Science Mapping of Publications in Natural Language Processing in the Philippines: 2006 to 2020

Science Mapping of Publications in Natural Language Processing in the Philippines: 2006 to 2020 Rachel Edita O. Roxas, Joseph Marvin R. Imperial, Angelica H. De La Cruz National University, Manila Manila, Philippines {reoroxas,jrimperial,ahdelacruz}@national-u.edu.ph Abstract This study presents the quantitative and qual- itative analyses of natural language process- ing (NLP) scientific literature during a 15- year period, from 2006 to 2020, of an un- derrepresented region, the Philippines. This aims to identify the current state of research work and further highlight research gaps for research development in the country. From the 217 Scopus-indexed publications which were extracted during June 2021, 185 publications were manually identified and screened for fur- ther analyses. The analyses reveal the collec- tive profiles of these documents, research pro- ductivity, modalities, language polarity, hu- man languages considered, and NLP areas considered in the studies. Conclusions and recommendations for the Philippine research community are presented. 1 Introduction Language (otherwise known as natural language or human language) is an important means of commu- nication, and natural language processing (NLP) al- lows the automatic processing of these forms of in- teractions. With the daily exponential increase in conversations as facilitated by the internet, the field of NLP is becoming a vital means of using comput- ers to make sense of these exchanges. Thus, it is imperative to determine the current state of NLP re- search work for the continual progress of this field of study, and further propel the development of the field to the next level in the coming years. The current state of scientific literature through quantita- tive means such as bibliometric analyses have been conducted in the area of NLP (Lopez-Martinez and Sierra, 2020; Radev et al., 2016). These studies con- sistently reveal an increasing trend in NLP research within the past 20 years, highlighting the most in- fluential authors, institutions, and countries in NLP research, and the main topics and actors in NLP re- search. Quantitative reviews on NLP have been wide and varied in terms of the area of investigation, as well as in terms of the NLP tasks considered. How- ever, such reviews tend to be biased towards the main actors (such as researchers, research institu- tions, and funding agencies), prominent researchers and institutions, and natural languages considered in the scientific literature, which are mostly of the majority. Scientific literature from underrepresented regions that may consider indigenous languages do not tend to appear in these studies, and they tend to be ignored in these broad scoped and general studies (Bender et al., 2021). There appears to be a potential to investigate particular regions or ge- ographic areas where there is a rich collection of human languages, which are mostly minority lan- guages. Some bibliometric studies have been un- dertaken to focus on specific geographic locations. For instance, research work by (Yu et al., 2020) fo- cused on China, while (S¸ahin and Candan, 2018) investigated research work within the Turk region. For NLP research literature, a bibliometric study on NLP in Russia (Barakhnin et al., …

Content truncated to 500 words