Back

Speaker "Keisuke Inoue" Details Back

 

Topic

Conversation Graph: Understanding Consumer Sentiment, Messaging Topics, and Content Sharing

Abstract

The goal of this research initiative is to build a technology that optimally suggests relevant, fun, and unique content (i.e. contextual stickers and GIFs) to messaging apps users to be shared within their peer-to-peer conversations. The Conversation Graph processes the consumer sentiment, messaging topics, and content sharing behaviors. This research expands and utilizes the sentiment analysis, machine learning, and natural language processing work that was previously presented by Keisuke at the 2017 Sentiment Symposium. The talk will include: 1) an overview of the data architecture of the Conversation Graph -- how the data is collected, processed and analyzed, 2) a case study analyzing the context and reactions to content sharing on messaging apps and 3) an experiment to build a predictive model of content sharing based on the users' behavioral and contextual data. The research and case study is based on the data that has been collected via Emogi’s partnerships with various mobile application partners such as Whisper, TextNow, TextPlus, and more. Content data revolves around the original emoji, stickers and GIFs created by the Emogi Content Studio, including branded examples for leading brands like McDonald's, IKEA, Universal Pictures, Moet Hennessy among others. The goal of this research initiative is to build a technology that optimally suggests relevant, fun, and unique content (i.e. contextual stickers and GIFs) to messaging apps users to be shared within their peer-to-peer conversations. The Conversation Graph processes the consumer sentiment, messaging topics, and content sharing behaviors. The talk will include: 1) an overview of the data architecture of the Conversation Graph -- how the data is collected, processed and analyzed, 2) a case study analyzing the context and reactions to content sharing on messaging apps and 3) an experiment to build a predictive model of content sharing based on the users' behavioral and contextual data. The research and case study is based on the data that has been collected via Emogi’s partnerships with various mobile application partners such as Whisper, TextNow, TextPlus, and more. Content data revolves around the original emoji, stickers and GIFs created by the Emogi Content Studio, including branded examples for leading brands like McDonald's, IKEA, Universal Pictures, Moet Hennessy among others.

Profile

As VP of Data Science, Keisuke's primary responsibilities include research and development of natural language processing (NLP) and machine learning (ML) technologies to better understand the meanings, intentions and emotions expressed in mobile messaging. Keisuke received his PhD in Information Science and Technologies at Syracuse University in 2013 on detecting discourse-level semantics in online chat messages using NLP and ML and received multiple research awards from renowned library and information foundations. During his PhD career, he served as the mentor for multiple companies at Syracuse Student Sandbox, one of the top college business incubators in the United States.