Multimodal data and Deep Learning
Multimodal data and Deep Learning
Utilizing advanced deep learning techniques to analyze diverse data sources, including street view imagery, Flickr, and Google Reviews, to gain nuanced insights into urban environments.
AI + Voice-based
AI + Voice-based is a structured dataset derived from raw spoken content through advanced speech recognition, capturing not only text but also acoustic features such as tone, pitch, and rhythm. Users can accurately transcribe speech, extract keywords, semantic clusters, and dialogue structures, while also detecting speaker emotions and communicative intentions beyond literal words.
AI + image-based data
Image data from social media platforms is the most intuitive way to observe people’s interest in urban vitality, and the variety of image content on social media platforms reflects the diversity of vitality. It includes geo-referenced information that enables the visualization of the distribution of different activities in space
AI + Text-based Data
Using sentiment analysis methods, we analyzed the frequency of nouns, adjectives, and negative, neutral, and positive words in the text-based review data. We also identified the most frequently occurring key phrases, which reveal the behavioral preferences of urban residents.