Over the past six years, CIMS — with help from IBM — has developed a powerful, relatively low cost, platform for Unstructured Text Analysis (UTA) that can help organizations make better, more informed strategic decisions in several important areas.
One of these UTA applications — helping a large pharmaceutical company assess the impact on its business of “personalized medicine” for the treatment of cancerous tumors — was described by NC State University Professors Paul Mugge and Dick Kouri in the Sept-Oct 2014 Innovation Management Report (“Why Big Data Is Not All Hype: The Power of Unstructured Text Analytics,” pp.1-6).
Finding funding partners for an NGO combating infectious disease is another UTA application the CIMS faculty has carried out successfully. Prof. Kouri tells the story below.
NC State University’s biosciences management program is collaborating with a large non-government organization (Healthcare NGO) that is focused on advancing healthcare associated with HIV, tuberculosis, malaria, and other infectious diseases in 28 Low-Income Countries. The groups were working together to develop a Directed Big Data Analytics (DBDA) capability for several infectious diseases.
The collaborators agreed to expand the partnership to use DBDA to identify potential and novel philanthropic organizations to fund the work of the Healthcare NGO. The team developed four guiding principles or criteria:
- The prospect had a demonstrated interest in global development or global health and a similar mindset to that of the Healthcare NGO. That is, the prospect had a giving behavior aligned to the mission of the Healthcare NGO.
- The prospect was consistently donating over $5 million and was currently investing.
- The Healthcare NGO had not solicited money from the prospect nor received any money in any fashion.
- Executives of either party had no previous dealings.
The Search Begins
With these criteria established, we were ready to tackle the strategic question: Identify potential donors with the ability to give the Healthcare NGO over $5 million.
We began the DBDA work by creating dictionaries under each of the main areas of interest: people and organizations with money, demonstrated interest in emerging economies, potential partner philanthropic organizations, and funding history. For example, for people and organizations with money, these dictionaries contained the Fortune 500 companies and their foundations, the biggest NGOs in the world, prior top donors to the Healthcare NGO, wealthy individuals in general, and government organizations.
We ended with 20 dictionaries that read on our strategic question. From these we selected around 25 websites that were likely to contain information pertinent to the search question. Some were disease-oriented, others were sites for NGOs that worked in similar areas and might be sources of money, others were advocacy or policy organizations, and still others were blogs directed at global health.
Some websites were foundation review organizations that listed all the philanthropic organizations working in various areas. We used these as seed URLs, because they contain numerous links to other URLs. In fact, the critical information usually comes from these linked URLs. We have data showing that every seed URL we pick correctly, will touch somewhere between 5,000 and 10,000 additional URLs.
By starting with around 25 seed URLs, we are likely to touch between 125,000 to 250,000 additional URLs with each of these URLs linked to hundreds of web pages. As a result, it was easy for us to collect approximately 1.6 million files of information, each of which contained an average of about 50 web pages, or roughly 80 million pages total. These were files that included “white” papers, blogs, journal articles, newsletters, etc.— all the information we might need.
Finding and Linking Key Words
Our next step in applying Unstructured Text Analytics was to search these 1.6 million files and look for the frequency and pattern of occurrence of the relevant key words in these documents. Examples of some test queries are shown in the table below.
We then linked these key words to a company or foundation that had previously given money as defined above. After finding nearly 200 that had been involved in that way, we began prioritizing them according to our criteria.
Interestingly, we found that some of these philanthropic organizations were very much involved with infrastructure; that is, they were building bridges, highways or water depots. Such as, the Caterpillar Foundation which is building bridges, hospitals and water purification systems in various countries. This gives us the opportunity to share a common goal that links infrastructure improvements in water purification and sewage treatment with work to improve healthcare associated with TB and HIV. Using various criteria against the most promising candidates, we narrowed our list down to the sixteen highest-priority organizations.
NCSU worked with the Healthcare NGO to further prioritize these 16 organizations based on additional criteria such as relevant technical experience, commitment to global health or global development, available donation funds and donation activity, and donations to countries of interest to the Healthcare NGO.
Then, with the Healthcare NGO’s help, we identified eight high-priority institutions on which to focus additional attention. Our goal was to identify key people responsible for making funding decisions within these eight organizations. From this and other efforts, the Healthcare NGO is completing brief white papers on each of these eight high-priority institutions with support from NCSU. These will be presented to the Healthcare NGO’s board to define rules of engagement.
Finding the Next Donors
The next step is to use feedback from the Healthcare NGO’s board and subsequent engagement with the eight high-priority philanthropic organization to refine our DBDA approach to find even better prospective donors. One topic we plan to address is the evolution of a donor’s giving focus. As an example, Google changes its giving priorities yearly. To be proactive with this type of organization, our DBDA approach has to anticipate what donors are thinking by analyzing their public presence and how that will impact the focus of their giving.
What It Took
The team included one part-time domain expert and three part-time MBA students from our business school, and roughly two staff from the Healthcare NGO. The entire project was completed in approximately three months. The goal is to reduce this time to a third.
Like many organizations, the Healthcare NGO has sought funding in very classic ways. This approach however relies, perhaps too heavily, on what you know and who you know. Nevertheless, the Healthcare has been successful to date: but as it grows, its need for funds grows. To its credit, the Healthcare NGO is looking to an evidence-based approach to identify and rank similarly-minded philanthropic organization to help advance its work. Furthermore, it believes that DBDA and its partnership with NCSU are central to the solution.
Richard Kouri is Executive Director, BioScience Management Initiative, Jenkins Graduate School of Management, Poole College of Management, NC State University; email@example.com