Browsing by Subject "Recommender systems (Information filtering)"

Now showing 1 - 5 of 5
  • Thumbnail Image
    Item
    Dark Reactions: Recommender Guided Materials Discovery
    (2014) Raccuglia, Paul; Friedler, Sorelle
    We present an exploration of data mining and machine learning techniques applied to a materials science dataset, with the goal of improving a lab's efficiency when running experiments. The primary product of our work is two tools to help chemists' better explore the space of possible reactions: a recommender system which we hope will increase the serendipitous discovery of interesting reactions that the chemists would not have thought to explore; and a seed-based ranking system which helps chemists prioritize which reactions to run, and with what parameters. We present a number of different techniques for tuning our recommender system, as well as presenting an automated approach to evaluating recommender systems in contexts where labels are expensive to learn (time, resources, equipment). Reactions are given a label in f1; 2; 3; 4g, where 4 corresponds to successful formation of a crystalline product, 3 corresponds to mostly successful formation of a crystalline product and 1 and 2 correspond to different failure cases. Using SVM we are able to achieve 65% accuracy on a 4-category classification on a held-out test set of 30% of our data set. Preliminary empirical results suggest a significant improvement in efficiency: observed rate of observing a 3 or a 4 increased from 65% (n=5486) without our system to 86% (n=190) using recommendations from our system. Our system is available at http://darkreactions.haverford.edu/.
  • Thumbnail Image
    Item
    From Tapestry to SVD: A Survey of the Algorithms That Power Recommender Systems
    (2009) Huttner, Joseph; Lindell, Steven
    This paper is a survey of the algorithms that power recommender systems. To start, the social and monetary relevance of recommender systems is outlined. Then we delve into the specifics of how the first recommender system, Tapestry, coined the idea of numerically defining customer similarity. Moving forward, we show how this central concept of similarity is re-hashed in present day recommender systems, namely that of Amazon.com. Specifically, we examine the complexity of a user-based approach in a large scale system such as Amazon's, identify its weaknesses, and see how these weaknesses are overcome using an item-based approach. The last component of this paper focuses on the Netflix Prize™ and investigates the single most important algorithm in the contest so far: an incremental approach to finding the singular value decomposition (SVD) of a mostly-blank matrix.
  • Thumbnail Image
    Item
    How Acquiring the “Perfect Match” within Clothing Recommendation Systems Relies on Human Judgement: Literature Review
    (2019) Quintero, Amanda-Lynn; Dougherty, John P.
    Recommendation systems exist in almost every aspect of online business. Whether the system is recommending books, movies, clothes, etc., an algorithm is running behind the scenes. These algorithms are hard to acquire in pseudocode because they are generally proprietary but I will analyze one algorithm found in my research that could hint at a recommendation algorithm’s general nature. Many such recommender algorithms implement some form of collaborative filtering, which utilizes historical user data, clustered product data, and user spending patterns to identify specific recommendations. Clothing recommendation systems are becoming increasingly more complex as they not only have to account for matching clothing products to user patterns and trends, but also the ability to analyze the product images themselves to match attributes of color, style, and clothing type to the user’s preferences. I will be focusing on clothing recommender systems with an ultimate suggestion on a new format that could potentially solve the problems of cold start (i.e. problem when algorithm does not have enough user data to begin with) and fashion style consistency (i.e. problem when fashion genres or clothing types are inconsistent skewing the algorithm’s ability to recommend products that match).
  • Thumbnail Image
    Item
    Social Recommender systems: Improving recommendations through personalization
    (2011) Surti, Tanvi; Lindell, Steven
    The vision for Web 3.0 (popularly referred to as the Semantic Web) is the ability to create meaning out of a deluge of qualitative data. This paper explores a very specific instance of the Semantic Web – Social Recommender Systems. This paper discusses the possibility of converting social crowdsourced data into quantitative information and using this information to power social recommendations. Over the course of this paper, we discuss five important recommender algorithms. This paper first outlines the importance of recommendations and elaborates the different types of recommendation algorithms used widely. We then discuss the potential to further personalize these recommendations by trying to identify user ‘taste’ by capitalizing on the social data available about the user. Next, we discuss the availability and applicability of data from social networks and how this data may be processed into quantitative input. This data is then used as input for two different social algorithms and their merits are discussed. And lastly this paper covers the topic of creating hybrid systems out of a wide range of recommendation algorithms so as to create social systems which are able to give diverse and personalized recommendations.
  • Thumbnail Image
    Item
    Who Can Use What? Proposals for the Development of a Recommender System for Selecting the Most Appropriate Text Input Method for a Given User Based on the User’s Physical Abilities.
    (2013) Sigmond, Carl; Dougherty, John P.
    Those who cannot use computers are severely disenfranchised by society’s present and future reliance on digital technology. Persons with disabilities comprise a significant portion of this disenfranchised group. For persons with disabilities, one of the biggest obstacles to computer use is the input of text. This thesis aims to be the basis for the development of a recommender system to aid users with a wide array of physical abilities in selecting the most applicable text input method for them. We present a survey of text input devices both for people with disabilities and for people in the general population. We then propose a framework for the development of a full scale recommender system for selecting the most appropriate text input method for a given user based on that user’s physical abilities.