Using Multi-Modal Topic Modeling in National Culture Resources: Methods and Applications

Abstract

In field of multi-modal data modeling, semantic analysis method has been widely applied for solving the problem of semantic gap, and one of the leading approaches is based on topic modelling. From a computational method perspective, the national culture data is a typical example of multi-modal data, which combines information from different sources. This paper reviews the development of multi-modal topic modeling and discusses several possible applications of multi-modal topic modeling in national culture resource system, such as cross-media retrieval, automatic annotation, and recommendation system. However, the factors of multi-lingual and inadequate training data give rise to an emerging demand to study and explore the improvement of existing multi-modal topic models. The summation of this paper lays the foundation for the future researches of multi-modal topic modeling applied in national culture resources.