What really are recommendations engines nowadays?
This presentation introduces the foundations of recommendation algorithms, and covers common approaches as well as some of the most advanced techniques. Although more focused on efficiency than theoretical properties, basics of matrix algebra and optimization-based machine learning are used through the presentation.
Table of Contents:
1. Collaborative Filtering
1.1 User-User
1.2 Item-Item
1.3 User-Item
* Matrix Factorization
* Stochastic Gradient Descent (SGD)
* Truncated Singular Value Decomposition (SVD)
* Alternating Least Square (ALS)
* Deep Learning
2. Content Extraction
* Item-Item Similarities
* Deep Content Extraction: NLP, CNN, LSTM
3. Hybrid Models
4. In Production
4.1 Problematics
4.2 Solutions
4.3 Tools
In this lecture, I will first cover the recent advances in neural recommender systems such as autoencoder-based and MLP-based recommender systems. Then, I will introduce the recent achievement for automatic playlist continuation in music recommendation.
What really are recommendations engines nowadays?
This presentation introduces the foundations of recommendation algorithms, and covers common approaches as well as some of the most advanced techniques. Although more focused on efficiency than theoretical properties, basics of matrix algebra and optimization-based machine learning are used through the presentation.
Table of Contents:
1. Collaborative Filtering
1.1 User-User
1.2 Item-Item
1.3 User-Item
* Matrix Factorization
* Stochastic Gradient Descent (SGD)
* Truncated Singular Value Decomposition (SVD)
* Alternating Least Square (ALS)
* Deep Learning
2. Content Extraction
* Item-Item Similarities
* Deep Content Extraction: NLP, CNN, LSTM
3. Hybrid Models
4. In Production
4.1 Problematics
4.2 Solutions
4.3 Tools
In this lecture, I will first cover the recent advances in neural recommender systems such as autoencoder-based and MLP-based recommender systems. Then, I will introduce the recent achievement for automatic playlist continuation in music recommendation.
Recommender Systems represent one of the most widespread and impactful applications of predictive machine learning models.
Amazon, YouTube, Netflix, Facebook and many other companies generate an important fraction of their revenues thanks to their ability to model and accurately predict users ratings and preferences.
In this presentation we cover the following points:
→ introduction to recommender systems
→ working with explicit vs implicit feedback
→ content-based vs collaborative filtering approaches
→ user-based and item-item methods
→ machine learning and deep learning models
→ pros & cons of the methods: scalability, accuracy, explainability
Overview of the Recommender system or recommendation system. RFM Concepts in brief. Collaborative Filtering in Item and User based. Content-based Recommendation also described.Product Association Recommender System. Stereotype Recommendation described with advantage and limitations.Customer Lifetime. Recommender System Analysis and Solving Cycle.
Recommendation systems, also known as recommendation engines, are a type of information system whose purpose is to suggest, or recommend items or actions to users.
The recommendations may consist of:
-> retail items (movies, books, etc.) or
-> actions, such as following other users in a social network.
It can be said that, Recommendation engines are nothing but an automated form of a “shop counter guy”. You ask him for a product. Not only he shows that product, but also the related ones which you could buy. They are well trained in cross selling and up selling. So, does our recommendation engines.
Recommender systems are software tools and techniques providing suggestions for items to be of interest to a user. Recommender systems have proved in recent years to be a valuable means of helping Web users by providing useful and effective recommendations or suggestions.
Recommender systems support the decision making processes of customers with personalized suggestions. These widely used systems influence the daily life of almost everyone across domains like ecommerce, social media, and entertainment. However, the efficient generation of relevant recommendations in large-scale systems is a very complex task. In order to provide personalization, engines and algorithms need to capture users’ varying tastes and find mostly nonlinear dependencies between them and a multitude of items. Enormous data sparsity and ambitious real-time requirements further complicate this challenge. At the same time, deep learning has been proven to solve complex tasks like object or speech recognition where traditional machine learning failed or showed mediocre performance.
Join Marcel Kurovski to explore a use case for vehicle recommendations at mobile.de, Germany’s biggest online vehicle market. Marcel shares a novel regularization technique for the optimization criterion and evaluates it against various baselines. To achieve high scalability, he combines this method with strategies for efficient candidate generation based on user and item embeddings—providing a holistic solution for candidate generation and ranking.
The proposed approach outperforms collaborative filtering and hybrid collaborative-content-based filtering by 73% and 143% for MAP@5. It also scales well for millions of items and users returning recommendations in tens of milliseconds.
Event: O'Reilly Artificial Intelligence Conference, New York, 18.04.2019
Speaker: Marcel Kurovski, inovex GmbH
Mehr Tech-Vorträge: inovex.de/vortraege
Mehr Tech-Artikel: inovex.de/blog
With the explosive growth of online information, recommender system has been an effective tool to overcome information overload and promote sales. In recent years, deep learning's revolutionary advances in speech recognition, image analysis and natural language processing have gained significant attention. Meanwhile, recent studies also demonstrate its efficacy in coping with information retrieval and recommendation tasks. Applying deep learning techniques into recommender system has been gaining momentum due to its state-of-the-art performance. In this talk, I will present recent development of deep learning based recommender models and highlight some future challenges and open issues of this research field.
Recommender Systems represent one of the most widespread and impactful applications of predictive machine learning models.
Amazon, YouTube, Netflix, Facebook and many other companies generate an important fraction of their revenues thanks to their ability to model and accurately predict users ratings and preferences.
In this presentation we cover the following points:
→ introduction to recommender systems
→ working with explicit vs implicit feedback
→ content-based vs collaborative filtering approaches
→ user-based and item-item methods
→ machine learning and deep learning models
→ pros & cons of the methods: scalability, accuracy, explainability
Overview of the Recommender system or recommendation system. RFM Concepts in brief. Collaborative Filtering in Item and User based. Content-based Recommendation also described.Product Association Recommender System. Stereotype Recommendation described with advantage and limitations.Customer Lifetime. Recommender System Analysis and Solving Cycle.
Recommendation systems, also known as recommendation engines, are a type of information system whose purpose is to suggest, or recommend items or actions to users.
The recommendations may consist of:
-> retail items (movies, books, etc.) or
-> actions, such as following other users in a social network.
It can be said that, Recommendation engines are nothing but an automated form of a “shop counter guy”. You ask him for a product. Not only he shows that product, but also the related ones which you could buy. They are well trained in cross selling and up selling. So, does our recommendation engines.
Recommender systems are software tools and techniques providing suggestions for items to be of interest to a user. Recommender systems have proved in recent years to be a valuable means of helping Web users by providing useful and effective recommendations or suggestions.
Recommender systems support the decision making processes of customers with personalized suggestions. These widely used systems influence the daily life of almost everyone across domains like ecommerce, social media, and entertainment. However, the efficient generation of relevant recommendations in large-scale systems is a very complex task. In order to provide personalization, engines and algorithms need to capture users’ varying tastes and find mostly nonlinear dependencies between them and a multitude of items. Enormous data sparsity and ambitious real-time requirements further complicate this challenge. At the same time, deep learning has been proven to solve complex tasks like object or speech recognition where traditional machine learning failed or showed mediocre performance.
Join Marcel Kurovski to explore a use case for vehicle recommendations at mobile.de, Germany’s biggest online vehicle market. Marcel shares a novel regularization technique for the optimization criterion and evaluates it against various baselines. To achieve high scalability, he combines this method with strategies for efficient candidate generation based on user and item embeddings—providing a holistic solution for candidate generation and ranking.
The proposed approach outperforms collaborative filtering and hybrid collaborative-content-based filtering by 73% and 143% for MAP@5. It also scales well for millions of items and users returning recommendations in tens of milliseconds.
Event: O'Reilly Artificial Intelligence Conference, New York, 18.04.2019
Speaker: Marcel Kurovski, inovex GmbH
Mehr Tech-Vorträge: inovex.de/vortraege
Mehr Tech-Artikel: inovex.de/blog
With the explosive growth of online information, recommender system has been an effective tool to overcome information overload and promote sales. In recent years, deep learning's revolutionary advances in speech recognition, image analysis and natural language processing have gained significant attention. Meanwhile, recent studies also demonstrate its efficacy in coping with information retrieval and recommendation tasks. Applying deep learning techniques into recommender system has been gaining momentum due to its state-of-the-art performance. In this talk, I will present recent development of deep learning based recommender models and highlight some future challenges and open issues of this research field.
Recommendations are everywhere : music, movies, books, social medias, e-commerce web sites… The Web is leaving the era of search and entering one of discovery. This quick introduction will help you to understand this vast topic and why you should use it.
Modern Perspectives on Recommender Systems and their Applications in MendeleyKris Jack
Presentation given for one of Pearson's Data Research teams. It motivates the use of recommender systems, describes common approaches to building and evaluating them and gives examples of how they are used in Mendeley. Thanks to Maya Hristakeva for creating some of the slides.
Big data certification training mumbaiTejaspathiLV
ExcelR offers 160 hours classroom training on Business Analytics / Data Scientist / Data Analytics. We are considered as one of the best training institutes on Business Analytics in Mumbai. “Faculty and vast course agenda is our differentiator”. The training is conducted by alumni of premier institutions such as IIT & ISB who has extensive experience in the arena of analytics. They are considered to be one of the best trainers in the industry. The topics covered as part of this Data Scientist Certification program is on par with most of the Master of Science in Analytics (MS in Business Analytics / MS in Data Analytics) programs across the top-notch universities of the globe.
ExcelR is a proud partner of Universiti Malaysia Saravak (UNIMAS), Malaysia’s 1st public University and ranked 8th top university in Malaysia and ranked among top 200th in Asian University Rankings 2017 by QS World University Rankings. Participants will be awarded Data Science international certification from UNIMAS.
ExcelR is the best Data Science training institute in Hyderabad which offers services from training to placement as part of the Data Science training program with over 400+ participants placed in various multinational companies including E&Y, Panasonic, Accenture, VMWare, Infosys, IBM, etc….and the staff is from NIT’s & IIT’s
ExcelR is a proud partner of Universiti Malaysia Saravak (UNIMAS), Malaysia’s 1st public University and ranked 8th top university in Malaysia and ranked among top 200th in Asian University Rankings 2017 by QS World University Rankings. Participants will be awarded Data Science international certification from UNIMAS.
A presentation that gives an overview about latest machine learning and deep learning techniques and use-cases that are prevalent in the eCommerce industry
case based recommendation approach for market basket datamniranjanmurthy
Recommender systems have become an important part of various applications in e-commerce, supporting both customers and providers in their decision-making processes. However, these systems still must overcome limitations that reduce their performance, like recommendations overspecialization, less popular item providing, and difficulties when items with unequal probability distribution appear or recommendations for sets of items are asked. A novel approach, addressing the above issues through a case-based recommendation methodology is presented here. The scope of the presented approach is to generate meaningful recommendations based on items' co-occurring patterns and to provide more insight into customers' buying habits. In contrast to current recommendation techniques that recommend items based on users' ratings or history, and to most case-based item recommenders that evaluate items' similarities, the implemented recommender uses a hierarchical model for the items and searches for similar sets of items, in order to recommend those that are most likely to satisfy a user.
GraphSummit Singapore | The Future of Agility: Supercharging Digital Transfor...Neo4j
Leonard Jayamohan, Partner & Generative AI Lead, Deloitte
This keynote will reveal how Deloitte leverages Neo4j’s graph power for groundbreaking digital twin solutions, achieving a staggering 100x performance boost. Discover the essential role knowledge graphs play in successful generative AI implementations. Plus, get an exclusive look at an innovative Neo4j + Generative AI solution Deloitte is developing in-house.
Epistemic Interaction - tuning interfaces to provide information for AI supportAlan Dix
Paper presented at SYNERGY workshop at AVI 2024, Genoa, Italy. 3rd June 2024
https://alandix.com/academic/papers/synergy2024-epistemic/
As machine learning integrates deeper into human-computer interactions, the concept of epistemic interaction emerges, aiming to refine these interactions to enhance system adaptability. This approach encourages minor, intentional adjustments in user behaviour to enrich the data available for system learning. This paper introduces epistemic interaction within the context of human-system communication, illustrating how deliberate interaction design can improve system understanding and adaptation. Through concrete examples, we demonstrate the potential of epistemic interaction to significantly advance human-computer interaction by leveraging intuitive human communication strategies to inform system design and functionality, offering a novel pathway for enriching user-system engagements.
UiPath Test Automation using UiPath Test Suite series, part 5DianaGray10
Welcome to UiPath Test Automation using UiPath Test Suite series part 5. In this session, we will cover CI/CD with devops.
Topics covered:
CI/CD with in UiPath
End-to-end overview of CI/CD pipeline with Azure devops
Speaker:
Lyndsey Byblow, Test Suite Sales Engineer @ UiPath, Inc.
Essentials of Automations: The Art of Triggers and Actions in FMESafe Software
In this second installment of our Essentials of Automations webinar series, we’ll explore the landscape of triggers and actions, guiding you through the nuances of authoring and adapting workspaces for seamless automations. Gain an understanding of the full spectrum of triggers and actions available in FME, empowering you to enhance your workspaces for efficient automation.
We’ll kick things off by showcasing the most commonly used event-based triggers, introducing you to various automation workflows like manual triggers, schedules, directory watchers, and more. Plus, see how these elements play out in real scenarios.
Whether you’re tweaking your current setup or building from the ground up, this session will arm you with the tools and insights needed to transform your FME usage into a powerhouse of productivity. Join us to discover effective strategies that simplify complex processes, enhancing your productivity and transforming your data management practices with FME. Let’s turn complexity into clarity and make your workspaces work wonders!
The Art of the Pitch: WordPress Relationships and SalesLaura Byrne
Clients don’t know what they don’t know. What web solutions are right for them? How does WordPress come into the picture? How do you make sure you understand scope and timeline? What do you do if sometime changes?
All these questions and more will be explored as we talk about matching clients’ needs with what your agency offers without pulling teeth or pulling your hair out. Practical tips, and strategies for successful relationship building that leads to closing the deal.
Encryption in Microsoft 365 - ExpertsLive Netherlands 2024Albert Hoitingh
In this session I delve into the encryption technology used in Microsoft 365 and Microsoft Purview. Including the concepts of Customer Key and Double Key Encryption.
Smart TV Buyer Insights Survey 2024 by 91mobiles.pdf91mobiles
91mobiles recently conducted a Smart TV Buyer Insights Survey in which we asked over 3,000 respondents about the TV they own, aspects they look at on a new TV, and their TV buying preferences.
Sudheer Mechineni, Head of Application Frameworks, Standard Chartered Bank
Discover how Standard Chartered Bank harnessed the power of Neo4j to transform complex data access challenges into a dynamic, scalable graph database solution. This keynote will cover their journey from initial adoption to deploying a fully automated, enterprise-grade causal cluster, highlighting key strategies for modelling organisational changes and ensuring robust disaster recovery. Learn how these innovations have not only enhanced Standard Chartered Bank’s data infrastructure but also positioned them as pioneers in the banking sector’s adoption of graph technology.
Generative AI Deep Dive: Advancing from Proof of Concept to ProductionAggregage
Join Maher Hanafi, VP of Engineering at Betterworks, in this new session where he'll share a practical framework to transform Gen AI prototypes into impactful products! He'll delve into the complexities of data collection and management, model selection and optimization, and ensuring security, scalability, and responsible use.
SAP Sapphire 2024 - ASUG301 building better apps with SAP Fiori.pdfPeter Spielvogel
Building better applications for business users with SAP Fiori.
• What is SAP Fiori and why it matters to you
• How a better user experience drives measurable business benefits
• How to get started with SAP Fiori today
• How SAP Fiori elements accelerates application development
• How SAP Build Code includes SAP Fiori tools and other generative artificial intelligence capabilities
• How SAP Fiori paves the way for using AI in SAP apps
A tale of scale & speed: How the US Navy is enabling software delivery from l...sonjaschweigert1
Rapid and secure feature delivery is a goal across every application team and every branch of the DoD. The Navy’s DevSecOps platform, Party Barge, has achieved:
- Reduction in onboarding time from 5 weeks to 1 day
- Improved developer experience and productivity through actionable findings and reduction of false positives
- Maintenance of superior security standards and inherent policy enforcement with Authorization to Operate (ATO)
Development teams can ship efficiently and ensure applications are cyber ready for Navy Authorizing Officials (AOs). In this webinar, Sigma Defense and Anchore will give attendees a look behind the scenes and demo secure pipeline automation and security artifacts that speed up application ATO and time to production.
We will cover:
- How to remove silos in DevSecOps
- How to build efficient development pipeline roles and component templates
- How to deliver security artifacts that matter for ATO’s (SBOMs, vulnerability reports, and policy evidence)
- How to streamline operations with automated policy checks on container images
GDG Cloud Southlake #33: Boule & Rebala: Effective AppSec in SDLC using Deplo...James Anderson
Effective Application Security in Software Delivery lifecycle using Deployment Firewall and DBOM
The modern software delivery process (or the CI/CD process) includes many tools, distributed teams, open-source code, and cloud platforms. Constant focus on speed to release software to market, along with the traditional slow and manual security checks has caused gaps in continuous security as an important piece in the software supply chain. Today organizations feel more susceptible to external and internal cyber threats due to the vast attack surface in their applications supply chain and the lack of end-to-end governance and risk management.
The software team must secure its software delivery process to avoid vulnerability and security breaches. This needs to be achieved with existing tool chains and without extensive rework of the delivery processes. This talk will present strategies and techniques for providing visibility into the true risk of the existing vulnerabilities, preventing the introduction of security issues in the software, resolving vulnerabilities in production environments quickly, and capturing the deployment bill of materials (DBOM).
Speakers:
Bob Boule
Robert Boule is a technology enthusiast with PASSION for technology and making things work along with a knack for helping others understand how things work. He comes with around 20 years of solution engineering experience in application security, software continuous delivery, and SaaS platforms. He is known for his dynamic presentations in CI/CD and application security integrated in software delivery lifecycle.
Gopinath Rebala
Gopinath Rebala is the CTO of OpsMx, where he has overall responsibility for the machine learning and data processing architectures for Secure Software Delivery. Gopi also has a strong connection with our customers, leading design and architecture for strategic implementations. Gopi is a frequent speaker and well-known leader in continuous delivery and integrating security into software delivery.
Observability Concepts EVERY Developer Should Know -- DeveloperWeek Europe.pdfPaige Cruz
Monitoring and observability aren’t traditionally found in software curriculums and many of us cobble this knowledge together from whatever vendor or ecosystem we were first introduced to and whatever is a part of your current company’s observability stack.
While the dev and ops silo continues to crumble….many organizations still relegate monitoring & observability as the purview of ops, infra and SRE teams. This is a mistake - achieving a highly observable system requires collaboration up and down the stack.
I, a former op, would like to extend an invitation to all application developers to join the observability party will share these foundational concepts to build on:
GraphRAG is All You need? LLM & Knowledge GraphGuy Korland
Guy Korland, CEO and Co-founder of FalkorDB, will review two articles on the integration of language models with knowledge graphs.
1. Unifying Large Language Models and Knowledge Graphs: A Roadmap.
https://arxiv.org/abs/2306.08302
2. Microsoft Research's GraphRAG paper and a review paper on various uses of knowledge graphs:
https://www.microsoft.com/en-us/research/blog/graphrag-unlocking-llm-discovery-on-narrative-private-data/
Climate Impact of Software Testing at Nordic Testing DaysKari Kakkonen
My slides at Nordic Testing Days 6.6.2024
Climate impact / sustainability of software testing discussed on the talk. ICT and testing must carry their part of global responsibility to help with the climat warming. We can minimize the carbon footprint but we can also have a carbon handprint, a positive impact on the climate. Quality characteristics can be added with sustainability, and then measured continuously. Test environments can be used less, and in smaller scale and on demand. Test techniques can be used in optimizing or minimizing number of tests. Test automation can be used to speed up testing.
6. What are
Recommender Engines?
• Personalized Information Agents
• Predict items a user may be interested in
• Engines use different techniques utilizing
different knowledge sources
• Knowledge can be derived from user
features, item features and user-item ratings
8. Role of Recommender
Engines in E-Commerce
• Integral part of many e-businesses
• Are an unique feature of e-businesses
• Websites can track customer’s behaviours
• E-businesses can offer huge stocks, that
make personalization necessary
• Recommender engines allow personalization
9. Benefits provided by
Recommender Engines
• Make users purchase more items
• Help gaining customer loyalty
• Build a “value-added relationship” between
the e-business and the customer
• Promote older and low-demand items
10. Recommender Engines
• What are Recommender Engines?
• Recommendation Techniques
• Examples in E-Commerce
11. Demographic
Recommendation
• Use information about the users
(demographics)
• Find users with similar features
• Recommend items that are preferred by
similar users
14. Demographic
Recommendation
• Advantages:
• No user-item ratings needed - new users can
immediately get recommendations
• No knowledge about item features needed
• Problems:
• Users have privacy concerns about disclosing
information about themselves
• Demographic data is too crude for highly
personalized recommendations
15. Content-based
Recommendation
• Use information about the items (e.g.
keywords, genres)
• Find items similar to the items preferred in
the past
• Recommend the items with the highest
similarity
18. Content-based
Recommendation
• Works well if items can be properly
represented as a set of features
• Problems:
• Content analysis is necessary
• New user cold start problem: users with
little or no user-item ratings are difficult to
categorize
19. User-based
Collaborative Filtering
• Use user-item ratings matrix
• Make user-to-user correlations
• Find highly correlated users
• Recommend items preferred by those users
23. User-based
Collaborative Filtering
• Advantage:
• No knowledge about item features needed
• Problems:
• New user cold start problem
• New item cold start problem: items with few ratings
cannot easily be recommended
• Sparsity problem: depends on overlap in ratings across
users - has difficulties when space of ratings is sparse
• Does not scale well for large data sets
24. Item-based
Collaborative Filtering
• Use user-item ratings matrix
• Make item-to-item correlations
• Find items that are highly correlated with
the known preferred ones
• Recommend items with highest correlation
28. Item-based
Collaborative Filtering
• Advantages:
• No knowledge about item features needed
• Better scalability, because correlations between limited
number of items instead of very large number of users
• Reduced sparsity problem
• Problems:
• New user cold start problem
• New item cold start problem
29. Hybrid Approaches
• Combine collaborative and demographic/
content-based techniques
• Collaborative filtering systems achieve
better predictions but have cold start and
sparsity problems
• Demographic/content-based systems work
without rating data and can therefore
compensate these drawbacks
30. Recommender Engines
• What are Recommender Engines?
• Recommendation Techniques
• Examples in E-Commerce
31. Amazon.com
• Recommender engine based on item-based
collaborative filtering
• A item-to-item matrix is calculated offline
through a similarity function using item pairs
that customers tend to purchase together
• Thus the online lookup of similar items for
recommendations is very fast and scales
independent of catalog size and number of
customers
32.
33.
34.
35.
36.
37. Digg.com
• Recommender engine is based on user-based
collaborative filtering
• Correlation coefficient is calculated between
users than “dugg” the same stories
• Upcoming stories that have been “dugg” by highly
correlated users are recommended
• Engine uses a custom graph-database and runs in
real-time without prediction model and batch
processing
38.
39.
40. Conclusion
• Recommender engines are a powerful
technology for personalization
• Provide benefits for businesses and consumers
• Item-based collaborative filtering in a hybrid
system with content-based recommendation is
state of the art
• But which technique works best always
depends on the concrete use case
* My topic are recommender engines
* We all know them from websites like ...
... Amazon (the \"Customers who bought this also bought\" feature) ...
... Digg.com (“Diggers Like You” recommend upcoming stories) ...
... or last.fm (Music, Events, and Videos matching your music taste)
* First I’m talking about what recommender engines are in general and e-commerce
* Second I will talk about the different techniques used by recommender engines
* And finally I’m going to show some examples of recommender engines in e-commerce
* Recommender engines are Personalized Information Agents, which means, they provide personalized information to individual users
* Particularly they recommend items to users by predicting which items out of a large pool a user may be interested in
* Recommender engines use a variety of techniques that use different knowledge sources for their predictions
* These knowledge sources can be information about the users, about the items and information about users’ preferences for items. These are called ratings.
* Here we have the concept again in an illustration
* We have a pool of items and a group of users
* The recommender engine uses knowledge about the users, the items, and the user-item preferences
* Based on this knowledge it recommends a specific item to a specific user
* The user-item preferences are often called user-item ratings and can be derived from many different sources like the purchase history, explicit item ratings or preference questionnaires.
* So what role do recommender engines play in e-commerce?
* Recommender engines are used by more and more e-commerce businesses
* They are a tool for to personalize websites for customers
* Personalization is something only e-businesses can do in contrast to real stores
* E-Businesses have no storage space constrains, in contrast to real world stores
* Therefore they can have very large catalogs of items
* But these are unable to be looked through manually, which makes personalization a necessity
* What benefits do e-businesses have from using recommender engines?
* Recommender Engines present customers with items that are interesting for them, but that they didn’t plan to buy
* This makes them to buy more items, which increases the sales of the e-business
* These unplanned purchases are common in real world stores but don’t happen that often in e-businesses
* Engines make it easier and faster to find new items, which makes the customers come back more often. Customer loyalty is increased.
* And the often customers come back and the more they purchase, the more the recommendation system knows about them and their preferences.
* The accuracy of the recommendations increases which increases the value of the website for the user.
* Last but not least E-businesses can promote specific items through the recommendation system.
* Now I’m going to talk about the recommendation techniques with their strengths and weaknesses
* First we have non-personalized recommendations
* First we have demographic recommendations
* As said before the techniques are classification based on which knowledge sources they use
* Demographic recommendation uses the knowledge about the users
* This are features like age, gender, profession, income, or location
* To make recommendation the system first looks for users with similar features to the one it wants to find recommendations for
* And then recommends items that are preferred by these similar users
* Here is the process again in an illustration
* We have three users and three items
* User 2 likes Item A and User 3 likes Item C
* User 1 and User 3 have similar demographic features
* So the system recommends Item C to User 1
* The advantages of demographic recommendation are:
* The technique does not need rating data for the user it recommends items to
* So a new user without any ratings can immediately get recommendations as long as he has given information about himself
* The techniques also doesn’t need to know anything about the items the system deals with
* The drawbacks are:
* Users generally don’t like to give away information about themselves, but this is the only knowledge source the system has
* Demographic data is too crude to make good recommendations. The system will make generalisations that are problematic especially when the system deals with cultural items like books, music, or movies.
* Not all 30 year old females like the same movies.
* Users with unusual taste will get no good recommendations
* The next technique are content-based recommendations
* This technique uses information about the items
* This are keywords or genres. For movies for example this could be year, title, director, actors and so on.
* The system first looks at the items a user has preferred in the past
* And then searches the item database for similar items that have the same features
* From these search results the engine then recommends the items that have the highest similarity
* Here is how it works in an illustration
* We have a user and three items
* Items A and C have the same features
* So the engine recommends Item C to the user
* This technique works well if the items can be properly represented as a set of features.
* If the items or the item descriptions can be analysed with automatic content analysis the features can be extracted automatically.
* But if this is not possible it has to be done manually, which takes a lot of resources.
* The technique has problems if new users have rated none or only a few items. Because then the system doesn’t know what items to search for.
* The next techniques are collaborative filtering techniques.
* These techniques use the user-item rating data matrix as knowledge source.
* First we look at user-based collaborative filtering.
* This technique makes correlations between users based on their item preferences.
* The system first finds highly correlated users based on their rating profiles, that means users that have the shared item preferences
* Then it recommends items that the highly correlated users also preferred
* In the illustration it looks like this
* We have three users, and three items
* User 1 likes items A and B
* User 2 likes item C
* and User 3 likes all three items
* So User 1 and 3 share their preference for items A and B and have therefore a high correlation
* So the recommender system recommends item C to User 1 as there is a high probability User 1 will like it
* In the user-item rating matrix this looks like this
* We have 6 users, 5 items and ratings from 1-10
* User U3 has very similar ratings to User A, so there is a high correlation
* As user U3 also gives item I5 a high rating, we can recommend this item to User a
* The advantage of all collaborative filtering techniques is, that they don’t need to know anything about the items the system deals with, just like demographic recommendation
* The problems are:
* We have the new user cold start problem that we also had with the content-based recommendation
* We also have a cold start problem for new items, which means that new items that have none or only a few ratings cannot be recommended easily
* Sparse data in the rating matrix is a problem, as collaborative filtering techniques depend on an overlap in ratings across users. If few users have rated the same items the quality of the recommendations gets bad.
* User-based collaborative filtering also does not scale well for large data sets. The user-dimension of the matrix usually becomes much larger than the item-dimension, which makes drawing of user-to-user correlations very expensive.
* The next technique is item-based collaborative filtering
* Here we don’t draw correlations between users but between items
* The algorithm looks at an item and then finds other items that are preferred by the same users
* So it builds item-to-item correlations based on the shared appreciation of users
* The highly correlated items are then recommended
* In the diagram it looks like this
* We again have three users and three items
* User 1 likes Item A
* User 2 and User 3 like items A and C
* So the Items A and C are correlated as they are preferred by the same users
* So the engine recommends item C to User 1
* In the matrix it looks like this
* Items I and I5 have high ratings from the same users, so I5 is recommended
* The advantages of this technique are:
* No knowledge about the items is needed as in all collaborative filtering approaches
* The approach scales better than the user-based approach, as the number of items is usually smaller than the number of users, and items are easier to categorize.
* Because of this the sparsity problem is also reduced.
* But the approach also has the cold start problems for new users and new items
* The last approach I’m going to present are hybrid recommender systems
* This is not a new technique but rather a combination of techniques
* As we have seen all of the techniques have benefits and drawbacks
* Hybrid approaches try to combine complementary techniques so that the drawbacks will be neutralized
* This works especially for the combination of collaborative filtering techniques with demographic or content-based techniques
* Collaborative filtering techniques achieve a high prediction accuracy but have problems when the available rating data is limited
* Here demographic and content-based can help out because they don’t rely on rating data
* Concluding I’m going to show two examples of websites that use recommender engines
* What techniques do they use and how are the recommendations used on the websites
* Recommender engines are developed and run on the one hand by independent technology vendors
* Also many e-businesses have their own engines developed
* ChoiceStream is one of the leading independent vendors
* The first one is amazon.com
* Amazon is maybe the most famous example for an e-business utilizing a recommender engine
* They use the recommendation very extensively to personalize their website
* The recommender engine is based on item-based collaborative filtering
* Their system consists out of an online and an offline component
* The offline component calculates a item-to-item matrix with the correlation coefficients
* It does this by using a similarity function that gets applied to pairs of items that customers tend to purchase together
* The online component then only has to lookup similar items in this matrix when recommendations are needed
* Through this separation Amazon is able to deal with their huge data set of several million catalog items and tens of millions of users.
* The expensive calculations are performed offline and the online component scales independent on catalog size and number of users. It only depends on the number of items a user has purchased and rated.
* Now let’s see how the recommendation engine is used on the website.
* Of course there is the famous “Customers who bought this also bought” feature shown on every item detail page ...
.. and also on the shopping cart page.
* Then there is the “Your Recommendations” page, that shows all the recommendations for you.
* You can mark recommendations as already owned or not interested in and also rate them in order to influence your recommendations.
* It is also shown why an item is recommended, that is which purchase is correlated to the recommended item.