A category of product is not objective or a priori existence. In other words, a boundary or members of category is not determined in advance. In researches, there are some approaches to concept of category. To classify concepts, one is static approach, another is dynamic approach. In static approach, a category is regarded as bundles of properties. On the other hand, in dynamic approach, a category is constructed impromptu in decision making. Both of these researches, a concept of category is the point at issue because of this concept is important theme as consumer's decision making in consumer behavior researches. In detail, consumers are affected categories of products as consideration set in decision making processes. In this research,we consider category of concepts in innovation diffusion processes. In innovation diffusion, meanings of product categories transit one after another. This article makes this transition visible with using of text-mining. There are some advantages in using text-mining in this research. First, a text-mining tool allows us to review all terms that describe products and to examine the referential relationship among those terms or the co-occurrence relations among the terms. By abstracting the semantic dimension of all terms, we can avoid the problem of whether the terms used by parties are comprehended by observers in the same way. Second, a text-mining tool is able to identify the commonality of co-occurrence relations among all terms because such a tool allows massive amounts of data to be identified and quantified. Third, a text mining tool can uncover the commonality of co-occurrence relations between more than two terms. Because a greater number of collocated terms imply a narrower interpretation, it is more likely that parties and observers will share technology values. In concrete, we use co-occurrence network analysis. Co-occurrence network is described based on term's co-occurrence relationships. In short, this analysis is a summary of writings (in this article, electric words of mouth which is written at kakaku. com). Then we can describe clusters on co-occurrence networks. We regard this clusters on co-occurrence networks as levels of meaning. Level of meaning affects consumer's cognition and then consumer's decision making processes. Using this approach, a manufacturer can perform and make an incremental innovation ahead of others.