Semantic recommendations and topic modeling based on the chronology of romanian literary life

Abstract

As part of the Romanian Academy’s effort aimed at underlining the importance of events centered on national authors and writings across time, the Chronology of Romanian Literary Life (also referred to as CVLR) is a centralized text repository which contains all important literature-related events that occurred after World War II. The current work presents an approach to capture topics’ evolution across time and helps learners by recommending events from the chronology on a given topic, based on a subset of 24 years of the CVLR. Our method combines techniques from information retrieval, topic modeling using LDA (Latent Dirichlet Allocation), and recommender systems to improve e-learning centered on Romanian literature. The most frequent topics in each year are ranked in order to identify and visualize the main interests in literature across time periods. Recommendations are performed in order to facilitate the exploration of the chronology, as it is currently indexed only by event dates.