Co-link analysis is an effective analytical method in webometrics; however it is susceptible to the restrictions of the network search engines. Prior studies have showed that Web keyword analysis could be an effective alternative to co-link analysis in many industrial fields. On this basis, this study continued to explore the relationship between Web co-word analysis, co-assignee analysis, and co-citation analysis in patent bibliometrics. This study searched patent documents about the fuel cell technology in Derwent patent database, selected 30 companies that had the largest number of patents and co-citation, and developed a co-assignee matrix and a co-citation matrix respectively. After that, the selected companies' names were used as keywords for pair-wise search in Google blog sites, so as to build the corresponding co-word matrixes. By means of multidimensional scaling (MDS) analysis and comparison between each set of MDS maps (co-assignee vs. co-word and co-citation vs. co-word), this study found that each set of maps presented similar distribution of company clusters. At last, this study summarized the characteristics of the network co-word analysis method in corresponding to the co-assignee and co-citation analysis methods. The co-word analysis can be used as an alternative to co-assignee and co-citation analyses in the study of company clusters in patent bibliometrics, as well as be implemented in actual industrial analysis.