Presented results of thisexperiment are based on ENRON, TREC2005 databases, emailexchange among employees of Russian company. We haveoverseen conditions under which the innovation detection isstatistically significant and made the conclusion about thefuture of above-listed hypothesis of innovation detection.