Landscape assessment of Mountain Qingxiu using internet landscape discourse and LSTM

Abstract

The internet era has provided a wealth of data sources for landscape assessment by updating the methods of information exchange in public. Landscape assessment using internet discourse and computer language can provide guidance for the rational use of landscape and the scientific management of scenic areas. By taking the internet discourse of travels and reviews about Mountain Qinxiu as research objects, this study extracted the landscape assessment objects and analyze the sentiment responds of landscape using LSTM recurrent neural network. And the sentiment characteristics were analyzed by the sentiment-heat IPA quadrant. According to the results of the study, the public’s landscape assessment is both discrete and convergent. The landscape assessment of Mountain Qinxiu is mainly positive; There is no regular correlation between the landscape thermal index and the landscape sentiment assessment; The sentiment scores of landscape elements indicates that the plant landscapes and meteorological landscapes are natural landscape elements with high recognition, while the cultural features, buildings, and landscapes in human landscape elements are attractions with potential for development. The sentiment scores of attractions indicates that the botanical landscapes are natural attractions with high recognition, and the architectural and garden landscapes in humanistic attractions are attractions with potential for development. It can be found that the sentiment-based landscape assessment of Mountain Qingxiu using LSTM and internet discourse can effectively complement traditional methods of landscape assessment and provide a better opportunity for the public to participate in the management of landscapes.

A Study on the Landscape Elements and Distribution Characteristics of Mount Tai Appearing in Poems

“…Tsukada et al (2013) analyzed the landscape impressions of mountains in Japan through text minning based on the frequency of mountain names, images and landscape elements appearing in school songs. In addition, Wang et al (2020) analyzed the image of Mount Qingxiu through text minning based on the frequency of appearance of landscape elements, using network discourse as research materials. The above-mentioned studies used poems, stele pavilions, school songs, and network discourse as research materials to analyze landscape features or impressions based on the frequency of landscape elements appearing in the texts, but basically focused on the texts without temporal distinction, and therefore judged that the analysis of temporal changes was insufficient.…”