An investor evaluates a proposal. She considers the proposed budget, the potential earnings, the time needed for return on her investment. Decision models use these numerical inputs to evaluate the probability of success for the proposal, guiding the investor’s decision. However, in reality there is more influencing her decision than the numbers alone.

“Decision science models must incorporate qualitative factors such as interpersonal communication influences, including explicit and implicit efforts to persuade,” says Michelle Seref, assistant professor of business information technology.

Seref has a Ph.D. in operations management and a master’s degree in industrial and systems engineering, but she also acquired an additional master’s degree in rhetoric. “Rhetorical theory has a rich history in political science and communication studies, but I believe it has very relevant applications to business problems.”

Rhetoric mining

Seref is developing a new method called “rhetoric mining,” which she says can be used to quantify more complicated language expressions of persuasion. “Many text mining techniques use frequency analysis of simple word lists or basic language components,” she says. Text mining is used in business research to analyze patterns in language as a component of decision models. Seref’s new method advances text mining techniques by capturing more sophisticated language phrases identified as rhetorical moves.

“Rhetoric is the intentional, strategic use of language for the purpose of persuading an audience to accept a particular idea or take a particular action,” says Seref. She explains that rhetoric mining involves taking a small sample of an available text resource and examining it for different kinds of rhetorical moves. These moves involve using trust, logic, emotion, storytelling, ideological references, or metaphors as persuasion tools.

The method then uses a sequence alignment algorithm that can quickly identify equivalent phrases. “This research is an interdisciplinary approach, including theories and methods from rhetoric, digital humanities, machine learning, data mining, text mining, and computational biology,” she says.

Seref elaborates on the investment decision example: “The proposal may include background statements on the project engineers’ experience (an ‘ethos’ move) or present stories of successful experiments (a ‘narrative’ move). Identifying specific phrases as intentional language choices for the purpose of persuading the investor allows us to incorporate an important qualitative factor into decision analytics models.”

How does persuasive language influence decision-making?

Seref is using her rhetoric mining method to study the influence of persuasive language on crowd-funding decisions, product-purchase decisions, investment decisions, and other business applications. “I am very excited to be working with several colleagues in Pamplin on a variety of business data,” says Seref. “I believe our research team is producing high-quality work with innovative methods and new insights to decision analytics.”

In addition to her rhetoric mining research, Seref is also working with other researchers on a new context component network model to visualize the connection of language phases indicating context patterns. This research enhances decision models by incorporating context patterns, an improvement to current topic models used in text mining.

An award-winning teacher, Seref has worked with Alan Abrahams to develop the BIT Showcase event and the department’s undergraduate teaching assistantship program. The showcase, held every semester, allows company sponsors to view junior and senior student projects from decision support systems classes. Sponsorship of the event supports the latter program, in which top students from previous semesters serve as teaching assistants, offering other students additional help and mentorship.

Seref has also won the Pamplin Outstanding Diversity Award and has chaired the Pamplin Diversity Case Competition for multiple years. She has received Safe Zone diversity training. Diversity and inclusion are important issues for her, she says.

– By Dan Radmacher
– Photo by Jessamine Kane-Wiseley