Quantifying Semantic Shift Visually on a Malay Domain Specific Corpus Using Temporal Word Embedding Approach

Abstract

In this study, we propose an alternative approach to analyzing a domain-specific time series corpus for detecting word evolution. The method trains a target corpus in time series into a temporal word embedding (TWE) model. The advantage of TWE is that one can see how the meaning of a word changes over time. We have chosen the TWEC approach to model a Malay domain-specific time-series corpus, the Malaysian Hansard Corpus (MHC), to a TWE model and called the model as MHC-TWEC. Two primary analyses, i.e., self-similarity analysis and user-defined method analysis, were performed to validate the effectiveness of the MHC-TWEC model in quantifying semantic shift on MHC visually. From those analyses, we visually find out that the TWE model can capture the semantic shift in the temporal corpus (the MHC).

Domain-specific Stop Words in Malaysian Parliamentary Debates 1959 – 2018

“…Even though the data used is from parliament, it is also relevant to general use of the Malay language. In addition, many previous studies have utilised the MHC (Nor Fariza, Anis Nadiah, Azhar, Imran & Sabrina, 2019;Norsimah, Azhar, Anis Nadiah & Imran, 2019;Sabrina, Nor Fariza, Azhar & Anis Nadiah, 2020;Sabrina, Saidah et al, 2020). It is expected that the production of Malay stop words relating to the corpus will assist future research in terms of stop word removal and text processing in general.…”