Literature review of masonry structures under earthquake excitation utilizing machine learning algorithms

Abstract

This work aims to analyze and reveal critical features of the papers published since 1990 on the topic of masonry structures under earthquake loading. In particular, detailed information for nearly three thousand papers (exactly 2909) was extracted from the Scopus database [1], and investigated in two stages. Initially, the papers were analyzed in terms of simple statistics and keyword time series – as either raw or normalized data – in order to describe the evolution of the relevant research during the past twenty-seven years (1990-2016, inclusive). In a second phase, bibliometric maps of the papers were developed, regarding their similarities with respect to a variety of the papers' characteristics such as: author keywords and author names. The resulting diagrams constitute comprehensive maps of the relevant literature, with respect to the associations among the particular characteristics. The bibliometric maps were constructed based on a rigorous methodology, which converts each item (for example, keyword) to a two-dimensional (x, y) point on the bibliometric map. These distances between items reflect the dissimilarities between them, for a particular characteristic. The numerical procedure involved in the construction of the map is a constrained optimization problem which was formulated and solved with an efficient methodology.