Data Miing and Knowledge Discvoery - Web Data Mining
Overview of Web Mining and E-Commerce Data Analytics Bamshad BamshadMobasher Mobasher DePaul DePaulUniversity University Why Data Mining Increased Availability of Huge Amounts of Data point-of-sale customer data (Walmart: 60M transactions per day) E-commerce transaction data digitization of text, images, video, voice, etc. World Wide Web and Online collections usage/navigation data (Yahoo: 20 terabytes of clickstream data per day) Data Too Large or Complex for Classical or Manual Analysis
number of records in millions or billions high dimensional data (too many fields/features/attributes) often too sparse for rudimentary observations high rate of growth (e.g., through logging or automatic data collection) heterogeneous data sources Business Necessity e-commerce high degree of competition personalization, customer loyalty, market segmentation 2 From Data to Wisdom Data The raw material of information Wisdom Information Data organized and
presented by someone Knowledge Knowledge Information read, heard or seen and understood and integrated Information Data Wisdom Distilled knowledge and understanding which can lead to decisions The Information Hierarchy 3 What is Data Mining What do we need? Extract interesting and useful knowledge from the data Find rules, regularities, irregularities, patterns, constraints hopefully, this will help us better compete in business, do research, learn concepts, make money, etc.
Data Mining: A Definition The Thenon-trivial non-trivialextraction extractionof ofimplicit, implicit,previously previouslyunknown unknownand and potentially potentiallyuseful usefulknowledge knowledgefrom fromdata datain inlarge largedata datarepositories repositories Non-trivial: obvious knowledge is not useful implicit: hidden difficult to observe knowledge previously unknown
potentially useful: actionable; easy to understand 4 Data Minings Virtuous Cycle 1. Identifying the business problem* 2. Mining data to transform it into actionable information 3. Acting on the information 4. Measuring the results * Textbook interchanges problem with opportunity 5 1. Identify the Business Opportunity First Step: clearly identify the business problem that requires a solution Then translate this problem into a data mining problem
Many business processes are good candidates: New product introduction / eliminating a product line Direct marketing campaign Understanding customer attrition/churn Evaluating the results of a test market Measurements from past DM efforts: What types of customers responded to our last campaign? Where do the best customers live? Are long waits in check-out lines a cause of customer attrition? What products should be promoted with our XYZ product? 6 2. Mining data to transform it into actionable information Success is making business sense of the data
Need to identify the right data mining tasks that can address the specified problem Numerous data issues: Bad data formats (alpha vs numeric, missing, null, bogus data) Confusing data fields (synonyms and differences) Lack of functionality (I wish I could) Legal ramifications (privacy, etc.) Organizational factors (unwilling to change our ways) Lack of timeliness 7 3. Acting on the Information This is the purpose of Data Mining with the hope of adding value What type of action?
Interactions with customers, prospects, suppliers Modifying service procedures Adjusting inventory levels Consolidating Expanding Etc 8 4. Measuring the Results Assesses the impact of the action taken Often overlooked, ignored, skipped Planning for the measurement should begin when analyzing the business opportunity, not after it is all over Assessment questions (examples): Did this ____ campaign do what we hoped? Did some offers work better than others? Did these customers purchase additional products? Tons of others 9
The Knowledge Discovery Process Data Mining v. Knowledge Discovery in Databases (KDD) DM and KDD are often used interchangeably actually, DM is only part of the KDD process - The KDD Process 10 What Can Data Mining Do Two kinds of knowledge discovery: directed and undirected Directed Knowledge Discovery Purpose: Explain value of some field in terms of all the others (goal-oriented) Method: select the target field based on some hypothesis about the data; ask the algorithm to tell us how to predict or classify new instances Examples: what products show increased sale when cream cheese is discounted which banner ad to use on a web page for a given user coming to the site Undirected Knowledge Discovery Purpose: Find patterns in the data that may be interesting (no target field) Method: clustering, affinity grouping Examples: which products in the catalog often sell together market segmentation (groups of customers/users with similar characteristics) 11
What Can Data Mining Do Many Data Mining Tasks often inter-related often need to try different techniques for each task each tasks may require different types of knowledge discovery What are some of data mining tasks Classification Prediction Characterization Discrimination Affinity Grouping Clustering Sequence Analysis Description 12 Some Applications of Data mining
Business data analysis and decision support Marketing focalization Recognizing specific market segments that respond to particular characteristics Return on mailing campaign (target marketing) Customer Profiling Segmentation of customer for marketing strategies and/or product offerings Customer behavior understanding Customer retention and loyalty Mass customization / personalization 13 Some Applications of Data mining Business data analysis and decision support (cont.) Market analysis and management Provide summary information for decision-making Market basket analysis, cross selling, market segmentation. Resource planning Risk analysis and management
"What if" analysis Forecasting Pricing analysis, competitive analysis Time-series analysis (Ex. stock market) 14 Some Applications of Data mining Fraud detection Detecting telephone fraud: Telephone call model: destination of the call, duration, time of day or week Analyze patterns that deviate from an expected norm British Telecom identified discrete groups of callers with frequent intra-group calls, especially mobile phones, and broke a multimillion dollar fraud scheme Detection of credit-card fraud Detecting suspicious money transactions (money laundering) Text mining: Message filtering (e-mail, newsgroups, etc.) Newspaper articles analysis Text and document categorization Web Mining . . . 15 What is Web Mining From its very beginning, the potential of extracting valuable
knowledge from the Web has been quite evident Web mining is the collection of technologies to fulfill this potential. Web Mining Definition application applicationof ofdata datamining miningand andmachine machinelearning learning techniques techniquesto toextract extractuseful usefulknowledge knowledgefrom fromthe thecontent, content, structure, structure,and andusage usageof ofWeb Webresources. resources.
16 Types of Web Mining Web WebMining Mining Web WebContent Content Mining Mining Web WebUsage Usage Mining Mining 17 Web WebStructure Structure Mining Mining
Types of Web Mining Web WebMining Mining Web WebContent Content Mining Mining Web WebUsage Usage Mining Mining Extracting useful knowledge from the contents of Web documents or other semantic information about Web resources 18 Web WebStructure
Structure Mining Mining Types of Web Mining Web WebMining Mining Web WebContent Content Mining Mining Web WebUsage Usage Mining Mining Content data may consist of text, images, audio, video, structured records from lists and tables, or item attributes from backend databases.
19 Web WebStructure Structure Mining Mining Types of Web Mining Web WebMining Mining Web WebContent Content Mining Mining Web WebUsage Usage Mining Mining Applications: document clustering or categorization
topic identification / tracking concept discovery focused crawling content-based personalization intelligent search tools 20 Web WebStructure Structure Mining Mining Types of Web Mining Web WebMining Mining Web WebContent Content Mining Mining Web WebUsage Usage Mining
Mining Extracting interesting patterns from user interactions with resources on one or more Web sites 21 Web WebStructure Structure Mining Mining Types of Web Mining Web WebMining Mining Web WebContent Content Mining Mining Web
WebUsage Usage Mining Mining Applications: user and customer behavior modeling Web site optimization e-customer relationship management Web marketing targeted advertising recommender systems 22 Web WebStructure Structure Mining Mining Types of Web Mining Web WebMining Mining Web WebContent
Content Mining Mining Web WebUsage Usage Mining Mining Web WebStructure Structure Mining Mining Discovering useful patterns from the hyperlink structure connecting Web sites or Web resources 23 Types of Web Mining Web WebMining Mining
Web WebContent Content Mining Mining Web WebUsage Usage Mining Mining Web WebStructure Structure Mining Mining Data sources include the explicit hyperlink between documents, or implicit links among objects (e.g., two objects being tagged using the same keyword). 24
Types of Web Mining Web WebMining Mining Web WebContent Content Mining Mining Web WebUsage Usage Mining Mining Web WebStructure Structure Mining Mining Applications: document retrieval and ranking (e.g., Google) discovery of hubs and authorities
discovery of Web communities social network analysis 25 Web Content Mining :: common approaches and applications Basic notion: document similarity Most Web content mining and information retrieval applications involve measuring similarity among two or more documents Vector representation facilitates similarity computations using vector-space operations (such as Cosine of the angle between two vectors) Examples Search engines: measure the similarity between a query (represented as a vector) and the indexed document vectors to return a ranked list of relevant documents Document clustering: group documents based on similarity or dissimilarity (distance) among them Document categorization: measure the similarity of a new document to be classified with representations of existing categories (such as the mean vector representing a group of document vectors) Personalization: recommend documents or items based their similarity to a representation of the users profile (may be a term vector representing concepts or terms of interest to the user) 26 Web Content Mining
:: example clustered search results Can drill down within clusters to view subtopics or to view the relevant subset of results 27 Web Content Mining :: example personalized content delivery Google's personalized news is an example of a content-based recommender system which recommends items (in part) based on the similarity of their content to a users profile (gathered from search and click
history) 28 Web Structure Mining :: graph structures on the Web The structure of a typical Web graph Web pages as nodes hyperlinks as edges connecting two related pages Hyperlink Analysis Hyperlinks can serve as a tool for pure navigation But, often they are used to point to pages with authority on the same topic as the source page (similar to a citation in a publication) Some interesting Web structures * 29 Web Structure Mining :: example Googles PageRank algorithm Basic idea:
Illustration of PageRank propagation Rank of a page depends on the ranks of pages pointing to it Out Degree of page is the number of edges pointing away from it used to compute the contribution of the page to those to which it points The final PageRank value represents the probability that a random surfer will reach the page d is the prob. that a random surfer chooses the page directly rather than getting there via navigation 30 Web Structure Mining :: example Hubs and Authorities Basic idea Authority comes from in-edges Being a hub comes from out-edges Mutually re-enforcing relationship A good authority is a page that is pointed to by many good hubs.
A good hub is a page that points to many good authorities. Together they tend to form a bipartite graph This idea can be used to discover authoritative pages related to a topic HITS algorithm Hypertext Induced Topic Search 31 Hubs Authorities Web Structure Mining :: example online communities Basic idea Community 2 Web communities are collections of Web pages such that each member node has more hyperlinks (in either direction) within the community than outside the community. Community 1
Typical approach: Maximalflow model * Source node Ex: separate the two subgraphs with any choice of source node (left subgraph) and sink node (right subgraph), removing the three dashed links sink * Source: G. Flake, et al. Self-Organization and Identification of Web Communities, IEEE Computer, Vol. 35, No. 3, pp. 66-71, March 2002 . 32 Web Usage Mining The Problem: analyze Web navigational data to Find how the Web site is used by Web users Understand the behavior of different user segments Predict how users will behave in the future
Target relevant or interesting information to individual or groups of users Increase sales, profit, loyalty, etc. Challenge Quantitatively capture Web users common interests and characterize their underlying tasks 33 Applications of Web Usage Mining Electronic Commerce design cross marketing strategies across products evaluate promotional campaigns target electronic ads and coupons at user groups based on their access patterns predict user behavior based on previously learned rules and users profiles present dynamic information to users based on their interests and profiles: Web personalization Effective and Efficient Web Presence
determine the best way to structure the Web site identify weak links for elimination or enhancement prefetch files that are most likely to be accessed enhance workgroup management & communication Search Engines Behavior-based ranking 34 Web Usage Mining :: data sources Typical Sources of Data: automatically generated Web/application server access logs e-commerce and product-oriented user events (e.g., shopping cart changes, product clickthroughs, etc.) user profiles and/or user ratings meta-data, page content, site structure User Transactions sets or sequences of pageviews possibly with associated weights a pageview is a set of page files and associated objects that contribute to a single display in a Web Browser 35
Typical Fields in a Log File Entry client clientIP IPaddress address base url base url date/time date/time http httpmethod method file fileaccessed accessed protocol protocolversion version status code status code bytes bytestransferred transferred referrer referrerpage
page user useragent agent 126.96.36.199 188.8.131.52 maya.cs.depaul.edu maya.cs.depaul.edu 2006-02-01 2006-02-01 00:08:43 00:08:43 GET GET /classes/cs589/papers.html /classes/cs589/papers.html HTTP/1.1 HTTP/1.1 200 200 (successful (successfulaccess) access) 9221 9221 http://dataminingresources.blogspot.com/ http://dataminingresources.blogspot.com/ Mozilla/4.0+(compatible;+MSIE+6.0;+Windows+NT+5.1; Mozilla/4.0+(compatible;+MSIE+6.0;+Windows+NT+5.1;
+SV1;+.NET+CLR+2.0.50727 +SV1;+.NET+CLR+2.0.50727) ) In addition, there may be fields corresponding to login information client-side cookies (unique keys, issued to clients in order to identify a repeat visitor) session ids issued by the Web or application servers 37 Basic Entities in Web Usage Mining User (Visitor) - Single individual that is accessing files from one or more Web servers through a Browser Page File - File that is served through HTTP protocol Pageview - Set of Page Files that contribute to a single display in a Web Browser User Session - Set of Pageviews served due to a series of HTTP requests from a single User across the entire Web. Server Session - Set of Pageviews served due to a series of HTTP requests from a single User to a single site Transaction (Episode) - Subset of Pageviews from a single User or Server Session 38
Main Challenges in Data Collection and Preprocessing Main Questions: what data to collect and how to collect it; what to exclude how to identify requests associated with a unique user sessions (HTTP is stateless) how to identify/define user transactions (within each session) how to identify what is the basic unit of analysis (e.g., pageviews, items purchased) how to integrate e-commerce data with usage data Problems: user ids are usually suppressed due to security concerns individual IP addresses are sometimes hidden behind proxy servers; may not be unique client-side & proxy caching makes server log data less reliable data must be integrated from multiple sources (e.g., server logs, content data, e-commerce applications servers, customer demographic data, etc.) Standard Solutions/Practices:
user registration, cookies, server extensions and URL re-writing, cache busting heuristic approaches to session/user identification and path completion 39 Usage Data Preparation Tasks Data cleaning remove irrelevant references and fields in server logs remove references due to spider navigation add missing references due to client-side caching Data integration synchronize data from multiple server logs integrate e-commerce and application server data integrate meta-data Data Transformation pageview identification identification of unique users sessionization partitioning each users record into multiple sessions or transactions (usually representing different visits) mapping between user sessions and topics or categories Associating weights with object/pageviews in one session or transaction 40 Conceptual Representation of User
354 0 0 15 This is the typical representation of the data, after preprocessing, that is used for input into data mining algorithms. Raw weights may be binary, based on time spent on a page, or other measures of user interest in an item. In practice, need to normalize or standardize this data. 41 Web Usage Mining as a Process 42 E-Commerce Data Integrating E-Commerce and Usage Data Needed for analyzing relationships between navigational patterns of visitors and business questions such as profitability, customer value, product placement, etc. E-business / Web Analytics E.g., tracking and analyzing conversion of browsers to buyers E-Commerce v. Simple Usage Data E-commerce data is product oriented while usage data is pageview oriented Usage events (pageviews) are well defined and have consistent meaning across all Web sites E-commerce events are often only applicable to specific domains, and the
definition of certain events can vary from site to site Major difficulty for Usage events is getting accurate preprocessed data Major difficulty for E-commerce events is defining and implementing the events for a particular site 43 Why We Need Web Analytics Are we attracting new people to our site? Is our site sticky? Which regions in it are not? What is the health of our lead qualification process? How adept is our conversion of browsers to buyers? What behavior indicates purchase propensity? What site navigation do we wish to encourage? How can profiling help use cross-sell and up-sell? How do customer segments differ?
What attributes describe our best customers? Can we target other prospects like them? What makes customers loyal? How do we measure loyalty? 44 Three Skill Sets Required Technology How do we get the data? Are we collecting the right data? Data DataCollection Collection//Preprocessing Preprocessing//Integration Integration Analytics How do we turn the data into insightful information? Analysis AnalysisTools, Tools,OLAP, OLAP,Data DataMining Mining Business Management What action do we take? How do we measure the impact of that action?
E-Metrics E-Metrics 45 Using Analytics for E-Business Management Navigation Calibration Calculating Content Popularity Refresh rate <1? Freshness Visit Frequency Stickiness / Slipperiness / Leakage Stimulus - Inducement Conversion Quotient
Interaction Computation Customer Service Assessment Customer Experience Evaluation Branding 46 Web Usage and E-Business Analytics Different Different Levels Levels of of Analysis Analysis Session Analysis Static Aggregation and Statistics OLAP Data Mining 47 Session Analysis Simplest form of analysis: examine individual or groups of server sessions and e-commerce data. Advantages: Gain insight into typical customer behaviors. Trace specific problems with the site. Drawbacks: LOTS of data.
Difficult to generalize. 48 Static Aggregation (Reports) Most common form of analysis. Data is aggregated by predetermined units such as days or sessions. Generally gives most bang for the buck. Advantages: Gives quick overview of how a site is being used. Minimal disk space or processing power required. Drawbacks: No ability to dig deeper into the data. Page View Home Page Catalog Ordering Shopping Cart Number of Sessions 50,000 500 9000 49
Average View Count per Session 1.5 1.1 2.3 Online Analytical Processing (OLAP) Allows changes to aggregation level for multiple dimensions. Generally associated with a Data Warehouse. Advantages & Drawbacks Very flexible Requires significantly more resources than static reporting. Page View Kid's Stuff Products Number of Sessions 2,000 Page Number of View Sessions Kid's Stuff Products Electronics
Educational 63 Radio-Controlled 93 50 Average View Count per Session 5.9 Average View Count per Session 2.3 2.5 Data Mining: Going Deeper Frequent Itemsets and Association Rules The Donkey Kong Video Game and Stainless Steel Flatware Set product pages are accessed together in 1.2% of the sessions. When the Shopping Cart Page is accessed in a session, Home Page is also accessed 90% of the time. When the Stainless Steel Flatware Set product page is accessed in a session, the Donkey Kong Video page is also accessed 5% of the time.
30% of clients who accessed /special-offer.html, placed an online order in /products/software/ Sequential Patterns Add an extra dimension to frequent itemsets and association rules - time x% of the time, when AB appears in a transaction, C appears within z transactions) 40% of people who bought the book How to cheat IRS booked a flight to South America 6 months later The Video Game Caddy page view is accessed after the Donkey Kong Video Game page view 50% of the time. This occurs in 1% of the sessions. 15% of visitors followed the path home > * > software > * > shopping cart > checkout 51 Data Mining: Going Deeper Clustering: Content-Based or Usage-Based Customer/visitor segmentation Categorization of pages and products Classification
Classifying users into behavioral groups (browser, likely to purchase, loyal customer, etc.) Examples: Cusotmers who access Video Game Product pages, have income of 50K+, and have 1 or more children, should get a banner ad for Xbox in their next visit. Customers who make at least 4 purchases in one year should be categorized as loyal Load applicants in 45K-60K income range, low debt, and good-excellent credit should be approved for a new mortgage. 52 Example: Path Analysis for Ecommerce Visit 10% 90% No Search Search (64% successful) Avg sale per visit: $X Avg sale per visit: 2.2X 70%
30% Last Search Failed Last Search Succeeded Avg sale per visit: 0.9X Avg sale per visit: 2.8X 53 Example: Association Analysis for Ecommerce Product Fully Reversible Mats Association Egyptian Cotton Towels Lift
456 Confidence 41% Website Recommended Products J Jasper Towels Confidence 1.4% White Cotton T-Shirt Bra Plunge T-Shirt Bra 246 25% Black embroidered underwired bra
Confidence 1% Confidence: 41% who purchased Fully Reversible Mats also purchased Egyptian Cotton Towels Lift: People who purchased Fully Reversible Mats were 456 times more likely to purchase the Egyptian Cotton Towels compared to the general population 54 Web Usage Mining: clustering example Transaction Clusters: Clustering similar user transactions and using centroid of each cluster as a usage profile (representative for a user segment) Sample cluster centroid from dept. Web site (cluster size =330) Support URL Pageview Description 1.00 /courses/syllabus.asp?course=45096-303&q=3&y=2002&id=290 SE 450 Object-Oriented Development class syllabus
0.97 /people/facultyinfo.asp?id=290 Web page of a lecturer who thought the above course 0.88 /programs/ Current Degree Descriptions 2002 0.85 /programs/courses.asp? depcode=96&deptmne=se&courseid =450 SE 450 course description in SE program 0.82 /programs/2002/gradds2002.asp M.S. in Distributed Systems program description
55 Site Content Content Analysis Module Web/Application Server Logs Basic Basic Framework Framework for for E-Commerce E-Commerce Data DataAnalysis Analysis Data Cleaning / Sessionization Module Data Integration Module
Integrated Sessionized Data E-Commerce Data Mart Usage Analysis OLAP Tools OLAP Analysis Data Cube Site Map customers orders products Site Dictionary Operational Database Data Mining
Engine Pattern Analysis Components of E-Commerce Data Analysis Framework Content Analysis Module extract linkage and semantic information from pages potentially used to construct the site map and site dictionary analysis of dynamic pages includes (partial) generation of pages based on templates, specified parameters, and/or databases (may be done in real time, if available as an extension of Web/Application servers) Site Map / Site Dictionary site map is used primarily in data preparation (e.g., required for pageview identification and path completion); it may be constructed through content analysis and/or analysis of usage data (e.g., from referrer information) site dictionary provides a mapping between pageview identifiers / URLs and content/structural information on pages; it is used primarily for content labeling both in sessionized usage data as well as integrated e-commerce data 57 Components of E-Commerce Data Analysis Framework Data Integration Module used to integrate sessionized usage data, e-commerce data (from application servers),
and product/user data from databases user data may include user profiles, demographic information, and individual purchase activity e-commerce data includes various product-oriented events, including shopping cart changes, purchase information, impressions, click-throughs, and other basic metrics primarily used for data transformation and loading mechanism for the Data Mart E-Commerce Data mart this is a multi-dimensional database integrating data from a variety of sources, and at different levels of aggregation can provide pre-computed e-metrics along multiple dimensions is used as the primary data source in OLAP analysis, as well as in data selection for a variety of data mining tasks (performed by the data mining engine 58
Fish's Diversity Initiatives . Affinity Groups - A set of internal resource groups for legal staff with common interests and backgrounds, which foster a sense of community and inclusion - Asian Legal Staff, Black Legal Staff, GLBT and Hispanic Latino...
The OSI model describes the processes of encoding, formatting, segmenting, and encapsulating data for transmission over the network. The TCP/IP protocol suite is an open standard protocol that has been endorsed by the networking industry and ratified, or approved, by...
Experimental Design Criteria for Establishing a Causal Relationship Concomitant variation Temporal variation Control over other possible causal factors Concomitant Variation Temporal Ordering Causes must precede effects. Control over other possible causal factors Randomization: Randomly assign respondents to treatment conditions.
Our Mission Statement. Established as a 501 (c)(3) non-profit in 1995, Gateway's mission is to actively address the issue of hunger and seeks to safeguard and protect our most vulnerable community members by providing, food, clothing, housing assistance, referrals, and...
Marcy Gutierrez is Senior Counsel in Lozano Smith's Sacramento office. She began her career in education as a public school teacher, and has been advising and representing school districts and other education entities for over ten years. Her areas of...
Times New Roman Arial Symbol Courier New Default Design MS Organization Chart 2.0 Electrical Communications Systems ECE.09.331 Spring 2008 Plan ECOMMS: Topics AM Broadcast Transmitter Amplifier Classification BJT Collector Modulator BJT Collector Modulator AM Receiver Tuning Circuits Series Tuned Circuit...
61% of bankers say a customer-centric business model is "very important". Only 17% are "very prepared" for it. "Fewer Heart Transplants, More Bypasses" ... As a bonus from last month, the storefront play from retail holds true for branch banking....
Ready to download the document? Go ahead and hit continue!