Title

Using IBM Watson to Improve Non-Specialist Audience Understandability of Research Articles

School Name

Governor's School for Science and Mathematics

Grade Level

12th Grade

Presentation Topic

Computer Science

Presentation Type

Mentored

Abstract

Research data is growing at an alarming rate, and non-specialists in a field of study will struggle to stay up-to-date on current research findings. However, artificial intelligence may offer a solution to help understand the overwhelming amount of data available. Artificial intelligence methods, such as the Neural Network and machine learning, work similarly to logical human thinking because of the complexity and efficiency of the human brain. However, these algorithms are quicker and often times more accurate, allowing them to perform classification and other tasks even more efficiently. IBM Watson is a supercomputer with multiple deep neural networks, machine learning programs, and other tools available. Using IBM Watson’s available tools, this study creates and scores extractive summaries of research articles to condense the amount of information and make them easier to read. An extractive summary involves creating summaries based on sentences and phrases already present in a given piece of text, so this study focuses on the goal of summarizing by reducing the amount of details and information present. The results showed that when trained with the abstracts, introductions, and discussions of different articles, Watson was able to create relevant summaries using sentences already present, condensing articles to its sentences with more general information, as shown by high Rouge scores when compared to parts of the article. This study showed that Watson’s tools are a promising method to extract information from journal articles and present the general ideas and topics in a summary.

Location

Neville 206

Start Date

4-14-2018 8:30 AM

Presentation Format

Oral and Written

COinS
 
Apr 14th, 8:30 AM

Using IBM Watson to Improve Non-Specialist Audience Understandability of Research Articles

Neville 206

Research data is growing at an alarming rate, and non-specialists in a field of study will struggle to stay up-to-date on current research findings. However, artificial intelligence may offer a solution to help understand the overwhelming amount of data available. Artificial intelligence methods, such as the Neural Network and machine learning, work similarly to logical human thinking because of the complexity and efficiency of the human brain. However, these algorithms are quicker and often times more accurate, allowing them to perform classification and other tasks even more efficiently. IBM Watson is a supercomputer with multiple deep neural networks, machine learning programs, and other tools available. Using IBM Watson’s available tools, this study creates and scores extractive summaries of research articles to condense the amount of information and make them easier to read. An extractive summary involves creating summaries based on sentences and phrases already present in a given piece of text, so this study focuses on the goal of summarizing by reducing the amount of details and information present. The results showed that when trained with the abstracts, introductions, and discussions of different articles, Watson was able to create relevant summaries using sentences already present, condensing articles to its sentences with more general information, as shown by high Rouge scores when compared to parts of the article. This study showed that Watson’s tools are a promising method to extract information from journal articles and present the general ideas and topics in a summary.