This invention helps to measure engagement of digital content by combining signals from various sources including web traffic, search traffic, and social traffic. In addition, revenue and customer retention factors are included to provide a holistic view of the value of a piece of content while taking into account where it is placed on the website.
US Patent No. 20190384857
Automated food recognition has great value in terms of health benefits, as it provides a more efficient and accurate method of recording an individual's diet. It is the high variability within the same food groups and yet, the subtle distinctions between different food groups that makes food recognition a difficult task.
In this work, we evaluated and compared three popular image descriptors for their performance on two challenging food data sets: 50 Chinese Foods and ImageNet Foods, the latter of which we collected from ImageNet. We also proposed a method to automatically discover food related attributes to give rise to a recipe-based food descriptor. Our food descriptors were able to improve the performance of the baseline descriptors by encoding attributes such as ingredients, cooking method, and nutritional content. We also demonstrated that our food descriptors are capable of knowledge transfer for zero-shot learning. [Github Repo]
Keywords: image recognition, classification, fine-grained recognition
Recommendation algorithms are either based on content or driven by collaborative filtering. We study different ways to incorporate content information directly into the matrix factorization approach of collaborative filtering. These content-boosted matrix factorization algorithms not only improve recommendation accuracy, but also provide useful insights about the contents, as well as make recommendations more easily interpretable.
Nguyen, Jennifer, and Mu Zhu. "Content‐boosted matrix factorization techniques for recommender systems." Statistical Analysis and Data Mining: The ASA Data Science Journal 6.4 (2013): 286-301. [PDF]
Keywords: recommender systems, collaborative filtering, matrix factorization
This study aims to find specific contexts in which users in social networks influence each other. With these contexts, applications such as recommendation engines can focus on a specific context for making recommendations. We pose the problem of finding contexts of social influence where the influence is similar across all items in the context. We present a full-space clustering algorithm and a subspace clustering algorithm to find these contexts, and demonstrate that our algorithms are capable of finding meaningful contexts of influence in addition to rediscovering the predefined categories specific to the Digg news site.
Nguyen, Jennifer H., et al. "Finding contexts of social influence in online social networks." Proceedings of the 7th Workshop on Social Network Mining and Analysis. ACM, 2013. [PDF]
Keywords: social networks, clustering, knowledge discovery
Are you interested in becoming a data scientist and not sure where to begin? In this guide, I share with you the skills needed to begin a career in data science and how to get them.