AI summarizer assistant
Diaz Diaz, Eliecer (2021)
Diaz Diaz, Eliecer
2021
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi:amk-2021101818754
https://urn.fi/URN:NBN:fi:amk-2021101818754
Tiivistelmä
This thesis was done in cooperation with the Joint United Nations Program on HIV/AIDS UNAIDS. The mission of UNAIDS is to inspire the world in achieving universal access to HIV support. UNAIDS collects data from 193 United Nations’ member states, and tracks progress towards global AIDS targets writing annual reports. UNAIDS manifested the need for assistance on this task seeking expertise in implementing efficient AI and NLP solutions to improve working with vast amounts of unstructured data, which is typically highly laborious. A solution in the form of Proof of Concept POC was built using Artificial Intelligence AI tools such as Natural Language Processing NLP and Optical Character Recognition OCR. These tools work by summarizing text and extracting numerical information from tables embedded in reports, respectively. These functions were integrated with a user interface UI to facilitate the operation by UNAIDS experts. In addition, visualization functions were also developed to monitor monetary investments per country and the number of people living with HIV. Similarly, sentiment analysis and population pyramids charts were also included as visual capabilities in the solution. Finally, the critical functionality developed was the “integration” tab which puts the results from all the other functionalities in a logical output resembling a dashboard; experts can evaluate the HIV situation for a specific year and country. As a final deliverable result, a PoC was presented to UNAIDS experts. UNAIDS indicated that using this AI summarizer assistant solution and further AI implementations might save an estimated 30–40% of time spent on data analysis.