This article analyzes perceptions concerning cultural heritage sites along the central axis of Beijing from community, temporal, and spatial perspectives by extracting keywords, word frequency, term frequency-inverse document frequency (TF-IDF) weight, mutual information, posterior probability, and other features in microblogs, newspapers and magazines, and academic publications in 2012 and 2015. On the community dimension, through media information of characteristic groups, we found that different groups have different understanding of cultural heritage sites. The core sites of Beijing Central Axis cultural heritage, such as the Imperial Palace, Tiananmen, and Temple of Heaven are perceived relatively consistently by different communities. But the perceptions of the Bell and Drum Towers, Imperial Ancestral Temple, and Di'anmen are varied: officials are concerned with their administrative aspects, scholars are concerned with their historical values, and the public are concerned with their leisure and entertainment qualities. On the temporal dimension, changes of level of attention and perception on these cultural heritage sites are also observed. In 2015, the public paid more attention to the Forbidden City, Tiananmen, the Temple of Heaven, and the Imperial Ancestral Temple for their historical values as compared to 2012. Public perception, compared with that of officials and scholars, is more likely to change and more sensitive to significant events. On the spatial dimension, this research has examined the transfer of perception and correlation between cultural heritage sites. First, Tiananmen, Zhengyang Gate, and Zhengyang Avenue, which are connected in space, show higher two-way perceptions. Second, the posterior probability of the Imperial Palace, Tiananmen, and the Temple of Heaven is higher among the central axis cultural heritage sites, showing a cross space perception convergence model. Thus the analytical framework for perception of cultural heritage based on big data is an important supplement for traditional methods such as questionnaires, literature research, and interview analysis, as it increases the dimension and efficiency of analysis and aids to discover hidden knowledge and patterns. The conclusion of this study can provide important support for policy making in the rediscovery and protection of cultural heritage values.