This paper introduces a multiple-perspective co-citation analysis method to characterize and interpret the structure and dynamics of co-citation clusters. The method integrates network visualization, spectral clustering, automatic cluster labeling, and text summarization to facilitate analysis and interpretation. Co-citation networks are decomposed into clusters, which are interpreted using automatic labeling and summarization. The method focuses on the interrelations between cluster members and their citers. It is applied to a three-part analysis of the field of Information Science (1996-2008): 1) comparative author co-citation analysis (ACA), 2) progressive ACA of time series co-citation networks, and 3) progressive document co-citation analysis (DCA). Results show that the method increases the interpretability and accountability of ACA and DCA networks. The paper discusses the challenges of interpreting co-citation clusters, which require categorization, summarization, synthesis, and integration. Traditional methods focus on cited members, but focusing on citers provides deeper insights into research front dynamics. The new method reduces cognitive burden by automatically characterizing clusters using salient noun phrases and representative sentences. It also shifts the burden of interpretation to automatically generated structural and content cues. The paper reviews existing studies of co-citation analysis in information science, including ACA and DCA studies by White & McCain (1998), Zhao & Strotmann (2008a, 2008b), and Aström (2007). These studies identified two main camps in information science: information retrieval and literature. Recent studies suggest the field has evolved, with webometrics and visualization of knowledge domains becoming more prominent. The paper introduces a multiple-perspective method for ACA and DCA studies, including clustering, labeling, and sentence selection. It uses CiteSpace for visualization and analysis. The method is applied to a 12-journal dataset (1996-2008), revealing 40 ACA clusters and 50 DCA clusters. Results show that the method improves the interpretability of co-citation clusters by providing automatic labels and summaries. The paper discusses the challenges of interpreting co-citation clusters, including the need for domain knowledge and the complexity of synthesizing information. The new method addresses these challenges by integrating structural and content analysis, and by using automatic labeling and summarization. It also highlights the importance of considering both citers and cited entities in co-citation analysis. The paper concludes that the multiple-perspective method provides additional insights into co-citation structures and supports the interpretation of specialties in information science. It suggests that future research should focus on improving the robustness of co-citation analysis and exploring new methods for interpreting co-citation clusters.This paper introduces a multiple-perspective co-citation analysis method to characterize and interpret the structure and dynamics of co-citation clusters. The method integrates network visualization, spectral clustering, automatic cluster labeling, and text summarization to facilitate analysis and interpretation. Co-citation networks are decomposed into clusters, which are interpreted using automatic labeling and summarization. The method focuses on the interrelations between cluster members and their citers. It is applied to a three-part analysis of the field of Information Science (1996-2008): 1) comparative author co-citation analysis (ACA), 2) progressive ACA of time series co-citation networks, and 3) progressive document co-citation analysis (DCA). Results show that the method increases the interpretability and accountability of ACA and DCA networks. The paper discusses the challenges of interpreting co-citation clusters, which require categorization, summarization, synthesis, and integration. Traditional methods focus on cited members, but focusing on citers provides deeper insights into research front dynamics. The new method reduces cognitive burden by automatically characterizing clusters using salient noun phrases and representative sentences. It also shifts the burden of interpretation to automatically generated structural and content cues. The paper reviews existing studies of co-citation analysis in information science, including ACA and DCA studies by White & McCain (1998), Zhao & Strotmann (2008a, 2008b), and Aström (2007). These studies identified two main camps in information science: information retrieval and literature. Recent studies suggest the field has evolved, with webometrics and visualization of knowledge domains becoming more prominent. The paper introduces a multiple-perspective method for ACA and DCA studies, including clustering, labeling, and sentence selection. It uses CiteSpace for visualization and analysis. The method is applied to a 12-journal dataset (1996-2008), revealing 40 ACA clusters and 50 DCA clusters. Results show that the method improves the interpretability of co-citation clusters by providing automatic labels and summaries. The paper discusses the challenges of interpreting co-citation clusters, including the need for domain knowledge and the complexity of synthesizing information. The new method addresses these challenges by integrating structural and content analysis, and by using automatic labeling and summarization. It also highlights the importance of considering both citers and cited entities in co-citation analysis. The paper concludes that the multiple-perspective method provides additional insights into co-citation structures and supports the interpretation of specialties in information science. It suggests that future research should focus on improving the robustness of co-citation analysis and exploring new methods for interpreting co-citation clusters.