Trang 1
CS361 (Software Engineering Program) Artificial Intelligence II-Applied Machine Learning Lecture 6 Unsupervised Learning (Clustering) A Basic Introduction to Recommendation Systems Amr S. Ghoneim (Assistant Professor, Computer Science Dept.) Helwan University Fall 2019 Lecture is based on its counterparts in the following courses (and the following resources): o Recommender Systems: An introduction, lecture at the University of Szeged (Hungary), by Dietmar Jannach (Technische Universität "TU" Dortmund, Germany) o PV254 Recommender Systems, Masaryk University (Czech Republic), Department of Machine Learning and Data Processing – Faculty of Informatics o Recommender Systems: An Introduction. D. Jannach, M. Zanker, A. Felfernig, G. Friedrich, 2010. o Recommender Systems Handbook. F. Ricci, L. Rokach, B. Shapira, P. B. Kantor, 2015 (second edition).
Trang 2
Today’s Key Concepts o Machine Learning Approaches & their corresponding Applications o Motivation o the Jam experiment .. o Examples of Applications o Recommendation Systems & Education o Recommendation Systems o An “often-cited” Problem Characterization o Various Paradigms o Collaborative Filtering (CF) o User-based Nearest Neighbor CF o Pearson Correlation o Item-based CF o Content-based Recommendation o What is the “Content” ? o Content Representation o Item Similarities o Term-Frequency (𝑇𝐹) o Inverse Document Frequency (𝐼𝐷𝐹) o Example: TF–IDF Representation o Improving the Vector Space Model o Cosine Similarity o Recommending Items
Trang 5
Motivation & Examples of Applications o Information overload; o Manychoicesavailable.. o “the paradox of choice” (Choice Overload, .. “Whymoreisless”).. o Recommender System; o Provides aid.. o Set of items + user “context” ⇒ selection of items(predictedto be“good” for theuser) Movies, online videos, music, books, software (apps), products in general, people (dating, friends), services (restaurants, accommodation, .., etc.), research articles, jokes, .. etc.
Trang 8
Motivation & Examples of Applications Recommendation Systems & Education o Learningmaterials–directapplication o Problems,exercises: o Users∼Students o Items∼Problems o Ratings ∼ Performance (correctness of answers, problem solvingtimes) Personalization in Education: o Adaptive learning, Personalized learning, ..
Trang 9
Value of Recommendations .. Value for the customer: ▪ Find things that are interesting. ▪ Narrow down the set of choices. ▪ Help me explore the space of options. ▪ Discover new things. ▪ Entertainment. ▪ .. Value for the provider: ▪ Additional & probably unique personalized service for the customer. ▪ Increase trust and customer loyalty. ▪ Increase sales, click trough rates, conversion, .. etc. ▪ Opportunities for promotion, persuasion. ▪ Obtain more knowledge about customers. ▪ ..
Trang 10
Value of Recommendations .. ▪ Netflix: ~ 2/3 of the movies watched ▪ Amazon: ~ 35% of the sales ▪ Google News: Recommendations ⇒ ~ 38% more click through.
Trang 11
An “often-cited” Problem Characterization Given .. • The profile of the "active" user and possibly some situational context. Compute .. • A relevance (ranking) score for each recommendable item. The Profile: • Can include past user ratings (explicit or implicit), demographics and interest scores for item features. The Problem: • is to learn a function that predicts the relevance score for a given (typically unseen) item.
Trang 12
Models of Recommender Systems Recommender systems reduce information overload by estimating relevance. Usefulness of Recommendations; Is it worthwhile? It depends .. o Is there s “large” number of items? o Do users know exactly what are they looking for?
Trang 13
Models of Recommender Systems Personalized recommendations Implementing Personalized Systems is Difficult; (always) difficult debugging, testing, evaluation .. .. personalization ⇒ different behaviour for each user, thus, hard to distinguish bugs and surprising results ..
Trang 17
Models of Recommender Systems Hybrid: combinations of various inputs and/or composition of different mechanism ..
Trang 18
Recommender Systems .. versus .. Information Retrieval o Information retrieval (IR) is the activity of obtaining information resources relevant to an information need from a collection of information resources. (Wikipedia) o The goal of a Recommender System is to generate meaningful recommendations to a collection of users for items or products that might interest them. (Melville, Sindhwani) Recommender Systems and Information Retrieval are closely connected (many similar techniques), yet .. .. different goals: o Information Retrieval – “I know what I’m looking for” o Recommender Systems – “I’m not sure what I’m looking for”
Trang 19
Collaborative Filtering .. & .. Rating “tell me what’s popular among my peers ( = similar users)” ➢ one of the most often and successfully used techniques widely applicable, does not need any domain knowledge.
Trang 20
Collaborative Filtering .. & .. Rating Recommender systems (particularly collaborative filtering) rely on user “ratings” .. • Rating of item ∼ how much the user likes the item. There’re many different forms of ratings, for example: o Explicit o Like-it scale (5 stars), like/dislike require additional effort from users. o Implicit o Click through rate, buying an item, visiting a page, viewing a video, dwell time .. o Easier to collect, less precise .. o More “honest” (Netflix example: highly rated vs watched).
Hiển thị 20 trang đầu. Tải tài liệu để xem đầy đủ.
Bình luận (0)
Chưa có bình luận nào. Hãy là người đầu tiên!