Towards Predicting Trend of Scientific Research Topics using Topic Modeling

Abstract

This paper presents an alternative method to study and forecast the trend of research topics. Prediction of scientific research topics’ trend can be conducted using subjective judgments of experts or quantitative analysis. Due to the fact that subjective judgments of experts might be biased, researchers have been employing quantitative analysis such as bibliometrics, scientometrics, or informetrics. However, using these measures has limitations which require the use of an alternative approach. Results of the scientific research have been digitized and stored as news, scientific articles, and books. Such abundant information can be analyzed, taking the whole content of the papers into account. Hence, this paper proposes prediction of the trend of research topics using topic modeling. The proposed method was experimented using the proceedings of the International Conference on Computational Science (ICCS) which contains a total of 5982 papers over seventeen years (2001-2017). Non-negative Matrix Factorization (NMF) topic modeling method was utilized to discover topics. The result was structured as time series data and used to predict the trend of research topics. Auto-Regressive Integrated Moving Averages (ARIMA) prediction method was implemented and the performance of the model was evaluated using Root Mean Squared Error (RMSE). The proposed method may allow researchers, policy makers, funding agencies, and government to understand the current and the future state of research areas and take corrective actions.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 36 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.