Authorship Patterns in Computer Science Research in the Philippines

We studied patterns of authorship in computer science~(CS) research in the Philippines by using data mining and graph theory techniques on archives of scientific papers presented in the Philippine Computer Science Congresses from 2000 to 2010 involving 326papers written by 605authors. We inferred from these archives various graphs namely, a paper–author bipartite graph, a co-authorship graph, and two mixing graphs. Our results show that the scientific articles by Filipino computer scientists were generated at a rate of 33papers per year, while the papers were written by an average of 2.64authors (maximum=13). The frequency distribution of the number of authors per paper follows a power-law with a power of $\varphi=-2.04$ ($R^2=0.71$). The number of Filipino CS researchers increases at an annual rate of 60new scientists. The researchers have written an average of 1.42papers (maximum=20) and have collaborated with 3.70other computer scientists (maximum=54). The frequency distribution of the number of papers per author follows a power law with $\varphi=-1.88$ ($R^2=0.83$). This distribution closely agrees with Lotka’s {\em law of scientific productivity} having $\varphi\approx -2$. The number of co-authors per author also follows a power-law with $\varphi=-1.65$ ($R^2=0.80$). These results suggest that most CSpapers in the country were written by scientists who prefer to work alone or at most in small groups. These also suggest that few papers were written by scientists who were involved in large collaboration efforts. The productivity of the Philippines’ CS researchers, as measured by their number of papers, is positively correlated with their participation in collaborative research efforts, as measured by their number of co-authors (Pearson $r=0.7425$).

💡 Research Summary

This study investigates the authorship and collaboration patterns of Philippine computer science (CS) research by analysing the proceedings of the Philippine Computer Science Congresses (PCSC) from 2000 to 2010. The dataset comprises 326 peer‑reviewed papers authored by 605 distinct researchers. Using data‑mining techniques, the authors constructed a paper–author bipartite graph (PAG) and derived from it a co‑authorship graph (CAG) as well as two mixing graphs to explore assortativity.

Key quantitative findings include: an average of 33 papers produced per year; each paper is written by 2.64 authors on average (maximum 13). The distribution of authors per paper follows a power‑law with exponent φ = −2.04 (R² = 0.71). The CS community expands by roughly 60 new researchers annually. Individual researchers publish an average of 1.42 papers (maximum 20), and they collaborate with an average of 3.70 distinct co‑authors (maximum 54). The number of papers per author also obeys a power‑law (φ = −1.88, R² = 0.83), closely matching Lotka’s law (φ ≈ −2). Likewise, the number of co‑authors per author follows a power‑law (φ = −1.65, R² = 0.80).

A strong positive correlation (Pearson r = 0.7425) exists between a researcher’s productivity (papers authored) and their collaborative reach (co‑authors). Conversely, mixing analysis reveals low disassortative mixing: high‑productivity authors tend to collaborate with low‑productivity peers (r = −0.1015 for papers, r = −0.0398 for co‑authors). This suggests a strategic pattern where prolific scholars seek out less‑published collaborators, possibly to diversify expertise or mentor newcomers.

Methodologically, the authors parsed CD‑ROM and HTML archives, excluded poster and RIPS sessions due to incomplete author data, and built the bipartite graph in O(N log M) time by indexing authors in a balanced binary tree. The CAG was generated by replacing each paper node with a complete subgraph among its authors, enabling standard network‑science metrics (average degree, clustering, component size) to be computed.

Comparisons with international CS data (e.g., Newman’s 2011 study) show similar average authors per paper, but the Philippine CS community exhibits smaller collaboration clusters than high‑energy physics or large‑scale engineering fields, where average co‑author counts can exceed 100. The observed power‑law exponents indicate that most output is generated by a few highly connected individuals, while the majority of researchers contribute modestly and work in small teams.

The authors discuss implications for research policy: establishing a centralized, up‑to‑date author‑paper database could facilitate network analysis and funding decisions; grant mechanisms that explicitly reward multi‑institutional collaborations may increase average co‑author counts and, by extension, productivity; and mentorship programs that pair prolific authors with less‑published colleagues could exploit the observed disassortative mixing to accelerate knowledge diffusion.

Overall, the paper provides a comprehensive quantitative portrait of Philippine CS research activity, demonstrates that its authorship dynamics conform to well‑known scientometric laws, and offers data‑driven recommendations to strengthen collaborative practices and research output in the country.