Automated Text Analytics Tool for Extracting and Summarising Evidence from Academic Publications

Supporting Evidence-Based Interventions (SEBI) is a program at The Royal Dick School of Veterinary Studies funded by The Bill & Melinda Gates Foundation (BMGF). SEBI mobilises and applies data and evidence to help the livestock community make better investments that improve livelihoods for smallholders in low and middle-income countries. One of the internationally recognized challenges is the limited availability of reliable information on diseases and productivity impacting livestock and agriculture. SEBI is uniquely positioned to address this challenge through combined expertise in veterinary science, data science and their engagement with the global community of livestock data producers and users.
The purpose of this project is to build an automated text analytics software solution which would help SEBI and BMGF accelerate the extraction of relevant information and gathering research evidence from veterinary science publications. SEBI is using a systematic maps methodology to gather evidence in an unbiased way. This lends itself to text mining as the data collation and extraction process is documented. Currently, it takes 3 months for a researcher to extract the information and build a report on relevant livestock data for just one country. The first report was built for Ethiopia, but BMGF has requested reports for 11 more countries in Africa. Other funders, DFID and USAID, have expressed enthusiasm for the SEBI’s approach but have stated that they would like to see a much wider range of low and middle income countries (about 50 for DFID) reported on to meet their needs. This would require a significant input of time and resources.
The automated text analytics software tool would enable SEBI to deliver reports on more countries and allow for the timely update of the literature base, so that it is kept up to date with new research. With the current manual labor-intensive process, producing these reports in a timely manner is not feasible.
The proposed solution to automate this process would have an immense impact on the productivity of researchers, quality of data and resulting reports that will help inform decision making with the ultimate aim of improving livelihood for smallholders across Sub-Saharan Africa.