Catherine Havasi (born 1981) is an American scientist specializing in artificial intelligence (AI), known for her work in computational linguistics and common sense reasoning. She co-founded and served as CEO of Luminoso, a text analytics company, for eight years, and was a key contributor to the Open Mind Common Sense project at the MIT Media Lab, which led to the creation of ConceptNet. Havasi currently holds the position of Chief of Innovation and Technology Strategy at Babel Street, an AI-enabled data-to-knowledge platform.
Havasi's research has focused on bridging the gap between human common sense and machine understanding, particularly through the development of large-scale semantic networks. Her work has been widely cited in the fields of natural language processing and sentiment analysis, and she has been recognized as a leader in the AI industry.
Early life and education
Havasi grew up in Pittsburgh, Pennsylvania. Her interest in artificial intelligence was sparked by reading Marvin Minsky's 1986 book The Society of Mind. She attended the Massachusetts Institute of Technology (MIT), where she became involved with the MIT Media Lab and studied under Minsky. Havasi was an alumnus of the Science Talent Search 1999 and participated in the International Science and Engineering Fair in 1996, 1998, and 1999. She earned a S.B. and M.Eng from MIT and a PhD in computer science from Brandeis University.
Career and contributions to AI
In the 1990s, Havasi pioneered the use of crowdsourcing for artificial intelligence, a method that leverages collective human knowledge to train AI systems. In 1999, she joined the MIT project Open Mind Common Sense (OMCS) alongside Minsky and Push Singh. This project aimed to build a vast database of everyday knowledge contributed by volunteers. Havasi was part of the team that created ConceptNet, an open-source semantic network based on the OMCS database. ConceptNet organizes common sense knowledge into a graph of concepts and relations, enabling machines to make inferences about the world.
The development of ConceptNet represented a significant advancement in Artificial intelligence research, providing a resource that could be used for tasks such as analogy-making, text understanding, and reasoning. Havasi's work on ConceptNet led to several influential publications, including a 2007 paper on ConceptNet 3 and a 2012 paper on ConceptNet 5, which expanded the network's multilingual capabilities.
In 2010, Havasi co-founded Luminoso, a company that applied the principles of ConceptNet to text analytics. Luminoso's software used Machine learning techniques to analyze unstructured text data, helping businesses gain insights from customer feedback, surveys, and social media. Under Havasi's leadership as CEO, Luminoso grew into a notable player in the AI industry, serving clients across various sectors.
Recognition and leadership
Havasi's contributions to AI have been recognized by several prestigious honors. In 2014, she was named to the Boston Business Journal's "40 Under 40" list, which highlights business and civic leaders making a major impact in their fields. The following year, Fast Company included her in its "100 Most Creative People in Business 2015" listing, acknowledging her innovative approach to applying AI to real-world problems.
In 2019, the U.S. Embassy invited Havasi to Portugal to deliver a series of lectures on "Practical Natural Language Processing." These lectures drew on her work at MIT, covering topics such as transfer learning, meta learning, educational outreach, natural language understanding, and computational creativity. Her ability to communicate complex AI concepts to diverse audiences has made her a sought-after speaker and educator.
Research and publications
Havasi has co-authored seven peer-reviewed journal articles on AI and language, along with numerous peer-reviewed conference presentations. Her most cited publication is "New avenues in opinion mining and sentiment analysis," co-authored with Erik Cambria, Bjorn Schuller, and Yunqing Xia, published in the IEEE Journal of Intelligent Systems in 2013. This paper, cited over 700 times as of September 2018, explored new directions in analyzing public opinion and sentiment from text data.
Another significant work is "ConceptNet 3: a flexible, multilingual semantic network for common sense knowledge," presented at the Recent Advances in Natural Language Processing conference in 2007. This paper detailed the architecture and capabilities of ConceptNet, which has since become a foundational resource in the field.
Havasi also contributed to "Representing General Relational Knowledge in ConceptNet 5," published in the LREC proceedings in 2012. This work addressed the challenges of representing diverse relational knowledge in a scalable and interoperable format.
Her earlier research included "Digital Intuition: Applying Common Sense Using Dimensionality Reduction," published in the IEEE Journal of Intelligent Systems in 2009, and "AnalogySpace: Reducing the dimensionality of common sense knowledge," presented at AAAI in 2008. These papers introduced techniques for using dimensionality-reduction methods to improve the performance of common sense reasoning systems.
Impact on natural language processing
Havasi's work has had a lasting impact on the field of Natural language processing. By combining crowdsourced knowledge with computational techniques, she helped demonstrate the value of common sense in AI systems. ConceptNet has been used in various applications, including sentiment analysis, word sense disambiguation, and question answering. Her research on sentiment analysis, in particular, has influenced how companies and researchers approach the analysis of public opinion.
Her contributions to transfer learning and meta learning, as highlighted in her 2019 lectures, reflect a broader trend in AI toward developing systems that can learn from limited data and adapt to new tasks. These areas have become increasingly important in the era of Large language models, where efficient learning and generalization are critical.
Current role and ongoing work
As of the most recent information, Havasi serves as the Chief of Innovation and Technology Strategy at Babel Street, a company that provides AI-enabled data-to-knowledge platforms. In this role, she oversees the strategic direction of the company's technology, focusing on integrating advanced AI capabilities into products that help organizations make sense of vast amounts of data. Her work at Babel Street continues to build on her expertise in natural language understanding and common sense reasoning.
Havasi's career trajectory from academic research to industry leadership exemplifies the growing convergence of AI research and commercial applications. Her efforts have not only advanced the state of the art in computational linguistics but have also made AI more accessible and practical for businesses and organizations worldwide.
Legacy and influence
Havasi is considered a pioneer in the use of crowdsourcing for AI, a concept that has since become mainstream with the rise of platforms like Amazon Mechanical Turk and other human-in-the-loop systems. Her work on ConceptNet remains a valuable open-source resource, used by researchers and developers to incorporate common sense into their applications.
Her recognition in business and technology publications underscores her influence beyond academia, as she has successfully translated complex research into commercial products. Havasi's lectures and publications have inspired a new generation of AI researchers to explore the intersection of language, knowledge, and machine learning.
Throughout her career, Havasi has maintained a focus on the practical applications of AI, ensuring that her research addresses real-world challenges. Her contributions to sentiment analysis, semantic networks, and common sense reasoning have left a lasting mark on the field, and her ongoing work continues to shape the future of AI-driven data analysis.