David Zibriczky, profile picture
Uploaded byDavid Zibriczky
PDF, PPTX3,015 views

An introduction to Recommender Systems

Recommender systems are software agents that analyze a user's preferences through transactions and provide personalized recommendations accordingly. There are several recommendation paradigms including non-personalized rules, personalized rules based on user data, and transaction-based collaborative filtering that learns from user interactions. Context-based recommender systems also consider additional information like time, location, or device to provide adaptive recommendations. Common techniques used in recommender systems include content-based filtering that recommends similar items, collaborative filtering that finds users with similar tastes, and demographic-based recommendations.

Embed presentation

Download as PDF, PPTX
1 / 105
2 / 105
3 / 105
4 / 105
5 / 105
6 / 105
7 / 105
Most read
8 / 105
Most read
9 / 105
10 / 105
11 / 105
12 / 105
13 / 105
14 / 105
15 / 105
16 / 105
17 / 105
18 / 105
19 / 105
20 / 105
21 / 105
22 / 105
23 / 105
24 / 105
25 / 105
26 / 105
27 / 105
28 / 105
29 / 105
30 / 105
31 / 105
32 / 105
33 / 105
34 / 105
35 / 105
36 / 105
37 / 105
Most read
38 / 105
39 / 105
40 / 105
41 / 105
42 / 105
43 / 105
44 / 105
45 / 105
46 / 105
47 / 105
48 / 105
49 / 105
50 / 105
51 / 105
52 / 105
53 / 105
54 / 105
55 / 105
56 / 105
57 / 105
58 / 105
59 / 105
60 / 105
61 / 105
62 / 105
63 / 105
64 / 105
65 / 105
66 / 105
67 / 105
68 / 105
69 / 105
70 / 105
71 / 105
72 / 105
73 / 105
74 / 105
75 / 105
76 / 105
77 / 105
78 / 105
79 / 105
80 / 105
81 / 105
82 / 105
83 / 105
84 / 105
85 / 105
86 / 105
87 / 105
88 / 105
89 / 105
90 / 105
91 / 105
92 / 105
93 / 105
94 / 105
95 / 105
96 / 105
97 / 105
98 / 105
99 / 105
100 / 105
101 / 105
102 / 105
103 / 105
104 / 105
105 / 105
1 INTRODUCTION TO RECOMMENDER SYSTEMS 2014.12.04 David Zibriczky (ImpressTV)
2 About ImpressTV • Netflix Prize (2006-2009) • The Ensemble: Runner-up from 40K teams • Gravity team: Members of The Ensemble • Gravity R&D › Hungarian start-up, launched in 2007, B2B business › On-site personalization solution provider company for e-commerce, IPTV, OTT and classified media › 100M+ recommendations per day • ImpressTV Limited › IPTV and OTT was acquired from Gravity R&D by british investors in July 2014 › HQ in Budapest, international corporate clients • About me › Joined Gravity R&D in January 2010, transfered to ImpressTV Limited in July 2014 › Current position: Head of Data Science at ImpressTV Limited
3 Introduction
4 Information overload I.
5 • Too many existing items, too many options › YouTube videos, Amazon books, Netflix movies, Pandora musics, … • High item publishing intensity, hard to follow the flow of information › Google News, Facebook posts, Twitter tweets, … • Users are not qualified enough › They know what they want to watch/buy, but they don’t know whether a given item would satisfy their needs › E.g.: Grandma would like a new computer, knows what she wants to do with it, but has no idea of the computer parts and what they do • One may not know what they would like › First time in a Chinese restaurant, no experience about Chinese food. • How to improve the user satisfaction? Information overload II. – Consumers
6 • Challange in handling ever increasing number of contents and corresponding data • Challenge in handling transactional data • Increasing amount of inner information • What is useful? • How to determ the usefulness of data? • Competition in business market › keeping and converting consumers is essential › keeping the market adavantage › increasing revenue • How to exctract and use the collected information to improve business success? Information overload II. – Content Providers
7 „Recommender Systems (RS) are software agents that elicit the interests and preferences of individual consumers […] and make recommendations accordingly. They have the potential to support and improve the quality of the decisions consumers make while searching for and selecting products online.„ 1 What are Recommender Systems? 1 Xiao, Bo, and Izak Benbasat, 2007, E-commerce product recommendation agents: Use, characteristics, and impact, Mis Quarterly 31, 137-209. Recommender Systems
8 Recommender Systems as software agents Items Users Recommender System Recommend item X to user A
9 Examples / Netflix (Movies)
10 Examples / YouTube (Uploaded Videos)
11 Examples / Spotify+Last.fm (Music)
12 Examples / Amazon (Books)
13 Examples / LinkedIn (News, Jobs, Groups, People, …)
14 Examples / Twitter (Tweets, Peoples, Groups, Companies)
15 Examples / ScienceDirect (Papers)
16 Examples / StackOverflow (Answers)
17 Examples / Grocery (Foods, Drinks)
18 • News (What to read now?) • Dating sites (Which girl to contact?) • Coupon portals (Which coupon is good for me?) • Restaurants (Where to eat?) • Vacations (Where to travel and what to see?) • Retailers (How much products to supply?) • Financial assets (loan, portfolio) • Advertisements • …etc Other Examples
19 • For the consumers (users) › Helping the users to find useful contents to satisfy their needs › Reducing the time of content searching › Providing relevant information from the massive information flow › Exploring new preferences, trust in recommender system • For the business › Improving business success indicators (or key performance indicators, KPI) • Increasing revenue, CTR, watching/listening duration • Increasing conversion rate and user engagement, reducing churn • Cross-selling, upselling, advertisement › Reducing popularity effect, less popular contents are also consumed › Promotions, targeting, campaign Goals and benefits of Recommender Systems
20 • Goals for consumers not necessarily equal to business! • Simplified example 1: YouTube (free contents) › Goal of the business: Receive more income from advertisements How: More videos watched, more advertisements seen › Goal of the users: Having good/useful time by watching/listening videos How: Clicking on recommendations, using search engine › The goal is realized in the same way, more videos watched are good for both. • Simplified example 2: Netflix (DVD/Blu-ray or Video On Demand rental) › Goal of the business: Increase the income How: More expensive contents, improving user engagement › Goal for the users: Buying interesting movies, spending less time for searching How: Using recommendation engine to make it easier › The goal is different. Netflix wants the users pay more. The users basically don’t want to spend more, only if they find it worth. Difference between the goals of consumers and business
21 Recommendation paradigms
22 Recommender Systems as software agents Items Users Recommender System Recommend item X to user A
23 • Items: Entites that are recommended (movies, musics, books, news, coupons, restaurants, etc...) • Item data: Descriptive information about the items (e.g. genre, category, price) • Users: People to whom we recommend (Who is the user? Member, cookie, unidentified?) • Paradigm: Recommendations are calculated by predefined non-personalized rules • How: Setting item data based rules (recommending the latest movies for all users) • Properties: › Non-personalized static recommendations › Requires manual work (e.g editorial pick) Recommendation paradigms / Non-personalized rules Items Recommender System RecoItem data Users
24 • User data: Descriptive information about the users (e.g. age, gender, location) • Paradigm: Recommendations are calculated by predefined used data based rules • How: Setting item and user data based rules (men between 45-55  expensive cars) • Properties: › Semi-personalized › Requires manual work (e.g rule constructing) › Interpretable Recommendation paradigms / Personalized rules Items User data Recommender System Reco Item data Users
25 • Transactions: Interaction between users and items • Transaction types: Numerical ratings, ordinal ratings, binary ratings, unary ratings, textual reviews • Explicit feedback: The user quantifies his preference about an item (rating) • Implicit feedback: Events that indicates but not quantifies the preference of the user about an item › Positive: buy, watch, like, add to favourite › Negative: dislike, remove from favourites • In practice, implicit feedback is less valueable than explicit feedback, but significantly more provided Recommendation paradigms / Transaction based personalization I. Items User data Recommender System Reco Item data Users Transactions
26 • Paradigm: Recommendations are calculated based on the users interactions • How: Learning on interactional data (collaborative filtering) • Properties: › Personalized › Adaptive › Less interpretable Recommendation paradigms / Transaction based personalization II. Items User data Recommender System Reco Item data Users Transactions
27 • Context: Information that can be observed in the time of recommendation • Types: › Time-based temporal information (Recos of sports equipments should be different in summer and in winter) › Mood (Different TV programs should be recommended to the same person based on her current mood) › Device (Different content types based on device type, e.g. movie trailers in TV, music in phone) › Location › Event sequence information (If the user browses the TV category of a webshop, it is reasonable to recommend DVD players, but it is not if the user browses the laptop category) Recommendation paradigms / Context based personalization I. Items User data Recommender System Reco Item data Users Context Transactions
28 • Paradigm: Recommendations are calculated based on the users interactions and current context • How: Learning on contextual data (the user usually watches new in the morning, action at night) • Properties: › Context-sensitive, fully adaptive › More interpretable › More complex Recommendation paradigms / Context based personalization II. Items User data Recommender System Reco Item data Users Context Transactions
29 Cross-domain recommendation Task Goal Multi-domain Cross- selling Diversity Serendipity Linked-domain Accuracy Cross-domain Cold-start New users New items • Multi-domain: You watched movies and bought books, we recommend movies or books • Linked-domain: You bought books only, we recommend books using both movie and book consumption patterns • Cross-domain: You watched movies only, we recommend books, based on movie-book consumption relationship Domain Example Ratio Attribute Comedy Thriller 12% Type Movies   Books 9% Item Movies  Restaurants 55% System Netflix  MovieLens 24%
30 • Item to User › Conventional recommendation task: „You may like these items”. • Item to Item › Recommending items that are somewhat similar to the item currently viewed by the user › Personalized or non-personalized similarity • User to User › Recommending different users for the users based on metadata or activity › Social recommendations (who to follow, who to connect) or similar users • User to Item › Promoting items that the seller wants to sell to the most probable buyers (e.g. newsletters, notifications) › „Who would buy this item?” • Group recommendations › Recommending group of items or users Recommendation types
31 Recommendation techniques
32 • Content-based Filtering (CBF) › Recommend items that are similar to the ones that the user liked in the past. › Similarity based on the metadata of the items. › E.g.: If the user likes romantic movies, recommend her the like • Collaborative Filtering (CF) › Recommend items that are liked by users that have similar taste as the current user › Similarity betwen users is calculated by the transaction history of users › Only uses the transaction data  domain independent • Demographic › Recommendations made based on the demographic profile. › Realizes a simple personalization. › E.g.: Recommend computer parts to young people who study informatics. Recommendation techniques I.
33 • Knowledge-based › Uses extensive domain-specific knowledge to generate recs. › User requirements collected (problem), items as possible solutions to the specific problems › Example: user would like to find a digital camera, she provides her needs and the level of her skills, etc. › Pure knowledge based systems (without learning) tend to perform better in the beginning of deployment but they fall behind later. • Community-based › Recommendations based on the preferences of the user’s friends. › People tend to accept recommendations of their friends. › Using of social networks  social recommender systems. • Hybrid recommendation systems › Combination of the techniques above. › Trying to use the advantages and fix the disadvantages of the different techniques. Recommendation techniques II.
34 1. Setting the recommendation type (e.g. Item2user recommendation) 2. Requesting a recommendation 3. Selecting the recommendable items, filtering (e.g. new movies) 4. Selecting the algorithm for scoring these items (e.g. a collaborative filteirng algorithm) 5. The algorithm provides score for each items 6. Ordering the list by the scores 7. Post processing the input item list (e.g. randomizing topN, selecting one episode per series, etc…) 8. Selecting the first N items to send back as a response of recommendation request Recommendation data flow
35 • Key features of a good recommender systems, that should be considered › Accuracy (efficiency of modeling the user preference) › Adaptation, context awareness (ability to detect changes in user behavior) › Diversity, coverage (avoiding monotone recommendations, and preference cannibalization) › Novelty, serendipity (improving suprise factor, exploitation vs. exploration) › Trust, explanation (the users should understand and trust in recommender system) › Scalability, responsivity, availability (recommendations should be provided in reasonable time) • Tradeoffs in recommender systems › Accuracy vs. Diversity › Discovery vs. Continuation › Depth vs. Coverage › Freshness vs. Stability › Recommendations vs. Tasks Properties of recommender system
36 • Method › Splitting data set into disjunct train and test set › By time, random, users, user history › Training on train set, measuring on test set • Evaluation metrics › Accuracy: Rating (RMSE), TopN (nDCG@N, Recall@N, Precision@N) › Coverage: Ratio of the recommended items › Diversity: Entropy, Gini-index › Novelty: Ratio of the long tail items › Serendipity: Ratio of the less › Training time › Accuracy is the typical primary metric, others are secondary Offline Evaluation DataSet Train Test
37 • Method › Effects of recommender systems are measured › A/B testing: Splitting user base into equally valuable subsets, different recommendations for the subsets › A/B tests avoid › Online improvement can be measured • Evaluation metrics › Some effects cannot be evaluated before the recommender system set live › Click Though Rate (CTR, Number of clicks per recomendation) › Average Revenue Per User (ARPU), Total Revenue Increase › Page impression (number of page views) › Conversion Rate, churn Rate Online Evaluation
38 1. Understanding the business needs › Business goals, Key Performance Indicators for live recommender system › Recommendation scenarios, placeholders › Data understanding 2. Integration (for Vendors only) › Developing data integration method (the way the customer provide it’s data) › Setting up recommendations request/response interface (how the customer request a recommendation, and the vendor provides response, e.g. JSON or REST API) 3. Data preparation › Data enrichment from external sources (e.g. crawling additional item meta data) › Data transformation (shall we handle a complete series as an item, or a group of episodes?) Workflow of Recommender System integration
39 4. Data Mining and Offline Experiments Workflow of Recommender System integration
40 5. Online Evaluation / Deployment › Setting recommendation engine live (starting to provide recommendations for real end-users) › Measuring online performance metrics (e.g. CTR, ARPU or page impression) › Measuring response times (is the algorithm fast enough in live service?) › Analyzing correlation between the offline and online metrics (what to optimize in offline experiments) 6. Optimization › Implementing additional algorithms (Step 4) › A/B testing (is the new algoritm better than the original one?) › Statistical significance based A/B selection (e.g. T-test between the performance of algorithm A and B) 7. Reporting and follow-up › Reporting performance (weekly, monthly) › Monitoring response times and availability (if the algorithm not roboust enough, it may result outage) › Monitoring change in user behavior, size of user base › Adapting to new features (e.g. new placeholders in the website) Workflow of Recommender System integration
41 Content-based Filtering
42 • Content-based Filtering (CBF) › Recommend items that are similar to the ones that the user liked in the past › Similarity based on the metadata of the items › E.g.: If the user likes horror movies, recommend her horror movies • Collaborative Filtering (CF) • Demographic • Knowledge-based • Community-based • Hybrid recommendation systems Recommendation Techniques
43 Recommending an item to a user based upon a description of the item and a profile of the user’s interests • Items are represented by their metadata (e.g genre or description) • Users are represented by their transaction history (e.g. item A and B was rated by 5) • We generate user profiles from the user transaction data and the metadata of the items in the user’s transactions • The user profiles are compared to the representation of the items • The items similar to the user profile are recommended to the user CBF Method Twilight 5 5 ? ? Sci-Fi RomanceAdventure The Matrix The Matrix 2 The Matrix 3
44 Architecture
45 Content Analyzer • Preprocessing module • Input: Item meta data • Methods: › Text mining › Semantic Analysis › Natural Language Process › Feature extraction › Meta data enrichment › Auto tagging • Output: Item models
46 • We create terms from the item metadata › Standard text mining preprocessing steps • Filtering stopwords • Filtering too rare/common terms • Stemming • Items are represented by a (sparse) vector of terms › The vector of each item (=document) contains weights for each term. › Weighting is often done by the TF-IDF scheme • Rare terms are not less relevant than frequent terms (IDF). • Multiple occurences in a document are not less relevent than a single occurence (TF). • Invariant of the length of the document (normalization). • Similarity measurement › Most common: cosine similarity • Scalar product of the L2 normalized vectors Content Analyzer / Vector space model (VSM)                  T k k T k k T k kk ww ww iisim 1 2 2 1 2 1 1 21 21 ),(
47 • Semantic analysis in order to extract more information about the items • Domain-specific knowledge and taxonomy is used Content Analyzer / Semantic analysis, ontologies
48 • The meta data provided by clients is usually not enough • Meta data enrichment: Crawling additional data about the items from public web • Publicly available knowledge sources › Open Directory Project › Yahoo! Web Directory › Wikipedia › Freebase › Tribune › Etc… • Goals: › Fixing missing or wrong data › Better characterization of items › Improving the accuracy of CBF algorithms Content Analyzer / Meta data enrichment
49 Profile Learner • User preference modeling • Inputs: › Item models › User transactional data › User meta data • Methods: › Meta data weighting › Machine learning • Output: User profile
50 Profile Learner / Keyword-based profiles
51 • Negative and positive examples collected › Explicit case: Items rated under 3 counts negative, above positive (variants: above overall average, user average…) › Implicit case: Items viewed counts positive, others negative • Approach 1 › User profile: weighted average of item vectors (Rocchio’s algorithm) › Similarity between user profile and item vectors as relevance score • Approach 2 › Negative and positive user profiles › Items that match negative profile are filtered › Similarity between the positive profile and the item’s vector as relevance score • Approach 3 › Rule based classifiers (decision trees, decision lists) learning on the examples › New items are judged by the classifier • Approach 4 › Nearest Neighbor methods Profile Learner / User profile in VSM model
52 • The terms in the item metadata are handled independently • Semantic information is lost in the process, we have no idea of the meaning of the terms • It is hard to handle expressions containing more than one word • Data sparsity • Need diverse user feedbacks to improve the accuracy Profile Learner / Problems with the VSM model
53 Profile Learner / Phenomenon of polysemy
54 Profile Learner / Word sense disambiguation
55 Architecture / Filtering Component • Recommendation list filtering • Inputs: › User profile model › Item models › Recommendable items • Methods: › Filtering non-relevant items › Similarity based ranking • Output: Recommended items
56 • User independence › Profiles of the users are built independently › A user can not influence the recommendations for other users (no attacks on the recommender system) • Interpretable › Recommendations can be easily explained (we recommend this comedy because you watched another comedies) › Explanations can easily built, user profiles are easy to be understood (user keyword model) • Item cold start › Solving item cold start problem › Capable of recommending items that were never rated/viewed by anybody Advantages of Content-based Filtering
57 • Limited content analysis › Natural limit in the number and type of features that are associated with the recommended items › Domain knowledge is often needed • Over-specialization › Hard to improve serendipity (recommending some unexpected) › Recommendations mirrors the user history › Monotone recommendations: „The winner takes it all” effect › Exploitation over exploration • Meaning › word meaning disambiguation (apple = company or the fruit?) • User cold start › Needs some ratings before the profile learner can build an accurate user model › There will be no reliable recommendations when only a few ratings available Disadvantages of Content-based Filtering
58 Disadvantages / Word meaning disambiguation
59 • Deep learning (Sentiment Analysis, Paragraph Vector Model) • User generated contents (Tagging, Folksonomies) • Serendipity problem › Using randomness or genetic algorithms › Filtering too similar items › Balance between exploration and exploitation › Using poor similarity measures to produce anomalies and exceptions • Centralized knowledgebase for meta data enrichment Research topics in Content-Based Filtering
60 Collaborative Filtering
61 • Content-based Filtering (CBF) • Collaborative Filtering (CF) › Recommend items that are liked by users that have similar taste as the current user › Similarity between users is calculated by the transaction history of users › Only uses the transaction data  domain independent • Demographic • Knowledge-based • Community-based • Hybrid recommendation systems Recommendation Techniques
62 – Explicit Collaborative Filtering – Implicit Collaborative Filtering Explicit Collaborative Filtering
63 5 ? ? The Matrix The Matrix 2 Twilight The Matrix 3 ? • Classic item recommendation (Netflix) on explicit feedbacks (ratings) • Rating problem: The goal is to predict how the user would rate the items • Accuracy metrics: › Root Mean Squared (RMSE) › Mean Absolute Error (MAE)   test Rriu iu R rr RMSE test    ),,( 2 , ˆ test Rriu iu R rr MAE test    ),,( , ˆ Rating prediction problem
64 5 ? ? The Matrix The Matrix 2 Twilight The Matrix 3 ? • Global average (average of all ratings) • User average (average of user ratings) • Item average (average of ratings given to the item) Baseline Methods
65 5 5 4 5 5 ? ? The Matrix 3The Matrix The Matrix 2 Twilight 5 ? Explicit Collaborative Filtering
66 R Item1 Item2 Item3 Item4 User1 5 ? ? ? User2 5 User3 5 4 User4 5 5 Explicit CF / Rating Matrix Representation 5 5 4 5 5 ? ? Item4Item1 Item2 Item3 5 ?
67 Collaborative Filtering – Explicit Collaborative Filtering – Neighbor methods – Matrix Factorization – Implicit Collaborative Filtering Explicit Neighbor Methods
68 • User based neigbor methods › 1. Find users who have rated item i › 2. Select k from these users that are the most similar to u › 3. Calculate R(u,i) from their ratings on i • Item based neigbor methods › 1. Find items that have been rated by user u › 2. Select k from these items that are the most similar to i › 3. Calculate R(u,i) from their ratings by u Explicit CF / Neigbor methods
69 • Similarity between › Rows of the preference matrix (user based) › Columns of the preference matrix (item based) › Dimension reduction on user/item preference vectors  feature vectors • Similarity properties: › S(a,a) >= S(a,b) (often S(a,a) = 1 required) › S(a,b) = S(b,a) • Similarity measures › Cosine similarity (CS) › Pearson correlation (PC) › Adjusted cosine similarity (ACS) › Eucledian distance (EUC) Explicit CF / Neigbor methods / Similarity setup 2 2 uv u v ui vi i I ui vj i I j I r r CS r r               uvuv uv Ii vvi Ii uui Ii vviuui rrrr rrrr PC 22 )()( ))(( 2 2 ( )( ) ( ) ( ) uv uv uv ui i vi i i I ui i vi i i I i I r r r r ACS r r r r               uvIi viui rrEUC 2
70 • Advantages › Simplicity (easy to set up, only a few parameters) › Justifiability (recommendations can be explained) › Stability (not very sensitive to new items/users/ratings) › Good for item2item and user2user recommendation • Disadvantages › Computationally expensive in the recommending phase • Similarities have to be computed for recommendation • Similarity maxtrix might be computed before recommendations but then there is a costly training phase • Similarity matrix might not fit in the memory › Less accurate than model based methods for personalized recommendation Advantages and disadvantages of neighbor methods
71 Collaborative Filtering – Explicit collaborative filtering – Neighbor methods – Matrix Factorization – Implicit collaborative filtering Explicit Matrix Factorization
72 𝑹 = 𝑷𝑸 𝑻 𝑟 𝑢𝑖 = 𝒑 𝑢 𝑇 𝒒𝑖 𝑹 𝑵𝒙𝑴: rating matrix 𝑷 𝑵𝒙𝑲: user feature matrix 𝑸 𝑴𝒙𝑲: item feature matrix 𝑵: #users 𝑴: #items 𝑲: #features 𝑲 ≪ 𝑴, 𝑲 ≪ 𝑵 R Item1 Item2 Item3 … User1 User2 𝒓 𝑢𝑖 User3 … P 𝒑 𝑢 𝑇 QT 𝒒𝑖 Explicit CF / Matrix Factorization
73 • ALS (Alternating Least Squares) › Updating P and Q matrices by multivariate linear regression based solver • BRISMF (Biased Regularized Simultaneous Matrix Factorization) › Stochastic gradient descent based matrix factorization › Minimizing the prediction error by iterating on transactions and modifying factors › Using bias for user and item model › Prediction: • SVD (Singular Value Decomposition) › Decomposes matrix R to three matrices › S is a diagonal matrix, containing singular values of matrix R • NSVD1 › Decomposes matrix R to three matrices › W is a weight matrix Explicit CF / Matrix Factorization iu K k kiukiuiuui cbqpcbqpr  1 ˆ  T R PSQ T R PWQ
74 Explicit CF / Matrix Factorization / Model Visualization P John 1.1 0.5 Paul 0.6 0.9 Suzy 1.0 0.9 Item 1 Item 2 Item 3 Item 4 Item 5 QT 1.0 -0.9 0.5 -0.5 1.3 -0.3 1.6 0.7 1.6 -1.1
75 Explicit CF / Matrix Factorization / Model Visualization P John 1.1 0.5 Paul 0.6 0.9 Suzy 1.0 0.9 Item 1 Item 2 Item 3 Item 4 Item 5 QT 1.0 -0.9 0.5 -0.5 1.3 -0.3 1.6 0.7 1.6 -1.1 J P S 1 2 3 4 5
76 Explicit CF / Matrix Factorization / Clustering P John 1.1 0.5 Paul 0.6 0.9 Suzy 1.0 0.9 Item 1 Item 2 Item 3 Item 4 Item 5 QT 1.0 -0.9 0.5 -0.5 1.3 -0.3 1.6 0.7 1.6 -1.1 J P S 1 2 3 4 5 A B C
77 Memory based algorithms Model based algorithms Hierarchy of Collaborative Filtering methods Collaborative Filtering Matrix factorization Explicit feedback base algorithms Implicit feedback base algorithms Neighbor methods
78 – Explicit Collaborative Filtering – Implicit Collaborative Filtering Implicit Collaborative Filtering
79 ? ? The Matrix The Matrix 2 Twilight The Matrix 3 ? • Classic item recommendation on implicit feedbacks (view, buy, like, add to favourite) • Preference problem: The goal is to predict the probability that the user would choose that item • Accuracy metrics: › Ranking error instead of prediction error › Recall@N › Precision@N Preference prediction problem
80 ? ? The Matrix The Matrix 2 Twilight The Matrix 3 ? • Most popular items • Most recent items • Most popular items in the user’s favourite category Baseline Methods
81 ? ? The Matrix 3The Matrix The Matrix 2 Twilight ? Implicit Collaborative Filtering
82 Implicit CF / Preference Matrix ? ? Item4Item1 Item2 Item3 ? • Implicit Feedbacks: Assuming positive preference • Value = 1 • Estimation of unknown preference? • Sorting items by estimation  Item Recommendation R Item1 Item2 Item3 Item4 User1 1 User2 1 User3 1 1 User4 1 1
83 • Zero value for unknown preference (zero example). Many 0s, few 1s, in practice- • 𝒄 𝑢𝑖 confidence for known feedback (constant or function of the context of event) • Zero examples are less important, but important. R Item1 Item2 Item3 Item4 User1 1 0 0 0 User2 0 0 1 0 User3 1 1 0 0 User4 0 1 0 1 C Item1 Item2 Item3 Item4 User1 𝒄11 1 1 1 User2 1 1 𝒄23 1 User3 𝒄31 𝒄32 1 1 User4 1 𝒄42 1 𝒄44 Implicit CF / Confidence Matrix
84 Collaborative Filtering – Explicit Collaborative Filtering – Implicit Collaborative Filtering – Neighbor methods – Matrix Factorization Implicit Neighbor Methods
85 • User based neigbor methods › 1. Find users who have rated item i › 2. Select k from these users that are the most similar to u › 3. Calculate R(u,i) from their ratings on i • Item based neigbor methods › 1. Find items that have been rated by user u › 2. Select k from these items that are the most similar to i › 3. Calculate R(u,i) from their ratings by u Implicit CF / Neigbor methods
86 • Co-occurrence similarity algorithm (most often used in business) • The algorithm: 1. For all item i: 1. Count the number of users that interacted with the item i: 𝑠𝑢𝑝𝑝(𝑖) 2. For all user u that interacted with the item i: 1. For all item i iteracted by user u: 1. Increment the counter of the co-occurrence of item i and j: 𝑠𝑢𝑝𝑝 𝑖, 𝑗 2. The similarity between item i and item j: 𝑆 𝑖, 𝑗 = 𝑠𝑢𝑝𝑝(𝑖,𝑗) 𝑠𝑢𝑝𝑝 𝑖 +𝛾 1−𝛼 𝑠𝑢𝑝𝑝 𝑗 +𝛾 𝛼 • 𝛼: popularity factor • 𝛾: reguralization factor 3. The prediction for user u and item i: 𝑟 𝑢𝑖 = 𝑗∈𝐶 𝑢>0 𝑐 𝑢𝑗 𝑆(𝑖,𝑗) 𝑗∈𝐶 𝑢>0 𝑐 𝑢𝑗 Implicit CF / Item based neigbor methods
87 Collaborative Filtering – Explicit Collaborative Filtering – Implicit Collaborative Filtering – Neighbor methods – Matrix Factorization – Tensor Factorization Implicit Matrix Factorization
88 𝑹 = 𝑷𝑸 𝑻 𝑟 𝑢𝑖 = 𝒑 𝑢 𝑇 𝒒𝑖 𝑹 𝑵𝒙𝑴: preference matrix 𝑷 𝑵𝒙𝑲: user feature matrix 𝑸 𝑴𝒙𝑲: item feature matrix 𝑵: #users 𝑴: #items 𝑲: #features 𝑲 ≪ 𝑴, 𝑲 ≪ 𝑵 R Item1 Item2 Item3 … User1 User2 𝒓 𝑢𝑖 User3 … P 𝒑 𝑢 𝑇 QT 𝒒𝑖 Matrix Factorization
89 • Objective function: C Item1 Item2 Item3 Item4 User1 𝒄11 1 1 1 User2 1 1 𝒄23 1 User3 𝒄31 𝒄32 1 1 User4 1 𝒄42 1 𝒄44 𝒇 𝑷, 𝑸 = 𝑾𝑺𝑺𝑬 = (𝒖,𝒊) 𝒄 𝒖𝒊 𝒓 𝒖𝒊 − 𝒓 𝒖𝒊 𝟐 𝑷 = ? 𝑸 = ? R Item1 Item2 Item3 Item4 User1 1 0 0 0 User2 0 0 1 0 User3 1 1 0 0 User4 0 1 0 1 Weighted Sum of Squared Errors
90 Optimizer / Alternating Least Squares Ridge Regression • 𝑝 𝑢 = 𝑄 𝑇 𝐶 𝑢 𝑄 −1 𝑄 𝑇 𝐶 𝑢 𝑅 𝑟 𝑢 • 𝑞𝑖 = 𝑃 𝑇 𝐶 𝑖 𝑃 −1 𝑃 𝑇 𝐶 𝑖 𝑅 𝑐 𝑖 QT 0.1 -0.4 0.8 0.6 0.6 0.7 -0.7 -0.2 P -0.2 0.6 0.6 0.4 0.7 0.2 0.5 -0.2 R Item1 Item2 Item3 Item4 User1 1 0 0 0 User2 0 0 1 0 User3 1 1 0 0 User4 0 1 0 1
91 Optimizer / Alternating Least Squares Ridge Regression • 𝑝 𝑢 = 𝑄 𝑇 𝐶 𝑢 𝑄 −1 𝑄 𝑇 𝐶 𝑢 𝑅 𝑟 𝑢 • 𝑞𝑖 = 𝑃 𝑇 𝐶 𝑖 𝑃 −1 𝑃 𝑇 𝐶 𝑖 𝑅 𝑐 𝑖 R Item1 Item2 Item3 Item4 User1 1 0 0 0 User2 0 0 1 0 User3 1 1 0 0 User4 0 1 0 1 QT 0.3 -0.3 0.7 0.7 0.7 0.8 -0.5 -0.1 P -0.2 0.6 0.6 0.4 0.7 0.2 0.5 -0.2
92 Optimizer / Alternating Least Squares Ridge Regression • 𝑝 𝑢 = 𝑄 𝑇 𝐶 𝑢 𝑄 −1 𝑄 𝑇 𝐶 𝑢 𝑅 𝑟 𝑢 • 𝑞𝑖 = 𝑃 𝑇 𝐶 𝑖 𝑃 −1 𝑃 𝑇 𝐶 𝑖 𝑅 𝑐 𝑖 R Item1 Item2 Item3 Item4 User1 1 0 0 0 User2 0 0 1 0 User3 1 1 0 0 User4 0 1 0 1 QT 0.3 -0.3 0.7 0.7 0.7 0.8 -0.5 -0.1 P -0.2 0.7 0.6 0.5 0.8 0.2 0.6 -0.2
93 Optimizer / Alternating Least Squares • Complexity of naive solution: 𝚶 𝑰𝑲 𝟐 𝑵𝑴 + 𝑰𝑲 𝟑 𝑵 + 𝑴 𝑬: number of examples, 𝑰 : number of iterations • Improvement (Hu, Koren, Volinsky) – Ridge Regression: 𝑝 𝑢 = 𝑄 𝑇 𝐶 𝑢 𝑄 −1 𝑄 𝑇 𝐶 𝑢 𝑅 𝑟 𝑢 – 𝑄 𝑇 𝐶 𝑢 𝑄 = 𝑄 𝑇 𝑄 + 𝑄 𝑇 𝐶 𝑢 − 𝐼 𝑄 = 𝐶𝑂𝑉𝑄0 + 𝐶𝑂𝑉𝑄+, 𝚶(𝑰𝑲 𝟐 𝑵𝑴) is costly – 𝐶𝑂𝑉𝑄0 is user independent, need to be calculated at the start of the iteration – Calculating 𝐶𝑂𝑉𝑄+ needs only #𝑷(𝒖)+ steps. • #𝑷(𝒖)+ : number of positive examples of user u – Complexity: 𝜪 𝑰𝑲 𝟐 𝑬 + 𝑰𝑲 𝟑(𝑵 + 𝑴) = 𝜪 𝑰𝑲 𝟐(𝑬 + 𝑲(𝑵 + 𝑴) – Codename: IALS • Complexity issues on large dataset: – If 𝑲 is low: 𝜪(𝑰𝑲 𝟐 𝑬) is dominant – If 𝑲 is high: 𝑶(𝑰𝑲 𝟑(𝑵 + 𝑴)) is dominant
94 Collaborative Filtering – Explicit Collaborative Filtering – Implicit Collaborative Filtering – Neighbor methods – Matrix Factorization – Tensor Factorization Implicit Tensor Factorization
95 R1 Item1 Item2 Item3 … User1 1 … User2 1 0 … User3 … …. … … … … • Tensor Factorizaton • Different preferences during the day • Context: Time period • Time period 1: 06:00-14:00
96 R1 Item1 Item2 Item3 … User1 1 … User2 1 0 … User3 … …. … … … … R2 Item1 Item2 Item3 … User1 0 1 … User2 1 … User3 1 … …. … … … … • Tensor Factorizaton • Different preferences during the day • Context: Time period • Time period 2: 14:00-22:00
97 R1 Item1 Item2 Item3 … User1 1 … User2 1 0 … User3 … …. … … … … R2 Item1 Item2 Item3 … User1 0 1 … User2 1 … User3 1 … …. … … … … R3 Item1 Item2 Item3 … User1 1 … User2 … User3 1 1 … …. … … … … • Tensor Factorizaton • Different preferences during the day • Context: Time period • Time period 3: 22:00-06:00
98 R1 Item1 Item2 Item3 … User1 1 … User2 1 0 … User3 … …. … … … … R2 Item1 Item2 Item3 … User1 0 1 … User2 1 … User3 1 … …. … … … … R3 Item1 Item2 Item3 … User1 … User2 𝒓 𝑢𝑖𝑡 … User3 … …. … … … … QT q11 q21 q31 … q12 q22 q32 … P p11 p12 p21 p22 p31 p32 … … Tt11 t12 t21 t22 t31 t32 𝑹 𝑵𝒙𝑴: preference matrix 𝑷 𝑵𝒙𝑲: user feature matrix 𝑸 𝑴𝒙𝑲: item feature matrix 𝑻 𝑳𝒙𝑲: time feature matrix 𝑵: #users 𝑴: #items 𝑳: #time periods 𝑲: #features 𝒓 𝒖𝒊t = 𝒌 𝒑 𝒖𝒌 𝒒𝒊𝒌 𝒕𝒕𝒌 𝑹 = 𝑷° 𝑸° 𝑻• Tensor Factorizaton
99 • Relies on feedbacks › Explicit feedback is more reliable than implicit feedback › But explicit feedback is often not provided, just implicit feedbacks › Does not need heterogeneous data sources • Extracts latent information using consumtion behaviors • More accurate than content based filtering • Domain independent Advantages of Collaborative Filtering
100 • Cold start problem › Items without feedbacks cannot be recommended › CF algorithms cannot provide personalized recommendations for users without feedbacks • Warmup period › Requires numerous events to be able to characterize the user properly › Inaccurate for users with few feedbacks › Inaccurate for domains with weak collaboration • Vulnerable to attacks • Harder to explain the recommendations • Unable to provide cross-domain recommendations without overlapping data Disadvantages of Collaborative Filtering
101 Other applications
102 Other applications / Hybrid Filtering • Combination of collaborative and content-based filtering • Advantages › Transforms latent behavioral knowledge to meta data level › Solves cold start problem › Used for weighting meta data, able to improve the efficiency of CBF › Behavioral knowledge can be interpreted by meta data • Disadvantages › Requires both meta data and transactional data › More complex than CF and CBF › Less developed than CF and CBF, hot topic › Challenging to provide mixed recommendations with new and old items › Heterogeneous data issues
103 Other applications / Explanation • Approaches › Item explanation: Explanaining why those items were recommended › User explanation: Description about the user profile • Types of the item explanations › Non-personalized: „… because this content is trending” › Explicit features: „… because you bought horror movies” › Explicit user-to-user links: „… because your friends love this movie” › Explicit user relations: „.. because some users similar to you like this movie” › Implicit features: „… because actor X is similar to your preferred actor Y” • Types of user explanations › User tag cloud: „the user consumes 80% horror and 20% comedy” › User similarity: „the user is similar to another user who is a fan of Star Wars” • Goal: Increasing trust in recommender system • Difficulties: Hard to optimize how the recommendations should be explained
104 Other applications / Recommender strategies • The most common way to recommend from prediction scores is to order items by them • How to avoid monotonity and „the winner takes all” effect? • How to avoid preference cannibalization? • Recommender strategies: How to select items from the scored list? › Best match › One episode per series › At most N items per category › Different categories should follow each other • Exploration vs. Exploitation with Multi-Armed Bandits • Entropy maximalization per recommendaiton box for new users
105 Presented by: Contact: THANK YOU! www.impresstv.com David Zibriczky Head of Data Science, ImpressTV Limited david.zibriczky@impresstv.com

More Related Content

PPTX
Recommender Systems
byGirish Khanzode
 
PPTX
Recommendation System
byAnamta Sayyed
 
PPTX
Recommender system
byNilotpal Pramanik
 
PPTX
Recommender system introduction
byLiang Xiang
 
PDF
Recommender Systems
byCarlos Castillo (ChaTo)
 
PPTX
Recommender systems: Content-based and collaborative filtering
byViet-Trung TRAN
 
PDF
Recent advances in deep recommender systems
byNAVER Engineering
 
PDF
Recommender system algorithm and architecture
byLiang Xiang
 
Recommender Systems
byGirish Khanzode
 
Recommendation System
byAnamta Sayyed
 
Recommender system
byNilotpal Pramanik
 
Recommender system introduction
byLiang Xiang
 
Recommender Systems
byCarlos Castillo (ChaTo)
 
Recommender systems: Content-based and collaborative filtering
byViet-Trung TRAN
 
Recent advances in deep recommender systems
byNAVER Engineering
 
Recommender system algorithm and architecture
byLiang Xiang
 

What's hot

PDF
Introduction to Recommendation Systems
byTrieu Nguyen
 
PDF
Recommender Systems
byFrancesco Casalegno
 
PPTX
Recommendation system
byAkshat Thakar
 
PPT
A Hybrid Recommendation system
byPranav Prakash
 
PDF
Boston ML - Architecting Recommender Systems
byJames Kirk
 
PDF
How to build a recommender system?
byblueace
 
PPTX
Recommendation system
byRishabh Mehta
 
PDF
Matrix Factorization Techniques For Recommender Systems
byLei Guo
 
PPTX
Recommender Systems
byLior Rokach
 
PDF
Matrix Factorization In Recommender Systems
byYONG ZHENG
 
PDF
Recommendation System Explained
byCrossing Minds
 
PPTX
Collaborative Filtering Recommendation System
byMilind Gokhale
 
PPT
Recommendation system
byVikrant Arya
 
PPTX
Recommendation Systems
byRobin Reni
 
PPTX
Recommender systems using collaborative filtering
byD Yogendra Rao
 
PDF
Collaborative Filtering 1: User-based CF
byYusuke Yamamoto
 
PPTX
Recommendation Systems Basics
byJarin Tasnim Khan
 
PDF
Overview of recommender system
byStanley Wang
 
PDF
Deep Learning for Recommender Systems RecSys2017 Tutorial
byAlexandros Karatzoglou
 
PPTX
Collaborative Filtering using KNN
byŞeyda Hatipoğlu
 
Introduction to Recommendation Systems
byTrieu Nguyen
 
Recommender Systems
byFrancesco Casalegno
 
Recommendation system
byAkshat Thakar
 
A Hybrid Recommendation system
byPranav Prakash
 
Boston ML - Architecting Recommender Systems
byJames Kirk
 
How to build a recommender system?
byblueace
 
Recommendation system
byRishabh Mehta
 
Matrix Factorization Techniques For Recommender Systems
byLei Guo
 
Recommender Systems
byLior Rokach
 
Matrix Factorization In Recommender Systems
byYONG ZHENG
 
Recommendation System Explained
byCrossing Minds
 
Collaborative Filtering Recommendation System
byMilind Gokhale
 
Recommendation system
byVikrant Arya
 
Recommendation Systems
byRobin Reni
 
Recommender systems using collaborative filtering
byD Yogendra Rao
 
Collaborative Filtering 1: User-based CF
byYusuke Yamamoto
 
Recommendation Systems Basics
byJarin Tasnim Khan
 
Overview of recommender system
byStanley Wang
 
Deep Learning for Recommender Systems RecSys2017 Tutorial
byAlexandros Karatzoglou
 
Collaborative Filtering using KNN
byŞeyda Hatipoğlu
 

Similar to An introduction to Recommender Systems

PPTX
Mini-training: Personalization & Recommendation Demystified
byBetclic Everest Group Tech Team
 
PDF
Recommender Systems, Part 1 - Introduction to approaches and algorithms
byRamzi Alqrainy
 
PPTX
Recommender System _Module 1_Introduction to Recommender System.pptx
bySatyam Sharma
 
PPTX
advanced_recommendation_PPt_for_college.pptx
bykumarsourav8432
 
PDF
Personalized E-commerce based recommendation systems using deep-learning tech...
byIAESIJAI
 
PDF
Introduction to recommender systems
byAndrea Gigli
 
PDF
Recommandation systems -
byYousef Fadila
 
PPT
[livecast] Personalization on the Web
byVibhanshu Abhishek
 
PPTX
ADM6274 - Final (NEHA)
byNeha Gupta
 
PPT
JM Information Retrieval Techniques Unit V
byJeyamohanHAsstProfCS
 
PPT
Information Storage and Retrieval Techniques Unit V
byjeyamohan2519
 
PPTX
3e recommendation engines_meetup
byPranab Ghosh
 
PDF
REAL-TIME RECOMMENDATION SYSTEMS
byBigDataCloud
 
PPT
Recommender system in Information Retrieval system.ppt
byDrSSriDharan
 
PPT
Artififial intelligence and its appications in tourism.ppt
byshahzadkhan34
 
PPT
Artificial Intelligence and the basics of recommender systems
byshahzadkhan34
 
PPT
Introdcution to Artificial intelligence basic to advance
bycslight1
 
PPT
Culbert (1).pptmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmm
byMrMohammedZakirBELEC
 
PPTX
ai recomendation. how Ai is use to recommend
byrajchaudhary11223344
 
PDF
leewayhertz.com-How to build an AI-powered recommendation system.pdf
byrobertsamuel23
 
Mini-training: Personalization & Recommendation Demystified
byBetclic Everest Group Tech Team
 
Recommender Systems, Part 1 - Introduction to approaches and algorithms
byRamzi Alqrainy
 
Recommender System _Module 1_Introduction to Recommender System.pptx
bySatyam Sharma
 
advanced_recommendation_PPt_for_college.pptx
bykumarsourav8432
 
Personalized E-commerce based recommendation systems using deep-learning tech...
byIAESIJAI
 
Introduction to recommender systems
byAndrea Gigli
 
Recommandation systems -
byYousef Fadila
 
[livecast] Personalization on the Web
byVibhanshu Abhishek
 
ADM6274 - Final (NEHA)
byNeha Gupta
 
JM Information Retrieval Techniques Unit V
byJeyamohanHAsstProfCS
 
Information Storage and Retrieval Techniques Unit V
byjeyamohan2519
 
3e recommendation engines_meetup
byPranab Ghosh
 
REAL-TIME RECOMMENDATION SYSTEMS
byBigDataCloud
 
Recommender system in Information Retrieval system.ppt
byDrSSriDharan
 
Artififial intelligence and its appications in tourism.ppt
byshahzadkhan34
 
Artificial Intelligence and the basics of recommender systems
byshahzadkhan34
 
Introdcution to Artificial intelligence basic to advance
bycslight1
 
Culbert (1).pptmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmm
byMrMohammedZakirBELEC
 
ai recomendation. how Ai is use to recommend
byrajchaudhary11223344
 
leewayhertz.com-How to build an AI-powered recommendation system.pdf
byrobertsamuel23
 

More from David Zibriczky

PDF
Highlights from the 8th ACM Conference on Recommender Systems (RecSys 2014)
byDavid Zibriczky
 
PDF
Predictive Solutions and Analytics for TV & Entertainment Businesses
byDavid Zibriczky
 
PDF
Improving the TV User Experience by Algorithms: Personalized Content Recommen...
byDavid Zibriczky
 
PPTX
Recommender Systems meet Finance - A literature review
byDavid Zibriczky
 
PPTX
A Combination of Simple Models by Forward Predictor Selection for Job Recomme...
byDavid Zibriczky
 
PDF
Fast ALS-Based Matrix Factorization for Recommender Systems
byDavid Zibriczky
 
PDF
EPG content recommendation in large scale: a case study on interactive TV pla...
byDavid Zibriczky
 
PDF
Personalized recommendation of linear content on interactive TV platforms
byDavid Zibriczky
 
PDF
Data Modeling in IPTV and OTT Recommender Systems
byDavid Zibriczky
 
PDF
Entropy based asset pricing
byDavid Zibriczky
 
Highlights from the 8th ACM Conference on Recommender Systems (RecSys 2014)
byDavid Zibriczky
 
Predictive Solutions and Analytics for TV & Entertainment Businesses
byDavid Zibriczky
 
Improving the TV User Experience by Algorithms: Personalized Content Recommen...
byDavid Zibriczky
 
Recommender Systems meet Finance - A literature review
byDavid Zibriczky
 
A Combination of Simple Models by Forward Predictor Selection for Job Recomme...
byDavid Zibriczky
 
Fast ALS-Based Matrix Factorization for Recommender Systems
byDavid Zibriczky
 
EPG content recommendation in large scale: a case study on interactive TV pla...
byDavid Zibriczky
 
Personalized recommendation of linear content on interactive TV platforms
byDavid Zibriczky
 
Data Modeling in IPTV and OTT Recommender Systems
byDavid Zibriczky
 
Entropy based asset pricing
byDavid Zibriczky
 

Recently uploaded

PDF
[DSC Europe 25] Bojan Djuricic - Predictive Design Process.pdf
byDataScienceConferenc1
 
PPTX
Enhanced signature verification through the integration of divers' models, op...
bychWaqasZahid
 
PDF
Nepal's Health Policies: From 1991 to 2024 - A Complete Analysis of Nepal's H...
byPublic Health Concern Nepal
 
PPTX
Technical_Workflow_of_Indian_Startup_Funding_Trends_Project_1 (4).pptx
byAryanthakur98
 
PPTX
Child Safeguarding training ppt for one day.pptx
bymeasho Birhanu
 
PDF
Bratislava Tableau User group (BA_TUG) - 15_01_2026 - in-person meeting.pdf
byPavol Hromadka
 
PDF
Serverless Data Infrastructure (examples)
byMichal Gancarski
 
PDF
[DSC Europe 25] Andrzej Kowalczyk - AI - how to start small and grow in the f...
byDataScienceConferenc1
 
PPTX
[DSC Europe 25] Ivan Lukovic & Marija Djukic - From Data to Value: Why Maturi...
byDataScienceConferenc1
 
PPTX
Inventory Managemnet for111111111111111W2
bydrizwan89
 
PDF
BUKTI KEMENANGAN MEMBER KANCAH4D – WD PULUHAN JUTA DARI STARLIGHT PRINCESS!
bykancah4d
 
PDF
Best digital marketing agency in Delhi NCR - Bala Infotech
byBala Infotech
 
PDF
The Architecture of Tomorrow: Elon Musk’s Vision for a Post-Work Society
bySubtract 8
 
PDF
(UPDATE) JADWAL GEREJA OPS LILIN 2025.pdf
byMinlantasSumber
 
PDF
Torre Avianca- Modernity, Identity, and Technology in Colombian Architecture.pdf
bysanchezpipe1234
 
PDF
How To Bu'y Verifi'ed kraken Account in US.pdf
byuerubi95
 
PDF
Time Travel on ACID - A Gentle Introduction To Apache Iceberg
byMichal Gancarski
 
PDF
Semantic Technology
byRathachai Chawuthai
 
PDF
Applied_Data_Science_Unit_One_Notes_for_reference
bysanguleumeshit
 
PDF
Presentation - 5 Essential Steps for Successful Data Migration
byalexmartincaneda
 
[DSC Europe 25] Bojan Djuricic - Predictive Design Process.pdf
byDataScienceConferenc1
 
Enhanced signature verification through the integration of divers' models, op...
bychWaqasZahid
 
Nepal's Health Policies: From 1991 to 2024 - A Complete Analysis of Nepal's H...
byPublic Health Concern Nepal
 
Technical_Workflow_of_Indian_Startup_Funding_Trends_Project_1 (4).pptx
byAryanthakur98
 
Child Safeguarding training ppt for one day.pptx
bymeasho Birhanu
 
Bratislava Tableau User group (BA_TUG) - 15_01_2026 - in-person meeting.pdf
byPavol Hromadka
 
Serverless Data Infrastructure (examples)
byMichal Gancarski
 
[DSC Europe 25] Andrzej Kowalczyk - AI - how to start small and grow in the f...
byDataScienceConferenc1
 
[DSC Europe 25] Ivan Lukovic & Marija Djukic - From Data to Value: Why Maturi...
byDataScienceConferenc1
 
Inventory Managemnet for111111111111111W2
bydrizwan89
 
BUKTI KEMENANGAN MEMBER KANCAH4D – WD PULUHAN JUTA DARI STARLIGHT PRINCESS!
bykancah4d
 
Best digital marketing agency in Delhi NCR - Bala Infotech
byBala Infotech
 
The Architecture of Tomorrow: Elon Musk’s Vision for a Post-Work Society
bySubtract 8
 
(UPDATE) JADWAL GEREJA OPS LILIN 2025.pdf
byMinlantasSumber
 
Torre Avianca- Modernity, Identity, and Technology in Colombian Architecture.pdf
bysanchezpipe1234
 
How To Bu'y Verifi'ed kraken Account in US.pdf
byuerubi95
 
Time Travel on ACID - A Gentle Introduction To Apache Iceberg
byMichal Gancarski
 
Semantic Technology
byRathachai Chawuthai
 
Applied_Data_Science_Unit_One_Notes_for_reference
bysanguleumeshit
 
Presentation - 5 Essential Steps for Successful Data Migration
byalexmartincaneda
 

An introduction to Recommender Systems

  • 1.
    1 INTRODUCTION TO RECOMMENDERSYSTEMS 2014.12.04 David Zibriczky (ImpressTV)
  • 2.
    2 About ImpressTV • NetflixPrize (2006-2009) • The Ensemble: Runner-up from 40K teams • Gravity team: Members of The Ensemble • Gravity R&D › Hungarian start-up, launched in 2007, B2B business › On-site personalization solution provider company for e-commerce, IPTV, OTT and classified media › 100M+ recommendations per day • ImpressTV Limited › IPTV and OTT was acquired from Gravity R&D by british investors in July 2014 › HQ in Budapest, international corporate clients • About me › Joined Gravity R&D in January 2010, transfered to ImpressTV Limited in July 2014 › Current position: Head of Data Science at ImpressTV Limited
  • 3.
  • 4.
  • 5.
    5 • Too manyexisting items, too many options › YouTube videos, Amazon books, Netflix movies, Pandora musics, … • High item publishing intensity, hard to follow the flow of information › Google News, Facebook posts, Twitter tweets, … • Users are not qualified enough › They know what they want to watch/buy, but they don’t know whether a given item would satisfy their needs › E.g.: Grandma would like a new computer, knows what she wants to do with it, but has no idea of the computer parts and what they do • One may not know what they would like › First time in a Chinese restaurant, no experience about Chinese food. • How to improve the user satisfaction? Information overload II. – Consumers
  • 6.
    6 • Challange inhandling ever increasing number of contents and corresponding data • Challenge in handling transactional data • Increasing amount of inner information • What is useful? • How to determ the usefulness of data? • Competition in business market › keeping and converting consumers is essential › keeping the market adavantage › increasing revenue • How to exctract and use the collected information to improve business success? Information overload II. – Content Providers
  • 7.
    7 „Recommender Systems (RS)are software agents that elicit the interests and preferences of individual consumers […] and make recommendations accordingly. They have the potential to support and improve the quality of the decisions consumers make while searching for and selecting products online.„ 1 What are Recommender Systems? 1 Xiao, Bo, and Izak Benbasat, 2007, E-commerce product recommendation agents: Use, characteristics, and impact, Mis Quarterly 31, 137-209. Recommender Systems
  • 8.
    8 Recommender Systems assoftware agents Items Users Recommender System Recommend item X to user A
  • 9.
  • 10.
    10 Examples / YouTube(Uploaded Videos)
  • 11.
  • 12.
  • 13.
    13 Examples / LinkedIn(News, Jobs, Groups, People, …)
  • 14.
    14 Examples / Twitter(Tweets, Peoples, Groups, Companies)
  • 15.
  • 16.
  • 17.
    17 Examples / Grocery(Foods, Drinks)
  • 18.
    18 • News (Whatto read now?) • Dating sites (Which girl to contact?) • Coupon portals (Which coupon is good for me?) • Restaurants (Where to eat?) • Vacations (Where to travel and what to see?) • Retailers (How much products to supply?) • Financial assets (loan, portfolio) • Advertisements • …etc Other Examples
  • 19.
    19 • For theconsumers (users) › Helping the users to find useful contents to satisfy their needs › Reducing the time of content searching › Providing relevant information from the massive information flow › Exploring new preferences, trust in recommender system • For the business › Improving business success indicators (or key performance indicators, KPI) • Increasing revenue, CTR, watching/listening duration • Increasing conversion rate and user engagement, reducing churn • Cross-selling, upselling, advertisement › Reducing popularity effect, less popular contents are also consumed › Promotions, targeting, campaign Goals and benefits of Recommender Systems
  • 20.
    20 • Goals forconsumers not necessarily equal to business! • Simplified example 1: YouTube (free contents) › Goal of the business: Receive more income from advertisements How: More videos watched, more advertisements seen › Goal of the users: Having good/useful time by watching/listening videos How: Clicking on recommendations, using search engine › The goal is realized in the same way, more videos watched are good for both. • Simplified example 2: Netflix (DVD/Blu-ray or Video On Demand rental) › Goal of the business: Increase the income How: More expensive contents, improving user engagement › Goal for the users: Buying interesting movies, spending less time for searching How: Using recommendation engine to make it easier › The goal is different. Netflix wants the users pay more. The users basically don’t want to spend more, only if they find it worth. Difference between the goals of consumers and business
  • 21.
  • 22.
    22 Recommender Systems assoftware agents Items Users Recommender System Recommend item X to user A
  • 23.
    23 • Items: Entitesthat are recommended (movies, musics, books, news, coupons, restaurants, etc...) • Item data: Descriptive information about the items (e.g. genre, category, price) • Users: People to whom we recommend (Who is the user? Member, cookie, unidentified?) • Paradigm: Recommendations are calculated by predefined non-personalized rules • How: Setting item data based rules (recommending the latest movies for all users) • Properties: › Non-personalized static recommendations › Requires manual work (e.g editorial pick) Recommendation paradigms / Non-personalized rules Items Recommender System RecoItem data Users
  • 24.
    24 • User data:Descriptive information about the users (e.g. age, gender, location) • Paradigm: Recommendations are calculated by predefined used data based rules • How: Setting item and user data based rules (men between 45-55  expensive cars) • Properties: › Semi-personalized › Requires manual work (e.g rule constructing) › Interpretable Recommendation paradigms / Personalized rules Items User data Recommender System Reco Item data Users
  • 25.
    25 • Transactions: Interactionbetween users and items • Transaction types: Numerical ratings, ordinal ratings, binary ratings, unary ratings, textual reviews • Explicit feedback: The user quantifies his preference about an item (rating) • Implicit feedback: Events that indicates but not quantifies the preference of the user about an item › Positive: buy, watch, like, add to favourite › Negative: dislike, remove from favourites • In practice, implicit feedback is less valueable than explicit feedback, but significantly more provided Recommendation paradigms / Transaction based personalization I. Items User data Recommender System Reco Item data Users Transactions
  • 26.
    26 • Paradigm: Recommendationsare calculated based on the users interactions • How: Learning on interactional data (collaborative filtering) • Properties: › Personalized › Adaptive › Less interpretable Recommendation paradigms / Transaction based personalization II. Items User data Recommender System Reco Item data Users Transactions
  • 27.
    27 • Context: Informationthat can be observed in the time of recommendation • Types: › Time-based temporal information (Recos of sports equipments should be different in summer and in winter) › Mood (Different TV programs should be recommended to the same person based on her current mood) › Device (Different content types based on device type, e.g. movie trailers in TV, music in phone) › Location › Event sequence information (If the user browses the TV category of a webshop, it is reasonable to recommend DVD players, but it is not if the user browses the laptop category) Recommendation paradigms / Context based personalization I. Items User data Recommender System Reco Item data Users Context Transactions
  • 28.
    28 • Paradigm: Recommendationsare calculated based on the users interactions and current context • How: Learning on contextual data (the user usually watches new in the morning, action at night) • Properties: › Context-sensitive, fully adaptive › More interpretable › More complex Recommendation paradigms / Context based personalization II. Items User data Recommender System Reco Item data Users Context Transactions
  • 29.
    29 Cross-domain recommendation Task Goal Multi-domain Cross-selling Diversity Serendipity Linked-domain Accuracy Cross-domain Cold-start New users New items • Multi-domain: You watched movies and bought books, we recommend movies or books • Linked-domain: You bought books only, we recommend books using both movie and book consumption patterns • Cross-domain: You watched movies only, we recommend books, based on movie-book consumption relationship Domain Example Ratio Attribute Comedy Thriller 12% Type Movies   Books 9% Item Movies  Restaurants 55% System Netflix  MovieLens 24%
  • 30.
    30 • Item toUser › Conventional recommendation task: „You may like these items”. • Item to Item › Recommending items that are somewhat similar to the item currently viewed by the user › Personalized or non-personalized similarity • User to User › Recommending different users for the users based on metadata or activity › Social recommendations (who to follow, who to connect) or similar users • User to Item › Promoting items that the seller wants to sell to the most probable buyers (e.g. newsletters, notifications) › „Who would buy this item?” • Group recommendations › Recommending group of items or users Recommendation types
  • 31.
  • 32.
    32 • Content-based Filtering(CBF) › Recommend items that are similar to the ones that the user liked in the past. › Similarity based on the metadata of the items. › E.g.: If the user likes romantic movies, recommend her the like • Collaborative Filtering (CF) › Recommend items that are liked by users that have similar taste as the current user › Similarity betwen users is calculated by the transaction history of users › Only uses the transaction data  domain independent • Demographic › Recommendations made based on the demographic profile. › Realizes a simple personalization. › E.g.: Recommend computer parts to young people who study informatics. Recommendation techniques I.
  • 33.
    33 • Knowledge-based › Usesextensive domain-specific knowledge to generate recs. › User requirements collected (problem), items as possible solutions to the specific problems › Example: user would like to find a digital camera, she provides her needs and the level of her skills, etc. › Pure knowledge based systems (without learning) tend to perform better in the beginning of deployment but they fall behind later. • Community-based › Recommendations based on the preferences of the user’s friends. › People tend to accept recommendations of their friends. › Using of social networks  social recommender systems. • Hybrid recommendation systems › Combination of the techniques above. › Trying to use the advantages and fix the disadvantages of the different techniques. Recommendation techniques II.
  • 34.
    34 1. Setting therecommendation type (e.g. Item2user recommendation) 2. Requesting a recommendation 3. Selecting the recommendable items, filtering (e.g. new movies) 4. Selecting the algorithm for scoring these items (e.g. a collaborative filteirng algorithm) 5. The algorithm provides score for each items 6. Ordering the list by the scores 7. Post processing the input item list (e.g. randomizing topN, selecting one episode per series, etc…) 8. Selecting the first N items to send back as a response of recommendation request Recommendation data flow
  • 35.
    35 • Key featuresof a good recommender systems, that should be considered › Accuracy (efficiency of modeling the user preference) › Adaptation, context awareness (ability to detect changes in user behavior) › Diversity, coverage (avoiding monotone recommendations, and preference cannibalization) › Novelty, serendipity (improving suprise factor, exploitation vs. exploration) › Trust, explanation (the users should understand and trust in recommender system) › Scalability, responsivity, availability (recommendations should be provided in reasonable time) • Tradeoffs in recommender systems › Accuracy vs. Diversity › Discovery vs. Continuation › Depth vs. Coverage › Freshness vs. Stability › Recommendations vs. Tasks Properties of recommender system
  • 36.
    36 • Method › Splittingdata set into disjunct train and test set › By time, random, users, user history › Training on train set, measuring on test set • Evaluation metrics › Accuracy: Rating (RMSE), TopN (nDCG@N, Recall@N, Precision@N) › Coverage: Ratio of the recommended items › Diversity: Entropy, Gini-index › Novelty: Ratio of the long tail items › Serendipity: Ratio of the less › Training time › Accuracy is the typical primary metric, others are secondary Offline Evaluation DataSet Train Test
  • 37.
    37 • Method › Effectsof recommender systems are measured › A/B testing: Splitting user base into equally valuable subsets, different recommendations for the subsets › A/B tests avoid › Online improvement can be measured • Evaluation metrics › Some effects cannot be evaluated before the recommender system set live › Click Though Rate (CTR, Number of clicks per recomendation) › Average Revenue Per User (ARPU), Total Revenue Increase › Page impression (number of page views) › Conversion Rate, churn Rate Online Evaluation
  • 38.
    38 1. Understanding thebusiness needs › Business goals, Key Performance Indicators for live recommender system › Recommendation scenarios, placeholders › Data understanding 2. Integration (for Vendors only) › Developing data integration method (the way the customer provide it’s data) › Setting up recommendations request/response interface (how the customer request a recommendation, and the vendor provides response, e.g. JSON or REST API) 3. Data preparation › Data enrichment from external sources (e.g. crawling additional item meta data) › Data transformation (shall we handle a complete series as an item, or a group of episodes?) Workflow of Recommender System integration
  • 39.
    39 4. Data Miningand Offline Experiments Workflow of Recommender System integration
  • 40.
    40 5. Online Evaluation/ Deployment › Setting recommendation engine live (starting to provide recommendations for real end-users) › Measuring online performance metrics (e.g. CTR, ARPU or page impression) › Measuring response times (is the algorithm fast enough in live service?) › Analyzing correlation between the offline and online metrics (what to optimize in offline experiments) 6. Optimization › Implementing additional algorithms (Step 4) › A/B testing (is the new algoritm better than the original one?) › Statistical significance based A/B selection (e.g. T-test between the performance of algorithm A and B) 7. Reporting and follow-up › Reporting performance (weekly, monthly) › Monitoring response times and availability (if the algorithm not roboust enough, it may result outage) › Monitoring change in user behavior, size of user base › Adapting to new features (e.g. new placeholders in the website) Workflow of Recommender System integration
  • 41.
  • 42.
    42 • Content-based Filtering(CBF) › Recommend items that are similar to the ones that the user liked in the past › Similarity based on the metadata of the items › E.g.: If the user likes horror movies, recommend her horror movies • Collaborative Filtering (CF) • Demographic • Knowledge-based • Community-based • Hybrid recommendation systems Recommendation Techniques
  • 43.
    43 Recommending an itemto a user based upon a description of the item and a profile of the user’s interests • Items are represented by their metadata (e.g genre or description) • Users are represented by their transaction history (e.g. item A and B was rated by 5) • We generate user profiles from the user transaction data and the metadata of the items in the user’s transactions • The user profiles are compared to the representation of the items • The items similar to the user profile are recommended to the user CBF Method Twilight 5 5 ? ? Sci-Fi RomanceAdventure The Matrix The Matrix 2 The Matrix 3
  • 44.
  • 45.
    45 Content Analyzer • Preprocessingmodule • Input: Item meta data • Methods: › Text mining › Semantic Analysis › Natural Language Process › Feature extraction › Meta data enrichment › Auto tagging • Output: Item models
  • 46.
    46 • We createterms from the item metadata › Standard text mining preprocessing steps • Filtering stopwords • Filtering too rare/common terms • Stemming • Items are represented by a (sparse) vector of terms › The vector of each item (=document) contains weights for each term. › Weighting is often done by the TF-IDF scheme • Rare terms are not less relevant than frequent terms (IDF). • Multiple occurences in a document are not less relevent than a single occurence (TF). • Invariant of the length of the document (normalization). • Similarity measurement › Most common: cosine similarity • Scalar product of the L2 normalized vectors Content Analyzer / Vector space model (VSM)                  T k k T k k T k kk ww ww iisim 1 2 2 1 2 1 1 21 21 ),(
  • 47.
    47 • Semantic analysisin order to extract more information about the items • Domain-specific knowledge and taxonomy is used Content Analyzer / Semantic analysis, ontologies
  • 48.
    48 • The metadata provided by clients is usually not enough • Meta data enrichment: Crawling additional data about the items from public web • Publicly available knowledge sources › Open Directory Project › Yahoo! Web Directory › Wikipedia › Freebase › Tribune › Etc… • Goals: › Fixing missing or wrong data › Better characterization of items › Improving the accuracy of CBF algorithms Content Analyzer / Meta data enrichment
  • 49.
    49 Profile Learner • Userpreference modeling • Inputs: › Item models › User transactional data › User meta data • Methods: › Meta data weighting › Machine learning • Output: User profile
  • 50.
    50 Profile Learner /Keyword-based profiles
  • 51.
    51 • Negative andpositive examples collected › Explicit case: Items rated under 3 counts negative, above positive (variants: above overall average, user average…) › Implicit case: Items viewed counts positive, others negative • Approach 1 › User profile: weighted average of item vectors (Rocchio’s algorithm) › Similarity between user profile and item vectors as relevance score • Approach 2 › Negative and positive user profiles › Items that match negative profile are filtered › Similarity between the positive profile and the item’s vector as relevance score • Approach 3 › Rule based classifiers (decision trees, decision lists) learning on the examples › New items are judged by the classifier • Approach 4 › Nearest Neighbor methods Profile Learner / User profile in VSM model
  • 52.
    52 • The termsin the item metadata are handled independently • Semantic information is lost in the process, we have no idea of the meaning of the terms • It is hard to handle expressions containing more than one word • Data sparsity • Need diverse user feedbacks to improve the accuracy Profile Learner / Problems with the VSM model
  • 53.
    53 Profile Learner /Phenomenon of polysemy
  • 54.
    54 Profile Learner /Word sense disambiguation
  • 55.
    55 Architecture / FilteringComponent • Recommendation list filtering • Inputs: › User profile model › Item models › Recommendable items • Methods: › Filtering non-relevant items › Similarity based ranking • Output: Recommended items
  • 56.
    56 • User independence ›Profiles of the users are built independently › A user can not influence the recommendations for other users (no attacks on the recommender system) • Interpretable › Recommendations can be easily explained (we recommend this comedy because you watched another comedies) › Explanations can easily built, user profiles are easy to be understood (user keyword model) • Item cold start › Solving item cold start problem › Capable of recommending items that were never rated/viewed by anybody Advantages of Content-based Filtering
  • 57.
    57 • Limited contentanalysis › Natural limit in the number and type of features that are associated with the recommended items › Domain knowledge is often needed • Over-specialization › Hard to improve serendipity (recommending some unexpected) › Recommendations mirrors the user history › Monotone recommendations: „The winner takes it all” effect › Exploitation over exploration • Meaning › word meaning disambiguation (apple = company or the fruit?) • User cold start › Needs some ratings before the profile learner can build an accurate user model › There will be no reliable recommendations when only a few ratings available Disadvantages of Content-based Filtering
  • 58.
    58 Disadvantages / Wordmeaning disambiguation
  • 59.
    59 • Deep learning(Sentiment Analysis, Paragraph Vector Model) • User generated contents (Tagging, Folksonomies) • Serendipity problem › Using randomness or genetic algorithms › Filtering too similar items › Balance between exploration and exploitation › Using poor similarity measures to produce anomalies and exceptions • Centralized knowledgebase for meta data enrichment Research topics in Content-Based Filtering
  • 60.
  • 61.
    61 • Content-based Filtering(CBF) • Collaborative Filtering (CF) › Recommend items that are liked by users that have similar taste as the current user › Similarity between users is calculated by the transaction history of users › Only uses the transaction data  domain independent • Demographic • Knowledge-based • Community-based • Hybrid recommendation systems Recommendation Techniques
  • 62.
    62 – Explicit CollaborativeFiltering – Implicit Collaborative Filtering Explicit Collaborative Filtering
  • 63.
    63 5 ? ? The MatrixThe Matrix 2 Twilight The Matrix 3 ? • Classic item recommendation (Netflix) on explicit feedbacks (ratings) • Rating problem: The goal is to predict how the user would rate the items • Accuracy metrics: › Root Mean Squared (RMSE) › Mean Absolute Error (MAE)   test Rriu iu R rr RMSE test    ),,( 2 , ˆ test Rriu iu R rr MAE test    ),,( , ˆ Rating prediction problem
  • 64.
    64 5 ? ? The MatrixThe Matrix 2 Twilight The Matrix 3 ? • Global average (average of all ratings) • User average (average of user ratings) • Item average (average of ratings given to the item) Baseline Methods
  • 65.
    65 5 5 4 5 5 ? ? The Matrix 3TheMatrix The Matrix 2 Twilight 5 ? Explicit Collaborative Filtering
  • 66.
    66 R Item1 Item2Item3 Item4 User1 5 ? ? ? User2 5 User3 5 4 User4 5 5 Explicit CF / Rating Matrix Representation 5 5 4 5 5 ? ? Item4Item1 Item2 Item3 5 ?
  • 67.
    67 Collaborative Filtering – ExplicitCollaborative Filtering – Neighbor methods – Matrix Factorization – Implicit Collaborative Filtering Explicit Neighbor Methods
  • 68.
    68 • User basedneigbor methods › 1. Find users who have rated item i › 2. Select k from these users that are the most similar to u › 3. Calculate R(u,i) from their ratings on i • Item based neigbor methods › 1. Find items that have been rated by user u › 2. Select k from these items that are the most similar to i › 3. Calculate R(u,i) from their ratings by u Explicit CF / Neigbor methods
  • 69.
    69 • Similarity between ›Rows of the preference matrix (user based) › Columns of the preference matrix (item based) › Dimension reduction on user/item preference vectors  feature vectors • Similarity properties: › S(a,a) >= S(a,b) (often S(a,a) = 1 required) › S(a,b) = S(b,a) • Similarity measures › Cosine similarity (CS) › Pearson correlation (PC) › Adjusted cosine similarity (ACS) › Eucledian distance (EUC) Explicit CF / Neigbor methods / Similarity setup 2 2 uv u v ui vi i I ui vj i I j I r r CS r r               uvuv uv Ii vvi Ii uui Ii vviuui rrrr rrrr PC 22 )()( ))(( 2 2 ( )( ) ( ) ( ) uv uv uv ui i vi i i I ui i vi i i I i I r r r r ACS r r r r               uvIi viui rrEUC 2
  • 70.
    70 • Advantages › Simplicity(easy to set up, only a few parameters) › Justifiability (recommendations can be explained) › Stability (not very sensitive to new items/users/ratings) › Good for item2item and user2user recommendation • Disadvantages › Computationally expensive in the recommending phase • Similarities have to be computed for recommendation • Similarity maxtrix might be computed before recommendations but then there is a costly training phase • Similarity matrix might not fit in the memory › Less accurate than model based methods for personalized recommendation Advantages and disadvantages of neighbor methods
  • 71.
    71 Collaborative Filtering – Explicitcollaborative filtering – Neighbor methods – Matrix Factorization – Implicit collaborative filtering Explicit Matrix Factorization
  • 72.
    72 𝑹 = 𝑷𝑸𝑻 𝑟 𝑢𝑖 = 𝒑 𝑢 𝑇 𝒒𝑖 𝑹 𝑵𝒙𝑴: rating matrix 𝑷 𝑵𝒙𝑲: user feature matrix 𝑸 𝑴𝒙𝑲: item feature matrix 𝑵: #users 𝑴: #items 𝑲: #features 𝑲 ≪ 𝑴, 𝑲 ≪ 𝑵 R Item1 Item2 Item3 … User1 User2 𝒓 𝑢𝑖 User3 … P 𝒑 𝑢 𝑇 QT 𝒒𝑖 Explicit CF / Matrix Factorization
  • 73.
    73 • ALS (AlternatingLeast Squares) › Updating P and Q matrices by multivariate linear regression based solver • BRISMF (Biased Regularized Simultaneous Matrix Factorization) › Stochastic gradient descent based matrix factorization › Minimizing the prediction error by iterating on transactions and modifying factors › Using bias for user and item model › Prediction: • SVD (Singular Value Decomposition) › Decomposes matrix R to three matrices › S is a diagonal matrix, containing singular values of matrix R • NSVD1 › Decomposes matrix R to three matrices › W is a weight matrix Explicit CF / Matrix Factorization iu K k kiukiuiuui cbqpcbqpr  1 ˆ  T R PSQ T R PWQ
  • 74.
    74 Explicit CF /Matrix Factorization / Model Visualization P John 1.1 0.5 Paul 0.6 0.9 Suzy 1.0 0.9 Item 1 Item 2 Item 3 Item 4 Item 5 QT 1.0 -0.9 0.5 -0.5 1.3 -0.3 1.6 0.7 1.6 -1.1
  • 75.
    75 Explicit CF /Matrix Factorization / Model Visualization P John 1.1 0.5 Paul 0.6 0.9 Suzy 1.0 0.9 Item 1 Item 2 Item 3 Item 4 Item 5 QT 1.0 -0.9 0.5 -0.5 1.3 -0.3 1.6 0.7 1.6 -1.1 J P S 1 2 3 4 5
  • 76.
    76 Explicit CF /Matrix Factorization / Clustering P John 1.1 0.5 Paul 0.6 0.9 Suzy 1.0 0.9 Item 1 Item 2 Item 3 Item 4 Item 5 QT 1.0 -0.9 0.5 -0.5 1.3 -0.3 1.6 0.7 1.6 -1.1 J P S 1 2 3 4 5 A B C
  • 77.
    77 Memory based algorithms Model basedalgorithms Hierarchy of Collaborative Filtering methods Collaborative Filtering Matrix factorization Explicit feedback base algorithms Implicit feedback base algorithms Neighbor methods
  • 78.
    78 – Explicit CollaborativeFiltering – Implicit Collaborative Filtering Implicit Collaborative Filtering
  • 79.
    79 ? ? The Matrix TheMatrix 2 Twilight The Matrix 3 ? • Classic item recommendation on implicit feedbacks (view, buy, like, add to favourite) • Preference problem: The goal is to predict the probability that the user would choose that item • Accuracy metrics: › Ranking error instead of prediction error › Recall@N › Precision@N Preference prediction problem
  • 80.
    80 ? ? The Matrix TheMatrix 2 Twilight The Matrix 3 ? • Most popular items • Most recent items • Most popular items in the user’s favourite category Baseline Methods
  • 81.
    81 ? ? The Matrix 3TheMatrix The Matrix 2 Twilight ? Implicit Collaborative Filtering
  • 82.
    82 Implicit CF /Preference Matrix ? ? Item4Item1 Item2 Item3 ? • Implicit Feedbacks: Assuming positive preference • Value = 1 • Estimation of unknown preference? • Sorting items by estimation  Item Recommendation R Item1 Item2 Item3 Item4 User1 1 User2 1 User3 1 1 User4 1 1
  • 83.
    83 • Zero valuefor unknown preference (zero example). Many 0s, few 1s, in practice- • 𝒄 𝑢𝑖 confidence for known feedback (constant or function of the context of event) • Zero examples are less important, but important. R Item1 Item2 Item3 Item4 User1 1 0 0 0 User2 0 0 1 0 User3 1 1 0 0 User4 0 1 0 1 C Item1 Item2 Item3 Item4 User1 𝒄11 1 1 1 User2 1 1 𝒄23 1 User3 𝒄31 𝒄32 1 1 User4 1 𝒄42 1 𝒄44 Implicit CF / Confidence Matrix
  • 84.
    84 Collaborative Filtering – ExplicitCollaborative Filtering – Implicit Collaborative Filtering – Neighbor methods – Matrix Factorization Implicit Neighbor Methods
  • 85.
    85 • User basedneigbor methods › 1. Find users who have rated item i › 2. Select k from these users that are the most similar to u › 3. Calculate R(u,i) from their ratings on i • Item based neigbor methods › 1. Find items that have been rated by user u › 2. Select k from these items that are the most similar to i › 3. Calculate R(u,i) from their ratings by u Implicit CF / Neigbor methods
  • 86.
    86 • Co-occurrence similarityalgorithm (most often used in business) • The algorithm: 1. For all item i: 1. Count the number of users that interacted with the item i: 𝑠𝑢𝑝𝑝(𝑖) 2. For all user u that interacted with the item i: 1. For all item i iteracted by user u: 1. Increment the counter of the co-occurrence of item i and j: 𝑠𝑢𝑝𝑝 𝑖, 𝑗 2. The similarity between item i and item j: 𝑆 𝑖, 𝑗 = 𝑠𝑢𝑝𝑝(𝑖,𝑗) 𝑠𝑢𝑝𝑝 𝑖 +𝛾 1−𝛼 𝑠𝑢𝑝𝑝 𝑗 +𝛾 𝛼 • 𝛼: popularity factor • 𝛾: reguralization factor 3. The prediction for user u and item i: 𝑟 𝑢𝑖 = 𝑗∈𝐶 𝑢>0 𝑐 𝑢𝑗 𝑆(𝑖,𝑗) 𝑗∈𝐶 𝑢>0 𝑐 𝑢𝑗 Implicit CF / Item based neigbor methods
  • 87.
    87 Collaborative Filtering – ExplicitCollaborative Filtering – Implicit Collaborative Filtering – Neighbor methods – Matrix Factorization – Tensor Factorization Implicit Matrix Factorization
  • 88.
    88 𝑹 = 𝑷𝑸𝑻 𝑟 𝑢𝑖 = 𝒑 𝑢 𝑇 𝒒𝑖 𝑹 𝑵𝒙𝑴: preference matrix 𝑷 𝑵𝒙𝑲: user feature matrix 𝑸 𝑴𝒙𝑲: item feature matrix 𝑵: #users 𝑴: #items 𝑲: #features 𝑲 ≪ 𝑴, 𝑲 ≪ 𝑵 R Item1 Item2 Item3 … User1 User2 𝒓 𝑢𝑖 User3 … P 𝒑 𝑢 𝑇 QT 𝒒𝑖 Matrix Factorization
  • 89.
    89 • Objective function: CItem1 Item2 Item3 Item4 User1 𝒄11 1 1 1 User2 1 1 𝒄23 1 User3 𝒄31 𝒄32 1 1 User4 1 𝒄42 1 𝒄44 𝒇 𝑷, 𝑸 = 𝑾𝑺𝑺𝑬 = (𝒖,𝒊) 𝒄 𝒖𝒊 𝒓 𝒖𝒊 − 𝒓 𝒖𝒊 𝟐 𝑷 = ? 𝑸 = ? R Item1 Item2 Item3 Item4 User1 1 0 0 0 User2 0 0 1 0 User3 1 1 0 0 User4 0 1 0 1 Weighted Sum of Squared Errors
  • 90.
    90 Optimizer / AlternatingLeast Squares Ridge Regression • 𝑝 𝑢 = 𝑄 𝑇 𝐶 𝑢 𝑄 −1 𝑄 𝑇 𝐶 𝑢 𝑅 𝑟 𝑢 • 𝑞𝑖 = 𝑃 𝑇 𝐶 𝑖 𝑃 −1 𝑃 𝑇 𝐶 𝑖 𝑅 𝑐 𝑖 QT 0.1 -0.4 0.8 0.6 0.6 0.7 -0.7 -0.2 P -0.2 0.6 0.6 0.4 0.7 0.2 0.5 -0.2 R Item1 Item2 Item3 Item4 User1 1 0 0 0 User2 0 0 1 0 User3 1 1 0 0 User4 0 1 0 1
  • 91.
    91 Optimizer / AlternatingLeast Squares Ridge Regression • 𝑝 𝑢 = 𝑄 𝑇 𝐶 𝑢 𝑄 −1 𝑄 𝑇 𝐶 𝑢 𝑅 𝑟 𝑢 • 𝑞𝑖 = 𝑃 𝑇 𝐶 𝑖 𝑃 −1 𝑃 𝑇 𝐶 𝑖 𝑅 𝑐 𝑖 R Item1 Item2 Item3 Item4 User1 1 0 0 0 User2 0 0 1 0 User3 1 1 0 0 User4 0 1 0 1 QT 0.3 -0.3 0.7 0.7 0.7 0.8 -0.5 -0.1 P -0.2 0.6 0.6 0.4 0.7 0.2 0.5 -0.2
  • 92.
    92 Optimizer / AlternatingLeast Squares Ridge Regression • 𝑝 𝑢 = 𝑄 𝑇 𝐶 𝑢 𝑄 −1 𝑄 𝑇 𝐶 𝑢 𝑅 𝑟 𝑢 • 𝑞𝑖 = 𝑃 𝑇 𝐶 𝑖 𝑃 −1 𝑃 𝑇 𝐶 𝑖 𝑅 𝑐 𝑖 R Item1 Item2 Item3 Item4 User1 1 0 0 0 User2 0 0 1 0 User3 1 1 0 0 User4 0 1 0 1 QT 0.3 -0.3 0.7 0.7 0.7 0.8 -0.5 -0.1 P -0.2 0.7 0.6 0.5 0.8 0.2 0.6 -0.2
  • 93.
    93 Optimizer / AlternatingLeast Squares • Complexity of naive solution: 𝚶 𝑰𝑲 𝟐 𝑵𝑴 + 𝑰𝑲 𝟑 𝑵 + 𝑴 𝑬: number of examples, 𝑰 : number of iterations • Improvement (Hu, Koren, Volinsky) – Ridge Regression: 𝑝 𝑢 = 𝑄 𝑇 𝐶 𝑢 𝑄 −1 𝑄 𝑇 𝐶 𝑢 𝑅 𝑟 𝑢 – 𝑄 𝑇 𝐶 𝑢 𝑄 = 𝑄 𝑇 𝑄 + 𝑄 𝑇 𝐶 𝑢 − 𝐼 𝑄 = 𝐶𝑂𝑉𝑄0 + 𝐶𝑂𝑉𝑄+, 𝚶(𝑰𝑲 𝟐 𝑵𝑴) is costly – 𝐶𝑂𝑉𝑄0 is user independent, need to be calculated at the start of the iteration – Calculating 𝐶𝑂𝑉𝑄+ needs only #𝑷(𝒖)+ steps. • #𝑷(𝒖)+ : number of positive examples of user u – Complexity: 𝜪 𝑰𝑲 𝟐 𝑬 + 𝑰𝑲 𝟑(𝑵 + 𝑴) = 𝜪 𝑰𝑲 𝟐(𝑬 + 𝑲(𝑵 + 𝑴) – Codename: IALS • Complexity issues on large dataset: – If 𝑲 is low: 𝜪(𝑰𝑲 𝟐 𝑬) is dominant – If 𝑲 is high: 𝑶(𝑰𝑲 𝟑(𝑵 + 𝑴)) is dominant
  • 94.
    94 Collaborative Filtering – ExplicitCollaborative Filtering – Implicit Collaborative Filtering – Neighbor methods – Matrix Factorization – Tensor Factorization Implicit Tensor Factorization
  • 95.
    95 R1 Item1 Item2Item3 … User1 1 … User2 1 0 … User3 … …. … … … … • Tensor Factorizaton • Different preferences during the day • Context: Time period • Time period 1: 06:00-14:00
  • 96.
    96 R1 Item1 Item2Item3 … User1 1 … User2 1 0 … User3 … …. … … … … R2 Item1 Item2 Item3 … User1 0 1 … User2 1 … User3 1 … …. … … … … • Tensor Factorizaton • Different preferences during the day • Context: Time period • Time period 2: 14:00-22:00
  • 97.
    97 R1 Item1 Item2Item3 … User1 1 … User2 1 0 … User3 … …. … … … … R2 Item1 Item2 Item3 … User1 0 1 … User2 1 … User3 1 … …. … … … … R3 Item1 Item2 Item3 … User1 1 … User2 … User3 1 1 … …. … … … … • Tensor Factorizaton • Different preferences during the day • Context: Time period • Time period 3: 22:00-06:00
  • 98.
    98 R1 Item1 Item2Item3 … User1 1 … User2 1 0 … User3 … …. … … … … R2 Item1 Item2 Item3 … User1 0 1 … User2 1 … User3 1 … …. … … … … R3 Item1 Item2 Item3 … User1 … User2 𝒓 𝑢𝑖𝑡 … User3 … …. … … … … QT q11 q21 q31 … q12 q22 q32 … P p11 p12 p21 p22 p31 p32 … … Tt11 t12 t21 t22 t31 t32 𝑹 𝑵𝒙𝑴: preference matrix 𝑷 𝑵𝒙𝑲: user feature matrix 𝑸 𝑴𝒙𝑲: item feature matrix 𝑻 𝑳𝒙𝑲: time feature matrix 𝑵: #users 𝑴: #items 𝑳: #time periods 𝑲: #features 𝒓 𝒖𝒊t = 𝒌 𝒑 𝒖𝒌 𝒒𝒊𝒌 𝒕𝒕𝒌 𝑹 = 𝑷° 𝑸° 𝑻• Tensor Factorizaton
  • 99.
    99 • Relies onfeedbacks › Explicit feedback is more reliable than implicit feedback › But explicit feedback is often not provided, just implicit feedbacks › Does not need heterogeneous data sources • Extracts latent information using consumtion behaviors • More accurate than content based filtering • Domain independent Advantages of Collaborative Filtering
  • 100.
    100 • Cold startproblem › Items without feedbacks cannot be recommended › CF algorithms cannot provide personalized recommendations for users without feedbacks • Warmup period › Requires numerous events to be able to characterize the user properly › Inaccurate for users with few feedbacks › Inaccurate for domains with weak collaboration • Vulnerable to attacks • Harder to explain the recommendations • Unable to provide cross-domain recommendations without overlapping data Disadvantages of Collaborative Filtering
  • 101.
  • 102.
    102 Other applications /Hybrid Filtering • Combination of collaborative and content-based filtering • Advantages › Transforms latent behavioral knowledge to meta data level › Solves cold start problem › Used for weighting meta data, able to improve the efficiency of CBF › Behavioral knowledge can be interpreted by meta data • Disadvantages › Requires both meta data and transactional data › More complex than CF and CBF › Less developed than CF and CBF, hot topic › Challenging to provide mixed recommendations with new and old items › Heterogeneous data issues
  • 103.
    103 Other applications /Explanation • Approaches › Item explanation: Explanaining why those items were recommended › User explanation: Description about the user profile • Types of the item explanations › Non-personalized: „… because this content is trending” › Explicit features: „… because you bought horror movies” › Explicit user-to-user links: „… because your friends love this movie” › Explicit user relations: „.. because some users similar to you like this movie” › Implicit features: „… because actor X is similar to your preferred actor Y” • Types of user explanations › User tag cloud: „the user consumes 80% horror and 20% comedy” › User similarity: „the user is similar to another user who is a fan of Star Wars” • Goal: Increasing trust in recommender system • Difficulties: Hard to optimize how the recommendations should be explained
  • 104.
    104 Other applications /Recommender strategies • The most common way to recommend from prediction scores is to order items by them • How to avoid monotonity and „the winner takes all” effect? • How to avoid preference cannibalization? • Recommender strategies: How to select items from the scored list? › Best match › One episode per series › At most N items per category › Different categories should follow each other • Exploration vs. Exploitation with Multi-Armed Bandits • Entropy maximalization per recommendaiton box for new users
  • 105.
    105 Presented by: Contact: THANK YOU! www.impresstv.com DavidZibriczky Head of Data Science, ImpressTV Limited david.zibriczky@impresstv.com