Text miming with R packages

by Paul A Agbodza (June 30, 2022)

It was when covid-19 struck in early 2020 that more attention was paid to scientists, researchers, medical professionals, public health institutes, research institutions and scientific research on the African continent. Recent discourse on investment/spending on Research and Development (hence R&D) in Africa has shown that inadequate funds has been allocated to R&D on the continent. This is obvious. They have many concerns of providing food to the poor, patching several kilometers of man-holed roads, providing basic education for a predominantly young population, scientific research which does not bring immediate dividends ‘must always wait’. But the ‘covid-19 moment’ has exposed such skewed prioritization. To localize the state of R&D, Ghana’s situation will be discussed here. It would be demonstrated that some R&D activity is happening in Ghana but whether that level can sustain (future) productivity growth and become a dividend of scientific output is yet to be tested.

R&D in Africa in Charts

Ghana on R&D in the pages of budget statements 1999–2022

To reconstruct Ghana’s R&D investment/spending prioritization, text mining tools are employed to map the budget statements from 1999 to 2022. [Other form of textual data that could be used are the State of the Nation’s Address, ISSER reports and Bank of Ghana Annual Reports. A study on these would be released later]. Packages of R used were tidyverse, tidytext, tm, rvest, igraph, ggraph and forcats. ggplot2 was combined with ggthemes, papaja and hrbrthemes to visualize the information in a publishable format.

Most frequent terms that correlate with the term ‘research’ are first visualized below (Fig. 3). They show the economic policy priority the state gives to R&D in the various fields of engagement.

Some terms that correlate highly with the word ‘research’ are information, training, students, education, technology, legal, and some cities known for research institutions are mentioned. Other words include promote, investment and institutions. Other important correlates are operational, science, and environmental and environment. This is true of all budgets from 1999 to 2022. What can be said of each separate budget statement?

Similarly, a network graph to visualize the relationship between the word ‘research’ and related terms found in each budget statement is shown below (Fig. 4).

The network graph (Fig. 4) is a visualization of bigrams (pair of terms) that occur more than once. It is best viewed on a wider screen. Too few bigrams occur more than two counts so a plot of that only would have been clearer but would have swamped much of the information. There appears some clustering of budget years here. Budgets 2002, 2019 and 2020 have common concerns on research types. Budgets 1999, 2004 and 2008 but budgets 2005, 2006 and 2007 have unique interests. This graph helps to infer the engagement of the particular budget with research and its associated term. In a subsequent paper the topics around which the terms are used would be presented.

Scientific publications as proxy for R&D in Ghana

Fig. 5 is the research output on selected subject areas from research institutes and researchers in Ghana in the year 2020. Note that the y-axis scale is not the same for all subject areas so one cannot make a simple comparison. The Biological and Agricultural Science publications are conspicuously high, over 2,000 publications. Compared to the citations of the publications the H-index is significantly low. The H-index is the true metric that measures the productivity and citation impact of publications.

In the period 1996 and 2021, Computer Science papers gave some of the highest research output. Yet, H-index is still low. The scientific output of Ghana of 34018 is 0.0493% of world output and 2.86% of Africa’s output between 1996 and 2021. The volume of publications shown here point to the quantum of expenditure on R&D in Ghana. The sources of funding and accurate figures if available would have swelled the state’s expenditure/investment on R&D. A comparative analysis of Africa’s scientific output per R&D against the rest of the world is well documented.

Patents and Trademark

That patents and publications count as indices of R&D output is a stylized fact already stated. The Patents Act, 2003 (Act 657) and the Patents Regulations 1996 (L.I. 1616) have been promulgated to provide a legislative framework for the grant and protection of patents in Ghana. Fig. 7 (below) shows Ghana’s performance in innovation. From the data, patent applications by residents were 15 in 2017, 13 in 2018 and 12 in 2020

Conclusion

Addendum

Related posts:

Following two weeks of threats to end the Fourteenth Amendment’s “birth right citizenship,” Drumpf issued an executive order to end birth-right citizenship. He signed the order Friday night, after…

All Rights Reserved I lay here, my eyes wide open. My life it flashes before my life. My childhood. My teen years, and the start of the path I went down. My adult years and that final act that…

User Centered Design merupakan proses desain yang berfokus pada kebutuhan pengguna. Untuk mengetahui kebutuhan itu, hal yang paling mendasar yang harus kita lakukan yaitu dengan melakukan Interview…