Maximizing Engagement with a Knowledge-Based News Recommendation System: Leveraging Surprise and Suspense to Counter Fake News and Echo Chambers
Ganesh Iyer, Professor
Business, Haas School
Closed. This professor is continuing with Fall 2023 apprentices on this project; no new apprentices needed for Spring 2024.
A knowledge-based news recommendation system is significant because it can help to mitigate the negative effects of fake news, echo chambers, and opinion polarization. By presenting a diverse range of viewpoints and sources, a knowledge-based system can help to expose users to a greater range of ideas and prevent the formation of echo chambers. This can also help to reduce the spread of fake news, as it provides a more reliable and fact-based alternative to sensationalist or biased sources. Additionally, a knowledge-based system can help to preserve the element of surprise and suspense in news consumption, as it may recommend articles on topics that the user might not have otherwise encountered. Overall, a knowledge-based news recommendation system has the potential to promote a more informed and balanced understanding of current events.
A graph neural network (GNN) is a type of neural network that is specifically designed to process graph-structured data, such as a knowledge graph. A knowledge graph is a structured representation of real-world entities and the relationships between them. It can be thought of as a graph, with the entities as nodes and the relationships as edges.
GNN techniques can be applied to building a news recommendation system by using the knowledge graph to model the relationships between different news articles and the entities they mention. For example, a GNN could be used to identify articles that are related to a particular entity or topic, based on the connections in the knowledge graph. The GNN could then be trained to predict which articles a user might be interested in, based on their past reading history and the connections in the knowledge graph. This could be used to recommend articles that are related to the user's interests or that introduce them to new topics that they might not have encountered otherwise.
Surprise and suspense can affect user news consumption behaviors by making content more engaging and stimulating. When users encounter unexpected or intriguing information, they may be more likely to continue reading or watching in order to find out more. This can increase the overall enjoyment of consuming news, as it keeps the user engaged and on the edge of their seat.
To measure surprise and suspense in a knowledge graph, one approach could be to track the user's engagement with content that falls outside of their usual interests or beliefs. If a user typically consumes news on a particular set of topics, and they show a significantly higher level of engagement with content on a different topic, it could be an indication that the new content was surprising or suspenseful to them. Another approach could be to use natural language processing techniques to analyze the content itself and identify elements that may be surprising or suspenseful to the reader. This could include the use of words or phrases that convey a sense of mystery or uncertainty, or the inclusion of unexpected plot twists or surprising revelations.
We are seeking a highly skilled and motivated individual to join our team. In this role, you will be responsible for helping to develop a knowledge graph-based news recommendation system. You will work closely with the rest of the development team to design, implement, and maintain the knowledge graph and related systems.
Students with strong data analysis skills are also encouraged to apply for this position.
Role: Responsibilities:
·Design and implement a knowledge graph to represent news articles and their relationships
·Develop algorithms and models to analyze and reason over the knowledge graph to provide personalized news recommendations
·Collaborate with the rest of the development team to integrate the knowledge graph into the overall news recommendation system
·Stay up-to-date with the latest techniques and technologies in knowledge graph development
Qualifications: ·Strong background in computer science/data science
·Strong interest in knowledge graph development, including designing and implementing knowledge graphs and integrating data sources
·Proficiency in at least one programming language, Python preferred
·Excellent problem-solving and communication skills
·A team player with a strong desire to learn and grow
·Preferably works 9 or more hours per week
Day-to-day supervisor for this project: Yunhao Huang , Ph.D. candidate
Hours: to be negotiated
Social Sciences