Web Search and Mining

Web Search Overview & CrawlingWeb Search Overview & Crawling
By
SATHISHKUMAR G
(sathishsak111@gmail.com)
Web Search and Mining

Algorithmic results.
Paid
Search Ads

Search and Information Retrieval
 Search on the Web is a daily activity for many people
throughout the world
 Search and communication are most popular uses of
the computer
 Applications involving search are everywhere
 The field of computer science that is most involved
with R&D for search is information retrieval (IR)
Search & IR

Information Retrieval
 “Information retrieval is a field concerned with the
structure, analysis, organization, storage, searching,
and retrieval of information.” (Salton, 1968)
 General definition that can be applied to many types
of information and search applications
 Primary focus of IR since the 50s has been on text
and documents
IR

What is a Document?
 Examples:
 web pages, email, books, news stories, scholarly papers,
text messages, Word™, Powerpoint™, PDF, forum postings,
patents, etc.
 Common properties
 Significant text content
 Some structure (e.g., title, author, date for papers;
subject, sender, destination for email)
IR

Documents vs. Database Records
 Database records (or tuples in relational databases)
are typically made up of well-defined fields (or
attributes)
 e.g., bank records with account numbers, balances,
names, addresses, social security numbers, dates of birth,
etc.
 Easy to compare fields with well-defined semantics
to queries in order to find matches
 Text is more difficult
IR

Documents vs. Records
 Example bank database query
 Find records with balance > $50,000 in branches located in
Amherst, MA.
 Matches easily found by comparison with field values of
records
 Example search engine query
 bank scandals in western mass
 This text must be compared to the text of entire news
stories
IR

Comparing Text
 Comparing the query text to the document text and
determining what is a good match is the core issue of
information retrieval
 Exact matching of words is not enough
 Many different ways to write the same thing in a “natural
language” like English
 e.g., does a news story containing the text “bank director
in Amherst steals funds” match the query?
 Some stories will be better matches than others
IR

Dimensions of IR
 IR is more than just text, and more than just web
search
 although these are central
 People doing IR work with different media, different
types of search applications, and different tasks
IR

Other Media
 New applications increasingly involve new media
 e.g., video, photos, music, speech
 Like text, content is difficult to describe and compare
 text may be used to represent them (e.g. tags)
 IR approaches to search and evaluation are
appropriate
IR

Dimensions of IR
Content Applications Tasks
Text Web search Ad hoc search
Images Vertical search Filtering
Video Enterprise search Classification
Scanned docs Desktop search Question answering
Audio Forum search
Music P2P search
Literature search
IR

IR Tasks
 Ad-hoc search
 Find relevant documents for an arbitrary text query
 Filtering
 Identify relevant user profiles for a new document
 Classification
 Identify relevant labels for documents
 Question answering
 Give a specific answer to a question
IR

Big Issues in IR
 Relevance
 What is it?
 Simple (and simplistic) definition:
A relevant document contains the information that a
person was looking for when they submitted a query to
the search engine
 Many factors influence a person’s decision about what is
relevant: e.g., task, context, novelty, style
 Topical relevance (same topic) vs. user relevance
(everything else)
IR

Big Issues in IR
 Relevance
 Retrieval models define a view of relevance
 Ranking algorithms used in search engines are based on
retrieval models
 Most models describe statistical properties of text rather
than linguistic
 i.e., counting simple text features such as words instead of parsing
and analyzing the sentences
 Statistical approach to text processing started with Luhn in the 50s
 Linguistic features can be part of a statistical model
IR

Big Issues in IR
 Evaluation
 Experimental procedures and measures for comparing
system output with user expectations
 Originated in Cranfield experiments in the 60s
 Typically use test collection of documents, queries, and
relevance judgments
 Most commonly used are TREC collections
 Recall and precision are two examples of effectiveness
measures
IR

Big Issues in IR
 Users and Information Needs
 Search evaluation is user-centered
 Keyword queries are often poor descriptions of actual
information needs
 Interaction and context are important for understanding
user intent
 Query refinement techniques such as query expansion,
query suggestion, relevance feedback improve ranking
IR

IR and Search Engines
 A search engine is the practical application of
information retrieval techniques to large scale text
collections
 Web search engines are best-known examples, but
many others
 Open source search engines are important for research
and development
 e.g., Lucene, Lemur/Indri, Galago
 Big issues include main IR issues but also some
others
IR

IR and Search Engines
Relevance
-Effective ranking
Evaluation
-Testing and measuring
Information needs
-User interaction
Performance
-Efficient search and indexing
Incorporating new data
-Coverage and freshness
Scalability
-Growing with data and users
Adaptability
-Tuning for applications
Specific problems
-e.g. Spam
Information Retrieval Search Engines
IR

Search Engine Issues
 Performance
 Measuring and improving the efficiency of search
 e.g., reducing response time, increasing query throughput,
increasing indexing speed
 Indexes are data structures designed to improve search
efficiency
 designing and implementing them are major issues for search
engines
Search Engine

 Dynamic data (Incorporating new data)
 The “collection” for most real applications is constantly
changing in terms of updates, additions, deletions
 e.g., web pages
 Acquiring or “crawling” the documents is a major task
 Typical measures are coverage (how much has been indexed) and
freshness (how recently was it indexed)
 Updating the indexes while processing queries is also a
design issue
Search Engine

 Scalability
 Making everything work with millions of users every day,
and many terabytes of documents
 Distributed processing is essential
 Adaptability
 Changing and tuning search engine components such as
ranking algorithm, indexing strategy, interface for
different applications
Search Engine

 Spam
 For Web search, spam in all its forms is one of the major
issues
 Affects the efficiency of search engines and, more
seriously, the effectiveness of the results
 Many types of spam
 e.g. spamdexing or term spam, link spam, “optimization”
 New subfield called adversarial IR, since spammers are
“adversaries” with different goals
Search Engine

Architecture of SE
How do search engines like Google work?
Search Engine

Algorithmic results.
Paid
Search Ads
Search Engine

Architecture
The Web
Ad indexes
Web Results 1 - 10 of about 7,310,000 for miele. (0.12 seconds)
Miele, Inc -- Anything else is a compromise
At the heart of your home, Appliances by Miele. ... USA. to miele.com. Residential Appliances.
Vacuum Cleaners. Dishwashers. Cooking Appliances. Steam Oven. Coffee System ...
www.miele.com/ - 20k - Cached - Similar pages
Miele
Welcome to Miele, the home of the very best appliances and kitchens in the world.
www.miele.co.uk/ - 3k - Cached - Similar pages
Miele - Deutscher Hersteller von Einbaugeräten, Hausgeräten ... - [ Translate this
page ]
Das Portal zum Thema Essen & Geniessen online unter www.zu-tisch.de. Miele weltweit
...ein Leben lang. ... Wählen Sie die Miele Vertretung Ihres Landes.
www.miele.de/ - 10k - Cached - Similar pages
Herzlich willkommen bei Miele Österreich - [ Translate this page ]
Herzlich willkommen bei Miele Österreich Wenn Sie nicht automatisch
weitergeleitet werden, klicken Sie bitte hier! HAUSHALTSGERÄTE ...
www.miele.at/ - 3k - Cached - Similar pages
Sponsored Links
CG Appliance Express
Discount Appliances (650) 756-3931
Same Day Certified Installation
www.cgappliance.com
San Francisco-Oakland-San Jose,
CA
Miele Vacuum Cleaners
Miele Vacuums- Complete Selection
Free Shipping!
www.vacuums.com
Miele Vacuum Cleaners
Miele-Free Air shipping!
All models. Helpful advice.
www.best-vacuum.com
Web spider
Indexer
Indexes
Search
User
Search Engine

Indexing Process
Search Engine

Indexing Process
 Text acquisition
 identifies and stores documents for indexing
 Text transformation
 transforms documents into index terms or features
 Index creation
 takes index terms and creates data structures (indexes) to
support fast searching
Search Engine

Query Process
Search Engine

Query Process
 User interaction
 supports creation and refinement of query, display of
results
 Ranking
 uses query and indexes to generate ranked list of
documents
 Evaluation
 monitors and measures effectiveness and efficiency
(primarily offline)
Search Engine

Details: Text Acquisition
 Crawler
 Identifies and acquires documents for search engine
 Many types – web, enterprise, desktop
 Web crawlers follow links to find documents
 Must efficiently find huge numbers of web pages (coverage) and
keep them up-to-date (freshness)
 Single site crawlers for site search
 Topical or focused crawlers for vertical search
 Document crawlers for enterprise and desktop search
 Follow links and scan directories
Indexing Process

Web Crawler
 Starts with a set of seeds, which are a set of URLs given to it
as parameters
 Seeds are added to a URL request queue
 Crawler starts fetching pages from the request queue
 Downloaded pages are parsed to find link tags that might
contain other useful URLs to fetch
 New URLs added to the crawler’s request queue, or frontier
 Continue until no more new URLs or disk full
Indexing Process

Crawling picture
Web
URLs crawled
and parsed
URLs frontier
Unseen Web
Seed
pages
Indexing Process

Crawling the Web
Indexing Process

Text Acquisition
 Feeds
 Real-time streams of documents
 e.g., web feeds for news, blogs, video, radio, tv
 RSS is common standard
 RSS “reader” can provide new XML documents to search engine
 Conversion
 Convert variety of documents into a consistent text plus
metadata format
 e.g. HTML, XML, Word, PDF, etc. → XML
 Convert text encoding for different languages
 Using a Unicode standard like UTF-8
Indexing Process

Text Acquisition
 Document data store
 Stores text, metadata, and other related content for
documents
 Metadata is information about document such as type and
creation date
 Other content includes links, anchor text
 Provides fast access to document contents for search
engine components
 e.g. result list generation
 Could use relational database system
 More typically, a simpler, more efficient storage system is used
due to huge numbers of documents
Indexing Process

Text Transformation
 Parser
 Processing the sequence of text tokens in the document to
recognize structural elements
 e.g., titles, links, headings, etc.
 Tokenizer recognizes “words” in the text
 must consider issues like capitalization, hyphens, apostrophes,
non-alpha characters, separators
 Markup languages such as HTML, XML often used to specify
structure
 Tags used to specify document elements
 E.g., <h2> Overview </h2>
 Document parser uses syntax of markup language (or other
formatting) to identify structure
Indexing Process

Text Transformation
 Stopping
 Remove common words
 e.g., “and”, “or”, “the”, “in”
 Some impact on efficiency and effectiveness
 Can be a problem for some queries
 Stemming
 Group words derived from a common stem
 e.g., “computer”, “computers”, “computing”, “compute”
 Usually effective, but not for all queries
 Benefits vary for different languages
Indexing Process

Text Transformation
 Link Analysis
 Makes use of links and anchor text in web pages
 Link analysis identifies popularity and community
information
 e.g., PageRank
 Anchor text can significantly enhance the representation
of pages pointed to by links
 Significant impact on web search
Indexing Process

Text Transformation
 Information Extraction
 Identify classes of index terms that are important for some
applications
 e.g., named entity recognizers identify classes such as
people, locations, companies, dates, etc.
 Classifier
 Identifies class-related metadata for documents
 i.e., assigns labels to documents
 e.g., topics, reading levels, sentiment, genre
 Use depends on application
Indexing Process

Index Creation
 Document Statistics
 Gathers counts and positions of words and other features
 Used in ranking algorithm
 Weighting
 Computes weights for index terms
 Used in ranking algorithm
 e.g., tf.idf weight
 Combination of term frequency in document and inverse
document frequency in the collection
Indexing Process

Index Creation
 Inversion
 Core of indexing process
 Converts document-term information to term-document
for indexing
 Difficult for very large numbers of documents
 Format of inverted file is designed for fast query
processing
 Must also handle updates
 Compression used for efficiency
Indexing Process

Index Creation
 Index Distribution
 Distributes indexes across multiple computers and/or
multiple sites
 Essential for fast query processing with large numbers of
documents
 Many variations
 Document distribution, term distribution, replication
 P2P and distributed IR involve search across multiple sites
Indexing Process

User Interaction
 Query input
 Provides interface and parser for query language
 Most web queries are very simple, other applications may
use forms
 Query language used to describe more complex queries
and results of query transformation
 e.g., Boolean queries
 similar to SQL language used in database applications
 IR query languages also allow content and structure specifications,
but focus on content
Query Process

User Interaction
 Query transformation
 Improves initial query, both before and after initial search
 Includes text transformation techniques used for
documents
 Spell checking and query suggestion provide alternatives
to original query
 Query expansion and relevance feedback modify the
original query with additional terms
Query Process

User Interaction
 Results output
 Constructs the display of ranked documents for a query
 Generates snippets to show how queries match
documents
 Highlights important words and passages
 Retrieves appropriate advertising in many applications
 May provide clustering and other visualization tools
Query Process

Ranking
 Scoring
 Calculates scores for documents using a ranking algorithm
 Core component of search engine
 Basic form of score is ∑ qi di
 qi and di are query and document term weights for term i
 Many variations of ranking algorithms and retrieval
models
Query Process

Ranking
 Performance optimization
 Designing ranking algorithms for efficient processing
 Term-at-a time vs. document-at-a-time processing
 Safe vs. unsafe optimizations
 Distribution
 Processing queries in a distributed environment
 Query broker distributes queries and assembles results
 Caching is a form of distributed searching
Query Process

Evaluation
 Logging
 Logging user queries and interaction is crucial for
improving search effectiveness and efficiency
 Query logs and clickthrough data used for query
suggestion, spell checking, query caching, ranking,
advertising search, and other components
 Ranking analysis
 Measuring and tuning ranking effectiveness
 Performance analysis
 Measuring and tuning system efficiency
Query Process

How Does It Really Work?
 This course explains these components of a search
engine in more detail
 Often many possible approaches and techniques for a
given component
 Focus is on the most important alternatives
 i.e., explain a small number of approaches in detail rather than
many approaches
 “Importance” based on research results and use in actual search
engines
 Alternatives described in references

Thank you

Web Search and Mining

More Related Content

What's hot (20)

Similar to Web Search and Mining (20)

More from sathish sak (20)

Recently uploaded (20)

Web Search and Mining