Recommender System
Recommender systems support users by identifying
interesting products and services in situations where the
number and complexity of offers outstrips the user’s
capability to survey them and reach a decision.
docsity.com
Prepare for your exams
Study with the several resources on Docsity
Earn points to download
Earn points by helping other students or get them with a premium plan
Guidelines and tips
Prepare for your exams
Study with the several resources on Docsity
Earn points to download
Earn points by helping other students or get them with a premium plan
Recommender System-E Commerce-Lecture Slides, Slides of Fundamentals of E-Commerce
This is lecture for E-Commerce course. It was delivered by Prof. Abhra Honnenahalli at Acharya Nagarjuna University. It inlcudes: Recommender, System, Products, Services, Complexity, Survey, Tasks, Preference, Evaluating, Portfolio
Typology: Slides
2011/2012
1 / 19
This page cannot be seen from the preview
Don't miss anything!
Related documents
Partial preview of the text
Download Recommender System-E Commerce-Lecture Slides and more Slides Fundamentals of E-Commerce in PDF only on Docsity!
Recommender System Recommender^ systems^ support^ users
by^ identifying interesting products and services in situations where thenumber^ and^ complexity^ of
offers^ outstrips^ the^ user’s capability to survey them and reach a decision.
Tasks to Do for Building Recommender System Acquiring preference from customer’s input data Computing the recommendation Presenting the recommendation results Evaluating the recommender system
Input Data Customer Data Demographic data – name, age, gender, profession, DOB, hobbies, salary, education Rating data - rating scores: discrete multi-levels ratings and continuous rating; comments Behavior pattern data-^ duration of browsing, click times, links of web, save, print,scroll, refresh of webs, selection, search, bookmark, download of web content Transaction data - purchasing data, purchase quantity, price, discounting In early research first two were considered ascustomer data Web mining and other web techniques have beenused for capturing behavior patterns and transactiondata
Input Data Product Data Mainly the attributions of products for recommendation Example: Movie Recommender System Product data: topic, directors, actors, release years….
Recommendation Approaches Collaborative Filtering Content-based Filtering Hybrid Techniques Knowledge-based Techniques
Collaborative Filtering Nearest-neighbor^ method^ applied
to^ a^ ratings
matrix. Implements the idea of word-of-mouth promotion If two customers bought similar CDs and ratedthem similarly, the system would recommend toone customer CDs that the other customer boughtand rated positively Types of collaborative methods ^ Heuristic-based ^ Model-based
Collaborative Filtering Model-based Builds a model based on the training data using varioustechniques Check the model validity by using testing data Compute the production list or prediction rating for customer Different from heuristic-based In heuristic-based, entire database is used for computing results In model-based, just some inputs are given from customer tomodel and then prediction is used.
Collaborative Filtering Problems Cold starting :^ Initially there is very less information available about/from usersand about items ratings. Sparsity: Users can not rate millions of items available, so they rate very few. Gray sheep: There might exist users whose ratings are not consistently similarwith any group of users, and for this reason, they will rarely receive any accuraterecommendation.
Content-based Filtering Problems Limited content analysis because of limited keyword, overspecialization, and new user problems
Advantages/Disadvantages of Two Approaches user profiles are long-term models don’t^ exploit^ deep^ knowledge
about^ the^ product domain both^ are^ excellent^
for^ supporting^ the recommendation of simple products no^ additional^ knowledge
acquisition^ efforts^ are necessary if historical data is available
Knowledge-based Approaches An explicit representation of product, marketing, andsales knowledge is provided Useful for purchasing complex products such as financialservices, computers, or digital cameras Such a system helps to calculate solutions that fulfil certain quality requirements, explain solutions to a customer, and support customers when the system can’t find a solution. Implements explicit sales dialogues that help users selectitems Enables the validation of a recommender system’squality regarding calculated solutions.
Future Directions Construction of consumers’ profiles. Overcome the cold-starting, sparsity and scalabilityproblems Persuasion of recommender system Identify the factors that affect customers’ opinions –Relationship between customers and recommendersystems
Taxonomy for Personalized Recommendation Service